WO2006013690A1 - 画像復号装置 - Google Patents
画像復号装置 Download PDFInfo
- Publication number
- WO2006013690A1 WO2006013690A1 PCT/JP2005/012310 JP2005012310W WO2006013690A1 WO 2006013690 A1 WO2006013690 A1 WO 2006013690A1 JP 2005012310 W JP2005012310 W JP 2005012310W WO 2006013690 A1 WO2006013690 A1 WO 2006013690A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- image
- block
- reference image
- memory
- compensation
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/42—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation
- H04N19/436—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation using parallelised computational arrangements
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/127—Prioritisation of hardware or computational resources
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/156—Availability of hardware or computational resources, e.g. encoding based on power-saving criteria
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
- H04N19/176—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/42—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation
- H04N19/43—Hardware specially adapted for motion estimation or compensation
- H04N19/433—Hardware specially adapted for motion estimation or compensation characterised by techniques for memory access
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/44—Decoders specially adapted therefor, e.g. video decoders which are asymmetric with respect to the encoder
Definitions
- the present invention relates to an image decoding apparatus that decodes or expands a compressed image, and is particularly suitable for decoding a compressed image in accordance with the MPEG4 AV standard, Moving Picture Experts uroup phase 4 Advanced ideo Codings I3 ⁇ 4Ol4 496-10).
- the present invention relates to image decoding technology.
- MPEG Motion Picture Experts Group
- ISOZIEC18818-2 MPEG2 video standard
- ISOZIEC14496-2 MPEG4 visual standard
- Such an image decoding apparatus decodes a variable-length encoded stream related to a compressed image taken in a memory such as a large-capacity DRAM (Dynamic Random Access Memory) and reads the stream from the memory.
- Variable length decoding is used to extract motion vectors, block data, etc. for each macroblock (MB: Macroblock), and for each macroblock, block data is referenced with reference to the reference image specified by the motion vector in that memory.
- motion compensation processing equivalent to so-called motion detection inverse processing (hereinafter referred to as “motion compensation processing”) is performed, and the decoded image obtained as a result of the motion compensation processing is recorded in the memory.
- motion compensation processing equivalent to so-called motion detection inverse processing
- a conventional image decoding apparatus performs pipeline control in which each component that performs each process necessary for decoding is operated in parallel in units of macroblocks in order to decode compressed images at high speed. .
- FIG. 13 is a diagram illustrating an execution sequence of the knock line control in the conventional image decoding apparatus.
- a variable-length decoder (VLD) processing queue, motion compensation processing unit, and DMA (Direct Memory Access) controller operate in parallel.
- the DMA controller can perform stream transfer processing for DMA transfer of a variable-length encoded stream for one macroblock from the memory to the variable-length decoding processing unit, and motion compensation for the reference image corresponding to one macroblock from the memory.
- the reference image transfer processing for DMA transfer to the processing unit and the decoded image transfer processing for DMA transfer of the decoded image corresponding to one macroblock obtained by the motion compensation processing unit to the memory are performed as synchronization cycles in the noise control. Executed in time division within each cycle time. Here, each cycle time is also called a time slot (TS)!
- n is an arbitrary integer
- the reference image transfer process corresponding to is executed by the DMA controller in the (n + 2) second time slot, and the motion compensation processing power corresponding to the macro block + is executed by the motion compensation processing unit in the third time slot.
- the decoded image transfer process corresponding to the macroblock is executed by the DMA controller in the (n + 4) th time slot.
- the conventional image decoding device has the maximum time required for processing in units of one macroblock for each component that executes reference image transfer processing, variable length decoding processing, motion compensation processing, and the like.
- a cycle time that satisfies the decoding performance of one macroblock required by the image decoding device is determined, and pipeline control is performed so that each component executes processing in synchronization with each cycle time. is doing.
- reference image transfer which is a memory access for supplying a reference image corresponding to each macro block to the motion compensation processing unit, in order to optimally perform pipeline control based on the processing time of one macro block unit. Each process was performed in one cycle time.
- the MPEG4AVC standard (see Non-Patent Document 1) developed in recent years has scalability for target images, from small images such as QCIF (Quarter Common Intermediate Format) to HD (High Definition) images. This is a standard for content that is divided into multiple levels based on the size of the image.
- the maximum number of motion vectors corresponding to one macroblock was 4 in the MPEG2 video standard, but increased to 32 in the MPEG4AVC standard.
- Non-Patent Document 1 ISO / IEC 14496 10 Information technology-Coding of audiovisual objects ⁇ Part 10: Advanced video and oamg
- the conventional image decoding apparatus described above is designed to be able to decode an image compressed in accordance with the provisions for the level corresponding to a large image of the MPEG4AVC standard, the conventional image decoding device can be used within one cycle time.
- the memory bus bandwidth needs to be significantly increased to handle the possibility of transferring more than twice the reference image from the memory to the motion compensation processing unit. For this reason, a large-capacity memory for storing images needs to operate at a higher frequency, for example, and as a result, the manufacturing cost of the image decoding device increases greatly.
- the present application has been made in view of the above problems, and is an image decoding apparatus capable of decoding an image compressed in accordance with a level definition corresponding to a large image of the MPEG4AVC standard.
- An object of the present invention is to provide an image decoding device having a configuration for suppressing a large memory bus bandwidth.
- an image decoding device decodes a compressed image for each block, and performs motion compensation for a block compressed based on a reference image.
- An image decoding apparatus that performs decoding including compensation processing corresponding to the above, has a memory for storing a reference image group, and an input canister for storing the reference image, and is stored in an input buffer.
- Compensation means for sequentially performing compensation processing for one block with reference to a reference image, and within each cycle time, as long as there is a reference image corresponding to one block, all the reference images are read from the memory and Reference image transfer means for writing to the input buffer; and determination means for determining whether or not the total amount of reference images read from the memory by the reference image transfer means at each cycle time is greater than a predetermined reference amount Greater and within the determined cycle time by said determining means, other than the access for the reference image reading, characterized in that it comprises a suppression means for suppressing access to the memory.
- the cycle time is a period of a predetermined period of time, for example, a unit time in pipeline control.
- the image decoding apparatus suppresses other memory accesses for reading the reference image within the cycle time when the amount of reference image read from the memory is large.
- the maximum amount of data transferred in time can be reduced, which makes it possible to reduce the memory bus bandwidth.
- the maximum amount of memory between the memory and the cycle time is achieved. Need to transfer the amount of data transfer.
- the image according to the present invention Memory accesses temporarily inhibited within a certain cycle time A by the decoding device need only be performed at the next cycle time B when the data transfer amount for the reference image is reduced. Therefore, the image decoding apparatus according to the present invention does not cause any particular problem by suppressing memory access at a certain cycle time.
- the image decoding apparatus further includes acquisition means for acquiring a motion vector corresponding to one block that is sequentially compressed based on the reference image, and the compensation means refers to the reference image. Then, based on the motion vector acquired by the acquisition unit, a compensation process for a block is performed, and all the reference images corresponding to one block transferred by the reference image transfer unit are Specified by each motion vector corresponding to the block acquired by the means, and the determination means has a total amount of reference images corresponding to the blocks read from the memory at each cycle time larger than a predetermined reference amount. Whether or not the number of motion vectors corresponding to the block acquired by the acquisition unit is greater than a predetermined threshold. Even Rukoto!,.
- the image decoding apparatus further includes an output buffer for storing a decoded image for a block after compensation processing is performed by a compensation unit, and a decoded image for one block sequentially.
- a decoded image transfer means for reading from the output buffer and writing to the memory, wherein the suppression means suppresses access to the memory by the decoded image transfer means within a cycle time determined to be large by the determination means. It is good to do.
- VLD processing unit that shares the decoding of the compressed image by making the memory access for writing the decoded image into the memory a suppression control target. Between each unit such as motion compensation processing unit Therefore, the memory bus bandwidth can be controlled relatively easily without the need to change the processing time relationship.
- the capacity of the output buffer is a capacity capable of storing decoded images of two blocks, and the decoded image transfer means is within the cycle time next to the cycle time suppressed by the suppression means.
- the decoded image for two blocks in succession is read out and written into the memory, and the decoded image for one block is stored in the output buffer within the other cycle times. Read from and write to the memory.
- the decoded image suppressed in a certain cycle time is written to the memory in the next cycle time, which is a relatively small amount of transfer of the reference image, it is performed in units of frames.
- the compressed image can be decoded with almost no increase in the decoding time of the image.
- the memory includes block header information including block information and block data including information indicating whether or not each block is compressed based on the reference image and a motion vector.
- the image decoding apparatus further includes variable length decoding means for performing variable length decoding when an encoded stream is input, and an encoded stream from the memory.
- variable length decoding means for performing variable length decoding when an encoded stream is input, and an encoded stream from the memory.
- intra processing means for performing decoding processing without referring to a reference image for the block data of the sequentially transmitted block
- the motion vector is obtained from the result of variable length decoding by the variable length decoding means, and the result of variable length decoding by the variable length decoding means is obtained.
- the block data is transmitted to the compensation means for the block compressed based on the reference image, and is compressed based on the reference image for the block!
- the output buffer after the decoding processing by the intra processing means is transmitted to the intra processing means.
- the decoded image may be stored in the block! /.
- the memory includes block header information and a block including information indicating whether or not the power is a block compressed based on the reference image for each block and a motion vector.
- an encoded stream obtained by variable-length encoding data including data and the image decoding device further includes variable-length decoding means for performing variable-length decoding when encoded stream data is input.
- a stream transfer unit that sequentially reads an encoded stream from the memory and inputs the encoded stream to the variable length decoding unit; and an intra processing unit that performs a decoding process that does not refer to a reference image on block data of the sequentially transmitted block
- the obtaining means obtains the motion vector from the result of variable length decoding by the variable length decoding means, and further uses the block data resulting from variable length decoding by the variable length decoding means based on a reference image.
- the compressed block is transmitted to the compensation means, and is compressed based on the reference image.
- the block is previously compressed. Transmitted to the intra processing unit, the restraining means, the per cent, in the cycle time it is determined as large by the determining means Te, it is also possible to prevent access to the memory by the stream transfer unit.
- the image decoding apparatus decodes a compressed image for each block! A block compressed based on a reference image! It is an image decoding device that performs decoding including compensation processing corresponding to motion compensation, and stores a reference image group memory and a reference image for two blocks for storing the reference image. Compensation means that has an input buffer with sufficient capacity and performs compensation processing for one block sequentially with reference to the reference image stored in the input buffer, and for each of the two blocks sequentially within two consecutive cycle times Reference image transfer means for reading all reference images from the memory and writing them in the input buffer as long as there is a corresponding reference image.
- the image decoding apparatus further includes a processing unit that processes the decoded image after the compensation processing by the compensation unit is performed, and one of the processing unit and the compensation unit includes the reference image transfer The processing corresponding to one block for each cycle time is performed in parallel with the means, and the processing corresponding to the block started to be transferred within a certain cycle time by the reference image transfer means Just start running in a later cycle time!
- the image decoding method performs decoding on a block-by-block basis for a compressed image stored in a memory together with a reference image group, and uses a block compressed based on the reference image!
- An image decoding method that performs decoding including compensation processing corresponding to motion compensation, and sequentially reads all reference images from the memory as long as there is a reference image corresponding to one block within each cycle time.
- the access to the memory other than the access for reading the reference image is suppressed within the cycle time determined to be more in the determination step.
- suppression step characterized in that it comprises a compensation step for compensating processing for the plant constant for referring to a reference image stored in the buffer sequentially one block.
- the image decoding method performs decoding on a block-by-block basis for a compressed image stored in a memory together with a reference image group, and uses a block compressed based on the reference image!
- Image transfer that is read from the memory and written to a predetermined buffer.
- the transmission step is executed in parallel with a compensation step for sequentially performing compensation processing for one block with reference to the reference image stored in the predetermined buffer.
- the reference image Since the restriction to complete transfer within 1 cycle time is removed, the memory bus band allows the maximum data transfer amount of the reference image corresponding to 1 macroblock to be transferred within 1 cycle time. The width does not have to be determined, and the requirement for the memory bus bandwidth can be relaxed accordingly.
- the semiconductor integrated circuit decodes a compressed image for each block, and blocks compressed based on the reference image!
- the semiconductor integrated circuit performs decoding including compensation processing corresponding to motion compensation, and includes a memory storing a reference image group and an input buffer for storing the reference image.
- Compensation circuit section that performs compensation processing for one block sequentially with reference to the reference image stored in, and all reference images as long as there is a reference image corresponding to one block sequentially within each cycle time,
- a reference image transfer circuit unit that reads from the memory and writes the input image to the input buffer, and determines whether or not the total amount of reference images that the reference image transfer circuit unit reads from the memory at each cycle time is greater than a predetermined reference amount
- a deterrence circuit unit that deters access.
- the maximum data transfer amount within the cycle time is suppressed in order to suppress other memory accesses for reading the reference image during the cycle time when the amount of reference image read from the memory is large. This makes it possible to reduce the width of the memory bus band.
- the semiconductor integrated circuit decodes a compressed image for each block, and generates a block compressed based on the reference image!
- This is a semiconductor integrated circuit that performs decoding including compensation processing corresponding to motion compensation, and includes a memory storing a reference image group and two blocks of reference images for storing the reference image.
- An input buffer with enough storage capacity And a compensation circuit unit that sequentially performs compensation processing for one block with reference to a reference image stored in the input buffer, and a reference image corresponding to each of the two blocks sequentially within two consecutive cycle times.
- a reference image transfer circuit unit that reads out all reference images from the memory and writes them to the input canoffer as much as possible.
- FIG. 1 is a configuration diagram of an image decoding device 100 according to Embodiment 1 of the present invention.
- FIG. 2 is a diagram illustrating a structure of stream data of a compressed image.
- FIG. 3 is a diagram illustrating a relationship between a macroblock and a reference image.
- FIG. 4 is a flowchart showing control details of decoding by a control unit 110 for a macroblock.
- FIG. 5 is a diagram showing an execution sequence of pipeline control by the control unit 110.
- FIG. 6 is a diagram showing DMA transfer control in the image decoding apparatus 100.
- FIG. 7 is a configuration diagram of an image decoding device 500 according to Embodiment 2 of the present invention.
- FIG. 8 is a flowchart showing control details of decoding of macroblocks by control unit 510.
- FIG. 9 is a diagram showing an execution sequence of pipeline control by control unit 510.
- FIG. 10 is a flowchart showing DMA transfer instruction issue processing by control unit 510.
- FIG. 11 is a diagram showing DMA transfer control in the image decoding apparatus 500.
- FIG. 12 is a diagram showing an image of an image decoding device 100 in which a part other than the memory 120 is realized as the semiconductor integrated circuit 101.
- FIG. 13 is a diagram illustrating an execution sequence of pipeline control in a conventional image decoding apparatus.
- FIG. 1 is a configuration diagram of an image decoding device 100 according to Embodiment 1 of the present invention.
- the image decoding apparatus 100 includes a control unit 110, a memory 120, a DMA controller 130, a variable length decoding (VLD) processing unit 140, an inverse frequency conversion processing unit 150, and a motion compensation processing unit 160. , An intra processing unit 170, a deblock filter processing unit 180, and an output buffer 190.
- the memory 120 is a DRAM for storing an encoded stream in which a compressed image or the like is variable-length encoded, and for storing a reference image.
- image is used as a term indicating a concept including image data that expresses the contents of the displayed image!
- the DMA controller 130 has a queue for receiving an instruction accompanied by memory address designation or the like from the control unit 110, and according to the received instruction, the memory 120 and the VLD processing unit 140, the motion compensation processing unit 160, or the output buffer 190 Responsible for transferring data between the two.
- the VLD processing unit 140 performs variable-length decoding on the encoded stream, and the stream data power of the compressed image It has the function of extracting the type and motion vector and transmitting them to the control unit, and extracting the block data representing the so-called difference value, which is the substance of the macro block and frequency-converted, and transmitting it to the inverse frequency conversion processing unit 150.
- the compressed image stream data includes an I picture, a P picture, a B picture and! /, A slice header 210 indicating the picture type, etc., and a macro block header for each macro block. 220 and block data 230 are configured to be continuous.
- the macro block header 220 includes a macro block type indicating whether an inter macro block compressed in the inter mode based on the reference image, an intra macro block compressed in the intra mode without using the reference image, and a motion. Including vectors.
- the inverse frequency conversion processing unit 150 has a function of outputting a difference value obtained by performing inverse frequency conversion on the block data transmitted from the VLD processing unit 140.
- the motion compensation processing unit 160 has an input buffer 161, calculates the reference image power stored in the input buffer 161, an image with a quarter-pel accuracy, and the macro block output by the inverse frequency conversion processing unit 150. It has a function of executing a compensation process for reconstructing an image by adding the difference value, that is, a process corresponding to a so-called inverse process of motion compensation (here, “motion compensation process” ⁇ ⁇ ).
- the intra processing unit 170 has a function of performing intra processing, that is, a difference value for the macroblock output by the inverse frequency conversion processing unit 150 and the macroblock. It has a function of reconstructing an image by adding the peripheral image.
- the deblocking filter processing unit 180 has a function of performing deblocking filter processing on the reconstructed image output from the motion compensation processing unit 160 or the intra processing unit 170, that is, a deblocking filter for suppressing block noise. It has a function of obtaining a decoded image by sending it to the output buffer 190.
- the output canoffer 190 is a buffer memory having a capacity sufficient to store a decoded image for two macroblocks.
- the control unit 110 includes a processor, a ROM (Read Only Memory), a timer, and the like in terms of hardware, and the processor executes the program stored in the ROM, so that the image decoding device 100
- Each component implements pipeline control in which each process related to decoding in units of macroblocks is performed in parallel at predetermined cycle times.
- the DMA controller 130, the VLD processing unit 140, the reverse frequency Controls the number conversion processing unit 150, the motion compensation processing unit 160, the intra processing unit 170, and the deblock filter processing unit 180, and as a functional component, a stream transfer control unit 111, an acquisition unit 112, and a reference image transfer control Unit 113, decoded image transfer control unit 114, and inhibition determination unit 116.
- the stream transfer control unit 111 has a function of instructing the DMA controller 130 to transfer the encoded stream from the memory 120 to the VLD processing unit 140.
- the acquisition unit 112 acquires a motion vector and a macroblock type from the VLD processing unit 140, transmits them to the reference image transfer control unit 113, counts the number of motion vectors, and transmits the number of the motion vectors to the suppression determination unit 116.
- the reference image transfer control unit 113 identifies the position of each reference image according to each motion vector, specifies the address of each reference image, and sets each cycle time. 1 has a function of instructing the DMA controller 130 to transfer all reference images corresponding to one macroblock from the memory 120 to the input buffer 161 of the motion compensation processing unit 160.
- the decoded image transfer control unit 114 is a non-output information storage unit that stores information indicating whether or not a force is transmitted from the output buffer 190 to the decoded image memory 120 within one cycle time. 115, and has a function of instructing the DMA controller 130 to transfer the decoded image in the output buffer 190 to the memory 120.
- the suppression determination unit 116 includes a threshold storage unit 117 that stores a threshold value related to the number of motion vectors in advance, and compares the number of motion vectors transmitted from the acquisition unit 112 with the threshold value to determine the motion vector. It has a function of determining whether or not the number exceeds the threshold, and inhibiting the decoded image transfer control unit 114 from instructing transfer of the decoded image to the memory 120 according to the determination result.
- the determination of whether or not the number of motion vectors exceeds the threshold in the suppression determination unit 116 is to indirectly determine whether or not the data transfer amount of the reference image corresponding to the motion vector is greater than a predetermined amount. Therefore, when the data transfer amount of the reference image in a certain cycle time is larger than the predetermined amount, the suppression determination unit 116 suppresses the instruction to write the decoded image in the memory 120 at the cycle time.
- FIG. 3 is a diagram showing the relationship between the macroblock and the reference image.
- a plurality of macro blocks formed by dividing the compressed image 300 are composed of a luminance signal and a color difference signal.
- the luminance signal is shown.
- one macro block is 16 pixels ⁇ 16 pixels. This is a signal for pixels.
- the motion vector 301 is related to the macroblock 311 and the reference image.
- the difference of the spatial position from 321 is shown.
- the reference image 320 corresponds to a partial set of reference images such as the reference image 321.
- the threshold value storage unit 117 stores a threshold value S that is the smallest integer that satisfies the following formula 1.
- the function f (x) is a function indicating the total bit amount of the reference image when the number of motion vectors is the power
- V is the upper limit number of motion vectors over two consecutive macroblocks
- C is the total bit amount of decoded image for one macroblock.
- the determination based on the threshold value S makes it possible to determine whether or not the amount of reference image for DMA transfer is greater than a predetermined amount equal to or greater than (V ⁇ S) + C.
- the control unit 110 determines macroblocks to be sequentially decoded from among the macroblocks constituting the compressed image, and controls each component of the image decoding apparatus 100 within each cycle time by pipeline control.
- One of the six macroblocks is processed in order, but here we explain what kind of processing is performed on one macroblock.
- FIG. 4 is a flowchart showing the contents of decoding control performed on the macroblock by the control unit 110.
- the stream transfer control unit 111 of the control unit 110 issues a transfer instruction to the DMA controller 130 to transfer the encoded stream from the memory 120 to the VLD processing unit 140 (step S11).
- the DMA controller 130 that has received the transfer instruction of the encoded stream by the queue transfers the encoded stream for one macroblock.
- the control unit 110 activates the VLD processing unit 140 (step S12), and the VLD processing unit 140 extracts the motion vector and the macroblock type from the encoded stream force and transmits the motion vector and the macroblock type to the acquisition unit 112 and the block. Data is extracted and transmitted to the inverse frequency conversion processing unit 150.
- the control unit 110 activates the inverse frequency conversion processing unit 150 (step S13), and the inverse frequency conversion processing unit 150 performs inverse frequency conversion on the block data and outputs it.
- the control unit 110 determines whether the macroblock type of the macroblock is an inter macroblock or an intra macroblock (step S14). If the macroblock type is an inter macroblock, the acquisition unit 112 of the control unit 110 The motion vector is transmitted to the reference image transfer control unit 113, and the reference image transfer control unit 113 calculates the position of each reference image based on the position of each macroblock and each motion vector. (Step S15), specify that address, and issue an instruction to the DMA controller 130 to transfer the reference image from the memory 120 to the input buffer 161 (step S16). The motion compensation processing unit 160 is activated (step S17), and the deblock filter processing unit 180 is activated (step S18). The DMA controller 130 that has received the transfer instruction of the reference image in step S16 by the queue selects all the reference images corresponding to one macroblock. Transfer to input buffer 161.
- step S 17 the motion compensation processing unit 160 reconstructs the image based on the reference image in the input buffer 161 and the block data subjected to the inverse frequency conversion! sent to the block filtering process unit 180, also fat port Kkufiruta processing unit 180 as a result of step S 18 is to suppress the block noise of the reconstructed image is stored in the output Roh Ffa 1 9 ⁇ .
- step S14 determines whether the macroblock type for the macroblock is an intra macroblock.
- control unit 110 skips steps S15 to S17 and activates intra processing unit 170.
- Step S19 the deblocking filter processing unit 180 is activated (Step S18).
- the intra processing unit 170 reconstructs the image with the block data force subjected to the inverse frequency conversion and sends it to the deblock filter processing unit 180.
- the suppression determination unit 116 of the control unit 110 compares the number of motion vectors acquired by the acquisition unit 112 with the threshold S stored in the threshold storage unit 117, and the number of motion vectors sets the threshold S. Beyond! In the case of uttering (step S20), the decoded image transfer control unit 114 sends the decoded image to the memory 120 without issuing an instruction related to the DMA transfer, and the unoutput information in the non-output information storage unit 115 is Then, a setting is made to indicate that the decoded image has not been sent to the memory (step S21).
- the decoded image transfer control unit 114 refers to the non-output information and stores the decoded image for the previous macroblock in the memory. It is determined whether or not the transmission has been performed (step S22), and if the transmission has not been performed, the decoded image transfer control unit 114 sends the decoded image of a certain macroblock to the decoded image of the macro block.
- step S23 Issuing an instruction to transfer the decoded image of 2 macroblocks from the output buffer 190 to the memory 120 together with the decoded image of the macroblock (step S23), then clear the unoutput information, that is, not output
- the information is updated to indicate that there is no decoded image whose information has not been transmitted (step S24), and when the decoded image for the previous macroblock has not been transmitted, the decoded image transfer control unit 114
- the DMA controller 130 issues an instruction to transfer the decoded image of a macro-block from the output node Ffa 190 in the memory 120 (step S25), and non-output information Clear the information (step S24).
- step S23 or S25 block noise is suppressed by the deblocking filter processing unit 180, and the DMA transfer is performed to the decoded image power memory 120 stored in the output buffer 190.
- the processing for one macro block has been basically explained in order according to FIG. 4.
- the control unit 110 actually performs pipeline control, and in the flowchart of FIG.
- the operation shown in each step is first executed at each cycle time, which is the periodic time of the knock line control.
- the macroblocks to be processed are not the same repulsive force Step S11, Step S12, Step S13, Step S15 to S16, Step S17 or Step S19, Step S18, and Step S20 to S25 Each of these can be executed in any order as long as they are executed at the beginning of each cycle time.
- each unit by the control unit 110 has a significance of transmitting the start of each cycle, and each unit corresponds to a basic block corresponding to one macro block within each cycle time. Do it.
- FIG. 5 is a diagram illustrating an execution sequence of pipeline control by the control unit 110.
- time slot TS n (where n is an arbitrary integer) means the nth cycle time.
- MB #n indicates that the nth macroblock is a processing target.
- the stream transfer processing performed by the DMA controller 130 according to the instruction of the stream transfer control unit 111 and the VLD processing unit 140 are performed.
- Variable length decoding processing, inverse frequency conversion processing performed by the inverse frequency conversion processing unit 150, reference image transfer processing performed by the DMA controller 130 in accordance with instructions from the reference image transfer control unit 113, and a motion compensation processing unit The motion compensation processing performed by 160 or the intra processing performed by the intra processing unit 170, the deblocking filter processing performed by the deblocking filter processing unit 180, and the DMA controller 130 according to the instruction of the decoded image transfer control unit 114.
- the decoded image transfer process to be performed is performed in parallel.
- FIG. 5 illustrates the decoded image transfer process.
- the number of motion vectors exceeds the threshold S
- An example is shown assuming that the number has been exceeded.
- FIG. 6 is a diagram showing DMA transfer control in the image decoding apparatus 100.
- the DMA controller 130 performs a stream transfer process in which a variable-length encoded stream of one macroblock is DMA-transferred from the memory 120 to the VLD processing unit 140, and a reference image corresponding to one macroblock is motion-compensated from the memory 120.
- Each cycle time includes a reference image transfer process for DMA transfer to the input buffer 1 61 of the processing unit 160 and a decoded image transfer process for DMA transfer of the decoded image corresponding to one macroblock stored in the output buffer 190 to the memory 120. Execute in a time-sharing manner.
- symbol A indicates the transfer time of the encoded stream corresponding to one macroblock that is the target of the stream transfer process
- symbol B indicates the target of the reference image transfer process.
- An example of the transfer time of various amounts of reference images corresponding to one macroblock is shown.
- Symbol C indicates the transfer time of the decoded image corresponding to one macroblock that is the target of the decoded image transfer process. ing.
- the example in FIG. 6 corresponds to the example in FIG.
- the amount of the reference image transferred by DMA is large!
- the image decoding apparatus 100 according to Embodiment 1 described above has a configuration that performs DMA transfer of all reference images corresponding to one macro block within each cycle time.
- the image decoding apparatus 500 according to Embodiment 2 also allows the DMA transfer of all reference images corresponding to one macroblock to be completed within one cycle time, and each successive two macroblocks are allowed to complete. It has a configuration that controls so that DMA transfer of all corresponding reference images is completed within 2 cycle times.
- FIG. 7 is a configuration diagram of an image decoding device 500 according to Embodiment 2 of the present invention.
- the image decoding apparatus 500 includes a control unit 510, a memory 120, a DMA controller 130, a variable length decoding (VLD) processing unit 140, an inverse frequency conversion processing unit 150, a buffer 551, and motion compensation.
- a processing unit 560, an intra processing unit 170, a deblock filter processing unit 180, and an output buffer 590 are provided.
- the same constituent elements as those of the image decoding apparatus 100 shown in Embodiment 1 are denoted by the same reference numerals as those in FIG. A detailed explanation of is omitted.
- the motion compensation processing unit 560 includes an input buffer 561 having a capacity sufficient to store all reference images corresponding to two consecutive macroblocks, and the reference image data stored in the input buffer 561 is stored in the input buffer 561. In addition, it has a function of executing motion compensation processing for reconstructing an image by adding the image calculated by calculating a quarter-pel accuracy image and the difference value for the macroblock output by the inverse frequency conversion processing unit 150. .
- the output canoffer 590 is a buffer memory having a capacity sufficient to store a decoded image for one macroblock.
- control unit 510 includes a processor, a ROM (Read Only Memory), a timer, and the like in terms of hardware.
- the control unit 510 includes Each component implements pipeline control in which each process related to decoding in units of macroblocks is performed in parallel at predetermined cycle times.
- the DMA controller 130, the VLD processing unit 140, the reverse frequency The number conversion processing unit 150, the motion compensation processing unit 560, the intra processing unit 170, and the deblock filter processing unit 180 are controlled, and the stream transfer control unit 111, the acquisition unit 112, and the reference image transfer control are provided as functional components.
- the reference image transfer control unit 513 specifies the position of each reference image according to each motion vector, and each reference image And a function for instructing the DMA controller 130 to transfer each reference image from the memory 120 to the input buffer 561 of the motion compensation processing unit 560.
- This instruction is given at each cycle time start point.
- the DMA transfer corresponding to the instruction stored in the queue of the MA controller 130 does not necessarily start at the start of the cycle time. However, the DMA transfer corresponding to the DMA transfer must be completed within two cycle times after the instruction to transfer the reference image.
- the decoded image transfer control unit 514 has a function of instructing the DMA controller 130 to transfer the decoded image for one macroblock in the output buffer 590 to the memory 120 within one cycle time.
- the noffer 551 is a buffer memory for temporarily storing the data output from the inverse frequency conversion processing unit 150.
- the macro block processed at the cycle time in which the inverse frequency conversion processing unit 150 is processed is the next one. It is also provided for processing by the motion compensation processing unit 560 or the intra processing unit 170 in the next cycle time.
- Control unit 510 determines macroblocks to be sequentially decoded from among the macroblocks constituting the compressed image, and sequentially controls each of the constituent elements of image decoding apparatus 500 within each cycle time by pipeline control. Or, the power to process any of the seven macroblocks Here, we first explain what processing is performed on one macroblock.
- FIG. 8 is a flowchart showing the control contents of decoding of the macroblock by the control unit 510.
- the stream transfer control unit 111 of the control unit 510 issues a transfer instruction to the DMA controller 130 to transfer the encoded stream from the memory 120 to the VLD processing unit 140 (step S51).
- the DMA controller 130 that has received the transfer instruction of the encoded stream by the queue transfers the encoded stream for one macroblock.
- the control unit 510 activates the VLD processing unit 140 (step S52), and the VLD processing unit 140 extracts the motion vector and the macroblock type from the encoded stream force and acquires the 1 12 and the block data is extracted and transmitted to the inverse frequency conversion processing unit 150.
- Control unit 510 activates inverse frequency conversion processing unit 150 (step S53), and inverse frequency conversion processing unit 150 performs inverse frequency conversion on the block data and outputs the block data.
- Control unit 510 determines whether the macro block type for the macro block is an inter macro block or an intra macro block (step S54), and if it is an inter macro block, acquisition unit 112 of control unit 510 operates.
- the reference image transfer control unit 513 transmits the vector to the reference image transfer control unit 513, and the reference image transfer control unit 513 calculates the position of each reference image based on the position of each macroblock and each motion vector.
- Step S55 specify the address, issue an instruction to the DMA controller 130 to transfer the reference image from the memory 120 to the input buffer 561 (step S56), and the motion compensation processing unit 560 is started (step S57), and the deblock filter processing unit 180 is started (step S58).
- the DMA controller 130 that has received the reference image transfer instruction in step S56 by the queue transfers all the reference images corresponding to one macroblock to the input buffer 561.
- step S57 the motion compensation processing unit 560 reconstructs the image based on the reference image in the input buffer 561 and the block data subjected to inverse frequency conversion and deblocks it! Then, as a result of step S58, the fat block filter processing unit 180 suppresses block noise of the reconstructed image and stores it in the output buffer 590.
- step S54 if it is determined in step S54 that the macroblock type for the macroblock is an intra macroblock, control unit 510 skips steps S55 to S57 and activates intra processing unit 170. (Step S59), the deblocking filter processing unit 180 is activated (Step S58). As a result of this step S59, the intra processing unit 170 also reconstructs the image with the block data force subjected to inverse frequency conversion, and sends it to the deblock filter processing unit 180.
- the decoded image transfer control unit 514 of the control unit 510 stores the decoded image of the macroblock stored in the output buffer 590 as a result after step S58 is performed in the memory 120.
- An instruction is issued to the DMA controller 130 to transfer to (step S60).
- the processing for one macro block has been basically explained in order according to FIG. 8, but the control unit 510 is actually performing pipeline control, and the processing in the flowchart of FIG. The operation shown in each step is first executed at each cycle time, which is the periodic time of the knock line control.
- Step S51, Step S52, Step S53, Step S55 to S56, Step S57 or Step S59, Step S58, and Step S60 Is executed at the beginning of each site time. At this time, the execution order of steps S51, S55 to S56, and step S60 is determined as described later, but the other steps may be executed in any order. What! /
- each unit by the control unit 110 has a significance of transmitting the start of each cycle, and each unit corresponds to a basic block corresponding to one macro block within each cycle time. Do it.
- DMA transfer for all reference images corresponding to one macroblock by the DMA controller 130 does not have to be completed within one cycle time, but for all reference images corresponding to two consecutive macroblocks. DMA transfer power can be completed within one cycle time.
- FIG. 9 is a diagram showing an execution sequence of pipeline control by the control unit 510.
- time slot TS n (where n is an arbitrary integer) means the nth cycle time.
- MB #n indicates that the nth macroblock is a processing target.
- stream transfer processing performed by the DMA controller 130 in accordance with instructions from the stream transfer control unit 111 and variable length decoding performed by the VLD processing unit 140 Processing, reverse frequency conversion processing executed by the reverse frequency conversion processing unit 150, reference image transfer processing executed by the DMA controller 130 in accordance with instructions from the reference image transfer control unit 513, and executed by the motion compensation processing unit 160.
- the DMA controller 130 Performed by the DMA controller 130 according to the instruction of the decoded image transfer control unit 514, the intra processing performed by the motion compensation processing or intra processing unit 170 to be performed, the deblocking filter processing performed by the deblocking filter processing unit 180, and The decoded image transfer process is performed in parallel.
- FIG. 9 merely shows that the period during which the DMA transfer of the reference image corresponding to each macroblock can be performed is two cycle times, and is based on the MPEG4AVC standard.
- the upper limit of the total number of motion vectors for two consecutive macroblocks is fixedly determined by the definition of the level corresponding to a large image, so when decoding such a large image, it is actually Above, for example, if the transfer of the reference image corresponding to the nth macroblock and the transfer of the reference image corresponding to the n + 1st macroblock are viewed together, both will be completed within 2 cycle times. .
- FIG. 10 is a flowchart showing DMA transfer instruction issue processing by control unit 510.
- the stream transfer control unit 111 issues a DMA transfer instruction for the encoded stream corresponding to the nth macroblock (step S71), and the decoded image transfer control unit 514 converts the n-6th macroblock.
- the DMA transfer instruction for the corresponding decoded image is issued (step S72), and finally the DMA transfer instruction for all the reference images corresponding to the n-2nd macroblock is issued (step S72). S73).
- FIG. 11 is a diagram showing DMA transfer control in the image decoding apparatus 500.
- the DMA controller 130 performs a stream transfer process in which a variable-length encoded stream of one macroblock is DMA-transferred from the memory 120 to the VLD processing unit 140, and a reference image corresponding to one macroblock is motion-compensated from the memory 120.
- Each cycle time includes a reference image transfer process for DMA transfer to the input buffer 5 61 of the processing unit 560 and a decoded image transfer process for DMA transfer of the decoded image corresponding to one macroblock stored in the output buffer 590 to the memory 120. Execute in a time-sharing manner.
- symbol A indicates the transfer time of the encoded stream corresponding to one macroblock to be subject to stream transfer processing
- symbol B represents one macroblock to be subject to reference image transfer processing. Examples of transfer times for various amounts of reference images corresponding to are shown
- symbol C shows the transfer time of a decoded image corresponding to one macroblock to be subjected to the decoded image transfer process.
- DMA transfer instructions are issued in the order according to FIG. That is, as soon as the DMA transfer is completed, the DMA controller 130 takes out DMA transfer instructions from the queue in the order in which they are instructed, and performs the next DMA transfer.
- the upper limit of the total number of motion vectors for two consecutive macroblocks in a large image is fixedly set to 16 etc. due to the level definition corresponding to a large image in the MPEG4AVC standard.
- the image decoding device has been described based on the first and second embodiments.
- the image decoding device can be modified as follows, and the present invention is not limited to the image decoding device shown in the above-described embodiment. Of course.
- the motion compensation processing unit includes the input buffer.
- a buffer memory may be provided outside the motion compensation processing unit.
- the image decoding device is configured.
- Each unit formed may include an input buffer and an output buffer.
- Each unit may not necessarily be an independent individual, but a plurality of units may be integrally molded.
- the motion compensation processing unit executes the macroblock at the cycle time after DMA transfer from the memory of all the reference images corresponding to a macroblock to the input buffer is completed by the reference image transfer processing. However, as soon as the reference image is stored in the input buffer, the motion compensation processing unit may start the motion compensation processing for the macroblock even before the DMA transfer is completed. In some cases, for example, in the cycle time after the completion of the DMA transfer, the deblock filter processing unit may perform the deblock filter processing in response to the result of the motion compensation processing. That is, in the second embodiment, when the motion compensation processing unit starts the motion compensation process as soon as the reference image is acquired, the DMA transfer of the reference image corresponding to the macroblock within a certain cycle time is performed.
- the image decoding apparatus shown in the second embodiment may be modified so that the deblocking filter processing corresponding to the macroblock starts executing within the cycle time two times after the cycle time when started. ,
- Embodiment 1 when a DMA transfer from a memory of a reference image exceeding a predetermined amount within one cycle time is performed, DMA transfer to the memory of the decoded image within that cycle time is suppressed. It is sufficient to suppress any memory access other than DMA transfer of the reference image within the cycle time when the DMA transfer of the reference image exceeding the predetermined amount is performed. For example, DMA transfer from the memory of the encoded stream is performed. It may be deterred
- the VLD processing unit may be provided with an input buffer that can store an extra 1 macroblock.
- the threshold S shown in the first embodiment is a force determined to be the smallest integer satisfying Equation 1.
- the threshold S is not necessarily the smallest integer. However, the smallest integer can minimize the memory bus bandwidth.
- the configuration in which it is determined whether to temporarily suppress DMA transfer of the decoded image according to the result of comparing the threshold value S with the number of motion vectors is shown.
- a configuration may be adopted in which it is determined whether or not the DMA transfer of the decoded image is temporarily suppressed according to a result of comparing the data amount of the reference image to be transferred with a predetermined amount.
- Embodiments 1 and 2 an image decoding apparatus that performs decoding in units of macroblocks has been shown.
- the present invention which may perform decoding in units of further subdividing macroblocks
- the present invention can be applied when decoding is performed by dividing into blocks of an arbitrary size.
- Nipline control (see Fig. 9) as shown in the second embodiment is performed not on a macroblock basis but on a block basis so that the block unit is basically processed in one cycle time. May be.
- the block unit is, for example, a size of 4 ⁇ 4 pixels, which is 1Z16 of a macro block, or a size of 8 ⁇ 16 pixels, which is 1Z2. Even when pipeline control is performed in units of blocks, the restriction that the transfer for one block is executed in one cycle time is removed for the pipeline stage of the reference image transfer process.
- Each of the image decoding devices 100 and 500 shown in the first and second embodiments may be configured as a semiconductor integrated circuit on one semiconductor chip, and the semiconductor integrated circuit is integrated into one package.
- the memory 120 may exist outside the package.
- FIG. 12 is a diagram showing an image of the image decoding device 100 in which the memory integrated circuit 101 other than the memory 120 is realized.
- the image decoding apparatus can be used for image reproduction apparatuses such as DVD players and digital televisions that reproduce moving pictures compressed in accordance with the MPEG4 AVC standard.
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Computing Systems (AREA)
- Theoretical Computer Science (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
Description
Claims
Priority Applications (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US11/632,548 US8428126B2 (en) | 2004-08-04 | 2005-07-04 | Image decoding device with parallel processors |
| JP2006531327A JP4668914B2 (ja) | 2004-08-04 | 2005-07-04 | 画像復号装置 |
| EP05765215.8A EP1775961B1 (en) | 2004-08-04 | 2005-07-04 | Video decoding device and method for motion compensation with sequential transfer of reference pictures |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2004228433 | 2004-08-04 | ||
| JP2004-228433 | 2004-08-04 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2006013690A1 true WO2006013690A1 (ja) | 2006-02-09 |
Family
ID=35786985
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/JP2005/012310 Ceased WO2006013690A1 (ja) | 2004-08-04 | 2005-07-04 | 画像復号装置 |
Country Status (5)
| Country | Link |
|---|---|
| US (1) | US8428126B2 (ja) |
| EP (1) | EP1775961B1 (ja) |
| JP (1) | JP4668914B2 (ja) |
| CN (1) | CN100534193C (ja) |
| WO (1) | WO2006013690A1 (ja) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2007318517A (ja) * | 2006-05-26 | 2007-12-06 | Sony Corp | 画像データの処理方法、画像データの処理方法のプログラム、画像データの処理方法のプログラムを記録した記録媒体及び画像データの処理装置 |
| WO2012017858A1 (ja) * | 2010-08-03 | 2012-02-09 | ソニー株式会社 | 画像処理装置と画像処理方法 |
| JP2012195703A (ja) * | 2011-03-15 | 2012-10-11 | Fujitsu Ltd | トランスコード装置及びトランスコード方法 |
| JP2014078891A (ja) * | 2012-10-11 | 2014-05-01 | Canon Inc | 画像処理装置、画像処理方法 |
Families Citing this family (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE602006018868D1 (de) * | 2005-01-28 | 2011-01-27 | Thomson Licensing | Verfahren und vorrichtung zur framekodierung in echtzeit |
| EP2026585A4 (en) | 2006-05-24 | 2016-08-31 | Panasonic Ip Man Co Ltd | IMAGE ENCODING DEVICE AND METHOD, AND IMAGE ENCODING INTEGRATED CIRCUIT |
| JP2008072647A (ja) * | 2006-09-15 | 2008-03-27 | Toshiba Corp | 情報処理装置、デコーダおよび再生装置の動作制御方法 |
| CN101543038B (zh) * | 2007-05-17 | 2013-01-02 | 索尼株式会社 | 信息处理装置和方法 |
| KR20090011223A (ko) * | 2007-07-25 | 2009-02-02 | 삼성전자주식회사 | 방송처리장치 및 그 제어방법 |
| KR101439848B1 (ko) * | 2008-01-08 | 2014-09-17 | 삼성전자주식회사 | 움직임 보상 방법 및 장치 |
| JP5275454B2 (ja) * | 2009-03-31 | 2013-08-28 | パナソニック株式会社 | 画像復号装置 |
| US8186876B2 (en) * | 2009-04-20 | 2012-05-29 | Welch Allyn, Inc. | Calibrated assembly for IR thermometer apparatus |
| US8136985B2 (en) * | 2009-05-05 | 2012-03-20 | Welch Allyn, Inc. | IR thermometer thermal isolation tip assembly |
| US9877033B2 (en) * | 2009-12-21 | 2018-01-23 | Qualcomm Incorporated | Temporal and spatial video block reordering in a decoder to improve cache hits |
| WO2011161949A1 (ja) * | 2010-06-23 | 2011-12-29 | パナソニック株式会社 | 画像復号装置、画像復号方法、集積回路およびプログラム |
| CN104170386A (zh) * | 2012-03-16 | 2014-11-26 | 松下电器产业株式会社 | 图像解码装置以及图像解码方法 |
| CN103402086B (zh) * | 2013-07-22 | 2017-02-15 | 华为技术有限公司 | 一种用于视频编码系统的性能控制方法及编码器 |
| US10396827B2 (en) * | 2015-09-25 | 2019-08-27 | SK Hynix Inc. | Data storage device |
| US11177835B2 (en) | 2015-09-25 | 2021-11-16 | SK Hynix Inc. | Data storage device |
| US11611359B2 (en) | 2015-05-29 | 2023-03-21 | SK Hynix Inc. | Data storage device |
| US11515897B2 (en) | 2015-05-29 | 2022-11-29 | SK Hynix Inc. | Data storage device |
| KR102285940B1 (ko) | 2015-05-29 | 2021-08-05 | 에스케이하이닉스 주식회사 | 데이터 처리 회로, 데이터 처리 회로를 포함하는 데이터 저장 장치 및 그것의 동작 방법 |
| US11087428B2 (en) * | 2018-03-23 | 2021-08-10 | Canon Kabushiki Kaisha | Image processing apparatus, data processing apparatus, and image processing method |
| US11558637B1 (en) * | 2019-12-16 | 2023-01-17 | Meta Platforms, Inc. | Unified search window to support multiple video encoding standards |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH07240844A (ja) * | 1993-03-19 | 1995-09-12 | Mitsubishi Electric Corp | 画像データ処理装置および画像データ処理方法 |
| JPH07336689A (ja) * | 1994-06-07 | 1995-12-22 | Hitachi Ltd | 動き補償回路 |
| JPH0937249A (ja) * | 1995-07-20 | 1997-02-07 | Hitachi Ltd | 符号化映像信号の復号化処理方法及びそれを用いた復号化装置 |
| JPH10327412A (ja) * | 1997-03-26 | 1998-12-08 | Matsushita Electric Ind Co Ltd | 画像復号化装置 |
| JPH10341422A (ja) * | 1997-04-07 | 1998-12-22 | Matsushita Electric Ind Co Ltd | 映像音声処理装置 |
| JPH11252549A (ja) * | 1998-02-27 | 1999-09-17 | Toshiba Corp | 画像符号化/復号化装置 |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5293486A (en) * | 1991-06-28 | 1994-03-08 | Digital Equipment Corporation | Deterministic method for allocation of a shared resource |
| JP3605826B2 (ja) * | 1993-06-10 | 2004-12-22 | ソニー株式会社 | ビデオおよびオーディオ信号処理システムにおける補助情報のための合理的入力バッファの回路配列 |
| JP3824678B2 (ja) * | 1995-05-09 | 2006-09-20 | 株式会社ルネサステクノロジ | 画像復号表示装置 |
| DE69738878D1 (de) * | 1996-04-12 | 2008-09-11 | Sony Corp | Bildkodierung mit Pufferunterlaufvermeidung |
| US6212236B1 (en) * | 1997-03-26 | 2001-04-03 | Matsushita Electric Industrial Co., Ltd. | Image decoding apparatus |
| CN1156171C (zh) * | 1997-04-07 | 2004-06-30 | 松下电器产业株式会社 | 提高处理效率的图象声音处理装置 |
-
2005
- 2005-07-04 WO PCT/JP2005/012310 patent/WO2006013690A1/ja not_active Ceased
- 2005-07-04 JP JP2006531327A patent/JP4668914B2/ja not_active Expired - Lifetime
- 2005-07-04 US US11/632,548 patent/US8428126B2/en active Active
- 2005-07-04 CN CNB2005800262624A patent/CN100534193C/zh not_active Expired - Lifetime
- 2005-07-04 EP EP05765215.8A patent/EP1775961B1/en not_active Expired - Lifetime
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH07240844A (ja) * | 1993-03-19 | 1995-09-12 | Mitsubishi Electric Corp | 画像データ処理装置および画像データ処理方法 |
| JPH07336689A (ja) * | 1994-06-07 | 1995-12-22 | Hitachi Ltd | 動き補償回路 |
| JPH0937249A (ja) * | 1995-07-20 | 1997-02-07 | Hitachi Ltd | 符号化映像信号の復号化処理方法及びそれを用いた復号化装置 |
| JPH10327412A (ja) * | 1997-03-26 | 1998-12-08 | Matsushita Electric Ind Co Ltd | 画像復号化装置 |
| JPH10341422A (ja) * | 1997-04-07 | 1998-12-22 | Matsushita Electric Ind Co Ltd | 映像音声処理装置 |
| JPH11252549A (ja) * | 1998-02-27 | 1999-09-17 | Toshiba Corp | 画像符号化/復号化装置 |
Non-Patent Citations (1)
| Title |
|---|
| NIEN-TSU WANG: "processing and storage models for MPEG-2 main level and high level video coding - A block-level pipeline approach", PHD THESIS, pages 40 - 118 |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2007318517A (ja) * | 2006-05-26 | 2007-12-06 | Sony Corp | 画像データの処理方法、画像データの処理方法のプログラム、画像データの処理方法のプログラムを記録した記録媒体及び画像データの処理装置 |
| WO2012017858A1 (ja) * | 2010-08-03 | 2012-02-09 | ソニー株式会社 | 画像処理装置と画像処理方法 |
| JP2012195703A (ja) * | 2011-03-15 | 2012-10-11 | Fujitsu Ltd | トランスコード装置及びトランスコード方法 |
| JP2014078891A (ja) * | 2012-10-11 | 2014-05-01 | Canon Inc | 画像処理装置、画像処理方法 |
Also Published As
| Publication number | Publication date |
|---|---|
| US20070206675A1 (en) | 2007-09-06 |
| EP1775961B1 (en) | 2017-06-14 |
| CN100534193C (zh) | 2009-08-26 |
| CN1993992A (zh) | 2007-07-04 |
| US8428126B2 (en) | 2013-04-23 |
| EP1775961A4 (en) | 2012-02-08 |
| EP1775961A1 (en) | 2007-04-18 |
| JPWO2006013690A1 (ja) | 2008-05-01 |
| JP4668914B2 (ja) | 2011-04-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2006013690A1 (ja) | 画像復号装置 | |
| JP2801560B2 (ja) | 動き画像符号化のための改善された低遅延符号化方法 | |
| JP2008529412A (ja) | 中間ループフィルタデータを格納するスクラッチパッド | |
| JP2011097617A (ja) | デジタルビデオ符号化処理における可変精度ピクチャ間タイミング指定方法及び装置 | |
| CN101166277B (zh) | 在用于处理运动图像的设备中访问存储器的方法 | |
| US9509940B2 (en) | Image output device, image output method, and recording medium | |
| US9386310B2 (en) | Image reproducing method, image reproducing device, image reproducing program, imaging system, and reproducing system | |
| JP2012085001A5 (ja) | ||
| JPH11168729A (ja) | 画像復号化装置及び画像復号化方法 | |
| JP2002112268A (ja) | 圧縮画像データ復号装置 | |
| US20110122952A1 (en) | Motion estimation device | |
| JP5115549B2 (ja) | デコード方法、デコーダ及びデコード装置 | |
| KR102171119B1 (ko) | 복수개의 블록 기반의 파이프라인을 이용한 데이터 처리 속도 개선 장치 및 그 동작 방법 | |
| JP2007318615A (ja) | 動画再生装置、動画再生方法およびプログラム | |
| JP4894793B2 (ja) | デコード方法、デコーダ及びデコード装置 | |
| WO2009085788A1 (en) | System, method and device for processing macroblock video data | |
| US20050031042A1 (en) | Method and system for providing a high speed multi-stream MPEG processor | |
| JP2000324484A (ja) | 画像データ処理装置 | |
| KR20170052143A (ko) | 동영상 디코더에 적용되는 메모리 기반의 루프 필터 | |
| JP2014075708A (ja) | 動画像符号化装置及び動画像符号化方法 | |
| JP2012019462A (ja) | 画像復号装置および画像復号方法 | |
| JP2011077609A (ja) | 再生装置および再生方法 | |
| JP2007288443A (ja) | 画像データ復号装置 | |
| WO2010150465A1 (ja) | AV(Audio Visual)データ再生回路、AVデータ再生装置、集積回路およびAVデータ再生方法 | |
| JP2004236321A (ja) | ビデオデータ圧縮のための装置及び方法 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| AK | Designated states |
Kind code of ref document: A1 Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BW BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE EG ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KM KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NA NG NI NO NZ OM PG PH PL PT RO RU SC SD SE SG SK SL SM SY TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW |
|
| AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): BW GH GM KE LS MW MZ NA SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LT LU LV MC NL PL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG |
|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
| REEP | Request for entry into the european phase |
Ref document number: 2005765215 Country of ref document: EP |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 2005765215 Country of ref document: EP |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 11632548 Country of ref document: US Ref document number: 2007206675 Country of ref document: US |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 2006531327 Country of ref document: JP |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 200580026262.4 Country of ref document: CN |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| WWP | Wipo information: published in national office |
Ref document number: 2005765215 Country of ref document: EP |
|
| WWP | Wipo information: published in national office |
Ref document number: 11632548 Country of ref document: US |