WO2015087712A1 - Tv映像向け加速超解像処理方法及び同方法によるtv映像向け加速超解像処理装置、第1~6加速超解像処理プログラム、並びに第1~2記憶媒体 - Google Patents
Tv映像向け加速超解像処理方法及び同方法によるtv映像向け加速超解像処理装置、第1~6加速超解像処理プログラム、並びに第1~2記憶媒体 Download PDFInfo
- Publication number
- WO2015087712A1 WO2015087712A1 PCT/JP2014/081474 JP2014081474W WO2015087712A1 WO 2015087712 A1 WO2015087712 A1 WO 2015087712A1 JP 2014081474 W JP2014081474 W JP 2014081474W WO 2015087712 A1 WO2015087712 A1 WO 2015087712A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- luminance distribution
- psf
- image
- initial value
- restored image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/73—Deblurring; Sharpening
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N9/00—Details of colour television systems
- H04N9/64—Circuits for processing colour signals
- H04N9/646—Circuits for processing colour signals for image enhancement, e.g. vertical detail restoration, cross-colour elimination, contour correction, chrominance trapping filters
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/14—Picture signal circuitry for video frequency region
- H04N5/20—Circuitry for controlling amplitude response
- H04N5/205—Circuitry for controlling amplitude response for correcting amplitude versus frequency characteristic
- H04N5/208—Circuitry for controlling amplitude response for correcting amplitude versus frequency characteristic for compensating for attenuation of high frequency components, e.g. crispening, aperture distortion correction
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/14—Picture signal circuitry for video frequency region
- H04N5/21—Circuitry for suppressing or minimising disturbance, e.g. moiré or halo
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/44—Receiver circuitry for the reception of television signals according to analogue transmission standards
- H04N5/57—Control of contrast or brightness
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
Definitions
- the present invention relates to image processing of TV images, and in particular, to remove deterioration information such as optical blur and blur included in TV images by mathematical arithmetic processing based on Bayesian probability theory, and to restore TV images before deterioration.
- the present invention relates to an accelerated super-resolution processing method for TV video, an accelerated super-resolution processing device for TV video, a first to sixth accelerated super-resolution processing program, and first and second storage media.
- a TV video is composed of still images called 30 frames or more per second, and each frame is not blurred as well as digital or analog, but includes deterioration information such as optical blur and blur. There is a problem of being out.
- FIG. 1 shows an example of deterioration information included in one frame of an actual TV image.
- the left image is one frame of a TV video composed only of the Y (luminance) component of the X-ray pinhole camera
- the right image is the prior art invented by the present inventor.
- Patent Documents 1 and 2 reduce the degradation information by super-resolution processing of the image on the left side of FIG. Comparing the two images in FIG. 1, it can be seen that the actual TV image includes deterioration information such as optical blur and blur.
- Patent Documents 1 and 2 The image restoration technique invented by the present inventor (Patent Documents 1 and 2) is based on the fact that iterative calculations are repeated in accordance with Bayse probabilistic formulas from information on a single still image including deterioration information such as optical blur and blur.
- the maximum likelihood degradation factor and the maximum likelihood restored image that is, the image after super-resolution processing, which is most likely to be related to the luminance distribution of the image, is obtained by numerical calculation. There was a problem that it was difficult to handle TV images.
- Patent Documents 1 and 2 The image restoration technique invented by the inventor (Patent Documents 1 and 2) has a problem in that it has a large computation scale and is not suitable for real-time processing because complex numbers are targeted for computation.
- the present inventor changed the calculation target from a complex number to a real number, and further realized the real-time processing by implementing the hardware using an FPGA (Field Programmable Gate Array). Made possible.
- FPGA Field Programmable Gate Array
- One of the other conventional super-resolution techniques for TV video is to superimpose a plurality of frames by superimposing the same subject by focusing on the same subject in a plurality of frames constituting the TV video.
- Patent Documents 3 and 4 propose a “reconstruction type super-resolution” method for performing the resolution.
- the “reconstruction super-resolution” method can be used. There was a problem that resolution became difficult.
- Non-Patent Document 1 Bayes statistical processing is performed on the basis of a plurality of continuous still images obtained by a video camera, which are slightly different in viewpoint, and super-resolution still images are obtained. There is a way to get a picture.
- this method always requires a large amount of memory for storing still images including a plurality of pieces of degradation information, and in order to obtain a single super-resolution still image, a plurality of memories are always required. Since it is necessary to process a single still image, not only a large amount of memory is required, but also the calculation becomes enormous, and there is a problem that TV images cannot be processed.
- Patent Documents 1 and 2 the operation object of the image restoration technique invented by the present inventor (Patent Documents 1 and 2) is changed from a complex number to a real number, and further implemented by hardware using an FPGA.
- Patent Documents 1 and 2 Invented and applied for a patent for a super-resolution processing method for TV video and a super-resolution processing device for TV video that can be processed.
- This super-resolution processing device for TV video is an FPGA and LSI (Large Scale Integrated circuit).
- LSI Large Scale Integrated circuit
- the present invention solves the above-mentioned problems, an accelerated super-resolution processing method for TV video for super-resolution processing of a TV video, an accelerated super-resolution processing device for TV video by the same method, and first to It is an object to provide a six-acceleration super-resolution processing program and first and second storage media.
- an accelerated super-resolution processing method for TV video using an acceleration algorithm newly invented by the present inventor an accelerated super-resolution processing device for TV video using the same method, a first -6 acceleration super-resolution processing program and first and second storage media are provided.
- This acceleration algorithm requires only a few iterations in order to perform iterative computations at an accelerated rate.
- the number can be greatly reduced.
- an accelerated super-resolution processing method for a TV image that reduces optical deterioration from a frame included in the TV image signal for one frame and restores the TV image signal for one frame before the deterioration.
- the accelerated super-resolution processing method for TV video according to the first invention of the present invention includes (S1) a step of setting the maximum number of iterations, and (S2) degradation suitable for the degradation status of the TV video while watching the TV video.
- Deterioration indication designating process for designating the indication and (S3) luminance distribution of the first PSF (Point Spread Function) associated with the degradation indication, and this system, and this was organized in association with the number of iterations
- a PSF preparation step of preparing a luminance distribution of a series of PSFs (S4) a deteriorated image preparation step of preparing a luminance distribution of a deteriorated image made up of luminance distribution for one frame from a TV video signal for one frame, and (S5 )
- PSF size acquisition step for obtaining this PSF size
- S7 first reset step for setting 1 to the counter for counting the number of iterations
- S8 estimated brightness of the restored image initial value
- a first restored image initial value correcting step of copying the distribution and making it an estimated luminance distribution of the corrected restored image initial value, and further correcting the estimated brightness distribution of the corrected restored image initial value based on the PSF size
- S9 A PSF selection step of selecting one linked to the counter value from the luminance distribution of a series of PSFs and making this a PSF luminance distribution
- S10 the estimated luminance distribution of the corrected restored image initial value A step of convolving the luminance distribution to obtain a first function
- S11 a step of inverting the first function to obtain a second function
- S12 multiplying the luminance distribution of the degraded image by the second function.
- the third function (S13) a step of multiplying the estimated luminance distribution of the restored image initial value by multiplying the third function with the estimated luminance distribution of the restored image, (S14) a step of adding 1 to the counter, and (S15)
- the hypothesis that the counter value is equal to or greater than the maximum number of iterations is verified. If the verification result is false, the process proceeds to step (S16). If the verification result is true, the process proceeds to step (S18). (S16) replacing the estimated luminance distribution of the restored image with the estimated luminance distribution of the restored image initial value, (S17) returning to the steps of (S8), and (S18) estimating the estimated luminance distribution of the restored image to the maximum likelihood.
- a first image comprising a step of outputting as a luminance distribution of the restored image, a preparation step comprising steps (S19), (S1) to (S7), and steps (S20), (S8) to (S18).
- a restoration process (S 21)
- the iterations of the maximum number of iterations are completed by executing in order from the smallest S suffix in each step, and the luminance distribution of the maximum likelihood restored image is output.
- a first acceleration super-resolution processing step, and (S22) a TV image forming step of forming the luminance distribution of the maximum likelihood restored image as a TV video signal for one frame and outputting this as a super-resolution TV video signal.
- the second aspect of the present invention relates to a second aspect of the PSF preparation step that constitutes the accelerated super-resolution processing method for TV video according to the first aspect of the invention.
- the second aspect of the PSF preparation step is (S30 ) Using the degradation indication as search information, the PSF database created by associating the PSF luminance distribution with the degradation indication in a one-to-one correspondence is searched, and the luminance distribution of the hit PSF is the first one.
- the value of the second counter is n
- the step of making the luminance distribution of the maximum likelihood restored PSF the luminance distribution of the nth PSF the step of returning to the steps of (S37) and (S33), S38)
- the maximum number of iterations is set to n_max and the natural number less than n_max is set to n, the luminance distribution of the first PSF to the luminance distribution of the n_maxth PSF are connected in this order in one sequence.
- PSF luminance distribution, and the deterioration distribution number is labeled as a label to the luminance distribution of the series of PSFs, and the luminance distribution of the series of PSFs associated with the deterioration indication is formed.
- This second invention is the same as that described in claim 2.
- the third aspect of the present invention relates to a third aspect of the PSF preparation process that constitutes the accelerated super-resolution processing method for TV video according to the first aspect of the invention.
- the third aspect of the PSF preparation process is (S40). ) A step of setting the maximum number of iterations to 5; (S41) A PSF preparation step is executed in advance for all the pairs of brightness distributions of the PSFs associated with all the degradation indications in the PSF database.
- a fourth invention according to the present invention relates to a PSF restoration process constituting the second aspect of the PSF preparation process according to the second invention.
- This PSF restoration process is a step of substituting 6 for the maximum number of iterations (S50).
- S51 a step of obtaining the luminance distribution of the degraded PSF by regarding the luminance distribution of the PSF initial value as a deteriorated image, and
- S52 forming the luminance distribution of the PSF initial value as an estimated luminance distribution of the restored PSF initial value.
- a restored PSF initial value correcting step of correcting the estimated luminance distribution 21 of the corrected restored PSF initial value by performing paste on the right side and then the lower side at the end in the same way (S55) A step of convolution of the luminance distribution of the PSF initial value with the estimated luminance distribution of the corrected restored PSF initial value to obtain a fourth function, and (S56) a step of inverting the fourth function to obtain a fifth function. (S57) multiplying the fifth function by the luminance distribution of the deteriorated PSF to obtain a sixth function, and (S58) multiplying the sixth function by the estimated luminance distribution of the restored PSF initial value to restore the PSF.
- a fifth invention according to the present invention relates to a first restored image initial value correcting step constituting the accelerated super-resolution processing method for TV video according to the first invention, and the first restored image initial value correcting step comprises (S70). ) A step of making the estimated luminance distribution of the restored image initial value the estimated luminance distribution of the corrected restored image initial value; and (S71) correcting one of the luminance distributions of a series of PSFs based on the PSF size to the corrected restored image initial A step of calculating a calculation difficulty region occurring in the periphery of the estimated luminance distribution of the corrected restored image initial value when performing convolution calculation on the estimated luminance distribution of the value; and (S72) calculating the estimated luminance distribution of the corrected restored image initial value
- the pixels in the difficult area are copied, and the copied pixels are inverted so as to be mirror-symmetric with respect to the four sides of the estimated luminance distribution of the corrected restored image initial value, and further the 4 of the estimated luminance distribution of the corrected restored image initial value.
- Edge border copying the pixel at the upper left corner in the difficult calculation area of the estimated luminance distribution of the corrected restored image, and copying the copied pixel at the upper left corner.
- S74 the corrected restored image initial value The pixel in the upper right corner in the difficult calculation region of the estimated luminance distribution is copied, and the copied pixel in the upper right corner is rotated by 180 degrees with the vertex in the upper right corner as the rotation center, and the estimated luminance distribution of the corrected restored image initial value
- a sixth aspect of the present invention relates to a second aspect of the first image restoration step that constitutes the accelerated super-resolution processing method for TV video according to the first aspect of the invention, and the second aspect of the first image restoration step.
- S80 When n_max is the maximum number of iterations and n is a natural number less than n_max, the n-th PSF luminance distribution of the PSF luminance distribution is set as the PSF luminance distribution in the nth stage.
- PSF providing step to be provided to the first single image restoration step of the eye (S81) Iterative calculation based on Bayse probabilistic formula from PSF luminance distribution, estimated luminance distribution of restored image initial value and luminance distribution of degraded image
- Restoration process S81 (S82)
- the second restored image initial stage having the same processing procedure as the first restored image initial value correcting step for correcting the estimated brightness distribution of the restored restored image initial value based on the PSF size to obtain the estimated brightness distribution of the corrected restored image initial value.
- a value correction step (S83) a step of convolving the PSF luminance distribution with the estimated luminance distribution of the corrected restored image initial value to obtain a seventh function, and (S84) inversion of the seventh function to obtain an eighth function. (S85) multiplying the eighth function by the luminance distribution of the deteriorated image to obtain a ninth function, and (S86) multiplying the ninth function by the estimated luminance distribution of the restored image initial value.
- a step of obtaining an estimated luminance distribution of the restored image and (S87) a step of outputting the estimated luminance distribution of the restored image.
- step (S88) (S87) of the first single-part image restoring step S81-n of the n-th stage The output of step (n) is the (n + 1) th stage
- the number of connected first single image restoration steps S81 is repeated n_max times, and the estimated luminance distribution of the restored image output from the first single image restoration step S81-n_max at the n_max stage is the maximum likelihood. It is characterized in that it is output as a luminance distribution of the restored image.
- the sixth aspect of the invention is the same as that described in claim 6.
- a seventh aspect of the present invention relates to a third aspect of the first image restoration step that constitutes the accelerated super-resolution processing method for TV images according to the first aspect of the invention, and the third aspect of the first image restoration step.
- S90 The process of resetting by assigning 0 to the counter
- S91 The process of resetting by assigning 1 to the second counter
- S92 The hypothesis that the value of the counter is other than 0. If the verification result is false, the process proceeds to step (S93). If the verification result is true, the process jumps to step (S96), and (S93) stores the luminance distribution of the deteriorated image as a deteriorated image.
- the process of transferring to the image buffer and the restored image initial value buffer, the step of jumping to the steps of (S94) and (S96), and the estimated luminance distribution of the restored image in the steps of (S95) and (S102) A step of transferring to the initial value buffer; and (S96) a step of forming the luminance distribution of the mth PSF in the luminance distribution of a series of PSFs as the luminance distribution of the PSF when the value of the second counter is m. , (S97) a step of reading the estimated luminance distribution of the restored image initial value from the restored image initial value buffer, and (S98) correcting the estimated luminance distribution of the restored image initial value based on the PSF size and correcting it.
- a third restored image initial value correcting step consisting of the same processing procedure as the first restored image initial value correcting step formed with the estimated luminance distribution of (S99), and convolution of the PSF luminance distribution with the estimated luminance distribution of the corrected restored image initial value
- a step of obtaining a tenth function (S100) a step of inverting the tenth function to obtain an eleventh function; and (S101) reading the luminance distribution of the deteriorated image from the deteriorated image storage buffer.
- step (S102) multiplying the twelfth function with the estimated luminance distribution of the restored image initial value to obtain the estimated luminance distribution of the restored image, and (S103) A step of adding 1; (S104) a step of adding 1 to the second counter; and (S105) a hypothesis that the counter value exceeds the maximum number of iterations, and if the verification result is false Jump to step (S95), and if the verification result is true, proceed to step (S106), and (S106) output the estimated luminance distribution of the restored image as the luminance distribution of the maximum likelihood restored image.
- the third image restoration that outputs the maximum likelihood restored image that has been subjected to the maximum number of iterations by executing in order from the smallest S subscript in each step and outputs the maximum likelihood restored image It is characterized by being a process.
- the seventh aspect of the present invention is the same as that described in the seventh aspect.
- An eighth invention according to the present invention relates to a deteriorated image preparation process constituting the accelerated super-resolution processing method for TV images according to the first invention.
- the deteriorated image preparation process includes (S110) a TV image signal for one frame.
- RGB signal extracting step for extracting RGB signals constituting a frame from (S111), a delay step for delaying and outputting the remaining TV video signal from which RGB signals are extracted from one frame of TV video signal by one frame, and (S112) YUV conversion step of converting RGB signals into YUV signals, and (S113) Extracting the luminance distribution of the deteriorated image consisting only of the Y signal that is the luminance component of the YUV signal to extract the Y deteriorated image
- a Y-degraded image extraction step that holds the distribution of the U-degraded image consisting of only the remaining U signal and the distribution of the V-degraded image consisting of only the V signal, which is a luminance distribution (S11).
- Y deterioration consists brightness distribution of one frame performs de-gamma processing of the luminance distribution of the image it is characterized there in and a de-gamma processing step of outputting forms the luminance distribution of the degraded image.
- This eighth invention is the same as that described in claim 8.
- a ninth invention according to the present invention relates to a TV imaging process that constitutes the accelerated super-resolution processing method for TV video according to the first invention.
- This TV imaging process includes (S120) luminance distribution of the maximum likelihood restored image. 1 from the gamma processing step of performing the gamma processing of (2), (S121) the distribution of the U-degraded image and the distribution of the V-degraded image held in the Y-degraded image extraction step and the luminance distribution of the maximum likelihood restored image after the gamma processing consisting of the Y component.
- a restored image composition step for synthesizing the distribution of one YUV restored image (S122) an RGB conversion step for performing RGB conversion of the distribution of the YUV restored image to form an RGB restored image distribution, and (S123) an RGB restored image distribution.
- the ninth aspect of the invention is the same as that described in the ninth aspect.
- the tenth aspect of the present invention relates to the luminance distribution of the PSFs constituting the first to seventh aspects of the accelerated super-resolution processing method for TV images, and the luminance distribution of the PSFs is the same in a square shape having no edges. It is composed of pixels of a size, and its luminance distribution is a two-dimensional normal distribution with the brightest center, and is characterized in that its size is 5 ⁇ 5 pixels.
- the tenth aspect of the invention is the same as that described in the tenth aspect.
- An eleventh invention according to the present invention is a first accelerated super-resolution processing program for causing a computer to execute a preparation step and a first image restoration step constituting the accelerated super-resolution processing method for TV video according to the first aspect of the invention. It is.
- the eleventh aspect of the invention is the same as that described in claim 11.
- a twelfth aspect of the present invention is a preparatory step that constitutes the accelerated super-resolution processing method for TV video according to the first invention, and a second step that constitutes the accelerated super-resolution processing method for TV video according to the sixth invention. It is the 2nd acceleration super-resolution processing program for making a computer perform an image restoration process.
- the twelfth aspect of the present invention is the same as that described in claim 12.
- a preparatory step constituting the accelerated super-resolution processing method for TV video according to the first invention, and a third step constituting the accelerated super-resolution processing method for TV video according to the seventh invention.
- It is a 3rd acceleration super-resolution processing program for making a computer perform an image restoration process.
- the thirteenth invention is the same as that described in claim 13.
- the first acceleration super-resolution processing program, the second acceleration super-resolution processing program, and the third acceleration super-resolution processing program are languages that can be read and executed by a computer, for example, C ++, XTML, HTML, and JAVA (registered). Trademark).
- C ++, XTML, HTML and JAVA registered trademark
- a fourteenth aspect of the present invention is a first accelerated super-resolution processing program according to the eleventh aspect, a second accelerated super-resolution processing program according to the twelfth aspect, and a third accelerated super-resolution according to the thirteenth aspect.
- the image processing programs are all encrypted, and the encrypted first acceleration super-resolution processing program, second acceleration super-resolution processing program, and third acceleration super-resolution processing program are stored.
- the first storage medium is characterized in that it can be connected to a computer and can be read by the computer. This fourteenth invention is the same as that described in claim 14.
- USB Universal Serial Bus
- a USB flash memory stick having a capacity of 8 GB or more and compatible with encrypted storage
- a CD Compact Disk
- DVD Digital Versatile Disk
- a flash memory card an external HDD (Hard Disk Drive), an HDD having a capacity of 8 Gbytes or more and compatible with encryption storage
- An external SDD Solidstate Disk Drive or the like can be used.
- optical degradation is performed from a frame included in a TV image signal for one frame.
- This invention relates to an accelerated super-resolution processing device for TV video that reduces and restores a TV video signal for one frame before deterioration.
- This accelerated super-resolution processing device for TV video uses (W1) means for setting the maximum number of iterations.
- (W2) a degradation index designation means for designating a degradation index suitable for the degradation status of the TV video while watching the TV video
- (W3) a luminance distribution of the first PSF associated with the degradation index
- PSF preparation means for preparing a luminance distribution of a series of PSFs arranged in association with the number of repetitive computations composed of this system
- (W4) the luminance distribution of the degraded image consisting of the luminance distribution for one frame.
- a deteriorated image preparation means provided, (W5) a restored image initial value preparation means for making the luminance distribution of the deteriorated image an estimated luminance distribution of the restored image, and (W6) the luminance distribution of a series of PSFs has the same image size.
- a PSF size acquisition means for obtaining the PSF size
- (W7) a first reset means for setting 1 to a counter for counting the number of repeated operations
- (W8) a restored image
- a first restored image initial value correction in which an estimated luminance distribution of an initial value is copied and made into an estimated luminance distribution of a corrected restored image initial value, and the estimated luminance distribution of the corrected restored image initial value is corrected based on the PSF size.
- PSF selection means for selecting one linked to the counter value from the luminance distribution of a series of PSFs and forming this as the PSF luminance distribution, and (W10) the first corrected restored image Means for obtaining a thirteenth function by convolving the luminance distribution of the PSF with the estimated luminance distribution of the value; (W11) means for obtaining the fourteenth function by inverting the thirteenth function; and (W12) the fourteenth function.
- (W13) means for multiplying the estimated luminance distribution of the restored image initial value to obtain the estimated luminance distribution of the restored image by multiplying the luminance distribution of the degraded image to obtain the fifteenth function; W14) The means for adding 1 to the counter and (W15) The hypothesis that the counter value is equal to or greater than the maximum number of iterations is verified. If the verification result is false, the process proceeds to (W16).
- preparation means composed of means of (W19) (W1) to (W7), and (W20) (W8) to (W18) (W21)
- the number of iterations of the maximum number of iterations is repeated by executing in order in ascending order of the subscripts of the means W.
- First acceleration super-resolution processing means for finishing the calculation and outputting the luminance distribution of the maximum likelihood restored image, (W22)
- the luminance distribution of the maximum likelihood restored image is formed into a TV video signal for one frame, and this is super-resolved. It has a feature in that it is provided with a TV imaging means for outputting as a TV video signal.
- the fifteenth aspect of the present invention is the same as that described in claim 15.
- a sixteenth aspect of the present invention relates to a second aspect of the PSF preparation means constituting the accelerated super-resolution processing apparatus for TV images according to the fifteenth aspect of the invention.
- the second aspect of the PSF preparation means is (W30 )
- the PSF database created by associating the PSF luminance distribution with the degradation indication in a one-to-one correspondence is searched, and the luminance distribution of the hit PSF is the first one.
- Means, (W33) means for adding 1 to the second counter, and (W34) the hypothesis that the value of the second counter has exceeded the maximum number of iterations, and if the verification result is false (W35) ) If the verification result is true, means for jumping to means (W38), (W35) means for obtaining the luminance distribution of the restored PSF by restoring the luminance distribution of the PSF initial value by the PSF restoration means, W36)
- means for making the luminance distribution of the restored PSF the luminance distribution of the nth PSF means for returning to the means of (W37) (W33), (W38) said maximum
- n_max is the number of iterations and n is a natural number less than n_max, the luminance distribution of the first PSF to the n_max luminance distribution are connected in this order to form a luminance distribution of one PSF.
- the sixteenth aspect of the present invention is the same as that described in the sixteenth aspect.
- a seventeenth aspect of the present invention relates to a third aspect of the PSF preparation means constituting the accelerated super-resolution processing apparatus for TV images according to the fifteenth aspect of the invention.
- the third aspect of the PSF preparation means is (W40 ) A means for setting the maximum number of iterations to 5; (W41) A PSF preparation means is executed in advance for all pairs of PSF brightness distributions associated with all the degradation indication numbers in the PSF database.
- An eighteenth aspect of the present invention relates to a PSF restoration means constituting the second aspect of the PSF preparation means according to the sixteenth invention, wherein the PSF restoration means is a means for substituting 6 for (W50) maximum number of iterations. And (W51) means for determining the luminance distribution of the PSF initial value as the luminance distribution of the deteriorated image and forming the luminance distribution of the deteriorated PSF, and (W52) forming the luminance distribution of the PSF initial value as the estimated luminance distribution of the restored PSF initial value.
- (W53) means for substituting 1 in the counter and resetting
- (W54) the estimated luminance distribution of the restored PSF initial value is made the estimated luminance distribution of the corrected restored PSF initial value, and further the luminance distribution of the PSF initial value
- a difficult calculation area generated in the periphery of the estimated luminance distribution of the corrected restored PSF initial value is represented by an image of the luminance distribution of the PSF initial value. It is calculated based on the noise, and the pixel of the difficult calculation area related to the upper boundary is copied, and the copied pixel is mirror-symmetrical with respect to the upper boundary outside the upper boundary of the estimated luminance distribution of the corrected restoration PSF initial value.
- (W59) means for adding 1 to the counter
- (W60) means for verifying the hypothesis that the counter value has exceeded the maximum number of iterations, and means (W61) if the verification result is false If the verification result is true, a means for jumping to the means (W63), a means (W61) for replacing the estimated luminance distribution of the restored PSF with the estimated brightness distribution of the restored PSF initial value, (W62) It is characterized by comprising means for jumping to means (W54) and means (W63) for outputting the estimated luminance distribution of the restored PSF as the luminance distribution of the maximum likelihood restored PSF.
- the eighteenth aspect of the invention is the same as that described in claim 18.
- a nineteenth invention according to the present invention relates to a first restored image initial value correcting means constituting the accelerated super-resolution processing apparatus for TV video according to the fifteenth invention, wherein the first restored image initial value correcting means comprises (W70 ) Means for making the estimated luminance distribution of the restored image initial value the estimated luminance distribution of the corrected restored image initial value; and (W71) correcting one of the luminance distributions of a series of PSFs based on the PSF size.
- Border Means for pasting and correcting so as to be arranged outside (W73) copying the pixel at the upper left corner in the difficult calculation region of the estimated luminance distribution of the corrected restored image initial value, and copying the copied pixel at the upper left corner And (W74) a corrected restored image initial value, and a means for correcting by pasting to a blank area generated at the upper left corner of the estimated luminance distribution of the corrected restored image initial value
- the pixel in the upper right corner in the difficult calculation region of the estimated luminance distribution is copied, and the copied pixel in the upper right corner is rotated by 180 degrees around the vertex of the upper right corner as the rotation center to estimate the estimated luminance distribution of the corrected restored image initial value Means for pasting and correcting in a blank area generated in the upper right corner of the image, and (W75) copying the pixel in the lower left corner in the difficult calculation area of the estimated luminance distribution of the corrected restored image initial value, and copying this lower left corner Corner Means for rotating the pixel 180 degrees around the top left corner
- a twentieth aspect of the present invention relates to a second aspect of the first image restoration means constituting the accelerated super-resolution processing apparatus for TV video according to the fifteenth aspect of the invention, and the second aspect of the first image restoration means.
- W80 When the maximum number of iterations is n_max and the natural number less than n_max is n, the n-th PSF luminance distribution of a series of PSF luminance distributions is designated as the PSF luminance distribution in the nth stage.
- PSF providing means provided to the first single-part image restoration means of the eye (W81) Iterative calculation based on Bayse probabilistic formula from PSF luminance distribution, estimated luminance distribution of restored image initial value, and luminance distribution of degraded image
- a first single-image restoration unit that performs an operation corresponding to one of the images, obtains an estimated luminance distribution of the restored image that is most likely to be a luminance distribution of the deteriorated image, and outputs the estimated luminance distribution;
- W82 a first restored image initial Value correction Means for obtaining an estimated luminance distribution of the corrected restored image initial value by the second restored image initial value correcting means having the same configuration as the stage; and (W83) convolving the luminance distribution of the PSF with the estimated luminance distribution of the corrected restored image initial value.
- Means for obtaining 19 functions (W84) means for inverting the 19th function to obtain the 20th function, and (W85) multiplying the 20th function by the luminance distribution of the deteriorated image to obtain the 21st function.
- Means for obtaining the estimated luminance distribution of the restored image by (W86) multiplying the twenty-first function by the estimated luminance distribution of the restored image initial value; (W87) means for outputting the estimated luminance distribution of the restored image;
- the output of the means (W87) of the first single image restoration means W81-n at the n-th stage is the output of the (W + 1) of the first single image restoration means W81- (n + 1) at the (n + 1) -th stage.
- a twenty-first aspect of the present invention relates to a third aspect of the first image restoration means constituting the accelerated super-resolution processing device for TV video according to the fifteenth aspect of the invention, and the third aspect of the first image restoration means.
- (W90) means to reset by assigning 0 to the counter
- (W91) means to reset by assigning 1 to the second counter
- (W92) the hypothesis that the value of the counter is non-zero.
- (W97) means for reading the estimated luminance distribution of the restored image initial value from the restored image initial value buffer; and (W98) correcting the estimated luminance distribution of the restored image initial value and correcting the estimated luminance distribution of the restored image initial value.
- a third restored image initial value correcting means having the same configuration as the first restored image initial value correcting means forming the distribution, and a (W99) PSF luminance distribution is convoluted with the estimated luminance distribution of the corrected restored image initial value.
- (W100) means for inverting the 22nd function to obtain the 23rd function
- (W101) means for multiplying the 23rd function by the luminance distribution of the deteriorated image to obtain the 24th function.
- (W102) Means for multiplying the function of 24 with the estimated luminance distribution of the restored image initial value to obtain the estimated luminance distribution of the restored image; (W103) means for adding 1 to the counter; and (W104) adding 1 to the second counter. (W105) The hypothesis that the value of the counter exceeds the maximum number of iterations is verified. If the verification result is false, the process jumps to (W95), and if the verification result is true ( (W106) means for outputting the estimated luminance distribution of the restored image as the luminance distribution of the maximum likelihood restored image, and (W107) in order from the smallest of the S suffix of each means.
- RGB signal extracting means for extracting RGB signals constituting a frame from (W111), delay means for delaying and outputting the remaining TV video signal from which RGB signals are extracted out of one frame of TV video signal by one frame (W112)
- YUV converting means for converting the RGB signal to YUV by converting it into a YUV signal
- Y-degraded image extraction means that holds the distribution of the U-degraded image consisting only of the remaining U signal and the distribution of the V-degraded image consisting of only the V signal, which is a luminance distribution
- Y deterioration consists brightness distribution of one frame performs de-gamma processing of the luminance distribution of the image it is characterized there in and a de-gamma processing means for outputting forms the luminance distribution of the degraded image.
- a TV image forming means constituting the accelerated super-resolution processing apparatus for TV images according to the fifteenth invention, wherein the TV image forming means comprises (W120) luminance distribution of maximum likelihood restored image. (W121) 1 from the luminance distribution of the maximum likelihood restored image after the gamma processing composed of the distribution of the U deteriorated image and the distribution of the V deteriorated image and the Y component held by the Y deteriorated image extracting means. (W123) RGB conversion means for performing RGB conversion of the distribution of the YUV restored image to form an RGB restored image distribution, and (W123) the distribution of the RGB restored image.
- RGB signal conversion means for outputting the read RGB signal, and (W124) whether the RGB signal is combined with the remaining TV video signal output by the delay means to generate a TV video signal for one frame. Consisting thing is featured and a TV video signal synthesizing means for outputting form a super-resolution TV video signal.
- the twenty-third invention is the same as that described in claim 23.
- a fourth accelerated super-resolution process for configuring and executing the preparation means and the first image restoration means constituting the accelerated super-resolution processing apparatus for TV images according to the fifteenth aspect of the invention. It is a program.
- the twenty-fourth aspect of the invention is the same as that described in claim 24.
- the twenty-fifth aspect of the invention is the same as that described in claim 25.
- a preparing means for constituting an accelerated super-resolution processing apparatus for TV images according to the fifteenth aspect of the invention and a third means for constituting an accelerated super-resolution processing apparatus for TV images according to the twenty-first aspect. It is a 6th acceleration super-resolution processing program for comprising and performing an image restoration means.
- the twenty-sixth aspect of the invention is the same as that described in claim 26.
- the fourth acceleration super-resolution processing program, the fifth acceleration super-resolution processing program, and the sixth acceleration super-resolution processing program are languages that can be read and executed by a computer, for example, C ++, XTML, HTML, and JAVA (registered). Trademark).
- C ++, XTML, HTML and JAVA registered trademark
- the fourth acceleration super-resolution processing program, the fifth acceleration super-resolution processing program, and the sixth acceleration super-resolution processing program are respectively encrypted
- the second storage medium that stores the fourth acceleration super-resolution processing program, the fifth acceleration super-resolution processing program, and the sixth acceleration super-resolution processing program that can be connected to the computer and can be read by the computer It is.
- the same storage medium as the first storage medium can be used as the second storage medium of the present invention.
- an apparatus that restores an image from only one frame of TV video information and continuously performs this to perform super-resolution processing of the TV video is a large and uneconomical scale of 1.5 million gates.
- the accelerated super-resolution method and apparatus for TV video according to the present invention it is possible to obtain a super-resolution image having the same quality as the conventional one by two iterations, the effect of reducing the number of steps and means, the speed-up effect, At least a part of the effect of the substantially real-time processing, the effect that the gate scale when LSI is realized can be reduced to 70,000 gates, which is about 3% of the conventional, and the economic effect that the cost when LSI is realized is low. Realized.
- the present invention may be anything regardless of the radiation source as long as it is a TV system image, it may be an image by an infrared camera or an X-ray camera, and an effect of wide application range and application range is realized.
- video contains.
- the flowchart which shows an example of the process sequence in the 2nd aspect of the PSF preparation process which is 2nd invention by this invention The flowchart which shows an example of the process sequence in the 3rd aspect of the PSF preparation process which is 3rd invention by this invention
- the flowchart which shows an example of the process sequence in the 3rd image restoration process which is 7th invention by this invention The flowchart which shows an example of the process sequence in the degradation image preparation process which is 8th invention by this invention.
- the flowchart which shows an example of the process sequence in the TV imaging process which is 9th invention by this invention The flowchart which shows an example of the luminance distribution of PSF which is 10th invention by this invention
- the figure which shows one example regarding the structure of the 2nd aspect of the PSF preparation means which is 16th invention by this invention The figure which shows one example regarding the structure of the 3rd aspect of the PSF preparation means which is 17th invention by this invention
- the figure which shows an example of the data in the PSF database by this invention The figure which shows an example of the data in the expansion PSF database by this invention
- restoration means which is 18th invention by this invention The figure which shows one example regarding the structure of the 1st decompression
- produces in the peripheral part of the estimated brightness distribution of the restored image initial value by a 1st restored image initial value correction means
- the figure which shows one example of the structure of the 1st acceleration super-resolution processing program by this invention, and the condition of installing this program in a computer The figure which shows one example regarding the structure of the super-resolution processing window by this invention
- the figure which shows one example regarding the structure inside the 1st set top box by this invention The figure which shows one example of the condition of the setup of the 1st set top box by this invention
- the degraded image, the PSF, and the restored image are configured by laying out square pixels of the same size without borders, and each pixel has a red primary color (R) having an 8-bit depth and an 8-bit depth.
- RGB color pixels composed of a green primary color (G) and a blue primary color (B) having a depth of 8 bits.
- PSF consists of only gray scale pixels.
- the degraded image, the PSF, and the restored image have the pixel at the upper left corner as the origin, the row where the origin exists, the axis parallel to the pixel row facing in the horizontal direction without changing the row as the x axis, and the origin
- the y-axis is an axis that is parallel to the pixel column that is oriented vertically without changing the column. All the pixels in the degraded image, the PSF, and the restored image can be specified by a two-dimensional coordinate (x, y).
- the deteriorated image and the restored image have the same image size and the same coordinates.
- the present invention does not handle the case where the image is so blurred that it is indistinguishable, so the peripheral portion of the PSF is substantially zero, and in order to reduce the number of calculations, to where in the degraded image and the restored image
- the assumption is that the luminance distribution of the PSF does not change even if it is performed, and a PSF size of 5 pixels by 5 pixels is used.
- the PSF, the deteriorated image, and the restored image are each handled only by the luminance component, and only the luminance component is used for the restoration calculation.
- the reason for this is that the number of operations is reduced and the hue does not change.
- the method of the present invention confirms that the super-resolution processing quality is not inferior to the case where R, G, and B are restored individually.
- the PSF since the PSF, the deteriorated image, and the restored image are each composed of a luminance component, the PSF is called a PSF luminance distribution, a deteriorated image luminance distribution, and a restored image estimated luminance distribution. Since the exact luminance distribution of the restored image is unknown, it is called an estimated luminance distribution.
- the image restoration is performed by the accelerated super-resolution processing method for TV video according to the present invention, the image is substantially converged in a state without optical deterioration and is substantially the same as the original image. Yes.
- FIG. 2 is a flowchart illustrating an example of a processing procedure in the accelerated super-resolution processing method for TV images according to the first invention of the present invention.
- a white square including a process number indicates a process other than the determination process
- a white diamond including a process number indicates a determination process
- an arrow includes a data input / output direction, and a number. If the white square is an element, for example 1, the maximum number of iterations is 1, the thick line indicates the flow of processing, the white circle indicates the combination and start and end of processing, and the black circle indicates the branch and combination of data. means.
- (S1) the step of setting the maximum number of iterations 1 is specified, and (S2) the deterioration indication number 2 suitable for the TV video deterioration status is specified while watching the TV video.
- (S3) The luminance distribution 14 of the first PSF associated with the degradation index 2 and the luminance distribution 3 of a series of PSFs composed of this system and arranged in association with the number of iterations.
- (S4) A deteriorated image preparation step of preparing a deteriorated image luminance distribution 4 composed of a luminance distribution for one frame from a TV video signal 5 for one frame, and (S5) a luminance distribution 4 of the deteriorated image.
- the luminance distribution 3 of a series of PSFs has the same image size, and when this image size is referred to as PSF size 7, Find PSF size 7 (S7) First reset step for setting 1 to a counter for counting the number of repetitive operations, (S8) Copy the estimated luminance distribution 6 of the restored image initial value and copy it to the corrected restored image initial value A first restored image initial value correction step for correcting the estimated brightness distribution 8 of the corrected restored image initial value based on the PSF size 7 in addition to the estimated brightness distribution 8, and (S9) from the brightness distribution 3 of a series of PSFs A PSF selection step of selecting one associated with the value of the counter and making it the PSF luminance distribution 9 (S10).
- the PSF luminance distribution 9 is convolved with the estimated luminance distribution 8 of the corrected restored image initial value.
- S11 a step of inverting the first function to obtain a second function
- S12 a step of multiplying the second function by the luminance distribution 4 of the deteriorated image to obtain a third function.
- S13 Restore the third function.
- the step of obtaining the estimated luminance distribution 10 of the restored image by multiplying the estimated luminance distribution 6 of the initial image value, the step of (S14) adding 1 to the counter, and (S15) the value of the counter being the maximum number of iterations of 1 or more. If the verification result is false, the process proceeds to step (S16). If the verification result is true, the process proceeds to step (S18).
- the image processing step includes (S22) a TV image forming step of forming the luminance distribution 11 of the maximum likelihood restored image as a TV video signal for one frame and outputting it as a super-resolution TV video signal 12. .
- the process starts from step S1 and ends at step S22. Since the accelerated super-resolution processing method for TV video in FIG. 2 is processing for one frame, it is necessary to continuously execute all the steps shown in FIG.
- Equation 1 is a rewrite of Formula 15 described in Patent Document 2 invented and registered by the present inventor so that real number processing and convolution calculation can be performed and for acceleration calculation.
- the method of Equation 1 uses PSF instead of OTF (Optical Transfer Function), which is a Fourier transform of PSF, and does not consider the phase, so that the calculation accuracy is reduced.
- OTF Optical Transfer Function
- the method of Formula 1 uses the PSF luminance distribution with a degree of restoration commensurate with the number of iterations starting from the same PSF, thereby convolving the calculation result in parentheses with the inversion function of the PSF.
- the calculation required in the method of Equation 15 is omitted. For this reason, not only the number of processes is reduced by 40%, but also acceleration calculation is possible, and the maximum likelihood restoration image that is not inferior to a substantially converged state (a state that is very close to the state before deterioration) with only a few iterations. Is obtained.
- Equation 1 F is the estimated luminance distribution of the restored image, F is the k-th value, G is the degraded image luminance distribution, H is the PSF luminance distribution, and H is the subscript.
- An asterisk mark in the circle means a convolution operation, which is the k-th value.
- K is a positive integer.
- F 1 is the estimated luminance distribution of the restored image initial value
- H1 is the luminance distribution of the first PSF
- F n is The estimated luminance distribution of the n-th restored image and H n mean the luminance distribution of the nth PSF.
- the estimated luminance distribution F 1 of the initial value of F in Equation 1 is the luminance distribution G of the degraded image. Is used.
- Equation 2 is an example of a general convolution integral arithmetic expression, which means that F (i, j) is convolved with H (M, N) and the result is G (i, j). is doing.
- F (i, j) is convolved with H (M, N) and the result is G (i, j). is doing.
- the data is converted into discrete values, and linear convolution is used for the calculation of convolution integration.
- Formula 3 is an example of a general linear convolution formula.
- Equations 2 to 3 i, j, m, n, M, and N are positive integers.
- an exclusion area that can be expressed by the largest integer that does not exceed half the size of the smaller size is generated in the larger size of F or H in which the convolution operation is performed. For example, when F is 100 ⁇ 100 pixels or more and H convolved with F is 3 ⁇ 3 pixel size, one peripheral pixel of F is an excluded area, and when H convolved with F is 5 ⁇ 5 pixel size, There is a problem that two pixels become an exclusion region.
- the number of pixels in the exclusion region is calculated according to the size of H to be used, and the outermost peripheral pixel of F existing in the exclusion region is mirrored outside the F region boundary. Copy and paste symmetrically to create a new pixel, and then change the position of the outermost circumference, that is, by changing the image of F and its size so that no excluded area occurs after the calculation. ing. At that time, by copying and pasting from the upper side to the right side for each side, newly generated pixels are transferred to the original pixels of F, so that areas that are not copied and pasted at the four corners are not generated.
- F when H is 5x5 size and F is WxL size, F is changed from WxL size to Wx (L + 2) size in the first copy and paste, and in the second copy and paste, F is changed from Wx (L + 2) size.
- F is changed from (W + 2) x (L + 2) size to (W + 2) x (L + 4) size
- F is ( The (W + 2) x (L + 4) size is changed to the (W + 4) x (L + 4) size, and all the pixels within the (W + 4) x (L + 4) size are filled.
- the luminance distribution 14 of the first PSF associated with the degradation index 2 is for use in the first iteration
- the luminance distribution of the second PSF is the second luminance distribution.
- the luminance distribution of the nth PSF is for use during the nth iteration.
- the luminance distribution of the second PSF is the one obtained by restoring the luminance distribution of the first PSF in the PSF restoration step
- the luminance distribution of the third PSF is the luminance distribution of the second PSF in the PSF restoration step.
- the luminance distribution of the nth PSF is the one obtained by restoring the luminance distribution of the (n ⁇ 1) th PSF in the PSF restoration step.
- all the luminance distributions after the second PSF luminance distribution are based on the luminance distribution 14 of the first PSF associated with the degradation index 2, and the luminance distribution 14 of the first PSF
- the luminance distribution 14 of the first PSF to the luminance distribution 17 of the nth PSF form a series of PSF luminance distributions 3 linked to the degradation index 2.
- FIG. 3 is a flowchart showing an example of the processing procedure in the second mode S3-2 of the PSF preparation step S3 according to the second aspect of the present invention.
- a white square including a process number indicates a process other than the determination process
- a white diamond including a process number indicates a determination process
- an arrow includes a data input / output direction, and a number. If the white square is an element, for example 1, the maximum number of iterations is 1, the thick line indicates the flow of processing, the white circle indicates the combination and start and end of processing, and the black circle indicates the branch and combination of data. means.
- the degradation indication number 2 is used as search information, and the luminance distribution of the PSF is associated with the degradation indication in a one-to-one correspondence.
- step (S35) The luminance distribution of the PSF initial value 15 PSF restoration The step of obtaining the luminance distribution 16 of the maximum likelihood restored PSF by restoring in step S63, (S36) When the value of the second counter is n, the luminance distribution 16 of the maximum likelihood restored PSF is the luminance distribution of the nth PSF.
- Step S17, Step S37, Step S33 Return Step S38
- the maximum number of iterations 1 is n_max and a natural number less than n_max is n
- Up to the luminance distribution 25 of the n_max-th PSF are connected in this order to form a luminance distribution 3 of one PSF
- the deterioration indication 2 is attached to the luminance distribution 3 of the one PSF as a label, and the deterioration indication 2 is linked. It is characterized in that it is composed of a process of forming a luminance distribution 3 of a series of PSFs.
- FIG. 4 is a flowchart showing an example of the processing procedure in the third aspect S3-3 of the PSF preparation step S3 according to the third invention of the present invention.
- a white square including a process number indicates a process other than the determination process
- a white diamond including a process number indicates a determination process
- an arrow includes a data input / output direction, and a number. If the white square is an element, for example 1, the maximum number of iterations is 1, the thick line indicates the flow of processing, the white circle indicates the combination and start and end of processing, and the black circle indicates the branch and combination of data. means.
- the third mode S3-3 of the PSF preparation step S3 in FIG. 4 includes (S40) a step of setting the maximum number of iterations 1 to 5, and (S41) all the degradation indications in the PSF database 13 and associated with them.
- the PSF preparation process is executed in advance for all the pairs of luminance distributions of the PSFs to obtain one series of PSF luminance distributions 3 for each deterioration indication, and one series of PSFs for each of these deterioration indications.
- FIG. 5 is a flowchart illustrating an example of the processing procedure in the PSF restoration step S63 according to the fourth aspect of the present invention.
- a white square including a process number indicates a process other than the determination process
- a white diamond including a process number indicates a determination process
- an arrow includes a data input / output direction, and a number. If the white square is an element, for example 1, the maximum number of iterations is 1, the thick line indicates the flow of processing, the white circle indicates the combination and start and end of processing, and the black circle indicates the branch and combination of data. means.
- the PSF restoration step S64 in FIG. 5 includes (S50) a step of substituting 6 for the maximum number of iterations 1, and (S51) the luminance distribution 15 of the PSF initial value as a deteriorated image luminance distribution 19 and the luminance distribution 19 of the deteriorated PSF. (S52) a step of forming the PSF initial value luminance distribution 15 with the estimated PSF initial value luminance distribution 20, (S53) a step of resetting 1 by substituting 1 into the counter, and (S54) estimating the restored PSF initial value.
- the luminance distribution 20 is formed with the estimated luminance distribution 21 of the corrected restored PSF initial value, and the corrected restored PSF initial value is estimated when the luminance distribution 15 of the PSF initial value is convolved with the estimated luminance distribution 21 of the corrected restored PSF initial value.
- the difficult calculation area generated in the periphery of the luminance distribution 21 is calculated based on the image size of the luminance distribution 15 of the PSF initial value, and the pixel of the difficult calculation area related to the upper boundary is copied. Then, the copied pixels are pasted so as to be arranged so as to be mirror-symmetric with respect to the upper side boundary outside the upper side boundary of the estimated luminance distribution 21 of the corrected restored PSF initial value, and in the same manner clockwise to the right side.
- the restoration PSF initial value correction step for correcting the estimated luminance distribution 21 of the corrected restored PSF initial value is performed at the end of the lower side, and (S55) the luminance distribution 15 of the PSF initial value is folded into the estimated luminance distribution 21 of the corrected restored PSF initial value.
- S58 Multiplying the sixth function by the estimated luminance distribution 20 of the restored PSF initial value to obtain the estimated luminance distribution 22 of the restored PSF,
- S59 adding 1 to the counter
- S6 The hypothesis that the value of the counter exceeds the maximum number of iterations 1 is verified.
- step (S61) If the verification result is false, the process proceeds to step (S61). If the verification result is true, (S63) A step of jumping to the step, (S61) a step of replacing the estimated luminance distribution 22 of the restored PSF with the estimated luminance distribution 20 of the restored PSF initial value, a step of jumping to the step of (S62) and (S54), and (S63) estimating the restored PSF. And a step of outputting the luminance distribution 22 as the luminance distribution 16 of the maximum likelihood restoration PSF.
- FIG. 6 is a flowchart illustrating an example of a processing procedure in the first restored image initial value correction step S8 according to the fifth aspect of the present invention.
- a white square including a process number indicates a process other than the determination process
- a white diamond including a process number indicates a determination process
- an arrow includes a data input / output direction, and a number. If the white square is an element, for example 1, the maximum number of iterations is 1, the thick line indicates the flow of processing, the white circle indicates the combination and start and end of processing, and the black circle indicates the branch and combination of data. means.
- (S70) a step of forming the estimated luminance distribution 6 of the restored image initial value as the estimated luminance distribution 8 of the corrected restored image initial value, and (S71) based on the PSF size 7.
- the upper left corner pixel in the area is copied, and the copied upper left corner pixel is rotated 180 degrees around the upper left corner vertex as the rotation center, and is generated at the upper left corner of the estimated luminance distribution 8 of the corrected restored image initial value (S74) Copy the pixel in the upper right corner in the difficult operation area of the estimated luminance distribution 8 of the corrected restored image initial value, and copy the copied pixel in the upper right corner to the upper right corner.
- (S75) Estimating a corrected restored image initial value by rotating by 180 degrees around the corner vertex and rotating a blank area generated at the upper right corner of the estimated luminance distribution 8 of the corrected restored image initial value
- the pixel in the lower left corner in the difficult-to-calculate area of the luminance distribution 8 is copied, and the copied pixel in the lower left corner is rotated 180 degrees with the vertex in the upper left corner as the rotation center, and the estimated luminance distribution 8 of the corrected restored image initial value is Lower left corner of
- (S76) Copying the pixel at the lower right corner in the difficult calculation area of the estimated luminance distribution 8 of the corrected restored image initial value, and copying the copied pixel at the lower right corner Is rotated 180 degrees with the vertex at the upper right corner as the center of rotation, and is provided with a step of pasting and correcting in a blank area generated at the lower right corner of the estimated luminance distribution 8 of the corrected restored image initial value.
- FIG. 7 is a flowchart illustrating an example of a processing procedure in the second image restoration step S88 as the second aspect of the first image restoration step S20 according to the sixth aspect of the present invention.
- a white square including a process number indicates a process other than the determination process
- a white diamond including a process number indicates a determination process
- an arrow includes a data input / output direction, and a number. If the white square is an element, for example 1, the maximum number of iterations is 1, the thick line indicates the flow of processing, the white circle indicates the combination and start and end of processing, and the black circle indicates the branch and combination of data. means.
- this first single image restoration step S81 corrects the estimated luminance distribution 6 of the restored image initial value based on the PSF size 7, thereby correcting the corrected restored image initial value. Guess A second restored image initial value correcting step comprising the same processing procedure as the first restored image initial value correcting step for obtaining the luminance distribution 8; (S83) the PSF luminance distribution 9 is convolved with the estimated restored luminance initial value luminance distribution 8; A step of obtaining a seventh function; (S84) a step of inverting the seventh function to obtain an eighth function; and (S85) multiplying the eighth function by the luminance distribution 4 of the deteriorated image to obtain a ninth function.
- the second image restoration step S88 in FIG. 7 has the same structure as the first single-part image restoration step S81-1 and the first single-part image restoration step S81 having the same structure as the first single-part image restoration step S81.
- the first single image restoration step S81-n_max at the n_max stage having the same configuration as the restoration step S81 is connected in series.
- the second single image restoration process S70-3 of the third stage to the first single image restoration process S81- (n_max-1) one stage before the last stage is omitted because of the same connection mode as the second stage. Yes.
- the second image restoration step S88 a process equivalent to the step S83 of the first single image restoration step S81-1 to the n_max-th n_max single image restoration step S88-n_max of the first step is provided by the PSF.
- the luminance distribution 9 of PSF suitable for the number of steps provided in step S80 is read. For example, if the first single image restoration process S81-1 in the first stage, the luminance distribution 14 of the first PSF is used, and if the first single image restoration process S81-2 in the second stage, the first single image restoration process S81-2.
- the luminance distribution 24 of the second PSF is the first single image restoration process S81-n of the nth stage
- the luminance distribution 17 of the nth PSF is restored to the first single image of the n_max stage.
- the luminance distribution 25 of the n_maxth PSF is read.
- a process equivalent to the step S85 of the first single-part image restoration step S81-1 to the n_max-th n_max single-part image restoration step S88-n_max of the first stage is a degraded image. Is read from the degraded image preparation step S4 of the preparation step S19. Further, a process equivalent to the process S82 of the first single image restoration process S81-1 in the first stage reads the estimated luminance distribution 6 of the restored image initial value from the restored image initial value preparation process S5.
- a step equivalent to the step S82 of the n-th first single image restoration step S81-n (2 ⁇ n ⁇ ) is the first (n ⁇ 1) -th first single image restoration step S81- ( n-1) Read the estimated luminance distribution 10 of the restored image output in a step equivalent to step S87 (2 ⁇ n ⁇ ). Also, in the n_max single image restoration step S88-n_max, a step equivalent to step S87 outputs the estimated luminance distribution 10 of the restored image as the luminance distribution 11 of the maximum likelihood restored image. Further, the second image restoration step S88 in FIG. 7 is equivalent to the first image restoration step S20 in FIG. 2 if the second single image restoration step S81 of the same number as the maximum number of iterations 1 is connected in series. Have the ability.
- FIG. 8 is a flowchart illustrating an example of a processing procedure in the third image restoration step S107 according to the seventh aspect of the present invention.
- a white square including a process number indicates a process other than the determination process
- a white diamond including a process number indicates a determination process
- an arrow includes a data input / output direction, and a number. If the white square is an element, for example 1, the maximum number of iterations is 1, the thick line indicates the flow of processing, the white circle indicates the combination and start and end of processing, and the black circle indicates the branch and combination of data. means.
- step S90 a step of resetting by assigning 0 to the counter
- step S91 a step of resetting by assigning 1 to the second counter
- step S92 the value of the counter being The hypothesis that it is other than 0 is verified, and if the verification result is false, the process proceeds to step (S93), and if the verification result is true, the process jumps to the process (S96), (S93) A step of transferring the luminance distribution 4 to the deteriorated image storage buffer 26 and the restored image initial value buffer 27, a step of jumping to the steps (S94) and (S96), and an estimated luminance of the restored image in the steps of (S95) and (S102).
- a step of transferring the distribution 10 to the restored image initial value buffer 27, (S96)
- the value of the second counter is m
- the m-th PS in a series of PSFs 3 (S97) a step of reading the estimated luminance distribution 6 of the restored image initial value from the restored image initial value buffer 27, and (S98) a restored image initial value based on the PSF size 7.
- a third restored image initial value correcting step comprising the same processing procedure as the first restored image initial value correcting step, wherein the estimated luminance distribution 6 is corrected to the estimated luminance distribution 8 of the corrected restored image initial value
- S99 A step of convolving the luminance distribution 9 with the estimated luminance distribution 8 of the corrected restored image initial value to obtain a tenth function
- S100 a step of inverting the tenth function to obtain an eleventh function
- S101 of the deteriorated image
- the luminance distribution 4 is read from the degraded image storage buffer 26 and multiplied by the eleventh function to obtain the twelfth function.
- the twelfth function is multiplied by the estimated luminance distribution 6 of the restored image initial value.
- FIG. 9 is a flowchart illustrating an example of the processing procedure in the degraded image preparation step S4 according to the eighth aspect of the present invention.
- a white square including a process number indicates a process other than the determination process
- a white diamond including a process number indicates a determination process
- an arrow includes a data input / output direction, and a number. If the white square is an element, for example 1, the maximum number of iterations is 1, the thick line indicates the flow of processing, the white circle indicates the combination and start and end of processing, and the black circle indicates the branch and combination of data. means.
- the degraded image preparation step S4 in FIG. 9 includes (S110) an RGB signal extraction step for extracting the RGB signal 28 constituting the frame from the TV video signal 15 for one frame, and (S111) the TV video signal 15 for one frame. (S112) YUV conversion step of converting the RGB signal 28 into YUV signal 30 by YUV conversion, and (S113) YUV signal 30 Among them, the luminance distribution 4 of the deteriorated image consisting only of the Y signal as the luminance component is extracted to form the luminance distribution 31 of the Y deteriorated image, and only the U deteriorated image distribution 32 consisting of only the remaining U signals and the V signal.
- Y deteriorated image extraction step for holding the V deteriorated image distribution 33 consisting of: (S114)
- De-gamma processing of the luminance distribution 31 of the Y deteriorated image is performed for one frame.
- De-gamma processing step of outputting forms the luminance distribution of the degraded image consisting degrees distribution 4, characterized is on it with a.
- FIG. 10 is a flowchart illustrating an example of a processing procedure in the TV imaging step S22 according to the ninth aspect of the present invention.
- a white square including a process number indicates a process other than the determination process
- a white diamond including a process number indicates a determination process
- an arrow includes a data input / output direction, and a number. If the white square is an element, for example 1, the maximum number of iterations is 1, the thick line indicates the flow of processing, the white circle indicates the combination and start and end of processing, and the black circle indicates the branch and combination of data. means.
- TV that is combined with the remaining TV video signal 29 to be output as a super-resolution TV video signal 12 composed of a TV video signal for one frame.
- Image signal combining step is characterized by comprising a.
- FIG. 11 shows an example of the PSF luminance distribution 9 according to the tenth aspect of the present invention.
- the luminance distribution 9 of the PSF in FIG. 11 is composed of square pixels having the same size without borders, and the luminance distribution is a two-dimensional normal distribution with the brightest center, and the size is 5 ⁇ 5 pixels. There is a feature. Since the PSF luminance distribution 9 is a two-dimensional normal distribution, it has point symmetry and is a shift invariant.
- the first accelerated super-resolution processing program 37 virtually creates all the steps in the preparation step S19 and the first image restoration step S20, and processes the steps of these steps. It is described.
- the second accelerated super-resolution processing program 38 virtually creates all the steps in the preparation step S19 and the second image restoration step S88, and processes the steps of these steps. It is described.
- the third accelerated super-resolution processing program 39 virtually creates all the steps in the preparation step S19 and the third image restoration step S107, and executes the processing procedures of these steps. It is described.
- the first acceleration super-resolution processing program 37, the second acceleration super-resolution processing program 38, and the third acceleration super-resolution processing program 39 are all encrypted.
- the encrypted first acceleration super-resolution processing program 37, second acceleration super-resolution processing program 38, and third acceleration super-resolution processing program 39 are stored, connectable to the computer, and read by the computer. This is the first storage medium 46 that is characterized in that it is possible.
- FIG. 12 shows an example of the configuration of the accelerated super-resolution processing apparatus 40 for TV images according to the fifteenth aspect of the present invention.
- the white squares containing the means numbers are means other than judgment and conditional branching
- the white diamonds containing the means numbers are judgment and conditional branching means
- the arrows are data
- a white square containing a number is an element, for example, 1 if it is 1, the maximum number of iterations is 1, a thick line is a flow of processing, a white circle is a combination of processing, and a black circle is a branch of data And bond respectively.
- the accelerated super-resolution processing device 40 for TV video in FIG. 12 specifies (W1) means for setting the maximum number of iterations 1 and (W2) specifies a degradation index 2 suitable for the TV video degradation status while watching the TV video. (W3) The luminance distribution 14 of the first PSF associated with the degradation index 2 and the luminance distribution 3 of a series of PSFs composed of this system and arranged in association with the number of iterations.
- (W4) Deteriorated image preparation means for preparing the luminance distribution 4 of the deteriorated image consisting of the luminance distribution for one frame from the TV video signal 5 for one frame, (W5) The luminance distribution 4 of the deteriorated image (W6) The luminance distribution 3 of a series of PSFs has the same image size, and when this image size is called PSF size 7, PSF size 7 PSF size obtaining means to be obtained; (W7) first resetting means for setting 1 to a counter for counting the number of iterations; (W8) copying the estimated luminance distribution 6 of the restored image initial value and using this as the corrected restored image initial value First restored image initial value correcting means for correcting the estimated brightness distribution 8 of the corrected restored image initial value based on the PSF size 7, which is the estimated brightness distribution 8, and (W9) a counter from the brightness distribution 3 of a series of PSFs PSF selection means for selecting one linked to the value of PSF and making it the PSF luminance distribution 9; (W10) The PSF luminance distribution 9 is convolve
- the estimated brightness of the restored image Means for replacing the distribution 10 with the estimated luminance distribution 6 of the restored image initial value, means for returning to the means (W17) and (W8), and (W18) outputting the estimated luminance distribution 10 of the restored image as the luminance distribution 11 of the maximum likelihood restored image.
- Means W1 9 and the first image restoration means W20 are executed in order in ascending order of the subscripts of W of the means to finish the iteration operation of the maximum number of iterations 1, and output the luminance distribution 11 of the maximum likelihood restored image.
- Accelerated super-resolution processing means, (W22) TV image forming means for forming the luminance distribution 11 of the maximum likelihood restored image as a TV video signal 41 for one frame and outputting it as the super-resolution TV video signal 12.
- FIG. 13 shows an example of the configuration of the second aspect W3-2 of the PSF preparation means W3 according to the sixteenth aspect of the present invention.
- the white squares containing the means numbers are means other than judgment and conditional branching
- the white diamonds containing the means numbers are judgment and conditional branching means
- the arrows are data
- a white square containing a number is an element, for example, 1 if it is 1, the maximum number of iterations is 1, a thick line is a flow of processing, a white circle is a combination of processing, and a black circle is a branch of data And bond respectively.
- the second mode W3-2 of the PSF preparation means W3 in FIG. 13 arranges (W30) the degradation indication number 2 as search information, and associates the PSF luminance distribution with the degradation indication number on a one-to-one basis.
- (W32) means for making the luminance distribution 14 of the first PSF the luminance distribution 15 of the initial PSF value
- (W33) means for adding 1 to the second counter
- (W34) the value of the second counter is the maximum iterative operation
- the hypothesis that the number of times 1 has been exceeded is verified, and if the verification result is false, proceed to the means of (W35), and if the verification result is true, the means of jumping to the means of (W38), (W35) PSF initial value
- the degradation index 2 is represented by an integer type variable BF, and the image file name of the nth luminance distribution 17 (1 ⁇ n ⁇ n_max) is PSF_BF_n.
- the nth luminance distribution 17 is received from the means of W36, a file name of PSF_BF_n.bmp is created from BF and n, and this file name is given
- the n-th luminance distribution 17 is stored in a large-capacity storage means of a computer, for example, an HDD (Hard Disk Drive), while the file name is stored in an array called SPSF (BF, n), so that the luminance distribution of a single PSF 3 can be created.
- FIG. 14 illustrates an example of the configuration of the third mode W3-3 of the PSF preparation means W3 according to the seventeenth aspect of the present invention.
- the white squares containing the means numbers are means other than judgment and conditional branching
- the white diamonds containing the means numbers are judgment and conditional branching means
- the arrows are data
- a white square containing a number is an element, for example, 1 if it is 1, the maximum number of iterations is 1, a thick line is a flow of processing, a white circle is a combination of processing, and a black circle is a branch of data And bond respectively.
- the third mode W3-3 of the PSF preparation means W3 of FIG. 14 is (W40) means for setting the maximum number of iterations 1 to 5, (W41) all the degradation indications in the PSF database 13 and associated with them.
- the PSF preparation means W3 is executed in advance for all pairs of luminance distributions of PSFs to obtain one series of PSF luminance distributions for each deterioration indication, and one series of PSFs for each of these deterioration indications.
- FIG. 15 shows an example of data in the PSF database 13 according to the present invention.
- the PSF database 13 in FIG. 15 is a CSV format table and is stored in the HDD.
- the degradation indication 2 of 256 levels is described in the first column from the left
- the image file name 23 of the PSF luminance distribution 9 corresponding to each degradation indication 2 is the second column from the left as a path. It is described in the eye.
- the image file of the PSF luminance distribution 9 exists under a directory called PSF in the C drive in the HDD
- the image file name of the PSF luminance distribution 9 corresponds to the degradation index 2.
- the degradation indication 2 is 1, the image file name C: ⁇ PSF ⁇ PSF_1.
- the means of W30 searches the PSF database 13 using the degradation indication number 2 as search information, acquires the image file name of the PSF luminance distribution 9 associated with the degradation indication number 2, and the PSF luminance distribution 9
- the luminance distribution 9 of PSF corresponding to the image file name is acquired by reading from the HDD.
- the PSF database 13 can be defined as a two-dimensional array other than the CSV format.
- FIG. 16 shows an example of data in the expanded PSF database 18 according to the present invention.
- the expanded PSF database 18 of FIG. 16 is a CSV format table and is stored in the HDD.
- 256-step deterioration indication 2 is written in the first column from the left, and the image file name 23 of the PSF luminance distribution 9 corresponding to each deterioration indication 2 is the second from the left as a path. It is described after the column.
- the second and subsequent columns from the left are associated with the number of iterations.
- the second column from the left lists the number of iterations 1, and the third column from the left is the iteration.
- the number of operations 1 is 2, and the last column lists the number of iterations n_max.
- the image file of the PSF luminance distribution 9 exists under the directory SPSF in the C drive in the HDD, and the image file name of the PSF luminance distribution 9 corresponds to the degradation indication number 2 and the number of iterations 1.
- the image file name 23 of the PSF luminance distribution 9 in the second column from the left of the top row of the expanded PSF database 18 is C: ⁇ SPSF ⁇ SPSF_1_1.
- bmp indicates the storage location of the image file of the PSF luminance distribution 9 in which the degradation indication number is 2 and the number of iterations is 1.
- the means of W42 searches the expanded PSF database 18 using the degradation indication number 2 as search information, identifies a row in the expanded PSF database 18 associated with the degradation indication number 2, and determines the PSF luminance distribution 9 in that row.
- a row in the expanded PSF database 18 associated with the degradation indication number 2 are obtained from the column corresponding to the number of iterations of 1 to the column of the last iterations of n_max, and stored in a two-dimensional array called SPSF (BF, n), thereby making a series of PSFs. Is obtained.
- SPSF (BF, n) is a character type two-dimensional array
- BF is a variable that means a degradation index of 2
- n is a variable that means the number of iterations.
- the means of W9 obtains the image file name 23 of the PSF luminance distribution 9 from SPSF (BF, n), assuming that the degradation index 2 is BF and the number of iterations is n, and this PSF luminance distribution 9
- the PSF luminance distribution 9 corresponding to the image file name 23 is obtained by reading from the HDD.
- the expanded PSF database 18 can be defined as a two-dimensional array other than the CSV format.
- FIG. 17 shows an example of the configuration of the PSF restoration means according to the eighteenth aspect of the present invention.
- the white squares containing the means numbers are means other than judgment and conditional branching
- the white diamonds containing the means numbers are judgment and conditional branching means
- the arrows are data
- a white square containing a number is an element, for example, 1 if it is 1, the maximum number of iterations is 1, a thick line is a flow of processing, a white circle is a combination of processing, and a black circle is a branch of data And bond respectively.
- the PSF restoration means in FIG. 17 is (W50) means for substituting 6 for the maximum number of iterations 1, and (W51) regards the brightness distribution 15 of the initial PSF value as a deteriorated image brightness distribution and forms the brightness distribution 19 of the deteriorated PSF.
- (W52) means for making the luminance distribution 15 of the PSF initial value the estimated luminance distribution 20 of the restored PSF initial value
- (W53) means for substituting 1 into the counter and resetting
- (W54) the estimated luminance of the restored PSF initial value
- the distribution 20 is formed with the estimated luminance distribution 21 of the corrected restored PSF initial value, and further, when the luminance distribution 15 of the PSF initial value is convolved with the estimated luminance distribution 21 of the corrected restored PSF initial value, the corrected restored PSF initial value is estimated.
- the difficult calculation area generated in the periphery of the luminance distribution 21 is calculated based on the image size of the luminance distribution 15 of the PSF initial value, and the pixel of the difficult calculation area related to the upper boundary is copied.
- the copied pixels are pasted so as to be arranged outside the upper boundary of the estimated luminance distribution 21 of the corrected restoration PSF initial value so as to be mirror-symmetrical with respect to the upper boundary, and in the same manner, clockwise and then the lower side.
- restoration PSF initial value correction means for correcting the estimated luminance distribution 21 of the corrected restored PSF initial value is performed on the left side.
- the luminance distribution 15 of the PSF initial value is convolved with the estimated luminance distribution 21 of the corrected restored PSF initial value.
- Means for obtaining the sixteenth function means for inverting the sixteenth function to obtain the seventeenth function, and (W57) multiplying the seventeenth function by the luminance distribution 19 of the deteriorated PSF to obtain the eighteenth function.
- Means for obtaining, (W58) means for multiplying the estimated luminance distribution 20 of the restored PSF initial value by the estimated luminance distribution 20 to obtain the estimated luminance distribution 22 of the restored PSF, and (W59) a step of adding 1 to the counter (W60) The hypothesis that the value of the counter exceeds the maximum number of iterations 1 is verified.
- (W63) means for jumping to the means
- (W61) means for replacing the estimated luminance distribution 22 of the restored PSF with the estimated luminance distribution 20 of the restored PSF initial value, means for jumping to the means of (W62) and (W54), (W63) It is characterized by comprising means for outputting the estimated luminance distribution 22 of the restored PSF as the luminance distribution 16 of the maximum likelihood restored PSF.
- FIG. 18 illustrates an example relating to the configuration of the first restored image initial value correcting means according to the nineteenth aspect of the present invention.
- the white squares containing the means numbers are means other than judgment and conditional branching
- the white diamonds containing the means numbers are judgment and conditional branching means
- the arrows are data
- a white square containing a number is an element, for example, 1 if it is 1, the maximum number of iterations is 1, a thick line is a flow of processing, a white circle is a combination of processing, and a black circle is a branch of data And bond respectively.
- the first restored image initial value correcting means W8 in FIG. 18 is based on (W70) PSF size 7, (W70) means for making the estimated luminance distribution 6 of the restored image initial value the estimated luminance distribution 8 of the corrected restored image initial value.
- Computation difficulty that occurs at the periphery of the estimated luminance distribution 8 of the corrected restored image initial value when convolving one of the luminance distributions 3 of the PSF into the estimated luminance distribution 8 of the corrected restored image initial value Means for calculating the area 42; (W72) copying the pixels in the difficult calculation area 42 of the estimated luminance distribution 8 of the corrected restored image initial value, and copying the copied pixels to the four sides of the estimated luminance distribution 8 of the corrected restored image initial value (W73) a corrected restored image initial value that is inverted so as to be mirror-symmetric with respect to each other and further pasted so as to be placed outside the boundary of the four sides of the estimated luminance distribution 8 of the corrected restored image initial value Estimated luminance distribution The upper left corner pixel in the difficult calculation area 42
- the pixel in the lower left corner in the difficult calculation area 42 of the estimated luminance distribution 8 of the initial image value is copied, and the copied pixel in the lower left corner is rotated by 180 degrees with the vertex in the upper left corner as the rotation center, so that the corrected restored image initial Value Means for pasting and correcting in a blank area generated at the lower left corner of the constant luminance distribution 8, (W76) copying the pixel at the lower right corner in the difficult operation area 42 of the estimated luminance distribution 8 of the corrected restored image initial value
- the copied lower right corner pixel is rotated by 180 degrees with the upper right corner vertex as the rotation center, and is pasted and corrected in the blank area generated in the lower right corner of the estimated luminance distribution 8 of the corrected restored image initial value.
- FIG. 19 illustrates an example of correction of the difficult-to-calculate area 42 generated in the peripheral portion of the estimated luminance distribution 6 of the restored image initial value by the first restored image initial value correcting unit W8.
- FIG. 19 shows a state where the estimated luminance distribution 8 of the corrected restored image initial value is generated from the estimated luminance distribution 6 of the restored image initial value by means of W70 to 72. In this state, there are blank portions at the four corners of the estimated luminance distribution 8 of the corrected restored image initial value. Due to W73, the blank portion at the upper left corner of the estimated luminance distribution 8 of the corrected restored image initial value copies the pixel at the upper left corner of the operation difficult region 42 of the estimated luminance distribution 6 of the restored image initial value, and this copied region is copied. It is compensated and corrected by rotating 180 degrees around the top left corner vertex and pasting.
- FIG. 20 illustrates an example of the configuration of the second image restoration unit W88 as the second mode of the first image restoration unit W20 according to the twentieth aspect of the present invention.
- the white squares that contain the means numbers are means other than judgment and conditional branching
- the white diamonds that contain the means numbers are judgment and conditional branching means
- the arrows are data
- a white square containing a number is an element, for example, 1 if it is 1, the maximum number of iterations is 1, a thick line is a flow of processing, a white circle is a combination of processing, and a black circle is a branch of data And bond respectively.
- the second image restoration means W88 in FIG. 20 (W80)
- the luminance distribution 3 of a series of PSFs of the nth PSF PSF providing means for providing the luminance distribution 17 as the PSF luminance distribution 9 to the first single image restoration means W81-n at the n-th stage, (W81) PSF luminance distribution 9, estimated luminance distribution 6 of the restored image initial value.
- an operation corresponding to one of the iterative calculations based on the Bayse probabilistic formula is performed to obtain the estimated luminance distribution 10 of the most likely restored image in the luminance distribution 4 of the deteriorated image.
- the first single image restoration unit W81 outputs (W82) second restored image initial value correction unit having the same configuration as the first restored image initial value correction unit W8.
- (W83) means for obtaining a nineteenth function by convolving the luminance distribution 9 of the PSF with the estimated luminance distribution 8 of the corrected restored image initial value; Means for inverting the function to obtain the twentieth function;
- (W85) means for multiplying the twentieth function by the luminance distribution of the degraded image to obtain the twenty-first function; and (W86) restoring the twenty-first function to the restored image initial value.
- (W87) means for outputting the estimated luminance distribution 10 of the restored image, and (W88) a first single-divided image at the nth stage.
- the output of the means (W87) of the restoration means W81-n is connected to the means (W82) of the (n + 1) th first single-part image restoration means W81- (n + 1) of n_max stages.
- Second image restoration hand consisting of serially connected body
- n_max iterations that are the number of first single-part image restoration means W81 connected in series are performed, and the first single-part image restoration means W81 ⁇ in the n_max stage is performed. It is characterized in that the estimated luminance distribution 10 of the restored image output from n_max is output as the luminance distribution 11 of the maximum likelihood restored image.
- FIG. 21 illustrates an example of the configuration of the third image restoration unit W107 as the third mode of the first image restoration unit W20 according to the twenty-first aspect of the present invention.
- the white squares containing the means numbers are means other than judgment and conditional branching
- the white diamonds containing the means numbers are judgment and conditional branching means
- the arrows are data
- a white square containing a number is an element, for example, 1 if it is 1, the maximum number of iterations is 1, a thick line is a flow of processing, a white circle is a combination of processing, and a black circle is a branch of data And bond respectively.
- the third image restoration means W107 in FIG. 21 is a means for resetting by assigning 0 to the (W90) counter, (W91) means for resetting by assigning 1 to the second counter, and the value of the (W92) counter is The hypothesis that it is other than 0 is verified, and if the verification result is false, proceed to the means of (W93), and if the verification result is true, the means of jumping to the means of (W96), (W93) Means for transferring the luminance distribution 4 to the degraded image storage buffer 26 and the restored image initial value buffer 27, means for jumping to the means (W94) and (W96), and estimated brightness of the restored image by means (W95) and (W102) Means for transferring the distribution 10 to the restored image initial value buffer 27.
- (W102) means for multiplying the estimated luminance distribution 6 of the restored image initial value to obtain the estimated luminance distribution 10 of the restored image;
- (W104) Means to add 1 to the second counter,
- (W104) Means to add 1 to the second counter,
- (W105) Verify the hypothesis that the counter value exceeds the maximum number of iterations 1, and if the verification result is false Jump to the means of (W95), and if the verification result is true, proceed to the means of (W106), (W106) Means to output the estimated luminance distribution 10 of the restored image as the luminance distribution 11 of the maximum likelihood restored image (W107)
- FIG. 22 shows an example of the configuration of the degraded image preparation means W4 according to the twenty-second aspect of the present invention.
- the white squares containing the means numbers are means other than judgment and conditional branching
- the white diamonds containing the means numbers are judgment and conditional branching means
- the arrows are data
- a white square containing a number is an element, for example, 1 if it is 1, the maximum number of iterations is 1, a thick line is a flow of processing, a white circle is a combination of processing, and a black circle is a branch of data And bond respectively.
- the degraded image preparation means W4 in FIG. 22 is (W110) an RGB signal extraction means for extracting an RGB signal 28 constituting a frame from one frame of the TV video signal 5, and (W111) one TV video signal 5 of one frame. (W112) YUV converting means for converting the RGB signal 28 into YUV signal 30 by YUV conversion, and (W113) YUV signal 30 Among them, the luminance distribution 4 of the deteriorated image consisting only of the Y signal as the luminance component is extracted to form the luminance distribution 31 of the Y deteriorated image, and only the U deteriorated image distribution 32 consisting of only the remaining U signals and the V signal.
- Y deteriorated image extracting means for holding the V deteriorated image distribution 33 composed of: (W114) Brightness of one frame by performing degamma processing of the luminance distribution 31 of the Y deteriorated image
- De-gamma processing means for outputting forms the luminance distribution 4 of the degraded image consisting of distribution, characterized there in that comprises a.
- FIG. 23 shows an example of the configuration of the TV visualization means W22 according to the twenty-third aspect of the present invention.
- the white squares containing the means numbers are means other than judgment and conditional branching
- the white diamonds containing the means numbers are judgment and conditional branching means
- the arrows are data
- a white square containing a number is an element, for example, 1 if it is 1, the maximum number of iterations is 1, a thick line is a flow of processing, a white circle is a combination of processing, and a black circle is a branch of data And bond respectively.
- a (G120) gamma processing unit that performs gamma processing of the luminance distribution 11 of the maximum likelihood restored image, and (W121) a U-degraded image distribution 32 and V-degradation held by the Y-degraded image extracting unit W113.
- Restored image composition means for synthesizing one YUV restored image distribution 34 from the luminance distribution 11 of the maximum likelihood restored image after the gamma processing composed of the image distribution 33 and the Y component, (W122) RGB of the distribution 34 of the YUV restored image RGB conversion means that performs conversion and forms an RGB restored image distribution 35, (W123) RGB signal conversion means that reads the RGB restored image distribution 35 and outputs an RGB signal 36, and (W124) an RGB signal 36 that is output by the delay means W111.
- the TV video signal that is combined with the remaining TV video signal 29 and output as a super-resolution TV video signal 12 composed of a TV video signal for one frame. Synthesis means, a feature is in the can with a.
- the fourth accelerated super-resolution processing program 43 describes a processing procedure for configuring and executing all the means in the preparation means W19 and the first image restoration means W20. is there.
- the fifth accelerated super-resolution processing program 44 describes a processing procedure for configuring and executing all the means in the preparation means W19 and the second image restoration means W88. is there.
- the sixth accelerated super-resolution processing program according to the twenty-sixth aspect of the present invention describes a processing procedure for configuring and executing the preparation means W19 and the third image restoration means W107.
- the fourth acceleration super-resolution processing program 43, the fifth acceleration super-resolution processing program 44, and the sixth acceleration super-resolution processing program 45 are respectively encrypted.
- the fourth acceleration super-resolution processing program 43, the fifth acceleration super-resolution processing program 44, and the sixth acceleration super-resolution processing program 45, which are encrypted, can be connected to the computer and read by the computer. This is a possible second storage medium 47.
- the first embodiment is an accelerated super-resolution processing system 49 for TV images in which an accelerated super-resolution processing device 40 for TV images according to the present invention is configured in a computer 48 by a first accelerated super-resolution processing program 37.
- FIG. 24 illustrates an example of the configuration of the accelerated super-resolution processing system 49 for TV video according to the first embodiment of the present invention.
- 24 is installed with a digital TV video signal input / output terminal 50 used to input / output a digital TV video signal and an accelerated super-resolution processing device 40 for TV video.
- the degraded image preparation means W4 for preparing the computer 48, the luminance distribution 4 of the deteriorated image from the TV image 67, the TV image input board 51 constituted by the FPGA, and the luminance distribution 11 of the maximum likelihood restored image are converted into a TV image and equivalent to one frame.
- the TV visualization means W22 for outputting the super-resolution TV video signal 12 is provided by the super-resolution TV video output board 52 constituted by the FPGA, the video display mode designated by the video display mode designation button 56 and the condition display designation button 57.
- Super-resolution processing mode switching control circuit board 5 for reflecting the specified presence / absence of superposition on the super-resolution TV video signal 12 , Digital TV video cable 54, and a bus cable 55,. In FIG. 24, the power cable is omitted.
- the degraded image preparation means W4 and the TV visualization means W22 of the accelerated super-resolution processing device 40 for TV video in the computer 48 constituting the accelerated super-resolution processing system 49 for TV video in FIG. 24 use FPGA. Since it is hardware, the degraded image preparation means W4 is configured as means for acquiring the luminance distribution 4 of the degraded image output from the TV video input board 51, and the TV imaging means W22 is applied to the super-resolution TV video output board 52. This is configured as means for transferring the luminance distribution 11 of the likelihood restored image.
- FIG. 25 illustrates an example of the configuration of the first accelerated super-resolution processing program 37 and an example of a situation in which the program is installed in the computer 48.
- the first acceleration super-resolution processing program 37 is in the first acceleration super-resolution processing means W21 in the TV video accelerated super-resolution processing device 40 that performs processing according to the TV video accelerated super-resolution processing method.
- the super-resolution processing window monitoring and coping program 63 that performs the processing for executing the above-described processing until the close button 79 in the super-resolution processing window 64 is pressed, and includes an accelerated super-resolution processing device program 61 for TV video,
- the resolution processing window creation program 62 and the super resolution processing window monitoring and handling program 63 are executed in this order.
- the first accelerated super-resolution processing program 37 stored in the first storage medium 46 is installed in the computer 48, and the TV 48
- An accelerated super-resolution processor icon 58 for TV video that is used to activate the accelerated super-resolution processor 40 for video is placed on the screen in the monitor 59 of the computer 48.
- the user 60 clicks on the accelerated super-resolution processing device icon 58 for TV video to activate the accelerated super-resolution processing device 40 for TV video.
- the super-resolution processing window 64 is displayed on the monitor 59 and is in a state of waiting for an action from the user 60.
- FIG. 26 illustrates an example relating to the configuration of the super-resolution processing window 64.
- a super-resolution processing window 64 in FIG. 26 is used to display a video window 69 for displaying a TV video 67 and a super-resolution TV video 68, a current date and time, a super-resolution condition, a message from the system, and the like.
- the information window 70 in which the vertical scroll bar is provided at the right end one of 256 levels from 0 to 255 for the deterioration indicator 2 is displayed as a pull-down menu (in the pull-down menu of the present invention, the default is displayed with a light blue background.
- Degradation indication setting button 71 for specifying and setting one number, and clicking this opens a combo box, so select a value in the list or select the maximum value from the default value displayed in the text box
- the maximum number of iterations setting button 72 used to directly set the number of iterations 1 and set the super-resolution processing.
- a super-resolution processing start button 73 that is clicked to start under the specified conditions
- a super-resolution processing stop button 74 that is clicked to interrupt the super-resolution processing
- Resolution processing restart button 75 super resolution processing stop button 76 that is clicked to stop the super resolution processing, a help button 77 for opening a help window to search for and displaying help contents, and reducing / reducing in the video window 69 Click and specify the image you want to enlarge, then press + in this button to enlarge the specified image. Press-to enlarge the specified image. Press 0 to return to the default magnification and display the specified image.
- Video enlarge / reduce button 78 video display mode in video window 69 is halved test mode, input video mode for displaying only TV video 67 or super Video display mode designation button 56 used to select one of the super-resolution TV video modes for displaying only the image TV video 68 from the pull-down menu, the maximum number of iterations 1 for the super-resolution processing condition, and the deterioration indication
- This is composed of a toggle switch condition display designation button 57 used to superimpose the numerical expression 2 on the upper right corner of the super-resolution TV image 68 and a close button 79 used to close the super-resolution processing window 64.
- FIG. 26 shows a video display state in the test mode.
- the computer 48 includes a 64-bit instruction set 32-bit 6-core / chip CPU (Central Processing Unit), a GPU (Graphic Processing Unit), a memory of 32 GB or more, an HDD (Hard Disk Drive) having a storage capacity of 1 Terabyte or more, 128 G Wireless communication modules such as SDD (Solidstate Disk Drive), 3 or more USB terminals, 1 or more LAN (Local Area Network) terminals, WiFi (Wireless Fidelity), Bluetooth (registered trademark), etc.
- a 64-bit instruction set 32-bit 6-core / chip CPU Central Processing Unit
- GPU Graphic Processing Unit
- HDD Hard Disk Drive
- 128 G Wireless communication modules such as SDD (Solidstate Disk Drive), 3 or more USB terminals, 1 or more LAN (Local Area Network) terminals, WiFi (Wireless Fidelity), Bluetooth (registered trademark), etc.
- Telephone terminal keyboard 65, mouse 66, FHD (Full High Definition) Spray, Windows (registered trademark) 8 O / S (Operating System) manufactured by Microsoft, Visual Studio 2010 (TM) (including Visual C ++ 2010) manufactured by Microsoft, Office 2013 (TM) manufactured by Microsoft, WiFi, It can communicate with other computers via Bluetooth (registered trademark), LAN, USB, and the Internet.
- O / S Operating System
- TM Visual C ++ 2010
- TM Office 2013
- WiFi WiFi
- It can communicate with other computers via Bluetooth (registered trademark), LAN, USB, and the Internet.
- the computer 48 a computer selected from many types of desktop computers mass-produced and distributed in the market is used, and a workstation having the same standard may be used.
- a workstation having the same standard may be used.
- a server may be sufficient.
- FIG. 27 shows an example of a procedure for executing super-resolution processing in the accelerated super-resolution processing system 49 for TV video as a transaction table.
- the transaction table in FIG. 27 includes numbered solid arrows indicating input processing by the user 60, dotted dotted arrows with letters indicating the response by the computer 48, start symbols, end symbols, and thick solid arrows indicating the time axis on the user 60 side. It is composed of a thick dotted line arrow indicating the time axis on the computer 48 side.
- the procedure until the super-resolution processing is executed in the TV video accelerated super-resolution processing system 49 is as follows: (1) In the “start-up” step, the user 60 performs the TV video accelerated super-resolution processing. When the device icon 58 is clicked, the computer 48 displays (A) the super-resolution processing window 64 on the monitor 59, and (2) in the “super-resolution processing condition input-1” step, the user 60 presses the deterioration indication setting button. 71, the computer 48 expands and displays a pull-down menu at the position of the (B) deterioration index setting button 71, and (3) in the “super-resolution processing condition input-2” step, the user 60 selects an appropriate value from the pull-down menu.
- the computer 48 selects (C) the deterioration index 2 and closes the pull-down menu, and (4) “Super-Resolution”.
- the user 60 clicks the maximum number of iterations setting button 72, and the computer 48 expands the combo box at the position of the (D) maximum number of iterations setting button 72, and (5)
- the “Resolving processing condition input-4” step the user 60 selects or inputs an appropriate numerical value from the combo box, and the computer 48 (E) captures the maximum number of iterations 1 and closes the combo box.
- the user 60 clicks the video display mode designation button 56, and the computer 48 expands a pull-down menu at the position of (F) video display mode designation button 56, and (7) “Super”
- the user 60 selects an appropriate mode from the pull-down menu, and The data 48 takes in the (G) video display mode designation button 56 and closes the pull-down menu.
- the user 60 clicks the condition display designation button 57 and 48 indicates that (H) the color of the condition display designation button 57 is inverted and the display setting is displayed.
- the super-resolution processing window 64 constantly monitors access to the buttons of the super-resolution processing window 64 even during the super-resolution processing, if the super-resolution processing condition is changed, the user 60 Click the degradation indication setting button 71, the maximum number of iterations setting button 72, the video display mode designation button 56, the condition display designation button 57, and reset the conditions so that the video remains displayed and the super-resolution processing conditions Changes are reflected. If the user 60 wishes to end the super-resolution processing, the user 60 clicks the close button 79. Then, in response to this, the computer 48 closes the super-resolution processing window 64 and further closes the accelerated super-resolution processing device 40 for TV video.
- step (1) the user 60 first executes step (1) to display the super-resolution processing window 64 on the monitor 59, and then executes step (2) to indicate deterioration.
- the pull-down menu is expanded and displayed at the position of the number setting button 71, and it is felt that the degree of deterioration of the TV image 67 is small while watching the terrestrial digital TV 80 on which the TV image 67 is displayed in step (3).
- 14 is selected as the default value 60) in 256 steps, and then step (4) is executed to expand the combo box at the position of the maximum number of iterations setting button 72, and the TV image is displayed in step (5).
- step (6) is executed to expand the pull-down menu at the position of the video display mode designation button 56, and in step (7), the super-resolution processing condition has not yet been determined.
- step (8) is executed to set the condition display designation button 57 to display setting, and then step (9) is executed to execute the super-resolution processing, so that the super-resolution processing window is displayed.
- the TV image 67 and the super-resolution TV image 68 are displayed side by side in the 64 video windows 69, and the user 60 operates the digital terrestrial TV 80 considering that the super-resolution processing conditions are acceptable.
- the input setting is set to video, and the video displayed in the video window 69 is displayed on the full screen of the terrestrial digital TV 80.
- the user 60 operates the video display mode designation button 56 to perform super-resolution TV. Change in image mode with the full-screen super-resolution TV image 68 in the image window 69 and digital terrestrial TV80, watched.
- FIG. 28 is a diagram illustrating an example of the state of super-resolution processing according to the first embodiment.
- one frame of the video displayed in the half state where the TV video 67 and the super-resolution TV video 68 in the test mode are horizontally arranged is displayed.
- the right side of the half is one frame of the TV video 67 before the super-resolution processing
- the left side of the half is one frame of the super-resolution TV video 68 obtained by super-resolution processing of the right frame of the half.
- the degradation index 2 is displayed with the symbol F
- the maximum number of iterations 1 is displayed with the symbol I
- the video display mode is the test mode.
- test Some are abbreviated as test. From FIG. 28, although the display start position of the TV video 67 is slightly shifted to the right from the super-resolution TV video 68, the super-resolution quality by the accelerated super-resolution processing system 49 for TV video according to the first embodiment of the present invention is shown. It can be seen that it has sufficient practicality.
- the first image restoration means W20 constituting the accelerated super-resolution processing apparatus 40 for TV video according to the present invention can be changed to the second image restoration means W88 by using the second accelerated super-resolution processing program 38, and although it can be changed to the third image restoration means W107 by using the 3-accelerated super-resolution processing program 39, the system configured in the computer 48 is an accelerated super-resolution processing system 49 for TV video.
- the second image restoration means W88 and the third image restoration means W107 are image restoration means suitable for hardware, but the TV video accelerated super-resolution processing system 49 is a software base that uses hardware for the signal processing portion. There is no significant difference in processing speed, and there is no difference in processing quality.
- Example 2 is a first set-top box 81 in which the second aspect of the accelerated super-resolution processing device 40 for TV video using the hardware-made second image restoration means W88 is boxed.
- all the preparation means W19 are hardware except that the PSF preparation means W3-3 uses the software-defined computer board 93, and the second image restoration means W88 also uses hardware. Therefore, everything is composed of hardware. For this reason, it is possible to operate at a higher speed than the accelerated super-resolution processing system 49 for TV video of the first embodiment.
- the super-resolution quality of the first set-top box 81 is the same as that of the accelerated super-resolution processing system 49 for TV images of the first embodiment.
- FIG. 29 is a diagram illustrating an example of the internal configuration of the first set top box 81.
- 29 includes a digital TV video signal input / output terminal 50, a TV video input board 51, a super-resolution TV video output board 52, and a super-solution used for inputting / outputting digital TV video signals.
- F setting means 82 used for setting the degradation indication 2 indicating the degree of degradation of the image processing mode switching control circuit board 53, the digital TV video cable 54, the bus cable 55, and the TV video 67, and the maximum number of iterations 1 are set.
- I setting means 83 used for the purpose, mode switch 84 for switching the super-resolution processing mode to the test or the main processing, super-resolution processing mode and whether the F setting means 82 and the value of the I setting means 83 are superimposed on the TV video signal LC for constantly monitoring the monitor switch 85 for specifying whether or not, the mode of the super-resolution processing, and the values of the F setting means 82 and the I setting means 83 (Liquid Crystal Display) monitor 86, power switch 87 used to turn on / off the first set-top box 81, LED (Light Emitting Diode) lamp 88, case 89 that lights only when the power switch 87 is on
- the second image restoration means W88 of the accelerated super-resolution processing device 40 for TV video automatically determines whether the digital TV video signal is an interlace system or a progressive TV video signal, using an F circuit board 90 manufactured using FPGA.
- a computer board 93 having a computer 92 pre-installed with PSF preparation means W3-3 for outputting a luminance distribution 3 of a series of PSFs, and a computer 48 constitute all means in the PSF preparation means W3-3.
- Optimal driving conditions from PSF preparation means configuration program 94 describing the processing procedure of these means HDD 95, power supply circuit board 96 for providing appropriate power to necessary ones in first set top box 81, and computer board 93 Is provided with a heat dissipation fan 97 that supports heat dissipation with an appropriate amount of air, a LAN terminal 98, USB terminals 99 to 101, a commercial AC single phase 100V power cable 102, a power cable 103, a signal line 104, and a signal line 105. ing.
- the super-resolution processing mode switching control circuit board 53 always reads and automatically determines the state of the mode switch 84 and the monitor switch 85, and (mode 1) if the mode switch 84 is in the main processing mode and the monitor switch 85 is not superimposed. 1 receives the TV video signal 5 for one frame from the up-converter circuit board 91, outputs it to the TV video input board 51, and further outputs one frame after super-resolution processing output from the super-resolution TV video output board 52.
- Mode 3 If the mode switch 84 is in the test mode and the monitor switch 85 is not superimposed, the TV video signal 5 for one frame is received from the up-converter circuit board 39 and duplicated, and one of them is input as the TV video. At the same time as the output to the substrate 51, the other is delayed by one frame, the delayed one is compressed to a horizontal width of 1 ⁇ 2 so that it becomes the right half of one screen, and further the super-resolution TV video output substrate 52 The super-resolution TV video signal 12 for one frame after super-resolution processing output from is compressed to 1 ⁇ 2 the horizontal width and synthesized so that it becomes the left half of one screen.
- the TV video signal for one frame adjusted so that the right half of the surface is the frame before the super-resolution processing and the left half is the frame after the super-resolution processing is output to the digital TV video signal input / output terminal 50; Mode 4) If the mode switch 84 is in the test mode and the monitor switch 85 is superimposed, the TV video signal 5 for one frame is received from the up-converter circuit board 91 and duplicated, and one is output to the TV video input board 51. At the same time, the other one is delayed by one frame, and the delayed one is compressed to 1/2 the horizontal width so that it becomes the right half of one screen, and is further output from the super-resolution TV video output board 52.
- the right half of one screen is obtained by compressing the super-resolution TV video signal 12 for one frame after the super-resolution processing so as to be compressed to 1/2 the horizontal width to be the left half of one screen.
- Super-resolution processing Digital TV video signal input / output terminal that superimposes the information read from the computer board 93 on the TV video signal for one frame adjusted so that the left half in the previous frame becomes the frame after super-resolution processing, in the upper right corner Output to 50.
- the F setting means 82 is used to set a deterioration index 2 corresponding to the optical deterioration degree of the TV image 67 projected on the TV monitor 106 of the terrestrial digital TV 80, and is an accelerated super-resolution process for TV images. It corresponds to the means W2 of the device 40.
- the F setting means 82 is a bit switch capable of setting a decimal number of three digits, and data and a power supply are performed via the bus cable 55, and a specification that can directly read the set value is preferable. Those with this specification are available on the market.
- the I setting means 83 is used to set the maximum number of iterations 1 and is a bit switch that can set a decimal number of 3 digits. Data and power are supplied via the bus cable 55 and set. A specification whose value can be read directly is preferable. Those with this specification are available on the market.
- the computer board 93 is the latest Windows (registered trademark) OS computer and is a PC motherboard that can use C ++ language.
- the PC motherboard has many external connection terminals such as various I / O (Input / Output interface), LAN terminal 98, USB terminals 99 to 101, microphone terminal, speaker terminal, etc., and the price is also so expensive. This is suitable for the second embodiment.
- I / O Input / Output interface
- LAN terminal 98 Local Area Network
- USB terminals 99 to 101 a USB terminals 99 to 101
- microphone terminal speaker terminal
- the C ++ language may be installed after purchasing the PC motherboard.
- the PSF preparation means configuration program 94 is a program written in the C ++ language and can be read and executed by the CPU mounted on the computer board 93.
- the PSF preparation means W3-3 installed in advance is installed in the computer board 93.
- an expanded PSF database 18 defined in the computer and referred to by the PSF preparation means W3-3 is configured in the HDD 95, and a control system for the cooling fan 97 is defined.
- the control system of the PSF preparation means W3-3 and the cooling fan 97 is automatically executed after initialization when the power switch 87 is turned on.
- the control system of the PSF preparation means W3-3 and the control system of the cooling fan 97 operates in an environment provided by the Windows (registered trademark) OS, and the Windows (registered trademark) OS provides.
- the HDD 95 includes a Windows (registered trademark) OS (not shown), C ++ (not shown), and the like.
- the HDD 95 has a capacity of 1 Tbytes or more, is capable of random high-speed reading and writing, is equipped with a bus interface and a buffer of 8 Mbytes or more, and transmits and receives data and signals and supplies power to the bus cable. Anything can be used as long as it is of the specification performed via the H.55. There are many types in the market, but a 2.5 inch size is desirable due to space limitations.
- the up-converter circuit board 91, the TV video input board 51, the super-resolution TV video output board 52, and the super-resolution processing mode switching control circuit board 53 need to be made by using FPGA, the up-converter circuit board 91 and the power source For the circuit board 96, it is possible to purchase an off-the-shelf product that satisfies the design specifications.
- FIG. 30 illustrates an example of the setup situation of the first set top box 81.
- the user 60 connects the digital signal cable 107 extending from the digital video signal / audio input / output terminal of the terrestrial digital TV 80 to the digital TV video signal input / output terminal 50 of the first set top box 81.
- the user 60 first turns on the power by turning on the power switch 87 of the first set-top box 81, then turns on the terrestrial digital TV 80, and uses the channel commander 108 of the terrestrial digital TV 80 to select a desired one.
- the mode switch 84 is switched to the test and the monitor switch 85 is switched to the superimposed state while checking the TV broadcast video on the TV monitor 106 of the terrestrial digital TV 80 according to the channel.
- the TV broadcast video before super-resolution processing flows in the right half of the TV monitor 106, and the TV broadcast video after super-resolution processing under the default setting conditions flows in the left half without delay.
- the user 60 uses the F setting means 82 to set the deterioration indication 2 to 60 to 30 in the default 256 steps, and uses the I setting means 83 to set the maximum number of iterations to the default because the frame deterioration is small.
- the change in image quality after the super-resolution processing by the substantially real-time processing was confirmed on the TV monitor 106. As a result, it has been found that there is no significant change in image quality, and it may be less than the default value.
- the user 60 searches the optimum image quality state by changing the F setting unit 82 and the I setting unit 83 while checking the change in the image quality after the super-resolution processing by the substantially real-time processing on the TV monitor 106. be able to.
- the mode switch 84 is switched and the TV after the super-resolution processing is performed on the entire screen of the TV monitor 106.
- the super-resolution processing conditions can be changed.
- FIG. 31 illustrates an example of the relationship between the number of iterations based on the development data of the TV video super-resolution processing method by the present inventor and the LSI scale.
- the standard for the number of iterations 1 is 2, which corresponds to the leftmost point in FIG.
- the second point from the left in FIG. 31 corresponds to the standard of the maximum number of iterations in the conventional super-resolution processing method for TV video by the present inventor, and the number is six. As is clear from FIG.
- the number of iterations is reduced to 1/3, and the image restoration means is converted to FPGA and LSI.
- the number of gates can be greatly reduced from 1.5 million gates to 70,000 gates.
- FIG. 32 is a diagram showing an example of comparison of super-resolution processing quality by the conventional method and the accelerated super-resolution processing method for TV video by the present inventor.
- the left image in FIG. 32 is an unprocessed deteriorated image, and is composed only of the luminance component of the color standard image.
- the central image in FIG. 32 is based on the accelerated super-resolution processing method for TV video, and is obtained by using the first set-top box 81 modified to process one frame instead of video, and the maximum iterative calculation.
- the number of times 1 is 2, and the degradation indication 2 is 30.
- the accelerated super-resolution processing method for TV video has better super-resolution processing quality than the conventional method by the present inventors. This result is consistent with many experimental results, and is an example that the super-resolution processing quality of the accelerated super-resolution processing method for TV video is equal to or higher than the conventional method by the present inventor.
- FIG. 33 is a diagram illustrating an example of the relationship between the degree of degradation of the standard image and the super-resolution processing quality based on the number of repeated operations. Accelerated super-resolution processing for TV video using a first set-top box 81 that has been modified to a single-frame process instead of video, preparing standard images with a degradation level corresponding to degradation index 2 of 30, 54, 74 According to the method, the super-resolution processing images obtained when the super-resolution processing is performed with the maximum number of iterations 1 being 2 and 3 are summarized for the degradation indication 2 and the maximum number of iterations 1.
- FIG. 33 From FIG.
- the degradation indication 2 is in the range of 30 to 74, the super-resolution processing quality is good if the maximum number of iterations 1 is at least 2, and the degradation indication 2 and the maximum number of iterations 1 It can be seen that the effects of can be ignored.
- the range in which the degradation index 2 is 30 to 74 corresponds to a considerably bad range from the standard in the current terrestrial digital TV video, and it can be seen that the accelerated super-resolution processing method for TV video is suitable for the processing of terrestrial digital TV video.
- the third embodiment applies the third aspect of the accelerated super-resolution processing device 40 for TV images in which the second image restoring means 88 of the accelerated super-resolution processing device 40 for TV images according to the present invention is changed to a third restoring means.
- the hardware configuration of the second set top box 109 is exactly the same as that of the first set top box 81.
- the difference between the first set top box 81 and the second set top box 109 is that the means configured on the F circuit board 90 is changed from the second image restoration means W88 to the third image restoration means. For this reason, FIG. 29 is not rewritten and shown again.
- the present invention can be applied to any application as long as it uses TV video, software such as precision equipment and electronic equipment, applications and games for developing and manufacturing video cameras and digital cameras. It can be used in development, medical equipment such as endoscopes and MRI, information equipment such as monitors, disaster prevention and security equipment such as surveillance cameras, and archiving.
- Super-resolution processing start button 74 ... Super-resolution processing stop button, 75 ... Super-resolution processing restart button, 76 ... Super Resolution processing stop button, 77 Help button, 78... Enlarge / reduce button, 79. Close button, 80. Terrestrial digital TV, 81... First set top box, 82... F setting means, 83. -I setting means, 84 ... mode switch, 85 ... monitor switch, 86 ... LCD monitor, 87 ... power switch, 88 ... LED lamp, 89 ... case, 90 ... F circuit board, 91 ... up-converter circuit board, 92 ... computer, 93 ... not shown, 94 ... PSF preparation means configuration program, 95 ... HDD, 96 ...
- power supply circuit board 97 Radiating fan, 98 ... LAN terminal, 99 ... USB terminal, 100 ... USB terminal, 101 ... USB terminal, 102 ... Commercial AC single phase 100V power cable DESCRIPTION OF SYMBOLS 103 ... Power cable, 104 ... Signal line, 105 ... Signal line, 106 ... TV monitor, 106 ... Not shown, 107 ... Digital signal cable, 108 ... Channel commander 109 ... second set top box, F ... degradation reading, FCN13 ... 13th function, FCN14 ... 14th function, FCN15 ... 15th function, FNC16 ... 16th function, FNC17 ... 17th function, FNC18 ... 18th function, FNC19 ... 19th function, FNC20 ...
- first image restoration step S21 ... first accelerated super-resolution processing step, S22 ... TV imaging step, S30 to S63 ... step, S64 ... PSF restoration step, S65 to S76 ... step, S80 ... PSF provision step, S81 ... first single image restoration step, S81-1 ... The first single image restoration process in the first stage, S81-2 ... the second stage S81-n_max... N_max stage first single image restoration process, S81-n... Nth stage first single image restoration process, S82 to S87. Step, S88 ... Second image restoration step, S90 to S106 ... Step, S107 ... Third image restoration step, S110 ... RGB signal extraction step, S111 ... Delay step, S112 ...
- first acceleration super-resolution processing means W22 ... TV imaging means, W30 to W63 ... means, W64 ... PSF restoration means, W70 to W80... Means, W81... First single image restoration means, W81-1... First stage first single image restoration means, W81-2.
- Second stage first single image restoration means W81-n_max... N_max stage first single image restoration means, W81-n... Nth stage first single image restoration means, W82 to W87 ... means, W88 ... second image restoration means, W90 to W106 ..., means W107 ... third image restoration means, W110 ... RGB signal extraction means, W111 ... delay means, W112 ... YUV conversion means, W113 ... Y deteriorated image extraction means, W114 ... De-gamma processing means, W120 to W124 ... means
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Picture Signal Circuits (AREA)
Abstract
Description
Claims (27)
- 1フレーム分のTV映像信号に含まれるフレームから光学的劣化を低減して劣化前の1フレーム分のTV映像信号を復元するTV映像向け加速超解像処理方法であって、
(S1)最大反復演算回数を設定する工程と、
(S2)前記TV映像を見ながら前記TV映像の劣化状況に適した劣化示数を指定する劣化示数指定工程と、
(S3)前記劣化示数に紐づいた第1番目のPSF(Point Spread Function)の輝度分布及びこの系から成り反復演算回数に紐づいて整理された1連のPSFの輝度分布を準備するPSF準備工程と、
(S4)1フレーム分の輝度分布から成る劣化画像の輝度分布を1フレーム分のTV映像信号から準備する劣化画像準備工程と、
(S5)前記劣化画像の輝度分布を復元画像初期値の推定輝度分布と成す復元画像初期値準備工程と、
(S6)前記1連のPSFの輝度分布は互いに同じ画像サイズであり、この画像サイズをPSFサイズと呼ぶ時、このPSFサイズを求めるPSFサイズ取得工程と、
(S7)反復演算回数を計数するカウンターに1を設定する第1リセット工程と、
(S8)前記復元画像初期値の推定輝度分布を複写してこれを補正復元画像初期値の推定輝度分布と成し更に前記補正復元画像初期値の推定輝度分布をPSFサイズを基にして補正する第1復元画像初期値補正工程と、
(S9)前記1連のPSFの輝度分布から前記カウンターの値に紐づいた1つを選択してこれをPSFの輝度分布と成すPSF選定工程と、
(S10)前記補正復元画像初期値の推定輝度分布に前記PSFの輝度分布を畳込み第1の関数を得る工程と、
(S11)前記第1の関数を反転して第2の関数を得る工程と、
(S12)前記第2の関数に前記劣化画像の輝度分布を掛け合わせて第3の関数を得る工程と、
(S13)前記第3の関数を前記復元画像初期値の推定輝度分布に掛け合わせて復元画像の推定輝度分布を得る工程と、
(S14)前記カウンターに1を加算する工程と、
(S15)前記カウンターの値が前記最大反復演算回数以上であるという仮説を検証し、もしも、検証結果が偽であれば(S16)の工程へ進むが、前記検証結果が真であれば、工程(S18)へ進む工程と、
(S16)前記復元画像の推定輝度分布を前記復元画像初期値の推定輝度分布と入れ替える工程と、
(S17)(S8)の工程へ戻る工程と、
(S18)前記復元画像の推定輝度分布を最尤復元画像の輝度分布として出力する工程と、
(S19)前記(S1)~(S7)の工程から構成される準備工程と、
(S20)前記(S8)~(S18)の工程から構成される第1画像復元工程と
を備え、
(S21)前記準備工程S19及び前記第1画像復元工程S20において、工程のSの添え数字が小さい順に順番通りに実行する事により前記最大反復演算回数の反復演算を終え、最尤復元画像の輝度分布を出力する第1加速超解像処理工程と、
(S22)前記最尤復元画像の輝度分布を1フレーム分のTV映像信号と成しこれを超解像TV映像信号として出力するTV映像化工程と
を備える事に特徴が有るTV映像向け加速超解像処理方法。 - 前記PSF準備工程の第2の態様は、
(S30)前記劣化示数を検索情報として、劣化示数に対してPSFの輝度分布を1対1で紐づけて整理して作成されたPSFデータベースを検索し、ヒットしたPSFの輝度分布を第1番目のPSFの輝度分布と成す工程と、
(S31)数を計数する第2カウンターに1を入力してリセットする工程と、
(S32)前記第1番目のPSFの輝度分布をPSF初期値の輝度分布と成す工程と、
(S33)前記第2カウンターに1を加算する工程と、
(S34)前記第2カウンターの値が前記最大反復演算回数を超えたという仮説を検証し、もしも検証結果が偽で有れば(S35)の工程へ進み、もしも検証結果が真で有れば(S38)の工程へジャンプする工程と、
(S35)PSF初期値の輝度分布をPSF復元工程により復元する事により最尤復元PSFの輝度分布を得る工程と、
(S36)前記第2カウンターの値をnとする時、前記最尤復元PSFの輝度分布を第n番目のPSFの輝度分布と成す工程と、
(S37)(S33)の工程へ戻る工程と、
(S38)前記最大反復演算回数をn_maxとし、更に前記n_max未満の自然数をnとする時、前記第1番目のPSFの輝度分布から前記n_max番目のPSFの輝度分布までをこの順番で連ねて1連のPSFの輝度分布とし、この1連のPSFの輝度分布に前記劣化示数をラベルとして付け、前記劣化示数に紐づいた前記1連のPSFの輝度分布と成す工程と
を備える事に特徴が有る請求項1に記載のTV映像向け加速超解像処理方法。 - 前記PSF準備工程の第3の態様は、
(S40)前記最大反復演算回数を5に設定する工程と、
(S41)前記PSFデータベース内の全ての劣化示数とこれに紐づいたPSFの輝度分布のペア全てに対して事前に前記PSF準備工程を実行して、前記劣化示数毎に1連のPSFの輝度分布を得、これ等の前記劣化示数毎の1連のPSFの輝度分布を編集する事によって、劣化示数に対して1連のPSFの輝度分布を紐づけて整理した拡大PSFデータベースを準備する工程と、
(S42)前記劣化示数を用いて前記拡大PSFデータベースを検索しヒットした1連のPSFの輝度分布を取得しこれを出力する工程と
を備える事に特徴が有る請求項1に記載のTV映像向け加速超解像処理方法。 - 前記PSF復元工程は、
(S50)前記最大反復演算回数に6を代入する工程と、
(S51)前記PSF初期値の輝度分布を劣化した画像の輝度分布と捉え劣化PSFの輝度分布と成す工程と、
(S52)前記PSF初期値の輝度分布を復元PSF初期値の推定輝度分布と成す工程と、
(S53)前記カウンターに1を代入してリセットする工程と、
(S54)前記復元PSF初期値の推定輝度分布を補正復元PSF初期値の推定輝度分布と成し、更に前記PSF初期値の輝度分布を前記補正復元PSF初期値の推定輝度分布に畳込み演算する際に前記補正復元PSF初期値の推定輝度分布の周辺部に発生する演算困難領域を、前記PSF初期値の輝度分布の画像サイズを基にして計算し、上辺境界に関する前記演算困難領域の画素をコピーし、このコピーした画素を前記補正復元PSF初期値の推定輝度分布の上辺境界の外側に、前記上辺境界に対してミラー対称に成る様に配置する様にペーストし、同様にして右回りに右辺次に下辺最後に左辺に関して行い、前記補正復元PSF初期値の推定輝度分布を補正する復元PSF初期値補正工程と、
(S55)前記PSF初期値の輝度分布を前記補正復元PSF初期値の推定輝度分布に畳込み第4の関数を得る工程と、
(S56)前記第4の関数を反転して第5の関数を得る工程と、
(S57)前記第5の関数を前記劣化PSFの輝度分布に掛け合わせて第6の関数を得る工程と、
(S58)前記第6の関数を前記復元PSF初期値の推定輝度分布に掛け合わせて復元PSFの推定輝度分布を得る工程と、
(S59)前記カウンターに1を加算する工程と、
(S60)前記カウンターの値が前記最大反復演算回数を超えたという仮説を検証し、もしも検証結果が偽で有れば(S61)の工程へ進み、もしも検証結果が真で有れば、(S63)の工程へジャンプする工程と、
(S61)前記復元PSFの推定輝度分布を前記復元PSF初期値の推定輝度分布と入れ替える工程と、
(S62)(S54)の工程へジャンプする工程と、
(S63)前記復元PSFの推定輝度分布を最尤復元PSFの輝度分布として出力する工程と
を備える事に特徴が有る請求項2に記載のTV映像向け加速超解像処理方法。 - 前記第1復元画像初期値補正工程は、
(S70)前記復元画像初期値の推定輝度分布を補正復元画像初期値の推定輝度分布と成す工程と、
(S71)前記PSFサイズを基に前記1連のPSFの輝度分布の内いずれか1つを前記補正復元画像初期値の推定輝度分布に畳込み演算する際に前記補正復元画像初期値の推定輝度分布の周辺部に発生する演算困難領域を計算する工程と、
(S72)前記補正復元画像初期値の推定輝度分布の前記演算困難領域内の画素をコピーし、このコピーした画素を前記補正復元画像初期値の推定輝度分布の4辺に対してミラー対称に成る様にそれぞれ反転させ更に前記補正復元画像初期値の推定輝度分布の4辺の境界の外側に配置する様にペーストして補正する工程と、
(S73)前記補正復元画像初期値の推定輝度分布の前記演算困難領域内の左上隅部の画素をコピーし、このコピーした左上隅部の画素を前記左上隅の頂点を回転中心として180度回転させ、前記補正復元画像初期値の推定輝度分布の左上隅部に発生する空白領域に、ペーストして補正する工程と、
(S74)前記補正復元画像初期値の推定輝度分布の前記演算困難領域内の右上隅部の画素をコピーし、このコピーした右上隅部の画素を前記右上隅の頂点を回転中心として180度回転させ、前記補正復元画像初期値の推定輝度分布の右上隅部に発生する空白領域に、ペーストして補正する工程と、
(S75)前記補正復元画像初期値の推定輝度分布の前記演算困難領域内の左下隅部の画素をコピーし、このコピーした左下隅部の画素を前記左上隅の頂点を回転中心として180度回転させ、前記補正復元画像初期値の推定輝度分布の左下隅部に発生する空白領域に、ペーストして補正する工程と、
(S76)前記補正復元画像初期値の推定輝度分布の前記演算困難領域内の右下隅部の画素をコピーし、このコピーした右下隅部の画素を前記右上隅の頂点を回転中心として180度回転させ、前記補正復元画像初期値の推定輝度分布の右下隅部に発生する空白領域に、ペーストして補正する工程と
を備える事に特徴が有る請求項1に記載のTV映像向け加速超解像処理方法。 - 前記第1画像復元工程の第2の態様は、
(S80)前記最大反復演算回数をn_maxとし、更に前記n_max未満の自然数をnとする時、1連のPSFの輝度分布うち、第n番目のPSFの輝度分布をPSFの輝度分布として第n段目の第1単分画像復元工程に提供するPSF提供工程と、
(S81)前記PSFの輝度分布、前記復元画像初期値の推定輝度分布及び前記劣化画像の輝度分布から、Bayse確率論的数式に基づく反復演算の内の1回分に相当する演算を行い、前記劣化画像の輝度分布に最も尤らしい復元画像の推定輝度分布を求めこれを出力する第1単分画像復元工程と
を備え、
この第1単分画像復元工程S81は、
(S82)前記PSFサイズに基づき前記復元画像初期値の推定輝度分布を補正して補正復元画像初期値の推定輝度分布を求める前記第1復元画像初期値補正工程と同じ処理手順から成る第2復元画像初期値補正工程と、
(S83)前記PSFの輝度分布を前記補正復元画像初期値の推定輝度分布に畳込み第7の関数を得る工程と、
(S84)前記第7の関数を反転して第8の関数を得る工程と、
(S85)前記第8の関数を前記劣化画像の輝度分布に掛け合わせて第9の関数を得る工程と、
(S86)前記第9の関数を前記復元画像初期値の推定輝度分布に掛け合わせて復元画像の推定輝度分布を得る工程と、
(S87)前記復元画像の推定輝度分布を出力する工程と
を備え、
(S88)第n段目の第1単分画像復元工程S81-nの(S87)の工程の出力を第(n+1)段目の第1単分画像復元工程S81-(n+1)の(S82)の工程に接続する事により構成される前記n_max段の直列連結体から成る第2画像復元工程であり、この第2画像復元工程S88において、直列接続された第1単分画像復元工程S81の数である前記n_max回の反復演算を行い、前記n_max段目の第1単分画像復元工程S81-n_maxから出力される復元画像の推定輝度分布を最尤復元画像の輝度分布として出力する事に特徴が有る請求項1に記載のTV映像向け加速超解像処理方法。 - 前記第1画像復元工程の第3の態様は、
(S90)前記カウンターに0を代入する事によりリセットする工程と、
(S91)第2カウンターに1を代入する事によりリセットする工程と、
(S92)前記カウンターの値が0以外であるという仮説を検証し、もしも検証結果が偽であれば(S93)の工程へ進み、もしも検証結果が真であれば(S96)の工程へジャンプする工程と、
(S93)前記劣化画像の輝度分布を劣化画像保存用バッファー及び復元画像初期値用バッファーに転送する工程と、
(S94)(S96)の工程へジャンプする工程と、
(S95)(S102)の工程の復元画像の推定輝度分布を前記復元画像初期値用バッファーに転送する工程と、
(S96)第2カウンターの値をmとする時、1連のPSFの輝度分布内の第m番目のPSFの輝度分布をPSFの輝度分布と成す工程と、
(S97)前記復元画像初期値用バッファーから復元画像初期値の推定輝度分布を読込む工程と、
(S98)前記PSFサイズに基づき前記復元画像初期値の推定輝度分布を補正しこれを補正復元画像初期値の推定輝度分布と成す前記第1復元画像初期値補正工程と同じ処理手順から成る第3復元画像初期値補正工程と、
(S99)前記PSFの輝度分布を前記補正復元画像初期値の推定輝度分布に畳込み第10の関数を得る工程と、
(S100)前記第10の関数を反転して第11の関数を得る工程と、
(S101)前記劣化画像の輝度分布を前記劣化画像保存用バッファーから読込これに前記第11の関数を掛け合わせて第12の関数を得る工程と、
(S102)前記第12の関数を前記復元画像初期値の推定輝度分布に掛け合わせて復元画像の推定輝度分布を得る工程と、
(S103)前記カウンターに1を加算する工程と、
(S104)前記第2カウンターに1を加算する工程と、
(S105)前記カウンターの値が前記最大反復演算回数を超えているという仮説を検証し、もしも検証結果が偽であれば(S95)の工程へジャンプし、もしも検証結果が真であれば(S106)の工程へ進む工程と、
(S106)前記復元画像の推定輝度分布を最尤復元画像の輝度分布として出力する工程と
を備え、
(S107)各工程のSの添え数字の小さい順に順番通りに実行する事により前記最大反復演算回数の反復演算を終え、最尤化された最尤復元画像を出力する第3画像復元工程である事に特徴が有る請求項1に記載のTV映像向け超解像処理方法。 - 前記劣化画像準備工程は、
(S110)1フレーム分のTV映像信号からフレームを構成するRGB信号を取り出すRGB信号抽出工程と、
(S111)前記1フレーム分のTV映像信号の内、RGB信号を取り出した残りのTV映像信号を1フレーム分遅延して出力する遅延工程と、
(S112)前記RGB信号をYUV変換してYUV信号と成すYUV変換工程と、
(S113)前記YUV信号の内、輝度成分であるY信号のみから成る劣化画像の輝度分布を抽出してY劣化画像の輝度分布と成し、残りのU信号のみから成るU劣化画像の分布及びV信号のみから成るV劣化画像の分布を保持するY劣化画像抽出工程と、
(S114)前記Y劣化画像の輝度分布のデガンマ処理を行い1フレーム分の輝度分布から成る劣化画像の輝度分布と成し出力するデガンマ処理工程と、
を備える事に特徴が有る請求項1に記載のTV映像向け加速超解像処理方法。 - 前記TV映像化工程は、
(S120)前記最尤復元画像の輝度分布のガンマ処理を行うガンマ処理工程と、
(S121)前記Y劣化画像抽出工程で保持した前記U劣化画像の分布及び前記V劣化画像の分布とY成分から成るガンマ処理後の最尤復元画像の輝度分布から1枚のYUV復元画像の分布を合成する復元画像合成工程と、
(S122)前記YUV復元画像の分布のRGB変換を行いRGB復元画像の分布と成すRGB変換工程と、
(S123)前記RGB復元画像の分布を読み取りRGB信号を出力するRGB信号変換工程と、
(S124)前記RGB信号を前記遅延工程が出力する前記残りのTV映像信号と合成して1フレーム分のTV映像信号から成る超解像TV映像信号と成し出力するTV映像信号合成工程と
を備える事に特徴が有る請求項1に記載のTV映像向け加速超解像処理方法。 - 請求項1~7に記載されたPSFの輝度分布は、縁の無い正方形状の同一サイズの画素から成るものであって、その輝度分布は中心が最も明るい2次元正規分布であり、そのサイズが5x5画素である事に特徴が有るTV映像向け加速超解像処理方法。
- 請求項1に記載された準備工程及び第1画像復元工程をコンピュータに実行させるための第1加速超解像処理プログラム。
- 請求項1に記載された準備工程及び請求項6に記載された第2画像復元工程をコンピュータに実行させるための第2加速超解像処理プログラム。
- 請求項1に記載された準備工程及び請求項7に記載された第3画像復元工程をコンピュータに実行させるための第3加速超解像処理プログラム。
- 前記第1加速超解像処理プログラム、前記第2加速超解像処理プログラム及び前記第3加速超解像処理プログラムはいずれも暗号化されたものであり、これ等暗号化された前記第1加速超解像処理プログラム、前記第2加速超解像処理プログラム及び前記第3加速超解像処理プログラムを記憶すると共に前記コンピュータに接続可能で且つ前記コンピュータに読み込ませる事が可能な事に特徴がある第1記憶媒体。
- 請求項1~5、8~10に記載されたTV映像向け加速超解像処理方法に従い、1フレーム分のTV映像信号に含まれるフレームから光学的劣化を低減して劣化前の1フレーム分のTV映像信号を復元するTV映像向け加速超解像処理装置であって、
(W1)最大反復演算回数を設定する手段と、
(W2)前記TV映像を見ながら前記TV映像の劣化状況に適した劣化示数を指定する劣化示数指定手段と、
(W3)前記劣化示数に紐づいた第1番目のPSFの輝度分布及びこの系から成り反復演算回数に紐づいて整理された1連のPSFの輝度分布を準備するPSF準備手段と、
(W4)1フレーム分の輝度分布から成る劣化画像の輝度分布を1フレーム分のTV映像信号から準備する劣化画像準備手段と、
(W5)前記劣化画像の輝度分布を復元画像初期値の推定輝度分布と成す復元画像初期値準備手段と、
(W6)前記1連のPSFの輝度分布は互いに同じ画像サイズであり、この画像サイズをPSFサイズと呼ぶ時、このPSFサイズを求めるPSFサイズ取得手段と、
(W7)反復演算回数を計数するカウンターに1を設定する第1リセット手段と、
(W8)前記復元画像初期値の推定輝度分布を複写してこれを補正復元画像初期値の推定輝度分布と成し更に前記補正復元画像初期値の推定輝度分布をPSFサイズを基にして補正する第1復元画像初期値補正手段と、
(W9)前記1連のPSFの輝度分布から前記カウンターの値に紐づいた1つを選択してこれをPSFの輝度分布と成すPSF選定手段と、
(W10)前記補正復元画像初期値の推定輝度分布に前記PSFの輝度分布を畳込み第13の関数を得る手段と、
(W11)前記第13の関数を反転して第14の関数を得る手段と、
(W12)前記第14の関数に前記劣化画像の輝度分布を掛け合わせて第15の関数を得る手段と、
(W13)前記第15の関数を前記復元画像初期値の推定輝度分布に掛け合わせて復元画像の推定輝度分布を得る手段と、
(W14)前記カウンターに1を加算する手段と、
(W15)前記カウンターの値が前記最大反復演算回数以上であるという仮説を検証し、もしも、検証結果が偽であれば(W16)の手段へ進むが、前記検証結果が真であれば、手段(W18)へ進む手段と、
(W16)前記復元画像の推定輝度分布を前記復元画像初期値の推定輝度分布と入れ替える手段と、
(W17)(W8)の手段へ戻る手段と、
(W18)前記復元画像の推定輝度分布を最尤復元画像の輝度分布として出力する手段と、
(W19)前記(W1)~(W7)の手段から構成される準備手段と、
(W20)前記(W8)~(W18)の手段から構成される第1画像復元手段と
を備え、
(W21)前記準備手段W19及び前記第1画像復元手段W20において、手段のWの添え数字が小さい順に順番通りに実行する事により前記最大反復演算回数の反復演算を終え、最尤復元画像の輝度分布を出力する第1加速超解像処理手段と、
(W22)前記最尤復元画像の輝度分布を1フレーム分のTV映像信号と成しこれを超解像TV映像信号として出力するTV映像化手段と
を備える事に特徴が有るTV映像向け加速超解像処理装置。 - 前記PSF準備手段の第2の態様は、
(W30)前記劣化示数を検索情報として、劣化示数に対してPSFの輝度分布を1対1で紐づけて整理して作成されたPSFデータベースを検索し、ヒットしたPSFの輝度分布を第1番目のPSFの輝度分布と成す手段と、
(W31)数を計数する第2カウンターに1を入力してリセットする手段と、
(W32)前記第1番目のPSFの輝度分布をPSF初期値の輝度分布と成す手段と、
(W33)前記第2カウンターに1を加算する手段と、
(W34)前記第2カウンターの値が前記最大反復演算回数を超えたという仮説を検証し、もしも検証結果が偽で有れば(W35)の手段へ進み、もしも検証結果が真で有れば(W38)の手段へジャンプする手段と、
(W35)PSF初期値の輝度分布をPSF復元手段により復元する事により最尤復元PSFの輝度分布を得る手段と、
(W36)前記第2カウンターの値をnとする時、前記最尤復元PSFの輝度分布を第n番目のPSFの輝度分布と成す手段と、
(W37)(W33)の手段へ戻る手段と、
(W38)前記最大反復演算回数をn_maxとし、更に前記n_max未満の自然数をnとする時、前記第1番目のPSFの輝度分布から前記n_max番目のPSFの輝度分布までをこの順番で連ねて1連のPSFの輝度分布とし、この1連のPSFの輝度分布に前記劣化示数をラベルとして付け、前記劣化示数に紐づいた前記1連のPSFの輝度分布と成す手段と
を備える事に特徴が有る請求項15に記載のTV映像向け加速超解像処理装置。 - 前記PSF準備手段の第3の態様は、
(W40)前記最大反復演算回数を5に設定する手段と、
(W41)前記PSFデータベース内の全ての劣化示数とこれに紐づいたPSFの輝度分布のペア全てに対して事前に前記PSF準備手段を実行して、前記劣化示数毎に1連のPSFの輝度分布を得、これ等の前記劣化示数毎の1連のPSFの輝度分布を編集する事によって、劣化示数に対して1連のPSFの輝度分布を紐づけて整理した拡大PSFデータベースを準備する手段と、
(W42)前記劣化示数を用いて前記拡大PSFデータベースを検索しヒットした1連のPSFの輝度分布を取得しこれを出力する手段と
を備える事に特徴が有る請求項15に記載のTV映像向け加速超解像処理装置。 - 前記PSF復元手段は、
(W50)前記最大反復演算回数に6を代入する手段と、
(W51)前記PSF初期値の輝度分布を劣化した画像の輝度分布と捉え劣化PSFの輝度分布と成す手段と、
(W52)前記PSF初期値の輝度分布を復元PSF初期値の推定輝度分布と成す手段と、
(W53)前記カウンターに1を代入してリセットする手段と、
(W54)前記復元PSF初期値の推定輝度分布を補正復元PSF初期値の推定輝度分布と成し、更に前記PSF初期値の輝度分布を前記補正復元PSF初期値の推定輝度分布に畳込み演算する際に前記補正復元PSF初期値の推定輝度分布の周辺部に発生する演算困難領域を、前記PSF初期値の輝度分布の画像サイズを基にして計算し、上辺境界に関する前記演算困難領域の画素をコピーし、このコピーした画素を前記補正復元PSF初期値の推定輝度分布の上辺境界の外側に上辺境界に対してミラー対称に成る様に配置する様にペーストし、同様にして右回りに右辺次に下辺最後に左辺に関して行い、前記補正復元PSF初期値の推定輝度分布を補正する復元PSF初期値補正手段と、
(W55)前記PSF初期値の輝度分布を前記補正復元PSF初期値の推定輝度分布に畳込み第16の関数を得る手段と、
(W56)前記第16の関数を反転して第17の関数を得る手段と、
(W57)前記第17の関数を前記劣化PSFの輝度分布に掛け合わせて第18の関数を得る手段と、
(W58)前記第18の関数を前記復元PSF初期値の推定輝度分布に掛け合わせて復元PSFの推定輝度分布を得る手段と、
(W59)前記カウンターに1を加算する手段と、
(W60)前記カウンターの値が前記最大反復演算回数を超えたという仮説を検証し、もしも検証結果が偽で有れば(W61)の手段へ進み、もしも検証結果が真で有れば、(W63)の手段へジャンプする手段と、
(W61)前記復元PSFの推定輝度分布を復元PSF初期値の推定輝度分布と入れ替える手段と、
(W62)(W54)の手段へジャンプする手段と、
(W63)前記復元PSFの推定輝度分布を最尤復元PSFの輝度分布として出力する手段と
を備える事に特徴が有る請求項16に記載のTV映像向け加速超解像処理装置。 - 前記第1復元画像初期値補正手段は、
(W70)前記復元画像初期値の推定輝度分布を補正復元画像初期値の推定輝度分布と成す手段と、
(W71)前記PSFサイズを基に前記1連のPSFの輝度分布の内いずれか1つを前記補正復元画像初期値の推定輝度分布に畳込み演算する際に前記補正復元画像初期値の推定輝度分布の周辺部に発生する演算困難領域を計算する手段と、
(W72)前記補正復元画像初期値の推定輝度分布の前記演算困難領域内の画素をコピーし、このコピーした画素を前記補正復元画像初期値の推定輝度分布の4辺に対してミラー対称に成る様にそれぞれ反転させ更に前記補正復元画像初期値の推定輝度分布の4辺の境界の外側に配置する様にペーストして補正する手段と、
(W73)前記補正復元画像初期値の推定輝度分布の前記演算困難領域内の左上隅部の画素をコピーし、このコピーした左上隅部の画素を前記左上隅の頂点を回転中心として180度回転させ、前記補正復元画像初期値の推定輝度分布の左上隅部に発生する空白領域に、ペーストして補正する手段と、
(W74)前記補正復元画像初期値の推定輝度分布の前記演算困難領域内の右上隅部の画素をコピーし、このコピーした右上隅部の画素を前記右上隅の頂点を回転中心として180度回転させ、前記補正復元画像初期値の推定輝度分布の右上隅部に発生する空白領域に、ペーストして補正する手段と、
(W75)前記補正復元画像初期値の推定輝度分布の前記演算困難領域内の左下隅部の画素をコピーし、このコピーした左下隅部の画素を前記左上隅の頂点を回転中心として180度回転させ、前記補正復元画像初期値の推定輝度分布の左下隅部に発生する空白領域に、ペーストして補正する手段と、
(W76)前記補正復元画像初期値の推定輝度分布の前記演算困難領域内の右下隅部の画素をコピーし、このコピーした右下隅部の画素を前記右上隅の頂点を回転中心として180度回転させ、前記補正復元画像初期値の推定輝度分布の右下隅部に発生する空白領域に、ペーストして補正する手段と
を備える事に特徴が有る請求項15に記載のTV映像向け加速超解像処理装置。 - 前記第1画像復元手段の第2の態様は、
(W80)前記最大反復演算回数をn_maxとし、更に前記n_max未満の自然数をnとする時、1連のPSFの輝度分布うち、第n番目のPSFの輝度分布をPSFの輝度分布として第n段目の第1単分画像復元手段に提供するPSF提供手段と、
(W81)前記PSFの輝度分布、前記復元画像初期値の推定輝度分布及び前記劣化画像の輝度分布から、Bayse確率論的数式に基づく反復演算の内の1回分に相当する演算を行い、前記劣化画像の輝度分布に最も尤らしい復元画像の推定輝度分布を求めこれを出力する第1単分画像復元手段と
を備え、
この第1単分画像復元手段は、
(W82)第1復元画像初期値補正手段と同じ構成から成る第2復元画像初期値補正手段により補正復元画像初期値の推定輝度分布を求める手段と、
(W83)前記PSFの輝度分布を前記補正復元画像初期値の推定輝度分布に畳込み第19の関数を得る手段と、
(W84)前記第19の関数を反転して第20の関数を得る手段と、
(W85)前記第20の関数を前記劣化画像の輝度分布に掛け合わせて第21の関数を得る手段と、
(W86)前記第21の関数を前記復元画像初期値の推定輝度分布に掛け合わせて復元画像の推定輝度分布を得る手段と、
(W87)前記復元画像の推定輝度分布を出力する手段と
を備え、
(W88)第n段目の第1単分画像復元手段W81-nの(W87)の手段の出力を第(n+1)段目の第1単分画像復元手段W81-(n+1)の(W82)の手段に接続する事により構成される前記n_max段の直列連結体から成る第2画像復元手段であり、
この第2画像復元手段W88において、直列接続された第1単分画像復元手段W81の数である前記n_max回の反復演算を行い、前記n_max段目の第1単分画像復元手段W81-n_maxから出力される復元画像の推定輝度分布を最尤復元画像の輝度分布として出力する事に特徴が有る請求項15に記載のTV映像向け加速超解像処理装置。 - 前記第1画像復元手段の第3の態様は、
(W90)前記カウンターに0を代入する事によりリセットする手段と、
(W91)第2カウンターに1を代入する事によりリセットする手段と、
(W92)前記カウンターの値が0以外であるという仮説を検証し、もしも検証結果が偽であれば(W93)の手段へ進み、もしも検証結果が真であれば(W96)の手段へジャンプする手段と、
(W93)前記劣化画像の輝度分布を劣化画像保存用バッファー及び復元画像初期値用バッファーに転送する手段と、
(W94)(W96)の手段へジャンプする手段と、
(W95)(W102)の手段の復元画像の推定輝度分布を前記復元画像初期値用バッファーに転送する手段と、
(W96)第2カウンターの値をmとする時、1連のPSFの輝度分布内の第m番目のPSFの輝度分布をPSFの輝度分布と成す手段と、
(W97)前記復元画像初期値用バッファーから復元画像初期値の推定輝度分布を読込む手段と、
(W98)前記復元画像初期値の推定輝度分布を補正しこれを補正復元画像初期値の推定輝度分布と成す第1復元画像初期値補正手段と同じ構成から成る第3復元画像初期値補正手段と、
(W99)前記PSFの輝度分布を前記補正復元画像初期値の推定輝度分布に畳込み第22の関数を得る手段と、
(W100)前記第22の関数を反転して第23の関数を得る手段と、
(W101)前記第23の関数を前記劣化画像の輝度分布に掛け合わせて第24の関数を得る手段と、
(W102)前記第24の関数を前記復元画像初期値の推定輝度分布に掛け合わせて復元画像の推定輝度分布を得る手段と、
(W103)前記カウンターに1を加算する手段と、
(W104)前記第2カウンターに1を加算する手段と、
(W105)前記カウンターの値が前記最大反復演算回数を超えているという仮説を検証し、もしも検証結果が偽であれば(W95)の手段へジャンプし、もしも検証結果が真であれば(W106)の手段へ進む手段と、
(W106)前記復元画像の推定輝度分布を最尤復元画像の輝度分布として出力する手段と
を備え、
(W107)各手段のSの添え数字の小さい順に順番通りに実行する事により前記最大反復演算回数の反復演算を終え、最尤化された最尤復元画像を出力する第3画像復元手段である事に特徴が有る請求項15に記載のTV映像向け超解像処理装置。 - 前記劣化画像準備手段は、
(W110)1フレーム分のTV映像信号からフレームを構成するRGB信号を取り出すRGB信号抽出手段と、
(W111)前記1フレーム分のTV映像信号の内、RGB信号を取り出した残りのTV映像信号を1フレーム分遅延して出力する遅延手段と、
(W112)前記RGB信号をYUV変換してYUV信号と成すYUV変換手段と、
(W113)前記YUV信号の内、輝度成分であるY信号のみから成る劣化画像の輝度分布を抽出してY劣化画像の輝度分布と成し、残りのU信号のみから成るU劣化画像の分布及びV信号のみから成るV劣化画像の分布を保持するY劣化画像抽出手段と、
(W114)前記Y劣化画像の輝度分布のデガンマ処理を行い1フレーム分の輝度分布から成る劣化画像の輝度分布と成し出力するデガンマ処理手段と
を備える事に特徴が有る請求項15に記載のTV映像向け加速超解像処理装置。 - 前記TV映像化手段は、
(W120)前記最尤復元画像の輝度分布のガンマ処理を行うガンマ処理手段と、
(W121)前記Y劣化画像抽出手段で保持した前記U劣化画像の分布及び前記V劣化画像の分布とY成分から成るガンマ処理後の最尤復元画像の輝度分布から1枚のYUV復元画像の分布を合成する復元画像合成手段と、
(W122)前記YUV復元画像の分布のRGB変換を行いRGB復元画像の分布と成すRGB変換手段と、
(W123)前記RGB復元画像の分布を読み取りRGB信号を出力するRGB信号変換手段と、
(W124)前記RGB信号を前記遅延手段が出力する前記残りのTV映像信号と合成して1フレーム分のTV映像信号から成る超解像TV映像信号と成し出力するTV映像信号合成手段と
を備える事に特徴が有る請求項15に記載のTV映像向け加速超解像処理装置。 - 請求項15に記載された準備手段及び第1画像復元手段を構成し且つ実行させるための第4加速超解像処理プログラム。
- 請求項15に記載された準備手段及び請求項20に記載された第2画像復元手段を構成し且つ実行させるための第5加速超解像処理プログラム。
- 請求項15に記載された準備手段及び請求項21に記載された第3画像復元手段を構成し且つ実行させるための第6加速超解像処理プログラム。
- 前記第4加速超解像処理プログラム、前記第5加速超解像処理プログラム及び前記第6加速超解像処理プログラムはそれぞれ暗号化されたものであり、これ等の暗号化された前記第4加速超解像処理プログラム、前記第5加速超解像処理プログラム及び前記第6加速超解像処理プログラムを記憶すると共に前記コンピュータに接続可能で且つ前記コンピュータに読み込ませる事が可能な事に特徴が有る第2記憶媒体。
Priority Applications (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR1020167015373A KR101807229B1 (ko) | 2013-12-11 | 2014-11-27 | Tv 영상용 가속 초해상 처리 방법 및 이 방법에 따른 tv 영상용 가속 초해상 처리 장치, 제1~6 가속 초해상 처리 프로그램과 제1~2 기억 매체 |
| CN201480067470.8A CN105814873B (zh) | 2013-12-11 | 2014-11-27 | 面向tv影像的加速超分辨率处理方法及装置、及存储介质 |
| US15/103,349 US9787962B2 (en) | 2013-12-11 | 2014-11-27 | Accelerated super-resolution processing method for TV video images, accelerated super-resolution processing device for TV video images that is used in same method, first to sixth accelerated super-resolution processing programs, and first to second storage media |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2013-255951 | 2013-12-11 | ||
| JP2013255951A JP6126523B2 (ja) | 2013-12-11 | 2013-12-11 | Tv映像向け加速超解像処理方法及び同方法によるtv映像向け加速超解像処理装置、第1〜6加速超解像処理プログラム、並びに第1〜2記憶媒体 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2015087712A1 true WO2015087712A1 (ja) | 2015-06-18 |
Family
ID=53371024
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/JP2014/081474 Ceased WO2015087712A1 (ja) | 2013-12-11 | 2014-11-27 | Tv映像向け加速超解像処理方法及び同方法によるtv映像向け加速超解像処理装置、第1~6加速超解像処理プログラム、並びに第1~2記憶媒体 |
Country Status (5)
| Country | Link |
|---|---|
| US (1) | US9787962B2 (ja) |
| JP (1) | JP6126523B2 (ja) |
| KR (1) | KR101807229B1 (ja) |
| CN (1) | CN105814873B (ja) |
| WO (1) | WO2015087712A1 (ja) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113228097B (zh) | 2018-12-29 | 2024-02-02 | 浙江大华技术股份有限公司 | 图像处理方法和系统 |
| CN112184835B (zh) * | 2020-10-16 | 2024-01-19 | 深圳阿图科技有限公司 | 一种钢琴键盘图像的形变恢复方法及系统 |
| CN114240752B (zh) * | 2021-12-17 | 2025-02-11 | 浙江大学 | 一种基于认知不确定性的单帧图像超分辨方法 |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2006041126A1 (ja) * | 2004-10-14 | 2006-04-20 | Lightron Co., Ltd. | 劣化情報復元方法と復元装置 |
| WO2011099244A1 (ja) * | 2010-02-10 | 2011-08-18 | パナソニック株式会社 | 画像処理装置および画像処理方法 |
| JP2013003610A (ja) * | 2011-06-10 | 2013-01-07 | Sony Corp | 画像処理装置および方法、プログラム、並びに記録媒体 |
| JP2013175003A (ja) * | 2012-02-24 | 2013-09-05 | Osaka Prefecture Univ | Psf推定方法とそれを用いた劣化画像の復元方法およびこれらを記録したプログラムとそれを実行するコンピュータ装置 |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1801753B1 (en) * | 2004-10-14 | 2016-12-07 | Lightron Co. Ltd. | Degradation information restoring method and device |
| JPWO2007122838A1 (ja) * | 2006-04-25 | 2009-09-03 | 国立大学法人 奈良先端科学技術大学院大学 | 階層ベイズ法に基づく超解像法および超解像プログラム |
| JP4454657B2 (ja) | 2007-01-12 | 2010-04-21 | 三洋電機株式会社 | ぶれ補正装置及び方法、並びに撮像装置 |
| JP2009100407A (ja) * | 2007-10-19 | 2009-05-07 | Toshiba Corp | 画像処理装置及びその方法 |
| JP5105286B2 (ja) | 2008-05-07 | 2012-12-26 | 国立大学法人東京工業大学 | 画像復元装置、画像復元方法及び画像復元プログラム |
| JP2009296410A (ja) * | 2008-06-06 | 2009-12-17 | Hitachi Ltd | 高解像度の映像再生装置 |
-
2013
- 2013-12-11 JP JP2013255951A patent/JP6126523B2/ja active Active
-
2014
- 2014-11-27 US US15/103,349 patent/US9787962B2/en active Active
- 2014-11-27 CN CN201480067470.8A patent/CN105814873B/zh active Active
- 2014-11-27 WO PCT/JP2014/081474 patent/WO2015087712A1/ja not_active Ceased
- 2014-11-27 KR KR1020167015373A patent/KR101807229B1/ko active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2006041126A1 (ja) * | 2004-10-14 | 2006-04-20 | Lightron Co., Ltd. | 劣化情報復元方法と復元装置 |
| WO2011099244A1 (ja) * | 2010-02-10 | 2011-08-18 | パナソニック株式会社 | 画像処理装置および画像処理方法 |
| JP2013003610A (ja) * | 2011-06-10 | 2013-01-07 | Sony Corp | 画像処理装置および方法、プログラム、並びに記録媒体 |
| JP2013175003A (ja) * | 2012-02-24 | 2013-09-05 | Osaka Prefecture Univ | Psf推定方法とそれを用いた劣化画像の復元方法およびこれらを記録したプログラムとそれを実行するコンピュータ装置 |
Also Published As
| Publication number | Publication date |
|---|---|
| JP6126523B2 (ja) | 2017-05-10 |
| JP2015115747A (ja) | 2015-06-22 |
| US9787962B2 (en) | 2017-10-10 |
| KR101807229B1 (ko) | 2017-12-08 |
| CN105814873A (zh) | 2016-07-27 |
| CN105814873B (zh) | 2019-02-19 |
| KR20160082705A (ko) | 2016-07-08 |
| US20160366387A1 (en) | 2016-12-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109272454B (zh) | 一种增强现实设备的坐标系校准方法及装置 | |
| JP6129759B2 (ja) | Simd型超並列演算処理装置向け超解像処理方法、装置、プログラム及び記憶媒体 | |
| US11740470B2 (en) | Low latency distortion unit for head mounted displays | |
| US20250086852A1 (en) | Green screen matting method, apparatus and electronic device | |
| TWI698834B (zh) | 用於圖形處理之方法及裝置 | |
| CN109767466A (zh) | 画面渲染方法、装置、终端及对应的存储介质 | |
| JP6155182B2 (ja) | Tv映像向け超解像処理方法および同方法によるtv映像向け超解像処理装置、第1〜第14超解像処理プログラム、並びに第1〜第4記憶媒体 | |
| CN112368737A (zh) | 将基于图像的跟踪操作从通用处理单元卸载到硬件加速器单元的系统和方法 | |
| JP6126523B2 (ja) | Tv映像向け加速超解像処理方法及び同方法によるtv映像向け加速超解像処理装置、第1〜6加速超解像処理プログラム、並びに第1〜2記憶媒体 | |
| WO2016043161A1 (ja) | 表示装置、表示システム、及び表示制御プログラム | |
| CN105051788B (zh) | 使用多个图元进行图形处理 | |
| US20180211429A1 (en) | Reactive overlays of multiple representations using augmented reality | |
| US11176720B2 (en) | Computer program, image processing method, and image processing apparatus | |
| CN111726566B (zh) | 一种实时校正拼接防抖的实现方法 | |
| US20250037325A1 (en) | Environment map upscaling for digital image generation | |
| US9384589B2 (en) | Anti-aliasing for geometries | |
| TWI744298B (zh) | 影像處理裝置、控制影像處理裝置之方法及電腦可讀取媒體 | |
| CN107194878A (zh) | 图像拼接方法及装置 | |
| CN108154473A (zh) | 用于处理图像的方法和装置 | |
| CN119028260A (zh) | 图像显示方法、装置、电子设备及计算机可读存储介质 | |
| JP5898109B2 (ja) | 画像合成装置、画像合成方法及び画像合成プログラム | |
| CN121462711A (en) | Method and device for mixing video resolution | |
| CN117152386A (zh) | 一种图像处理方法及装置 | |
| Popovic et al. | Computational Imaging Applications | |
| CN110969683A (zh) | 一种动画生成方法、装置及电子设备 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 14870572 Country of ref document: EP Kind code of ref document: A1 |
|
| ENP | Entry into the national phase |
Ref document number: 20167015373 Country of ref document: KR Kind code of ref document: A |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 15103349 Country of ref document: US |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 14870572 Country of ref document: EP Kind code of ref document: A1 |