JP2005260399A - Image processing apparatus and image forming apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide means for determining an appropriate feature amount extraction threshold value, and to make it less susceptible to a change in a gradation conversion table in feature amount extraction processing.
An image processing printer γ unit 409 performs gradation conversion of image data based on a printer γ table. The multi-tone image data subjected to the tone conversion is quantized by the tone processing unit 410 by an error diffusion method using a quantization threshold. The feature amount extraction unit 1701 extracts the feature amount of the image based on the feature amount threshold value from the image data after gradation conversion. Based on the feature value after the feature value is extracted, the threshold value selection unit 1702 uses an error diffusion method from a plurality of quantization threshold values that differ in at least the former among the presence or absence of periodic vibration and the magnitude of amplitude when there is vibration. Select the one used to execute The feature amount extraction threshold used for extracting the feature amount is processed by the printer γ table, and the feature amount extraction threshold after the processing is used for extracting the feature amount.
[Selection] FIG.

Description

  The present invention relates to an image processing apparatus that quantizes image data by an error diffusion method and an image forming apparatus including the image processing apparatus.

  Patent Documents 1 and 2 disclose a technique for quantizing image data by an error diffusion method using a quantization threshold.

  In the technique disclosed in Patent Document 1, the threshold pattern of error diffusion processing is switched depending on the edge degree of an image. The threshold pattern includes periodic vibrations, the vibration amplitude is reduced at the edge portion to provide simple low-value error diffusion, the vibration amplitude is increased at the flat portion, and the pattern is close to dither processing.

JP 2001-128004 A JP 2003-234893 A

  In the error diffusion processing technique for extracting the feature amount of the image from the image data after the gradation conversion by the gradation conversion table and switching the quantization threshold used in the error diffusion processing based on the feature amount extraction result, the hardware configuration Due to the above restrictions, even if the feature amount extraction threshold is fixed, the region from which the feature amount is extracted differs depending on the parameters of the tone conversion table, and the image finally obtained by the tone processing may differ.

  In this case, if the parameter of the feature amount extraction threshold is also changed according to the parameter of the gradation conversion table, the result of feature amount extraction can be made less susceptible to the influence of the gradation conversion table.

  However, in such a case, a means for determining an appropriate feature amount extraction threshold has not been known (see Patent Documents 1 and 2).

  An object of the present invention is to provide means for determining an appropriate feature amount extraction threshold value, and to make it less susceptible to the change in the gradation conversion table in the feature amount extraction processing.

  The present invention relates to a means for performing gradation conversion of image data based on a predetermined gradation conversion table, and error diffusion using a predetermined quantization threshold for the multi-gradation image data subjected to the gradation conversion. A means for quantizing by a method, a means for extracting a feature amount of an image from the image data after gradation conversion based on a feature amount threshold, and the presence or absence of periodic vibration based on the feature amount after extraction Means for selecting the one used for execution of the error diffusion method from among the plurality of quantization thresholds of which the former is different among the magnitudes of amplitudes when there is vibration; and a feature quantity extraction threshold used for extracting the feature quantity Is processed by the gradation conversion table, and a feature quantity extraction threshold after the process is used for extraction of the feature quantity.

  According to the present invention, an appropriate feature amount extraction threshold value can be determined by processing a feature amount extraction threshold value used for feature amount extraction using a gradation conversion table. Can be made less susceptible to changes.

  The best mode for carrying out the invention will be described.

  The following is an electrophotographic copying machine (hereinafter simply referred to as a copying machine) that implements the image forming apparatus of the present invention. First, the outline of the mechanism of the copying machine main body 101 according to the embodiment will be described with reference to the mechanism diagram shown in FIG. In FIG. 1, around the surface of organic photoconductor (OPC) drums 102a to 102d as image bearing members arranged side by side in the substantially central portion of the copying machine main body 101, the surfaces of the photoconductor drums 102a to 102d are disposed. Charging chargers 103a to 103d to be charged, laser optical systems 104a to 104d for forming an electrostatic latent image by irradiating the surface of uniformly charged photosensitive drums 102a to 102d with semiconductor laser light, and various colors to the electrostatic latent image It is formed on a black developing device 105 that supplies toner and develops toner images for each color, three color developing devices 106, 107, and 108 of yellow Y, magenta M, and cyan C, and photosensitive drums 102a to 102d. An intermediate transfer belt 109 for sequentially transferring toner images of respective colors, and bias rollers 110a to 110a for applying a transfer voltage to the intermediate transfer belt 109. 10d, cleaning devices 111a to 111d for removing the toner remaining on the surface of the photosensitive drum 102 after the transfer, neutralization units 112a to 112d for removing the charge remaining on the surface of the photosensitive drums 102a to 102d after the transfer, and the like. It is arranged. The intermediate transfer belt 109 has a transfer bias roller 113 for applying a voltage for transferring the transferred toner image to the transfer material, and a belt cleaning device 114 for cleaning the toner image remaining on the transfer material after transfer. Is arranged.

  A fixing device 116 that heats and pressurizes and fixes the toner image is disposed at the end of the conveyance belt 115 that conveys the transfer material peeled off from the intermediate transfer belt 109. A paper discharge tray 117 is attached to the section.

  Above the laser optical system 104, a contact glass 118 serving as a document placement table disposed above the copier body 101, an exposure lamp 119 for irradiating scanning light onto the document on the contact glass 118, and reflected light from the document Is guided to the imaging lens 122 by the reflecting mirror 121 and is incident on an image sensor array 123 of a CCD (Charge Coupled Device) which is a photoelectric conversion element. The image signal converted into an electrical signal by the CCD image sensor array 123 is controlled by an unillustrated image processing device to control the laser oscillation of the semiconductor laser in the laser optical system 104.

  Next, a control system built in the copying machine will be described. As shown in FIG. 2, the control system includes a main control unit (CPU) 130, and a predetermined ROM 131, RAM 132, NV-RAM (nonvolatile RAM) 183, and flash memory 184 are attached to the main control unit 130. Has been. Various parameters are stored in the NV-RAM 183, and a control program is stored in the flash memory. Further, the main control unit 130 includes, via an interface I / O 133, a laser optical system control unit 134, a power supply circuit 135, an optical sensor 136 installed in each YMCK image forming unit, and a toner installed in each YMCK developer. A density sensor 137, an environmental sensor 138, photoconductor surface potential sensors 139a to 139d, a toner supply circuit 140, an intermediate transfer belt drive unit 141, and an operation unit 142 are connected to each other. The laser optical system control unit 134 adjusts the laser output of the laser optical systems 104a to 104d, and the power supply circuit 135 gives a predetermined charging discharge voltage to the charging chargers 113a to 113d. At the same time, a developing bias having a predetermined voltage is applied to the developing devices 105, 106, 107, and 108, and a predetermined transfer voltage is applied to the bias rollers 110a to 110d and the transfer bias rollers 113a to 113d.

  The optical sensors 136 are opposed to the photoconductors 102a to 102d, respectively, are opposed to the optical sensor 136a and the transfer belt 109 for detecting the toner adhesion amount on the photoconductors 102a to 102d, and are attached to the toner on the transfer belt 109. The optical sensor 136b for detecting the amount and the optical sensor 136c for detecting the toner adhesion amount on the conveyor belt 115 while facing the conveyor belt 115 are shown. In practice, detection may be performed at any one of the optical sensors 136a to 136c.

The optical sensors 136a to 136c are formed of a light emitting element such as a light emitting diode and a light receiving element such as a photosensor, which are disposed in the vicinity of the areas after transfer of the photosensitive drums 102a to 102d, and are formed on the photosensitive drum 102. The toner adhesion amount in the toner image of the detection pattern latent image and the toner adhesion amount in the background portion are detected for each color, and so-called residual potential after static elimination on the photoreceptor is detected.
The detection output signals from the photoelectric sensors 136a to 136c are applied to a photoelectric sensor control unit (not shown). The photoelectric sensor control unit obtains a ratio between the toner adhesion amount in the detection pattern toner image and the toner adhesion amount in the background portion, compares the ratio value with a reference value, detects a change in image density, and detects each color of YMCK. The control value of the toner density sensor 137 is corrected.

  Further, the toner density sensor 137 detects the toner density in the developing devices 105 to 108 based on the change in magnetic permeability of the developer present in the developing devices 105 to 108. The toner density sensor 137 compares the detected toner density value with a reference value, and when the toner density falls below a certain value and becomes a toner shortage state, a toner replenishment signal having a magnitude corresponding to the shortage is supplied to the toner. A function of applying to the supply circuit 140 is provided. The potential sensor 139 detects the surface potential of each of the photoconductors 102a to 102d, which are image carriers, and the intermediate transfer belt drive unit 141 controls the drive of the intermediate transfer belt.

  The black developer 105 contains a developer containing black toner and a carrier, which is agitated by the rotation of the agent agitating member 202 and pumped onto the sleeve by the developer regulating member 202 on the developing sleeve 201B. Adjust the amount of developer produced. The supplied developer is magnetically carried on the developing sleeve 201B and rotates as a magnetic brush in the rotating direction of the developing sleeve 201B.

  Next, the image processing apparatus of the copying machine will be described based on the block diagram of FIG. 3, 400 is a scanner, 401 is a shading correction unit, 423 is an area processing unit, 402 is a scanner γ conversion unit, 403 is an image memory, 404 is an image separation unit, 405 is an MTF filter, and 406 is a color conversion UCR processing unit. 426 is an image compression / expansion unit, 407 is a scaling unit, 408 is an image processing (create) unit, 409 is an image processing printer γ conversion unit, 410 is a gradation processing unit, 411 is an interface I / F selector, Reference numeral 412 denotes an image forming unit printer γ (hereinafter referred to as “procon γ”) conversion unit, 413 a printer serving as a printer engine, 414 a ROM, 415 a CPU, 416 a RAM, 417 a system controller, 418 an external computer, 419 Is the printer controller, and 421 and 422 generate patterns respectively. , 425 is a NV-RAM (non-volatile RAM).

  A document to be copied by a copying machine is color-separated into R, G, and B by a color scanner 400 and is read as a 10-bit signal as an example. The read image signal is corrected for unevenness in the main scanning direction by the shading correction unit 401 and output as an 8-bit signal.

  The area processing unit 423 generates an area signal for distinguishing which area in the document the image data currently being processed belongs to. The parameters used in the subsequent image processing unit are switched according to the area signal generated by this circuit. These areas include color correction coefficients, spatial filters, and gradation conversion that are optimal for each original, such as text, silver halide photographs (photographic paper), printed originals, inkjets, highlighters, maps, and thermal transfer originals. Image processing parameters such as a table can be set according to the image area.

  An interface (I / F) 411 is used when an image read by the scanner 400 is output to the outside.

  In the scanner γ conversion unit 402, a read signal from the scanner 400 is converted from reflectance data to brightness data.

  The image memory 403 stores the image signal after the scanner γ conversion. The image separation unit 404 performs determination of a character part and a photograph part of an image, and chromatic / achromatic color determination.

  In the MTF filter 405, in addition to processing for changing the frequency characteristics of the image signal such as edge enhancement or smoothing according to the user's preference, such as a sharp image or a soft image, the edge according to the edge degree of the image signal Perform enhancement processing (adaptive edge enhancement processing). For example, so-called adaptive edge enhancement, in which edge enhancement is performed on a character edge and edge enhancement is not performed on a halftone image, is performed on each of the R, G, and B signals.

  FIG. 4 shows an example of the MTF filter unit 405. The image signal converted from reflectance linear to lightness linear by the scanner γ conversion 402 is smoothed by the smoothing filter unit 1101 in FIG. As an example, the coefficients shown in Table 1 are used for smoothing.

  Next, the differential component of the image data is extracted by the 3 × 3 Laplacian filter 1102 at the next stage. Specific examples of the Laplacian filter are shown in Table 2.

Of the 10-bit image signal that is not subjected to γ conversion by the scanner γ conversion, the edge detection is performed by the edge amount detection filter 1103 on the upper 8 bits (one example).
A specific example of the edge amount detection filter 1103 is shown in FIG.

Of the edge amounts obtained by the edge detection filter 1103, the maximum value is used as the edge degree. That is, the edge degree is smoothed by the subsequent smoothing filter 1104 as necessary.
This reduces the influence of the sensitivity difference between the even and odd pixels of the scanner.
As an example, the coefficients shown in Table 3 are used.

  Then, the table conversion unit 1105 converts the obtained edge degree into a table. The values of this table specify the darkness of lines and dots (including contrast and density) and the smoothness of the halftone dots. An example of the relationship between the edge degree of the image and the filter coefficient in this table is shown in FIG. The edge degree is the largest with black lines or dots on a white background, and the edge degree becomes smaller as the pixel boundaries become smoother, such as finely printed halftone dots, silver halide photographs, and thermal transfer originals.

  The product (image signal D) of the edge degree (image signal C) converted by the table conversion unit 1105 in FIG. 5 and the output value (image signal B) of the Laplacian filter 1102 is the image signal after smoothing (image signal). Is added to A), and is transmitted as an image signal E to a circuit subsequent to the MTF filter 405.

  The color conversion UCR processing unit 406 corrects the difference between the color separation characteristics of the input system and the spectral characteristics of the output system color material, and calculates the amount of color materials Y, M, and C necessary for faithful color reproduction. The UCR process for replacing the part where the three colors Y, M, and C overlap with Bk (black) is performed. The color correction process can be realized by performing a matrix operation as shown in the following equation.

  Here, s (R), s (G), and s (B) represent the R, G, and B signals of the scanner 400 after the scanner γ conversion processing. Hue represents each hue such as White, Black, Yellow, Red, Magenta, Blue, Cyan, and Green. This division of hue is an example, and it may be divided more finely. The matrix coefficient aij (hue) is determined for each hue described above according to the spectral characteristics of the input system and the output system (color material). Here, the primary masking equation is taken as an example, but by using a quadratic term such as s (B) × s (B), s (B) × s (G), or a higher order term. Therefore, color correction can be performed with higher accuracy. Further, the Neugebauer equation may be used. In any method, Y, M, and C can be obtained from the values of s (B), s (G), and s (R).

  The determination of the hue is performed as follows as an example.

  The relationship between the reading value of the scanner 400 and the colorimetric value is determined using predetermined coefficients bij (i, j = 1, 2, 3).

It is expressed.

  From the definition of colorimetric values,

Therefore, it is possible to determine which hue corresponds to a certain pixel of the document read from the R, G, and B signals of the scanner 400. FIG. 7 shows an example of the hue. Since the hue of FIG. 7 is generally well known, only the outline will be described. The center of the upper concentric circle is the L * a * b * color system, with a * = b * = 0 and an achromatic axis. The distance from the center of the circle in the radial direction is saturation c *, and the angle from a straight line of a *> 0 and b * = 0 to a certain point is the hue angle h *. Each hue of Yellow, Red, Magenta, Blue, Cyan, and Green has a saturation that satisfies saturation c * ≧ c0 * with respect to a reference value c0 * having saturation, and the hue angle is respectively
Yellow: H1 * ≦ h * <H6 *
Red: H2 * ≦ h * <H1 *
Magenta: H3 * ≦ h * <0 and 0 ≦ h * <H2 *
Blue: H4 * ≦ h * <H3 *
Cyan: H5 * ≦ h * <H4 *
Green: H6 * ≦ h * <H5 *
And so on (this is an example).

The vertical axis in the lower diagram of FIG. 7 represents L * (brightness), the saturation c * is c * ≦ c0 *,
White: L = 100
Black: L = 0
And so on.
Further, simply, the hue is determined from the ratio s (B): s (G): s (R) of each signal of s (B), s (G), and s (R) and the absolute value. Is also possible.

  On the other hand, UCR processing can be performed by calculating using the following equation.

Y ′ = Y−α · min (Y, M, C)
M ′ = M−α · min (Y, M, C)
C ′ = C−α · min (Y, M, C)
Bk = α · min (Y, M, C)
In the above equation, α is a coefficient that determines the amount of UCR, and when α = 1, 100% UCR processing is performed. α may be a constant value. For example, when α is close to 1 in the high density portion and close to 0 in the highlight portion (low image density portion), the image in the highlight portion of the image can be smoothed. The above-described color correction coefficient is changed to 12 hues obtained by further dividing the six hues of RGBYMC into two, and further to every 14 hues of black and white.

  The hue determination unit 424 in FIG. 3 determines to which hue the read image data is determined. Based on the determined result, a color correction coefficient for each hue is selected.

  The image compression / decompression unit 426 performs image compression / decompression, and the image memory 403 stores and calls the compressed image data.

  A scaling unit 407 performs vertical / horizontal scaling of the image, and an image processing (creating) unit 408 performs repeat processing and the like.

  An image processing printer γ unit 409 corrects an image signal in accordance with an image quality mode such as characters and photographs. In addition, it is possible to simultaneously perform background removal in an image. The image processing printer γ correction unit 409 includes a plurality of (for example, 10) gradation conversion tables (printer γ tables) that can be switched in accordance with the area signal generated by the area processing unit 423 described above. . In this gradation conversion table, a table for performing optimum gradation conversion for each original such as characters, silver halide photographs (printing paper), printed originals, ink jets, highlighter pens, maps, thermal transfer originals, etc. You can choose from parameters.

  The gradation processing unit 410 performs gradation processing on the image. This gradation processing is performed by a SIMD type processor.

  FIG. 8 is an explanatory diagram showing a schematic configuration of the SIMD type processor 1506. A SIMD (Single Instruction-stream Multiple Data-stream) processor 1506 executes a single instruction in parallel on a plurality of data, and is composed of a plurality of PEs (processor elements). Each PE stores a register (Reg) 2001 for storing data, a multiplexer (MUX) 2002 for accessing a register of another PE, a barrel shifter (ShiftExpand) 2003, a logical operation unit (ALU) 2004, and a logical result. An accumulator (A) 2005 and a temporary register (F) 2006 for temporarily saving the contents of the accumulator.

  Each register 2001 is connected to an address bus and a data bus (read line and word line), and stores an instruction code defining processing and data to be processed. The contents of the register 2001 are input to the logical operation unit 2004, and the operation processing result is stored in the accumulator 2005. In order to retrieve the result outside the PE, the result is temporarily saved in the temporary register 2006. By extracting the contents of the temporary register 2006, the processing result for the target data is obtained.

  The instruction code is given to each PE with the same contents, the processing target data is given in a different state for each PE, and the contents of the register 2001 of the adjacent PE are referred to in the multiplexer 2002, so that the operation result is processed in parallel. It is output to 2005.

  For example, if the content of one line of image data is arranged in the PE for each pixel and is processed by the same instruction code, the processing result for one line can be obtained in a shorter time than the sequential processing of each pixel. In particular, in the spatial filter processing, the instruction code for each PE is an arithmetic expression itself, and the processing can be performed in common for all the PEs.

  Next, a hardware configuration in the case of realizing the SIMD type image data processing and the sequential type image data processing used in the image processing apparatus of FIG. 3 will be described. FIG. 9 is a block diagram for explaining the hardware of the configurations of a SIMD type image data processing unit 1500 that executes SIMD type image data processing and a sequential image data calculation processing unit 1507 that executes sequential type image data processing. . In this embodiment, first, the SIMD type image data processing unit 1500 will be described, and then the sequential type image data processing unit 1507 will be described.

  The image data parallel processing unit 1500 and the image data sequential processing unit 1507 process an image as a plurality of pixel lines composed of a plurality of pixels arranged in one direction. FIG. 10 is a diagram for explaining pixel lines, and shows four pixel lines of pixel lines a to d. In addition, pixels indicated by hatching in the figure are target pixels to be processed this time.

  In the present embodiment, in the error diffusion process of the target pixel, the influence of surrounding pixels on the target pixel is considered for both pixels included in the same pixel line and pixels included in different pixel lines. Then, error diffusion processing between pixels included in a pixel line different from the target pixel is performed by the SIMD type image data processing unit 1500, and pixels included in the same pixel line as the target pixel (circled numbers in the figure). The sequential image data processing unit 1507 performs error diffusion processing between the pixels 1, 2, and 3.

  The SIMD type image data processing unit 1500 includes a SIMD type processor 1506, five data input / output buses 1501a to 1501e for inputting image data and control signals to the SIMD type image data processing unit 1500, and data input / output buses 1501a to 1501a. 1501e is switched to switch the image data and control signals input to the SIMD type processor 1506, and the bus switches 1502a to 1502c for switching the bus width of the connected bus, and data used for processing the input image data are stored. 20 RAMs 1503 that control the memory controller 1505a, memory controller 1505b, memory controller 1505a, or memory controller 1505b that control the corresponding RAM 1503 Thus it has four memory switches 1504a~1504d for switching the RAM 1503.

  In the above configuration, the memory controller controlled by the bus switches 1502a to 1502c is identified as the memory controller 1505b, and the memory controller that is not controlled by the bus switches 1502a to 1502c is identified as the memory controller 1505a.

  The SIMD type processor 1506 described above includes registers 0 (R0) to 23 (R23). Each of R0 to R23 functions as a data interface between the PE in the SIMD type processor 1506 and the memory controllers 1505a and 1505b. The bus switch 1502a switches the memory controller 1505b connected to R0 to R3 and inputs a control signal to the SIMD type processor.

  The bus switch 1502b switches the memory controller 1505 connected to R4 and R5 and inputs a control signal to the SIMD type processor 1506. The bus switch 1502c switches the memory controller 1505 connected to R6 to R9 and inputs a control signal to the SIMD type processor 1506. The bus switch 1502c switches the memory controller 1505b connected to R6 to R9 and inputs a control signal to the SIMD type processor 1506.

  The memory switch 1504a exchanges image data between the PE in the SIMD type processor 1506 and the RAM 1503 using a memory controller 1505b connected to R0 to R5. The memory switch 1504b exchanges image data between the PE in the SIMD type processor 1506 and the RAM 1503 using the memory controller 1505b connected to R6 and R7. The memory switch 1504c exchanges image data between the PE in the SIMD type processor 1506 and the RAM 1503 using the memory controller 1505a or the memory controller 1505b connected to R8 to R13.

  The memory switch 1504d exchanges image data between the PE in the SIMD type processor 1506 and the RAM 1503 using the memory controller 1505a connected to R14 to R19.

  The image data control unit 203 inputs a control signal for processing the image data together with the image data to the bus switches 1502a to 1502c via the data input / output buses 1501a to 1501e. The bus switches 1502a to 1502c switch the bus width of the connected bus based on the control signal. Further, the memory switches 1504a to 1504c are switched so that the memory controller 1505b connected indirectly or directly is controlled to take out data necessary for processing image data from the RAM 1503.

  When performing error diffusion processing, the SIMD type image data processing unit 1500 inputs image data via the image data control unit 1203 (see FIG. 11). Then, error data that is a difference between pixel data included in a pixel line (previous pixel line) processed before the pixel line (current pixel line) including the target pixel (previous pixel line) and a predetermined threshold value and the target pixel Add pixel data.

  The SIMD type image data processing unit 1500 uses the SIMD type processor 1506 to execute addition with error data in parallel for a plurality of target pixels. For this reason, a plurality of error data corresponding to the number of pixels processed in batch by the SIMD processor 1506 is stored in any of the RAMs 1503 connected to the SIMD processor 1506. In this embodiment, the SIMD processor 1506 collectively performs addition processing for one pixel line, and error data for one pixel line is stored in the RAM 1503.

  The addition value of the image data and error data for one pixel line collectively processed by the SIMD type processor 1506 is sent to the sequential image data processing unit 1507 from at least two of R20, R21, R23, and R22 one by one. Is output. The error data used in the above processing is calculated by a sequential image data processing unit 1507, which will be described later, and is input to the SIMD type processor 1506.

  On the other hand, the sequential image data processing units 1507a and 1507b are hardware that operates regardless of the control of the computer program. In FIG. 4, two sequential image data processing units 1507 are connected to the SIMD type processor 1506. However, the image processing unit of the present embodiment is dedicated to error diffusion processing in which 1507b is sequentially performed. The other sequential image data processing unit 1507 is specially designed to be used for table conversion such as γ conversion.

  A hardware configuration of the image processor 1204 constituting the image processing unit in FIG. 3 will be described. FIG. 11 is a block diagram illustrating an internal configuration of the image processor 1204. In the block diagram of FIG. 11, the image processor 1204 has a plurality of input / output ports 1401 for data input / output with the outside, and can arbitrarily set data input and output.

  Further, a bus switch / local memory group 1402 is provided inside so as to be connected to the input / output port 1401, and the memory control unit 1403 controls the memory area to be used and the path of the data bus. The input data and the data for output are assigned to the bus switch / local memory group 1402 as a buffer memory, stored in each, and the external I / F is controlled.

  Various processing is performed in the processor array unit 1404 on the image data stored in the bus switch / local memory group 1402, and the output result (processed image data) is stored again in the bus switch / local memory group 1402. . The processing procedure in the processor array unit 1404, parameters for processing, and the like are exchanged between the program RAM 1405 and the data RAM 1406.

  The recording data in the program RAM 1405 and the data RAM 1406 is downloaded from the process controller 211 to the host buffer 1407 via the serial I / F 1408. Further, the process controller 211 reads the contents of the data RAM 1406 and monitors the progress of processing.

  If the processing contents are changed or the processing mode required by the system is changed, the contents of the program RAM 1405 and data RAM 1406 referred to by the processor array unit 1404 are updated. In the configuration described above, the processor array unit 1404 corresponds to the SIMD type image data processing unit 1500 and the sequential type image data processing unit 1507 described above.

  FIG. 12 is a block diagram for explaining the sequential image data processing unit 1507. The sequential image data processing unit 1507b shown in the figure processes an error data calculation unit 1801, a multiplexer 1807 for selecting one from the error data calculated by the error data calculation unit 1801, and the error data selected by the multiplexer 1807. And an error data adding unit 1808 for adding to the data input from the SIMD type image data processing unit 1500.

  Further, the sequential image data processing unit 1507b inputs a signal required for selecting error data to the multiplexer 1807 and an error diffusion mode (2) set in advance for the sequential image data processing unit 1500. An error diffusion processing hardware register group 1805 that can set an operation coefficient used for error diffusion processing, or whether error diffusion is executed by any one of value error diffusion, three-value error diffusion, and four-value error diffusion). Yes. Further, the sequential image data processing unit 1507b includes a blue noise signal generation unit 1809, and is configured to be able to select whether or not to use blue noise for error diffusion processing by setting the error diffusion processing hardware register group 1805. ing.

  The error data calculation unit 1801 is configured to calculate error data that is a difference between pixel data of a pixel included in the current pixel line and a predetermined threshold value. The error data calculation means 1801 includes threshold value table groups 1810a to 1810c connected to three quantization reference value storage units 1803a to 1803c, three comparators 1804a to 1804c, and three multiplexers 1802a to 1802c, respectively. Yes.

  For example, six threshold value tables THxA to THxF (x = 0, 1, 2) are connected to the threshold value table groups 1810a to 1810c, respectively. This can be selected by setting the error diffusion processing hardware register group 1805. In the gradation processing in this embodiment, an image processor used for gradation processing of Magenta and Cyan image data, and Yellow and Black image data are used. Two image processors are used.

  In the following, an image processor for processing image data of Magenta and Cyan will be described as an example. THxA to THxC (x = 0, 1, 2) are used for Magenta, and THxD to THxF (x = 0, 1, 2) are used for Cyan. THxA to THxC (x = 0, 1, 2) used for Magenta can be selected so that each threshold value table can be selected according to the extraction results based on the feature values of images such as text, photos, and intermediates. Can be left as Simple error diffusion with a fixed threshold that does not depend on the main scanning or sub-scanning position in the character part, error diffusion with a dither threshold with a low number of lines in the photograph part, and a higher line than the photograph part in the intermediate part Error diffusion with a set threshold value can be performed, and a more preferable image can be formed. TH0A to TH2A are threshold values for pixels determined to have the same feature amount. The same applies to Cyan. The processor that processes the image data of Yellow and Black is the same as that described above with Magenta replaced with Yellow and Cyan replaced with Black.

  In this embodiment, the multiplexer 1802a connected to the quantization reference value storage unit 1803a, the comparator 1804a, and the threshold table group 1810a operates as a set. The multiplexer 1802b connected to the quantization reference value storage unit 1803b, the comparator 1804b, and the threshold table group 1810b operates as a set, and the multiplexer connected to the quantization reference value storage unit 1803c, the comparator 1804c, and the threshold table group 1810c. 1802c operates as a set.

  The sequential image data processing unit 1507 inputs an addition value (addition value data) between the image data and the error data from the SIMD type processor 1506. This image data is the image data of the target pixel processed this time, and the error data is the error data of the pixel processed before the target pixel.

  The input added value data is added with the value calculated by the error data adding unit 1808 based on the previously processed pixel error data, and is divided by 16 or 32 to reduce the calculation error. Further, the divided addition value data is input to all of the three comparators 1804a to 1804c of the error data calculation unit 1801. Note that the value calculated by the error data adding unit 1808 based on the error data of the pixel previously processed will be described later.

  The comparators 1804a to 1804c input threshold values from the multiplexers 1802a to 1802c connected to the connected threshold value table groups. Then, the threshold value is subtracted from the input addition value data to create image data. Further, a value obtained by subtracting the quantization reference value stored in each of the quantization reference value storage units 1803a to 1803c from the added value data is output to the multiplexer 1807 as error data. As a result, a total of three error data are simultaneously input to the multiplexer 1807. When blue noise is used for error diffusion processing, the blue noise signal generation unit 709 generates blue noise by turning on / off the blue noise data at a relatively high cycle. The threshold is subtracted from the blue noise before entering the comparators 1804a-1804c. By the process using blue noise, it is possible to prevent the occurrence of a unique texture in the image by giving an appropriate variation in the threshold value.

  Different threshold values are stored in the threshold value tables 1802a to 1802c. In this embodiment, the threshold value table 1802a stores the largest threshold value among the threshold value tables 1802a to 1802c, and then the threshold value stored in the order of the threshold value table 1802b and the threshold value table 1802c decreases. In addition, quantization reference values to be stored in the quantization standard value storage units 1804a to 1804c are set according to the connected threshold value tables 1802a to 1802c. For example, when the image data is represented by 256 values from 0 to 255, 255 is stored in the quantization reference value storage unit 1803a, 170 is stored in the quantization reference value storage unit 1803b, and quantization reference value storage unit 1803c. Is stored with 85.

  The comparators 1804a to 1804c output the created image data to the logic circuit 1806. The logic circuit 1806 selects the image data of the target pixel from these and inputs it to the multiplexer 1807. The multiplexer 1807 selects one of the three error data as the error data of the target pixel according to the input image data. The selected error data is input to one of the RAMs 1503 via the PE of the SIMD type processor 1506.

  Further, the image data output from the logic circuit 1806 is branched before being input to the multiplexer 1807 and input to one of the PEs of the SIMD type processor 1506. In the present embodiment, the image data is data represented by 2 bits of upper and lower bits. For this reason, the comparator 1804a is not used in this process. In the present embodiment, the image data of the target pixel is hereinafter referred to as pixel data.

  The selected error data is input to the error data adding unit 1808. The error data adding unit 1808 is the error data of the pixel indicated by the circled numbers 1, 2, and 3 in FIG. 10, that is, the pixel processed three times before the target pixel (the error data in FIG. Error data of the pixel processed two times before (denoted as error data 2 in FIG. 12), error data of the pixel processed immediately before (denoted as error data 1 in FIG. 12) ).

  The error data adding unit 1808 multiplies the error data 3 by 0 or 1 that is a calculation coefficient. Further, the error data 2 is multiplied by a calculation coefficient 1 or 2, and the error data 1 is multiplied by a calculation coefficient 2 or 4. Then, the three multiplied values are added together, and this value (weighting error data) is added to the added value data inputted next from the SIMD type processor 1506. As a result, a pixel closer to the target pixel has a larger influence on the error diffusion process of the target pixel, and the error of the pixel can be appropriately diffused to form an image closer to the original image.

The creation of the image data in the sequential image data processing unit 1507 described above is generally performed using a configuration called an IIR filter system. The arithmetic expression used in the IIR type filter system is as shown in FIG.
ODn = (1-K) * ODn-1 + K.IDn (1)
(However, ODn: pixel density after calculation, ODn-1: calculation result using previous pixel data, IDn: current pixel data, K: weighting factor)
It can be expressed as.

  As apparent from the equation (1) and FIG. 21, the density ODn after the calculation is obtained from the calculation result ODn-1 using the previous pixel data and the value of the current pixel data IDn. In general, an IIR filter system is a dedicated circuit for performing so-called sequential conversion, which performs an operation on a current pixel using an operation result using a pixel processed before the current pixel. The sequential image data processing unit 507 of the image processing apparatus according to the present embodiment can be used for the entire sequential conversion as shown in FIG. 13 regardless of the processing (described later) as shown in FIG. it can.

  Next, the gradation processing unit 410 shown in FIG. 3 will be described with reference to FIG. The gradation processing unit 410 includes a feature amount detection unit 1701, a threshold selection unit 1702, an error integration matrix 1704, and an error buffer 1705.

  The input image data is determined by the feature amount detection unit 1701 as to whether it is an image signal closer to a character, a character, or an image signal closer to a photograph. The threshold selection unit 1702 selects a threshold based on the extraction result of the feature amount detection unit 1701. The threshold value processing unit 1703 calculates the threshold value selected by the threshold value selection unit 1702, the quantized value corresponding to the threshold value, the difference between the image data and the quantized value, and the error between the selected image data, and outputs the error integration output matrix 1704. The output value is obtained from the result. The error integration output matrix 1704 reads out the error of the pixel of interest (* pixel in the error diffusion matrix shown in FIG. 15) and surrounding pixels from the error buffer 1705, and based on a predetermined coefficient (pixels a to l in FIG. 15). To accumulate.

  The configuration of the feature quantity extraction unit 1701 will be described with reference to FIG. 16 includes a primary differential filter 1711, an absolute value calculation unit 1712, a primary differential result maximum value selection unit 1713, a primary edge determination unit 1714, an altitude determination unit 1715, a secondary differential filter unit 1716, and a second differential filter unit 1716. A maximum value selection unit 1717 and a secondary edge determination unit 1718 for the secondary differential filter are included. The primary differential filter 1711 performs a filtering process on the primary differential filter coefficients (four types of FIGS. 5A to 5D).

  The absolute value calculation unit 1712 obtains the absolute value of the processing result of the primary differential filter 1711. The maximum value selection unit 1713 of the primary differential result obtains the maximum value of the four types of primary differential filter processing results calculated by the absolute value calculation unit 1712. The primary edge determination unit 1714 quantizes the value obtained by multiplying the result of the absolute value calculation unit 1712 by 1/8 into four levels (two bits) of edge degree using three quantization thresholds. The altitude determination unit 1715 determines input image data based on a high density determination threshold. The second-order differential filter unit 1716 performs filter processing according to the four types of second-order differential filter coefficients shown in FIGS. The secondary differential filter maximum value selection unit 1717 selects the maximum value of the filter processing result of the secondary differential filter unit 1716. The secondary edge determination unit 1718 determines a value obtained by multiplying the result of the maximum value selection unit 1717 of the secondary differential filter by 1/8 based on the secondary edge determination threshold.

  The function of the feature quantity extraction unit 1701 will be described with reference to FIG. Here, the function of feature amount extraction will be described. FIG. 17 includes graphs (a) to (e), in which the horizontal axis represents the position of a one-dimensional pixel, and the vertical axis represents the relative value of the output value of each unit.

  FIG. 17A shows input image data to the feature amount extraction unit 1701, and hatching represents a region where a determination region based on a high density threshold by the altitude determination unit 1715 is true.

  In FIG. 17B, the solid line indicates the output result of the primary differential filter 1711, and the wavy line indicates the result of converting the negative value portion of the output result of the primary differential filter 1711 into an absolute value by the absolute value calculation unit 1712.

  FIG. 17C illustrates the relationship with the threshold value quantized by the primary edge determination unit 1714 based on the selection result of the maximum value selection unit 1713 of the primary differentiation result. Hatching is a region where the primary differential determination threshold value 1 is true and the high density determination threshold value is true. Note that the result of FIG. 17B is already the processing result of the filter that obtains the maximum value, and thus the shape of the graph of the result of taking the absolute value of FIG.

  FIG. 17D shows the result of selecting the second derivative calculation result by the second derivative filter unit 1716 using the maximum value selecting unit 1717. Hatching represents a region in which determination by the secondary differentiation determination threshold performed by the secondary edge determination unit 1718 is true.

  FIG. 17E shows input image data, and hatching represents a determination region based on a primary differential determination threshold value, a high density determination region + secondary differential determination threshold value.

  Table 4 shows the relationship between the feature amount extraction result and the error diffusion quantization threshold.

  That is, as described above, the image data is subjected to gradation conversion in the image processing printer γ unit 409 based on the printer γ conversion table (tone conversion table), and the converted multi-gradation image data is The key processing unit 410 performs quantization using the quantization threshold. Also, the feature amount calculation unit 1701 of the gradation processing unit 410 determines whether the image is close to a character, in a character, or close to a photograph. The threshold selection unit 1702 selects a quantization threshold based on the extraction result of the feature amount detection unit 1701.

  Table 4 explains the selection of the quantization threshold by the threshold selection unit 1702. That is, in Table 4, the vertical item is the image type, and the horizontal item is the feature amount determination result and the amplitude of the selected quantization threshold. The threshold selection unit 1702 selects either a periodically oscillating quantization threshold or a fixed quantization threshold regardless of the pixel position according to the feature amount determination result, and periodically oscillating quantization. For the threshold value, the magnitude of the amplitude is also selected. Table 4 shows a case where amplitude is selected as large, small, or none as example 1, and a case where amplitude is selected as large, medium, or small as example 2. That is, an error diffusion quantization threshold having a large amplitude is selected for a photographic image or background. For a character on a halftone dot having a high appearance frequency in a map manuscript or the like, a quantization threshold having a small amplitude or a medium amplitude is selected. For character images, error diffusion with a fixed threshold value with no quantization threshold amplitude or error diffusion with a small amplitude value is performed.

  That is, in the error diffusion processing technology that extracts the image feature amount from the image data after gradation conversion by the printer γ table and switches the quantization threshold used in the error diffusion processing based on the feature amount extraction result, Due to structural limitations, even if the feature value extraction threshold used for feature value extraction is fixed, the area from which the feature value is extracted differs depending on the parameters of the printer γ table, and the image finally obtained by gradation processing May be different.

  Therefore, if the parameter of the feature quantity extraction threshold is also changed by the parameter of the printer γ table, the result of feature quantity extraction can be made less susceptible to the influence of the gradation conversion table.

  Here, the quantization threshold having a large amplitude is the quantization threshold described in FIGS. Examples of this threshold are 600 dpi and correspond to 168 lines and 144 lines (FIGS. 22 and 23), respectively.

  On the other hand, an example of the quantization threshold value without amplitude is a fixed threshold value regardless of the position of the pixel, as shown in th2 and th3 (threshold values 2 and 3 respectively) in FIG. The amplitude (small) is the threshold amplitude when the number of lines is increased from 300 lines corresponding to 600 dpi as shown by th1 in FIG. 24 and from 168 lines corresponding to FIGS. 18 to 22, or as indicated by th1 in FIG. Includes cases where the width is reduced.

  The parameters of the feature amount extraction threshold used for the feature amount extraction described above are the high density threshold value, the first derivative determination threshold values 1 to 3 and the second derivative determination threshold value, which are set according to the printer γ table set in the image processing printer γ 409. Tone conversion is performed, and the converted values are set in the high density determination unit 1715, the primary edge determination unit 1714, and the secondary edge determination unit 1718, respectively.

  That is, by converting the gradation of the parameter of the feature amount extraction threshold value using the printer γ table, it is possible to make the feature amount extraction process less susceptible to changes due to the printer γ table.

  In addition, the feature amount extraction parameter is obtained by performing processing using the parameters of the printer γ table on either the extraction threshold value of the primary differential filter, the extraction threshold value of the secondary differential filter, or the high density threshold value. It is possible to reduce the influence of gradation conversion parameters from the extraction result.

  Table 5 is an example of Black as the feature amount extraction threshold, and an optimal value is set for each color of Black, Cyan, Magenta, and Yellow. The set value with respect to the reference value is described.

  Regarding the high density threshold value in Table 5, since the output value for the input value 40h to the printer γ table in Table 3 is 50h, the set value 50h after conversion by the printer γ table for the high density threshold value in Table 5 is set. .

  Since the primary differentiation determination thresholds 1 to 3 are values of the primary differentiation, it is not always appropriate to obtain the output value by applying the threshold before conversion to the input value of the printer γ table. However, with regard to the primary differential determination thresholds 1 and 2, the background of the original is white (YMCK = (0,0,0,0)) and the characters are black (YMCK = (0,0, In the case of 0, k)) or the like, the first-order differential result is proportional to (k-0), so the output value D0h for the input value E0h to the printer γ table is set as the set value after conversion. On the other hand, the primary differentiation determination threshold 3 includes a case where pixels around a character are not white, such as a character in a halftone dot, so a reference value that is not converted in the printer γ table is set. A conceptual diagram of numerical value conversion by the printer γ table is shown in FIG. The horizontal axis of the graph is the input value, the vertical axis is the output value, and the output value is determined by converting the value input on the horizontal axis into a graph.

  The second-order differential determination threshold is related to the secondary change amount of the printer γ table and is not uniquely affected by the magnitude of the printer γ table value. To do.

  The above processing will be described based on the flowchart of FIG. First, it reads out from the memory in accordance with the color (Table 7) for setting the threshold and the image quality mode (Table 8) (step S301). Then, it is determined whether or not the threshold is to be converted by the printer γ table (step S302). In the case of the primary differentiation determination thresholds 1 and 2 and the high density threshold, the threshold is to be converted by the printer γ table (Y in step S302). Conversion is performed using the printer γ table (step S303), and the converted values of the primary differentiation determination thresholds 1 and 2 and the high density threshold are set in the threshold determination unit (step S304). In the case of the primary differential determination threshold value 3 and the secondary differential determination threshold value, since it is not necessary to convert with the printer γ table (N in step S302), it is set without conversion with the printer γ table (step S304). These settings are executed for each of the colors Y, M, C, and K (N in step S305), and when each color is completed (Y in step S305), a plurality of image quality modes are provided for each image area by area processing or the like. Is set (Y in step S306), the process returns to step S301 and is repeated according to the image quality mode to be used.

  The setting in step S304 is executed for each of the colors Y, M, C, and K (N in step S305) for the following reason. In other words, the relationship between the original density and the input value to the gradation process differs depending on the printer γ table and the color correction process before the gradation process, and the output density and the input value to the gradation process depend on the color of the toner. Therefore, the parameters common to YMCK may not be optimal for the image quality. Therefore, if the parameter for feature amount extraction can be set for each color, an image suitable for the user can be obtained.

  Further, it is desired to obtain an output image corresponding to the type of original such as a printed original, a silver halide photograph original, a map original, and an ink jet original, or to obtain an output image preferable for the user such as a character mode or a photo mode. Therefore, in order to obtain the output image according to the document type, such as a print document, a silver salt photograph document, a map document, and an inkjet document, and to obtain an output image preferable for the user such as a character mode and a photograph mode. In addition, when a plurality of image quality modes are set for each image area by area processing or the like (Y in step S306), the process returns to step S301 and is repeated according to the image quality mode to be used.

  FIG. 28 is an explanatory diagram for explaining the registers set in the error diffusion processing hardware register group 1805. The image processing apparatus according to the present embodiment is a mode for performing error diffusion processing by binary error diffusion (binary error diffusion mode) and a mode for performing error diffusion processing by ternary error diffusion (ternary) by setting the illustrated register. It is possible to select which of error diffusion processing and error diffusion processing is performed by quaternary error diffusion (quaternary error diffusion mode). In addition, a calculation coefficient used in the error data adding unit 1808 can be set. Furthermore, it is possible to select whether or not to use blue noise for error diffusion processing.

  The error diffusion processing hardware register group 1805 illustrated in FIG. 28 sets a register 2001 that sets a quantization reference value 0 of the quantization reference value storage unit 1803a and a quantization reference value 1 of the quantization reference value storage unit 1803b. A register 2002 and a register 2003 for setting the quantization reference value 2 of the quantization reference value storage unit 1803c are provided.

  Further, the error diffusion processing hardware register group 1805 is set in the register 2004 for setting the threshold 0 set in the threshold table 1802c, the register 2005 for setting the thresholds 10-17 set in the threshold table 1802b, and the threshold table 802a. A register 2006 for setting thresholds 20 to 27, a register 2007 for setting a blue noise value, and an error diffusion processing hardware control register 2008. Each register is assigned 8 bits, and the entire register has a data amount of 64 bits.

  In the binary error diffusion mode, the same value is set in all of the register 2001, the register 2002, and the register 2003. This can be realized by setting FFH in the registers 2004 and 2005. In the ternary error diffusion mode, the same value is set in the registers 2001 and 2002, and FFH is set in the register 2004. Further, in the binary error diffusion mode and the ternary error diffusion mode, the fixed threshold error diffusion process and the variable threshold error diffusion process are switched depending on whether the register 2005 and the register 2006 are set to the same value or different values. be able to.

  When blue noise is used for error diffusion processing, a value indicating that blue noise is used is set in the register 2007. Then, switching data indicating on / off of the blue noise data is set in the register 2005. When the switching data is 1, the blue noise value is added to each threshold value. When the switching data is 0, the threshold value is used as it is. Further, the calculation coefficient used in the error data adding unit 1808 can be selected by changing the set value of the error diffusion processing hardware control register.

  Next, error diffusion processing executed by the SIMD type processor 1506 will be described with reference to the flowchart of FIG. The SIMD processor 1506 first determines whether or not the current image data is the first line (step S2101). If it is the first line (Y in step S2101), the error addition value for the previous two lines is initialized. (Step S2101). Then, it is determined whether or not the image data to be subjected to the current error diffusion calculation is the first SIMD (step S2103). If it is the first SIMD (image data at the beginning of the current line) (Y in step S2103), an error is determined. The added value is initialized (step S2105). If it is not the first SIMD (N in step S2103), it is determined whether the error data after the error diffusion calculation in the previous SIMD is the same color as the currently calculated image data (step S2106). (N in step S2106), the previous SIMD calculation result is stored as a different color for the previous line (step S2107, process A2 in FIG. 30), and the reference position of the blue noise table is also stored (step S2109), the same color The blue noise reference position in the previous error diffusion calculation is called (step S2110). If they are the same color (Y in step S2106), they are stored as the 1 SIMD calculation result of the previous line of the same color (step S2108, process A1 in FIG. 30). For example, if it is Magenta color image data that is going to be used for error diffusion calculation, the image data of a different color is the Cyan image data. In some cases, it is judged as a different color, and in the case of the Magenta version, it is judged as the same color.

  The error added value data two lines before is stored as data one line before the previous SIMD (step S2111, processing B in FIG. 30), and the data for two lines before the current SIMD is called from the memory (step S2112, FIG. 30). Process D, E). After calling the current SIMD data from the current line (process C in FIG. 30), an error addition value is calculated (step S2113). Thereafter, the error diffusion process is calculated by the sequential processing calculation unit 1507b (step S2114).

  On the other hand, as shown in the flowchart of FIG. 31, the sequential image data processing unit 1507 inputs the added value data output from the SIMD type processor 1506 in step S2102 (step S2201). Then, the weighted error data generated by the error data adding unit 1808 is added to the input added value data (step S2202). The added value data to which the weighting error data is added is divided by 16 or 32 (step S2203) and input to the error data calculation unit 1801. The error data calculation unit 1801 generates error data and pixel data based on the input data (step S2204), and inputs the error data to the multiplexer 1807. Further, the pixel data is input to the logic circuit 1806 and the SIMD type processor 1506.

  The multiplexer 1807 selects one error data according to the image data input from the logic circuit 1806 (step S2205). The selected error data is output to the SIMD type processor 1506 and the error data adding unit 1808 (step S2206). The error data adding unit 1808 that has input the error data calculates weighted error data based on the error data (step S2207). The sequential image data processing unit 1507 repeatedly executes the above processing sequentially for the input addition value data.

  In this example, two SIMD processors 1506 having the sequential image data processing unit 507b shown in FIG. 12 are used, and Y (Yellow) image data and K (Black) image data are compared with YMCK image data. The processor 1506 having one sequential image data processing unit 507b is used, and two sets of image data of the image data of the C image signal M are used by using the SIMD processor 1506 having another sequential image data processing unit 507a. Perform gradation processing. Therefore, two image data (YK or CM) before gradation processing input to the SIMD processor 1506 and two image data (YK or CM) output from the SIMD processor 1506 are processed. To do.

  When performing the error diffusion process, the SIMD processor 1506 having one sequential processing operation unit is switched for each of the two pieces of input image data that can be subjected to SIMD processing.

  Next, a description will be given based on the state transition diagram of the image processor shown in FIG. As shown in FIG. 32, the processing state of the image processor loops as follows: command → main 1 (processing Magenta / Yellow image data) → main 2 (processing Cyan / Black image data) → command → main 1. ing.

  The operation of the image processor at the time of two inputs and two outputs will be described based on the flowchart of FIG. In the main process 1, Magenta or Yellow image data is processed, and in the main process 2, Cyan or Black image data is processed.

  An Magenta (Yellow) input is input to the SIMD processor 1506 using the data input / output bus 1501a and output using the data input / output bus 1501c. Similarly, input of Cyan (Black) image data is input using the data input / output bus 1501b and output using the data input / output bus 1501d. The data input / output bus 1501c is used for debugging output.

  If there is data input to the SIMD processor 1506 in the main process 1 (Y in step S2301), the process of taking image data into the memory 1503 is started (step S2302). When the capture of one line is completed (Y in step S2303), gradation processing (in this case, error diffusion processing) is started in units of image data that can be processed by the SIMD processing processor 1506 (step S2305). When the one-line processing is completed (Y in step S2305), one-line output is started (step S2306). In the memory data capture and output start processing such as steps S2302 and S2306, processing start commands to the respective memory controllers 1505a to 1505b are set in the registers, and the SIMD processor 1506 shifts to the next control (state transition). The start of gradation processing (error diffusion processing) (step S2305) is performed by writing a predetermined setting value corresponding to the start command of the error diffusion processing hardware control register 2008 as a start processing command to the sequential processing calculation unit 1507b. .

  The same applies to the main processing 2 (thus, detailed description is omitted).

  In the command processing, command reception processing from the control CPU is performed for the SIMD processor 1506.

  An image forming printer γ (process control γ) correction unit 412 shown in FIG. 3 converts an image signal from the interface 411 with a printer γ table and outputs the converted signal to a laser modulation circuit described later.

  In FIG. 3, a printer unit 452 is configured by an interface 411, an image forming printer γ 412, a printer 413, a controller 417, and the like, and can be used independently of the scanner / IPU unit 451. An image signal from the host computer 418 is input to the interface 411 through the printer controller 419, is subjected to gradation conversion by the image forming printer γ correction unit 412, and image formation is performed by the printer 413, so that it can be used as a printer.

  The image processing apparatus described above is controlled by the CPU 415. The CPU 415 is connected to the ROM 414, the RAM 416, and the BUS 418. The CPU 415 is connected to the system controller 417 through a serial I / F, and commands from an operation unit (not shown) are transmitted through the system controller 417. Various parameters are set in each of the image processing apparatuses described above based on the transmitted image quality mode, density information, region information, and the like. The pattern generation circuits 421 and 422 generate gradation patterns used in the scanner / IPU unit 451 and the printer unit 452, respectively.

1 is a configuration diagram of an entire copying machine according to an embodiment of the present invention. FIG. 2 is a block diagram of a control system of a copying machine. FIG. 1 is a block diagram illustrating an overall configuration of an image processing apparatus of a copying machine. It is a block diagram which shows an example of an adaptive type edge emphasis circuit. It is explanatory drawing of a primary differential edge detection filter. It is explanatory drawing of an adaptive type edge emphasis filter table. It is explanatory drawing explaining a hue. It is explanatory drawing explaining the structure of a SIMD type | mold processor. It is a block diagram which shows the structure of a SIMD type image data processing part and a sequential type image data processing part. It is explanatory drawing explaining the positional relationship of a pixel. It is explanatory drawing of the internal structure of an image processor. It is a block diagram explaining a sequential image data processing part. It is a block diagram of an IIR filter system. It is explanatory drawing of a gradation process part. It is explanatory drawing of an error diffusion matrix. It is explanatory drawing of a feature-value extraction part. It is explanatory drawing of a feature-value extraction part. It is explanatory drawing of a quantization threshold value. It is explanatory drawing of the example of a low image density. It is explanatory drawing of the example of medium image density. It is explanatory drawing of the example of high image density. It is explanatory drawing of a quantization threshold value. It is explanatory drawing of a quantization threshold value. It is explanatory drawing of a quantization threshold value. It is explanatory drawing of a quantization threshold value. It is a conceptual diagram of the process which converts a feature-value extraction threshold value with a printer gamma table. It is a flowchart of the process which converts a feature-value extraction threshold value. It is explanatory drawing of an error diffusion hardware register. It is a flowchart explaining the error diffusion process performed with a SIMD type | mold processor. It is explanatory drawing explaining a line shift. It is a flowchart explaining the procedure of the error diffusion process performed by a sequential type image data processing part. It is a state transition diagram of an image processor. It is a flowchart of an image processor.

Explanation of symbols

409, 410 Image processing device 413 Printer engine

Claims (6)

Means for performing gradation conversion of image data based on a predetermined gradation conversion table;
Means for quantizing the gradation-converted multi-gradation image data by an error diffusion method using a predetermined quantization threshold;
Means for extracting a feature amount of an image based on a feature amount threshold value from the image data after the gradation conversion;
Based on the feature value after extraction, the use of the error diffusion method from among a plurality of the quantization thresholds of which at least the former is different among the presence or absence of periodic vibration and the magnitude of amplitude when there is vibration Means for selecting
Means for processing a feature amount extraction threshold used for extracting the feature amount in the gradation conversion table;
With
The feature amount extraction threshold after the processing is used for the extraction of the feature amount.
Image processing device.   The image processing apparatus according to claim 1, wherein the feature amount extraction and error diffusion processing are executed by a programmable image processing processor capable of sequential processing.   The image processing apparatus according to claim 1, wherein the feature amount extraction threshold parameter includes one of an extraction threshold value of a first-order differential filter and a high density threshold value.   The image processing apparatus according to claim 1, wherein the feature amount extraction threshold parameter can be set for each of a plurality of color components constituting the image.   The image processing apparatus according to claim 1, wherein the feature amount extraction threshold parameter can be set for each image quality mode. A printer engine for forming an image on a medium;
An image processing device according to any one of claims 1 to 5,
An image forming apparatus.

Images