JP2011109390A - Device and method for encoding image, and device and method for decoding image - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a means to perform encoding/decoding processing of a high-definition video without needing a high-speed arithmetic processor by subjecting one frame to parallel processing without degrading encoding efficiency. <P>SOLUTION: The device for encoding an image includes: an encoding object signal storage part 1101; a block division part 1102 for dividing an encoding object signal into encoding unit blocks; an orthogonal conversion/quantization part 1103 for subjecting signals of the encoding unit blocks to orthogonal conversion to be quantized; a coefficient group storage part 1104 for storing quantization orthogonal conversion coefficients of one or more encoding unit blocks; a division position determination part 1105 for determining division positions for dividing the quantization orthogonal conversion coefficients to be subjected to entropy encoding from the stored quantization orthogonal conversion coefficients of the one or more encoding unit blocks; a coefficient division part 1106 for dividing the quantization orthogonal conversion coefficients of the encoding unit blocks into a plurality of regions at the division positions; and a means to subject the plurality of divided quantization orthogonal conversion coefficients to entropy encoding independently of one another. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to entropy coding of moving images, particularly arithmetic coding.

  With the widespread use of high-efficiency video coding represented by MPEG-4 AVC, it has become easier to enjoy high-definition HD (High Definition) video content. High definition video is further advanced, and HD video can be enjoyed with a small device such as a portable terminal. However, encoding and decoding of high-definition video requires a large amount of computation, and in order to reproduce without delay by a single arithmetic processing device, it is necessary to use a high-speed arithmetic processing device. High-speed arithmetic processing devices are expensive and there are obstacles to their introduction.

  For this, there is a measure of processing a high-definition video by operating a plurality of relatively low-speed arithmetic processing devices in parallel. MPEG-4 AVC has a mechanism for dividing one frame into a plurality of units called slices. Each slice has no dependency and cannot be referenced between slices. Therefore, parallel processing can be performed for each slice, and a high-definition video processing apparatus can be constructed at low cost.

  Of the tremendous amount of computation in MPEG-4 AVC, the ratio of arithmetic adaptation of context adaptive type called CABAC is large. CABAC holds a plurality of probability models of syntax elements called contexts for each syntax element. Each syntax element refers to a peripheral block to be processed among a plurality of held contexts, and determines one context corresponding to the coding status. The occurrence probability of each value that the target syntax element can take is calculated from the determined context. Then, the referenced context is updated based on the value taken now.

  Even in the context, reference between slices is not possible, and initialization must be performed at the head of each slice. In context initialization, one of three types of predetermined initial models is selected. For this reason, in the vicinity of the head of the slice, the optimal probability model for the video to be encoded is not used, but the state is shifted to the optimal probability model by proceeding with encoding.

JP 2008-11204 A

  As described above, in a multi-slice structure in which one frame is composed of a plurality of slices, there is no dependency between each slice. As a result, independent processing of each slice is possible, but reference between slices cannot be performed, so that intra-screen prediction efficiency decreases. Also in CABAC, dividing one frame into a plurality of slices leads to an increase in context initialization at the head of the slice, and the situation in which an optimal probability model can be used is greatly limited.

  Since reference between slices cannot be performed and initialization is performed at a slice boundary, processing using a true probability model of a probability model syntax element cannot be performed near the head of a slice.

  Due to the above two points, the encoding efficiency of the multi-slice structure is greatly reduced compared to the single-slice structure in which one frame is expressed by a single slice.

Patent Document 1 discloses a technique for solving a decrease in efficiency of a multi-slice structure by referring to a frame positioned immediately before a processing target frame. According to the technique of Patent Document 1,
-For blocks located at the slice boundary, improve the prediction efficiency by referring to the adjacent block of the previous frame.-The context at the beginning of the slice is initialized by taking over the final value of the previous slice of the previous frame. By avoiding this, it is possible to avoid a decrease in encoding efficiency even when a multi-slice structure is adopted.

However, in the technique of Patent Document 1,
・ Since the video is accompanied by motion, it cannot be said that information useful for encoding the block to be processed can be obtained by referring to the adjacent block of the previous frame. ・ Processing must be completed within the frame. The information of the immediately preceding frame cannot be used with a certain I frame. There is a problem that only the immediately preceding frame can be referred to in time, and a frame at a distant position cannot be referenced. It is difficult to perform parallel processing without reducing the coding efficiency with the above configuration.

  The present invention has been made in view of the above problems, and performs high-definition video encoding / decoding processing without requiring a high-speed arithmetic processing device by processing one frame in parallel without reducing the encoding efficiency. It provides a means for performing.

  An image encoding apparatus according to the present invention includes: means for storing a signal to be encoded; means for dividing the signal to be encoded into encoding unit blocks; and orthogonally transform and quantize the signal of the encoding unit block. Means for storing the quantized orthogonal transform coefficients of the one or more coding unit blocks, and the quantized orthogonal transform coefficients from the stored quantized orthogonal transform coefficients of the one or more coding unit blocks. Means for determining a division position for dividing and entropy coding; means for dividing the quantized orthogonal transform coefficient of the coding unit block into a plurality of regions at the division position; and a plurality of divided quantized orthogonal transform coefficients And independently entropy-encoding means.

  Also, the image decoding apparatus of the present invention stores a decoding target stream in which one block of orthogonal transformation coefficients is divided and multiplexed, and orthogonal transformation coefficients composed of two or more decoding target streams. A stream, a multiplexing / separating unit that separates the divided position stream, a divided position decoding unit that decodes the divided position stream and creates a divided position of an orthogonal transform coefficient of one block, and an orthogonal configuration including the two or more Quantized orthogonal transform coefficient entropy decoding means for independently decoding transform coefficient streams, and two or more quantized orthogonal transform coefficients subjected to entropy decoding are governed based on the division position, and one block of quantized orthogonal transform Means for creating transform coefficients, and an inverse quantization / inverse transform unit for inversely quantizing the governed quantized orthogonal transform coefficients to create a decoded signal. To.

  The image coding method of the present invention includes a step of storing a signal to be encoded, a step of dividing the signal to be encoded into encoding unit blocks, and orthogonally transforming the signal of the encoding unit block. A step of quantizing, a step of storing quantized orthogonal transform coefficients of one or more coding unit blocks, and a quantized orthogonal transform coefficient from the stored quantized orthogonal transform coefficients of the one or more coding unit blocks A division position for dividing and entropy encoding, a step of dividing the quantized orthogonal transform coefficient of the coding unit block into a plurality of regions at the division position, and a plurality of divided quantized orthogonal transform coefficients Independently entropy-encoding.

  The image decoding method of the present invention also includes a step of storing a decoding target stream in which one block of orthogonal transform coefficients is divided and multiplexed, and orthogonal transform coefficients configured by two or more decoding target streams. A stream, a demultiplexing step for separating the divided position stream, a step of decoding the divided position stream to create a divided position of one block of orthogonal transform coefficients, and an orthogonal transform coefficient stream composed of the two or more Quantized orthogonal transform coefficient entropy decoding step for decoding each independently, and entropy-decoded two or more quantized orthogonal transform coefficients are governed based on the division position to create one block of quantized orthogonal transform coefficients And a step of inversely quantizing the governed quantized orthogonal transform coefficient and performing inverse transform to generate a decoded signal.

  In addition, the image encoding apparatus of the present invention includes a block dividing unit that divides an encoding target image into encoding units, a reference frame buffer for performing inter-screen prediction, and an intra-screen reference code for performing intra-screen prediction. Encoding information created by means for storing encoding information, means for irreversibly predictive encoding a block to be encoded by the reference frame buffer and the intra-screen reference encoding information, and means for predictively encoding, Means for dividing each syntax element; lossless encoding means for individually performing lossless arithmetic coding on each syntax element divided by the dividing means for each syntax element; and each losslessly encoded by the lossless encoding means Multiplexing means for multiplexing lossless encoding information of syntax elements.

  In addition, the image coding method of the present invention includes a block dividing step for dividing an encoding target image into coding units, a step for storing intra-screen reference coding information for intra-screen prediction, and inter-screen prediction. A step of irreversibly predictive encoding a block to be encoded using a reference frame buffer and intra-screen reference encoding information, and a step of dividing the encoding information created in the predictive encoding step for each syntax element A lossless encoding step for individually lossless arithmetic encoding each syntax element divided in the division step for each syntax element, and lossless encoding information of each syntax element losslessly encoded in the lossless encoding step. And a multiplexing step for multiplexing.

  Also, the image decoding apparatus of the present invention individually reversibly encodes each of the lossless encoding information of each syntax element divided by the means for dividing the decoding target stream according to the syntax element and the means for dividing according to the syntax element. Lossless decoding means for arithmetic decoding, means for storing lossless decoding information in block units created by the lossless decoding means, a reference frame buffer for inter-screen prediction, and intra-screen reference decoding for intra-screen prediction Means for storing information, the reference frame buffer, and means for predictively decoding a decoding target block from intra-screen reference decoding information.

  The image decoding method of the present invention also includes a step of dividing a decoding target stream according to a syntax element, and lossless encoding information of each syntax element divided in the division step according to the syntax element individually. A lossless decoding step, a step of storing the block-based lossless decoding information created in the lossless decoding step, a step of storing intra-screen reference decoding information for intra-screen prediction, and an inter-screen prediction And a step of predictively decoding the decoding target block from the reference frame buffer and the intra-screen reference decoding information.

  According to the present invention, in one-frame encoding / decoding processing, encoding / decoding processing using different arithmetic encoding / decoding devices can be performed for each syntax element. Division into independent processing becomes possible. Therefore, it is possible to process CABAC in units of syntax elements in parallel without dividing one frame into a plurality of slices. As a result, it is possible to encode and decode one frame at a high speed without the disadvantages of the conventional multi-slice structure, that is, the prediction efficiency decreases due to a decrease in reference candidates and the entropy encoding efficiency decreases due to CABAC initialization. Become. Furthermore, in the present invention, the orthogonal transform coefficients can be divided and the amount of computation of each entropy encoding / decoding process can be made uniform by independent processing.

It is a block diagram which shows the encoding apparatus in the best embodiment of this invention. It is a block diagram which shows the decoding apparatus in the best embodiment of this invention. It is a figure which shows the example of the division | segmentation of an orthogonal transformation coefficient. It is a flowchart for demonstrating operation | movement of the encoding apparatus (FIG. 1) in the best embodiment of this invention. It is a flowchart for demonstrating operation | movement of the decoding apparatus (FIG. 2) in the best embodiment of this invention. It is a figure which shows the example of the stream structure produced by the encoding apparatus of this invention. It is a figure which shows an example of block division. It is a figure which shows another example of block division. It is a flowchart for demonstrating the procedure which encodes an encoding object signal. It is a flowchart for demonstrating the procedure which obtains a decoding signal. It is a block diagram of one Example of the encoding apparatus of this invention. It is a block diagram of one Example of the decoding apparatus of this invention. It is a flowchart for demonstrating the procedure which determines a division position. It is a flowchart for demonstrating an example of a division | segmentation procedure.

(Encoder)
FIG. 11 is a block diagram of an embodiment of the encoding apparatus of the present invention. FIG. 11 shows an encoding target signal storage unit 1101, a block division unit 1102, an orthogonal transform / quantization unit 1103, a coefficient group storage unit 1104, a division position determination unit 1105, a coefficient division unit 1106, and a coefficient first half. A partial entropy encoding unit 1107, a coefficient latter half entropy encoding unit 1108, a division position encoding unit 1109, and a multiplexing unit 1110 are included.

  The encoding target signal storage unit 1101 has a function of storing a signal to be encoded. The encoding target signal is a difference signal from the original image signal or the intra-screen / inter-screen prediction signal.

  The block dividing unit 1102 has a function of dividing the encoding target signal into blocks of a predetermined size in a predetermined order and creating an encoding target block. In this configuration, the block division order is the raster scan order, and the block size is 4 × 4.

  The orthogonal transform / quantization unit 1103 has a function of performing orthogonal transform / quantization on the coding target block created by the block dividing unit 1102 to create a quantized orthogonal transform coefficient.

  The coefficient group storage unit 1104 has a function of sequentially storing the quantized orthogonal transform coefficients created by the orthogonal transform / quantization unit 1103. In this configuration, the coefficient group storage unit 1104 stores quantized orthogonal transform coefficients for all regions to be encoded. However, for example, the configuration may be such that the quantized orthogonal transform coefficient for a partial region of the encoding target signal is stored, such as only one block line from the beginning of the encoding target signal.

  The division position determination unit 1105 has a function of referring to the quantized orthogonal transform coefficients for all the blocks stored in the coefficient group storage unit 1104 and determining the division positions of the quantized orthogonal transform coefficients.

  A procedure for determining the division position will be described with reference to the flowchart of FIG.

  The division position candidate array divPosition [i] (0 <= i <15) is initialized to 0 before processing.

  (Step S1301) If the process of step S1302 and subsequent steps is performed on the quantized orthogonal transform coefficients of all the blocks stored in the coefficient group storage unit 1104, the process proceeds to S1305. Otherwise, the process proceeds to S1302.

  (Step S1302) The division position determination unit 1105 acquires from the coefficient group storage unit 1104 the quantized orthogonal transform coefficient of the first block that has not been processed in the raster scan order. The absolute value sum AbsSumBlock (0, 15) of all elements of the acquisition block is obtained, and the process proceeds to step S1303. Here, AbsSumBlock (i, j) represents the sum of absolute values of the j-th element from the i-th element of the target block.

  (Step S1303) The first position i where the absolute value difference sum AbsSumBlock (0, i) from the beginning of the acquired block exceeds AbsSumBlock (0, 15) / 2 is obtained. Then, the process proceeds to step S1304.

  (Step S1304) The value of divPosition [i] is incremented by 1, and the process proceeds to Step S1301.

  (Step S1305) The maximum i in divPosition [i] is set as the dividing position for all i from i = 0 to i = 15.

  In this configuration, the first half and the second half of the quantized orthogonal transform coefficient are equalized with respect to the absolute value difference sum when entropy coding requiring a constant time is performed on one bit of the coefficient value. The processing load of each entropy encoding of the section can be made uniform.

  FIG. 7 shows an example of division with this configuration.

  Reference numeral 701 in FIG. 7 denotes a processing order when entropy coding is performed without dividing the quantized orthogonal transform coefficient.

  702 in FIG. 7 is a division when the division position i = 2. The first half 703 is the first three quantized orthogonal transform coefficients in the processing order of 701, and the second half 704 is the last 13 quantized orthogonal transform coefficients in the processing order of 701. Within each division, entropy encoding is performed in the same order as the processing order of 701.

  In FIG. 7, reference numeral 705 denotes a division when the division position i = 6. The first half 706 is the first six quantized orthogonal transform coefficients in the processing order of 701, and the second half 707 is the second half 10 quantized orthogonal transform coefficients in the processing order of 701.

  Reference numeral 708 in FIG. 7 denotes a division when the division position i = 0. The first half 709 is the first half quantized orthogonal transform coefficient in the processing order of 701, and the second half 710 is the second half 15 quantized orthogonal transform coefficient in the processing order of 701.

  The same applies when the entropy encoding processing order is the order shown in FIG. 8 instead of FIG. Reference numeral 802 denotes a state of division when i = 3, and division is performed as a first half 803 and a second half 804. Reference numeral 805 denotes a state of division when i = 5, which is divided into a first half 806 and a second half 807. Reference numeral 808 denotes a state of division when i = 1, which is divided into a first half 809 and a second half 810.

  Another example of the division procedure will be described with reference to the flowchart of FIG.

  The division position candidate array divPosition [i] (0 <= i <15) is initialized to 0 before processing.

  (Step S1401) If the processing of step S1402 and subsequent steps is performed on the quantized orthogonal transform coefficients of all the blocks stored in the coefficient group storage unit 1104, the process proceeds to S1405. Otherwise, the process proceeds to S1402.

  (Step S1402) The division position determination unit 1105 acquires the quantized orthogonal transform coefficient of the first block that has not been processed in the raster scan order from the coefficient group storage unit 1104. The number of non-zero coefficients NonZeroCount (0, 15) is obtained from all elements of the acquisition block, and the process advances to step S1403. Here, NonZeroCount (i, j) represents the number of non-zero coefficients from the i-th element to the j-th element of the target block.

  (Step S1403) The first position i where the number of non-zero coefficients NonZeroCount (0, i) from the beginning of the acquired block exceeds NonZeroCount (0, 15) / 2 is obtained. Then, the process proceeds to step S1404.

  (Step S1404) The value of divPosition [i] is incremented by 1, and the process proceeds to Step S1401.

  (Step S1405) The maximum i in divPosition [i] is set as the dividing position for all i from i = 0 to i = 15.

  In this configuration, the first half and the second half of the quantized orthogonal transform coefficients are made uniform with respect to the sum of absolute value differences, so when performing entropy coding that requires a constant time for the number of non-zero coefficients, The processing load of each entropy encoding in the latter half can be made uniform.

  The coefficient first half entropy encoding unit 1107 has a function of acquiring the first half of the quantized orthogonal transform coefficients divided by the coefficient dividing unit 1106 in the order of encoding target blocks, performing entropy coding, and creating a coefficient first half stream. Have. The basic configuration of entropy coding is the same as that of the prior art.

  The coefficient latter half entropy encoding unit 1108 has a function of acquiring the latter half of the quantized orthogonal transform coefficients divided by the coefficient dividing unit 1106 in the order of encoding target blocks, performing entropy coding, and creating a coefficient latter half stream. Have. The coefficient latter half entropy encoding unit 1108 determines whether all the coefficients belonging to the coefficient latter half of the processing target block are 0 or not. When all the values are 0, only the ALL_ZERO information is encoded, and the second half coefficient is not encoded. Otherwise, after sending NON_ZERO_EXIST information, the latter half is encoded. The basic configuration of the latter half of the entropy coding is the same as the conventional one.

  The division position encoding unit 1109 has a function of entropy encoding the coefficient division positions created by the coefficient division unit 1106 to create a coefficient division position stream.

  The multiplexing unit 1110 includes the coefficient first half stream created by the coefficient first half entropy coding unit 1107, the coefficient second half stream created by the coefficient second half entropy coding unit 1108, and the coefficient division created by the division position coding unit 1109. It has a function to acquire and multiplex a position stream.

  A procedure for encoding a signal to be encoded with this configuration will be described with reference to the flowchart of FIG.

  (Step 901) The block dividing unit 1102 acquires 4 × 4 size encoding target blocks from the encoding target signal storage unit 1101 in the raster scan order. Further, the orthogonal transform / quantization unit 1103 acquires the block to be encoded from the block division unit 1102, performs orthogonal transform / quantization, and creates a quantized orthogonal transform coefficient. A coefficient group storage unit 1104 stores the created quantized orthogonal transform coefficient. The above operation is repeated for all blocks of the encoding target signal. After the above operation is completed for all blocks, the process proceeds to step S902.

  (Step S902) The division position determination unit 1105 acquires the quantized orthogonal transform coefficient stored in the coefficient group storage unit 1104, and determines the division position. The division position determination procedure uses the procedure shown in FIG. 13 or FIG. The division position determination unit 1105 sends the first half of the quantized orthogonal transform coefficient shown from the head to the division position in the order shown in FIG. 701 to the coefficient first half entropy encoding unit 1107, and the final position from the position next to the division position. The quantized orthogonal transform coefficient latter half part shown above is sent to the coefficient latter half part entropy encoding part 1108. The process proceeds to step S903.

  (Step S903) The division position encoding unit 1109 performs entropy encoding on the determined division position. The process proceeds to step S904.

  (Step S904) If the encoding process has been completed for all the processing target blocks, the process proceeds to Step S907. Otherwise, the process proceeds to step S905.

  (Step S905) The coefficient first half entropy coding unit 1107 entropy codes the quantized orthogonal transform coefficient first half of the processing target block to create a coefficient first half stream. The process proceeds to step S906.

  (Step S906) The coefficient latter half entropy coding section 1108 entropy codes the quantized orthogonal transform coefficient latter half of the processing target block to create a coefficient latter half stream. When the values of the latter half of the quantized orthogonal transform coefficients are all 0, only the ALL_ZERO information is encoded, and the coefficients of the latter half are not encoded. Otherwise, after sending NON_ZERO_EXIST information, the latter half of the quantized orthogonal transform coefficient is encoded. By adopting such a configuration, it is possible to avoid a decrease in the coding efficiency of the latter half, particularly when the quantization step is increased and the latter half is composed of only 0. The process proceeds to step S904.

  (Step S907) The multiplexing unit 1110 multiplexes the division position stream, the coefficient first half stream, and the coefficient latter half stream and transmits them to the outside.

  During this operation, step S905 and step S906 have no processing dependency and can be processed independently. Processing in parallel enables high-speed decoding processing.

(decoder)
FIG. 12 is a block diagram of an embodiment of the decoding device of the present invention. FIG. 12 shows a stream buffer 1201, a demultiplexing unit 1202, an orthogonal coefficient first half entropy decoding unit 1203, an orthogonal transform coefficient second half entropy decoding unit 1204, an orthogonal transform coefficient governing unit 1205, an inverse quantization / inverse An orthogonal transform unit 1206, a decoded signal buffer 1207, and a division position decoding unit 1208 are configured.

  The stream buffer 1201 has a function of storing a decoding target stream.

  The multiplexing / separating unit 1202 has a function of separating the decoding target stream stored in the stream buffer 1201 into a division position information stream, an orthogonal transform coefficient first half stream, and an orthogonal transform coefficient latter half stream.

  The orthogonal coefficient first half entropy decoding unit 1203 has a function of entropy decoding the orthogonal transformation coefficient first half stream to create an orthogonal transformation coefficient first half.

  The orthogonal transform coefficient latter half entropy decoding unit 1204 has a function of entropy decoding the orthogonal transform coefficient latter half stream to create an orthogonal transform coefficient latter half. The orthogonal transform coefficient latter half entropy decoding unit 1204 decodes the block latter half non-zero coefficient presence information. When the block latter half non-zero coefficient existence information is ALL_ZERO information, the coefficient of the latter half of the target block is set to zero. When the block non-zero coefficient existence information is NON_ZERO_EXIST information, entropy decoding of the latter half of the quantized orthogonal transform coefficient is performed.

  The orthogonal transform coefficient governing unit 1205 has a function of governing the first half of the orthogonal transform coefficient and the second half of the orthogonal transform coefficient according to the division position created by the division position decoding unit 1208 and creating a quantized orthogonal transform coefficient of the processing target block. Have. 7 is associated with the first half of the orthogonal transform coefficient in the order indicated by 701 in FIG. 7 and is associated with the latter half of the orthogonal transform coefficient from the next position to the final position of the divided position.

  The inverse quantization / inverse orthogonal transform unit 1206 performs a predetermined inverse quantization / inverse transform process on the quantized orthogonal transform coefficient created by the orthogonal transform coefficient governing unit 1205 to create a decoded signal.

  The decoded signal buffer 1207 has a function of storing the decoded signal in units of blocks created by the inverse quantization / inverse orthogonal transform unit 1206 for one screen.

  The division position decoding unit 1208 has a function of entropy decoding the division position information stream and creating a division position.

  A procedure for decoding a decoding target stream with this configuration and obtaining a decoded signal will be described with reference to the flowchart of FIG.

  (Step S1001) The demultiplexing unit 1202 separates the decoding target stream into a division position information stream, an orthogonal transform coefficient first half stream, and an orthogonal transform coefficient latter half stream. The division position information stream is sent to the division position decoding unit 1208, the orthogonal transform coefficient first half stream is sent to the orthogonal coefficient first half entropy decoding unit 1203, and the orthogonal transform coefficient second half stream is sent to the orthogonal transform coefficient second half entropy decoding unit 1204. The process proceeds to step S1002.

  (Step S1002) The division position decoding unit 1208 entropy decodes the division position information stream, and sends the created division position to the orthogonal transform coefficient rule unit 1205. The process proceeds to step S1003.

  (Step S1003) If the processing in step S1004 and subsequent steps has been completed for all the blocks of the decoding target stream, the process proceeds to step S1008. Otherwise, the process proceeds to step S1004.

  (Step S1004) The orthogonal coefficient first half entropy decoding unit 1203 entropy decodes the orthogonal transformation coefficient first half stream to create the first half of the quantized orthogonal transformation coefficient of the processing target block. The process proceeds to step S1005.

  (Step S1005) The orthogonal transform coefficient latter half entropy decoding unit 1204 entropy decodes the orthogonal transform coefficient latter half stream to create the latter half of the quantized orthogonal transform coefficient of the processing target block. The orthogonal transform coefficient latter half entropy decoding unit 1204 decodes the block latter half non-zero coefficient presence information. When the block latter half non-zero coefficient existence information is ALL_ZERO information, the coefficient of the latter half of the target block is set to zero. When the block non-zero coefficient existence information is NON_ZERO_EXIST information, entropy decoding of the latter half of the quantized orthogonal transform coefficient is performed. The process proceeds to step S1006.

  (Step S1006) The orthogonal transform coefficient governing unit 1205 governs the first half of the quantized orthogonal transform coefficient and the second half of the quantized orthogonal transform coefficient of the processing target block according to the division position, and the quantized orthogonal transform coefficient of the processing target block Create 7 is associated with the first half of the orthogonal transform coefficient in the order indicated by 701 in FIG. 7 and is associated with the latter half of the orthogonal transform coefficient from the next position to the final position of the divided position. The process proceeds to step S1007.

  (Step S1007) The inverse quantization / inverse orthogonal transform unit 1206 performs inverse quantization / inverse transform on the quantized orthogonal transform coefficient of the processing target block to create a decoded signal. The decoded signal of the created target block is stored in the decoded signal buffer. The process proceeds to step S1003.

  (Step S1008) The decoded signal for one screen stored in the decoded signal buffer 1207 is output to the outside.

  During this operation, step S1005 and step S1006 have no processing dependency and can be processed independently. Processing in parallel enables high-speed decoding processing.

(Encoder)
FIG. 1 is a configuration diagram of an embodiment of an encoding apparatus according to the present invention.

  FIG. 1 illustrates a block division unit 101, a prediction encoding unit 102, a reference frame buffer 103, a syntax element division unit 104, a lossless encoding unit 105, a stream multiplexing unit 106, a stream buffer 107, and a reference. The information storage unit 108 is configured. Further, the lossless encoding unit 105 includes a block type information lossless encoding unit 109, a motion vector information lossless encoding unit 110, a DC component lossless encoding unit 111, a low frequency component lossless encoding unit 112, The component lossless encoding unit 113 is configured. The reference information storage unit 108 includes a block type information storage unit 114, a motion vector information storage unit 115, a DC component information storage unit 116, a low frequency component information storage unit 117, and a high frequency component information storage unit 118. The local decoding frame buffer 119 is configured. The operation of each block will be described below.

  The block division unit 101 performs predetermined division on the encoding target frame to obtain an image of the encoding target block. Then, the encoding target block image is sent to the sequential prediction encoding unit 102.

  The prediction encoding unit 102 is a general image such as intra prediction or / and inter prediction, orthogonal transform / quantization, inverse quantization / inverse orthogonal transform, on the encoding target block image acquired from the block dividing unit 101. Perform the encoding process. In encoding, the already-encoded information is referred to from the reference information storage unit 108 and / or the reference frame buffer 103. The details of the operation are general and will not be described. Encoding information generated by encoding the target block is sent to the reference information storage unit 108 and the syntax element division unit 104, respectively.

  The reference frame buffer 103 stores a reference frame for performing inter-screen prediction in the prediction encoding unit 102.

  The syntax element dividing unit 104 divides the encoded information created by the predictive encoding unit 102 and sends the divided information to the lossless encoding unit 105. In this embodiment, the encoded information is divided into block type information, motion vector information, DC component information, low frequency component information, and high frequency component information. The DC component information, the low-frequency component information, and the high-frequency component information are quantized orthogonal transform coefficients generated as a result of performing orthogonal transform / quantization processing on the prediction residual of the target block. The orthogonal transform coefficient division method in this configuration is shown at 304 in FIG. The orthogonal transform unit is defined as a 4 × 4 block, 301 in FIG. 3 is DC component information, 302 is low-frequency component information, and 303 is high-frequency component information. In this embodiment, the orthogonal transform coefficient is fixed to the division of 304 in FIG. 3, but the division may be further subdivided, for example, as indicated by 310 in FIG. 3, or the orthogonal transform coefficient is not divided. It may be handled as one piece of information. 310 in FIG. 3 is an example when the orthogonal transform coefficient is divided into five parts 305, 306, 307, 308, and 309. Further, it is possible to adaptively divide the orthogonal transform coefficient using, for example, a quantization parameter. In the division, it is desirable to divide so that the information amount of each component information is as uniform as possible.

  The lossless encoding unit 105 acquires each syntax element from the syntax element dividing unit 104, acquires target reference information from the reference information storage unit 108, and performs lossless encoding by CABAC. The processing of each block in the lossless encoding unit 105 is independent and performs encoding processing in parallel.

  The stream multiplexing unit 106 acquires the lossless encoding information of each syntax element from the lossless encoding unit 105, performs multiplexing, and sends the information to the stream buffer 107.

  The stream buffer 107 stores the multiplexing information acquired from the stream multiplexing unit 106 and outputs it to the outside.

  The reference information storage unit 108 holds the encoded information in the screen. In the encoding process of the predictive encoding unit 102, the reference already encoded information of the target block is sent to the predictive encoding unit 102, and the encoding information of the target block generated by the predictive encoding unit 102 is stored. In addition, in the encoding process of the lossless encoding unit 105, the reference already-encoded information of the target block is sent to the lossless encoding unit 105.

  The block type information lossless encoding unit 109 losslessly encodes the encoding block type (for example, intra prediction or inter prediction or intra prediction mode if intra prediction) in the encoding information of the target block. . In the encoding process, the block type of the peripheral reference area is acquired from the block type information storage unit 114, the context is determined, and encoding is performed.

  The motion vector information lossless encoding unit 110 operates only when the target block is inter-screen prediction, and performs lossless encoding of the motion vector information of the target block. In the encoding process, the motion vector information of the in-screen peripheral reference area is acquired from the motion vector information storage unit 115, the context is determined, and encoding is performed.

  The DC component lossless encoding unit 111 performs lossless encoding of the DC component (301) of the residual component of the target block.

  The low frequency component lossless encoding unit 112 performs lossless encoding of the low frequency component (302) of the residual component of the target block.

  The high frequency component lossless encoding unit 113 performs lossless encoding of the high frequency component (302) of the residual component of the target block.

  The block type information storage unit 114 stores block type information in the encoding target screen. In the encoding process of the predictive encoding unit 102, the block type of the peripheral reference region of the target block is sent to the predictive encoding unit 102, and the block type information of the target block generated by the predictive encoding unit 102 is stored. In addition, when the block type information lossless encoding unit 109 performs the encoding process, the reference block type information of the target block is sent to the block type information lossless encoding unit 109.

  The motion vector information storage unit 115 stores motion vector information in the encoding target screen. In the encoding process of the predictive encoding unit 102, motion vector information of the peripheral reference region of the target block is sent to the predictive encoding unit 102, and the motion vector information of the target block generated by the predictive encoding unit 102 is stored. Further, in the encoding process of the motion vector information lossless encoding unit 110, the reference motion vector information of the target block is sent to the motion vector information lossless encoding unit 110.

  The DC component information storage unit 116 stores DC component information of residual components in the encoding target screen. In the encoding process of the prediction encoding unit 102, DC component information of the residual component of the target block generated by the prediction encoding unit 102 is stored. In addition, when the DC component lossless encoding unit 111 performs the encoding process, the reference DC component information of the target block is sent to the DC component lossless encoding unit 111.

  The low frequency component information storage unit 117 stores the low frequency component information of the residual component in the encoding target screen. In the encoding process of the predictive encoding unit 102, the low frequency component information of the residual component of the target block generated by the predictive encoding unit 102 is stored. In addition, when the low-frequency component lossless encoding unit 112 performs the encoding process, the reference low-frequency component information of the target block is sent to the low-frequency component lossless encoding unit 112.

  The high frequency component information storage unit 118 stores high frequency component information of the residual component in the encoding target screen. In the encoding process of the predictive encoding unit 102, the high frequency component information of the residual component of the target block generated by the predictive encoding unit 102 is stored. Further, in the encoding process of the high frequency component lossless encoding unit 113, the reference high frequency component information of the target block is sent to the high frequency component lossless encoding unit 113.

  Next, a one-frame encoding procedure in this configuration will be described using the flowchart of FIG.

  The encoding target frame is divided into blocks as encoding units, and encoding processing is performed in the raster scan order of the blocks. When all the blocks have been processed, the encoding process is completed. If the processing target block remains, the process proceeds to step S402 (step S401).

  Predictive encoding processing is performed on the encoding target block. An encoding target block image is supplied from the block dividing unit 101 to the predictive encoding unit 102, and a reference frame is supplied from the reference frame buffer 103. The reference information storage unit 108 supplies already-encoded information (block type information, motion vector information, DC / low band / high band component information) of the peripheral reference area and a locally decoded image of the same frame. The predictive encoding unit 102 performs intra prediction or / and inter prediction and orthogonal transform / quantization processing based on the acquired information, and generates encoding information of the encoding target block. Furthermore, the inverse quantization / inverse orthogonal transform process is performed based on the generated encoded information to generate a locally decoded image of the encoding target block.

  The generated encoded information is sent to the syntax element dividing unit 104 and the reference information storage unit 108, and the locally decoded image is sent to the reference information storage unit 108. The syntax element division unit 104 obtains the encoding information of the block to be encoded from the predictive encoding unit 102, and converts each piece of encoded information into the block type information lossless encoding unit 109 and the motion vector information lossless in the lossless encoding unit 105. The data is sent to the encoding unit 110, the DC component lossless encoding unit 111, the low frequency component lossless encoding unit 112, and the high frequency component lossless encoding unit 113 (step S402).

  The block type information lossless encoding unit 109 acquires the block type of the target block from the syntax element dividing unit 104, and acquires block type information of the reference area from the block type information storage unit 114. Based on the acquired information, lossless encoding by CABAC is performed to generate a block type stream, which is sent to the stream multiplexing unit 106 (step S403).

  The motion vector information lossless encoding unit 110 acquires the motion vector information of the target block from the syntax element dividing unit 104, and acquires the motion vector information of the reference area from the motion vector information storage unit 115. Based on the acquired information, lossless encoding by CABAC is performed to generate a motion vector information stream, which is sent to the stream multiplexing unit 106 (step S404).

  The DC component lossless encoding unit 111 acquires the DC component information of the target block from the syntax element dividing unit 104, and acquires the DC component information of the reference region from the DC component information storage unit 116. Based on the acquired information, lossless encoding by CABAC is performed to generate a DC component stream, which is sent to the stream multiplexing unit 106 (step S405).

  The low frequency component lossless encoding unit 112 acquires the low frequency component information of the target block from the syntax element dividing unit 104, and acquires the low frequency component information of the reference region from the low frequency component information storage unit 117. Based on the acquired information, lossless encoding by CABAC is performed to generate a low-frequency component stream, which is sent to the stream multiplexing unit 106 (step S406).

  The high frequency component lossless encoding unit 113 acquires the high frequency component information of the target block from the syntax element dividing unit 104, and also acquires the high frequency component information of the reference region from the high frequency component information storage unit 118. Based on the acquired information, lossless encoding by CABAC is performed to generate a high-frequency component stream, which is sent to the stream multiplexing unit 106 (step S407).

  The stream multiplexing unit 106 acquires a stream from each lossless encoding unit of the lossless encoding unit 105, performs multiplexing, and stores the stream in the stream buffer 107. The stream buffer 107 outputs the stream to the outside at a predetermined timing (step S408).

  The processing from step S403 to step S407 of this operation procedure is independent. The processing procedure of each step may be mixed or processed in parallel.

  FIG. 6 shows an example of a stream generated by this configuration. 601 in FIG. 6 is a stream example of this configuration, and 607 is a conventional stream example. In 607, the block type, motion vector information, and orthogonal transform coefficient in the slice are arranged in series. In arithmetic coding / decoding processing by CABAC, it is necessary to refer to the immediately preceding processing result. All syntax elements must be processed and cannot be processed from the middle of the slice. On the other hand, the structure 601 according to this configuration has a slice header, a block type information stream 602, a motion vector information stream 603, a DC component information stream 604, a low frequency component information stream 605, and a high frequency component information stream 606. Can perform CABAC processing independently. Since each stream holds ID information, the corresponding stream can be uniquely identified. With this configuration, CABAC processing that requires a large amount of computation can be processed in parallel, and high-speed encoding / decoding processing can be performed without reducing prediction efficiency due to the multi-slice structure and lossless encoding efficiency due to initialization of context information. Can be performed.

(decoder)
FIG. 2 is a block diagram of an embodiment of the decoding apparatus of the present invention. 2 shows a stream buffer 201, a syntax element division unit 202, a lossless decoding unit 203, a lossless decoding information storage unit 204, a prediction decoding unit 205, a reference frame buffer 206, a frame buffer 207, and reference information storage. The unit 208 is configured. Further, the lossless decoding unit 203 includes a block type information lossless decoding unit 209, a motion vector information lossless decoding unit 210, a DC component lossless decoding unit 211, a low frequency component lossless decoding unit 212, and a high frequency component lossless decoding unit 213. Consists of. Further, the reference information storage unit 208 includes a block type information storage unit 214, a motion vector information storage unit 215, a DC component information storage unit 216, a low frequency component information storage unit 217, and a high frequency component information storage unit 218. Is done.

  The stream buffer 201 stores the decoding target bit stream and sends it to the syntax element division unit 202.

  The syntax element division unit 202 multiplexes and separates the decoding target bit stream acquired from the stream buffer 201, and performs a block type information stream, a motion vector information stream, a DC component information stream, a low frequency component information stream, and a high frequency Each component information stream is divided, and the divided stream is sent to the lossless decoding unit 203.

  The lossless decoding unit 203 performs a lossless decoding process using CABAC on the stream acquired from the syntax element dividing unit 202. The decoding result is sent to the lossless decoding information storage unit 204.

  The lossless decoding information storage unit 204 stores the lossless decoding information of the decoding target block that has been losslessly decoded by the lossless decoding unit 203, and sends the information to the prediction decoding unit 205.

  The predictive decoding unit 205 acquires the lossless decoding information of the decoding target block from the lossless decoding information storage unit 204, and also receives the decoding information of the peripheral reference area in the screen from the reference information storage unit 208 from the frame buffer 207. The decoded image is acquired from the reference frame buffer 206, respectively. Based on the acquired information, a prediction image is created and residual information by inverse quantization / inverse transformation is created, and a decoded image of the target block is created. The generated decoded image is sent to the frame buffer 207, and the decoded information is sent to the reference information storage unit 208.

  The reference frame buffer 206 stores a reference frame for performing inter-screen predictive decoding in the predictive decoding unit 205.

  The frame buffer 207 stores the decoded image of the target frame.

  The reference information storage unit 208 holds the decryption information in the screen. In the decoding process of the predictive decoding unit 205, the reference encoded information of the target block is sent to the predictive decoding unit 205, and the decoding information of the target block generated by the predictive decoding unit 205 is stored. Further, reference decoding information of the target block is sent to the lossless decoding unit 203 in the decoding process of the lossless decoding unit 203.

  The block type information lossless decoding unit 209 performs lossless decoding of the coding block type (for example, intra prediction or inter prediction, or intra prediction mode if intra prediction) in the encoding information of the target block. In the decoding process, the block type of the peripheral reference area is acquired from the block type information storage unit 214, the context is determined, and decoding is performed.

  The motion vector information lossless decoding unit 210 operates only when the target block is inter-screen prediction, and performs lossless decoding of the motion vector information of the target block. In the decoding process, the motion vector information of the in-screen peripheral reference area is acquired from the motion vector information storage unit 215, the context is determined, and decoding is performed.

  The DC component lossless decoding unit 211 performs lossless decoding of the DC component (301) of the residual component of the target block.

  The low frequency component lossless decoding unit 212 performs lossless decoding of the low frequency component (302) of the residual component of the target block.

  The high frequency component lossless decoding unit 213 performs lossless decoding of the high frequency component (302) of the residual component of the target block.

  The block type information storage unit 214 stores block type information in the decoding target screen. In the decoding process of the predictive decoding unit 205, the block type of the peripheral reference area of the target block is sent to the predictive decoding unit 205, and the block type information of the target block generated by the predictive decoding unit 205 is stored. Further, in the encoding process of the block type information lossless decoding unit 209, the reference block type information of the target block is sent to the block type information lossless decoding unit 209.

  The motion vector information storage unit 215 stores motion vector information in the decoding target screen. In the decoding process of the predictive decoding unit 205, the motion vector information of the peripheral reference region of the target block is sent to the predictive decoding unit 205, and the motion vector information of the target block generated by the predictive decoding unit 205 is stored. Further, in the decoding process of the motion vector information lossless decoding unit 210, the reference motion vector information of the target block is sent to the motion vector information lossless decoding unit 210.

  The DC component information storage unit 216 stores DC component information of residual components in the decoding target screen. In the decoding process of the predictive decoding unit 205, DC component information of the residual component of the target block generated by the predictive decoding unit 205 is stored. In addition, when the DC component lossless decoding unit 211 performs the decoding process, the reference DC component information of the target block is sent to the DC component lossless decoding unit 211.

  The low frequency component information storage unit 217 stores the low frequency component information of the residual component in the encoding target screen. In the decoding process of the predictive decoding unit 205, the low frequency component information of the residual component of the target block generated by the predictive decoding unit 205 is stored. In addition, in the decoding process of the low frequency component lossless decoding unit 212, the reference low frequency component information of the target block is sent to the low frequency component lossless decoding unit 212.

  The high frequency component information storage unit 218 stores high frequency component information of the residual component in the encoding target screen. In the decoding process of the predictive decoding unit 205, the high frequency component information of the residual component of the target block generated by the predictive decoding unit 205 is stored. In addition, in the decoding process of the high frequency component lossless decoding unit 213, the reference high frequency component information of the target block is sent to the high frequency component lossless decoding unit 213.

  Next, a one-frame decoding procedure in this configuration will be described using the flowchart of FIG.

  When the decoding process is completed for all the blocks of the decoding target frame, the process proceeds to step S509. Otherwise, the process proceeds to step S502 (step S501).

  The syntax element division unit 202 acquires a stream from the stream buffer 201 and divides the stream into a block type information stream, a motion vector information stream, a DC component information stream, a low frequency component information stream, and a high frequency component stream. Each stream is sent to the block type information lossless decoding unit 209, the motion vector information lossless decoding unit 210, the DC component lossless decoding unit 211, the low frequency component lossless decoding unit 212, and the high frequency component lossless decoding unit 213 of the lossless decoding unit 203, respectively. (Step S502).

  The block type information lossless decoding unit 209 acquires a block type information stream from the syntax element dividing unit 202 and acquires block type information of peripheral reference blocks from the block type information storage unit 214 of the reference information storage unit 208. Then, arithmetic decoding processing by CABAC is performed based on the acquired information, and block type information of the target block is created and sent to the lossless decoding information storage unit 204 (step S503).

  The motion vector information lossless decoding unit 210 acquires a motion vector information stream from the syntax element division unit 202, and acquires motion vector information of neighboring reference blocks from the motion vector information storage unit 215 of the reference information storage unit 208. Then, arithmetic decoding processing by CABAC is performed based on the acquired information, motion vector information of the target block is created, and sent to the lossless decoding information storage unit 204 (step S504).

  The DC component lossless decoding unit 211 acquires a DC component information stream from the syntax element dividing unit 202 and acquires DC component information of the peripheral reference block from the DC component information storage unit 216 of the reference information storage unit 208. Then, arithmetic decoding processing by CABAC is performed based on the acquired information, DC component information of the target block is created, and sent to the lossless decoding information storage unit 204 (step S505).

  The low frequency component lossless decoding unit 212 acquires the low frequency component information stream from the syntax element dividing unit 202, and also acquires the low frequency component information of the peripheral reference block from the low frequency component information storage unit 217 of the reference information storage unit 208. get. Then, arithmetic decoding processing by CABAC is performed based on the acquired information, and low frequency component information of the target block is generated and sent to the lossless decoding information storage unit 204 (step S506).

  The high frequency component lossless decoding unit 213 acquires the high frequency component information stream from the syntax element dividing unit 202, and also acquires the high frequency component information of the peripheral reference block from the high frequency component information storage unit 218 of the reference information storage unit 208. get. Then, arithmetic decoding processing by CABAC is performed based on the acquired information, and high frequency component information of the target block is generated and sent to the lossless decoding information storage unit 204 (step S507).

  The predictive decoding unit 205 acquires the lossless decoded block type, motion vector information, and orthogonal transform information from the lossless decoding information storage unit 204, acquires a reference frame from the reference frame buffer 206, and also acquires a frame buffer. The already decoded image of the encoding target frame is acquired from 207, and each decoding information of the neighboring blocks is acquired from the reference information storage unit 208. Based on the acquired information, intra-screen prediction or / and inter-screen prediction, and inverse quantization / inverse orthogonal transform processing are performed to obtain a decoded image of the target block. The decoding information of the target block is sent to the reference information storage unit 208, and the decoded image is sent to the frame buffer (step S508).

  When the decoding process is completed for all the blocks of the target frame, the decoded image held in the frame buffer 207 is output to the outside.

  As described above, the present invention can provide an encoding device / decoding device / stream configuration that enables CABAC parallel processing by independently encoding / decoding and transmitting stream syntax elements. it can.

In particular, for high-definition and high-quality images, the burden on CABAC processing is large. According to the present invention, the CABAC processing is performed in parallel, which occurs due to unnecessary slice division.
-Decrease in prediction efficiency due to a decrease in reference candidates-Avoid reduction in entropy encoding efficiency due to frequent occurrence of CABAC processing, and create an encoding device, decoding device, and stream that can be processed within the required time with high encoding efficiency It will be possible.

DESCRIPTION OF SYMBOLS 101 Block division part 102 Prediction encoding part 103 Reference frame buffer 104 Syntax element division part 105 Lossless encoding part 106 Stream multiplexing part 107 Stream buffer 108 Reference information storage part 109 Block type information lossless encoding part 110 Motion vector information lossless code Conversion unit 111 DC component lossless encoding unit 112 low frequency component lossless encoding unit 113 high frequency component lossless encoding unit 114 block type information storage unit 115 motion vector information storage unit 116 DC component information storage unit 117 low frequency component information storage unit 118 High-frequency component information storage unit 119 Local decoding frame buffer 201 Stream buffer 202 Syntax element division unit 203 Lossless decoding unit 204 Lossless decoding information storage unit 205 Predictive decoding unit 206 Reference frame buffer 207 Frame buffer 208 Reference information storage unit 209 Block type information lossless decoding unit 210 Motion vector information lossless decoding unit 211 DC component lossless decoding unit 212 Low band component lossless decoding unit 213 High band component lossless decoding unit 214 Block type information storage unit 215 Motion vector information storage unit 216 DC component information storage unit 217 Low frequency component information storage unit 218 High frequency component information storage unit 301 DC component 302 Low frequency component 303 High frequency component 304 Block division example 1
305 DC component 306 AC component 1
307 AC component 2
308 AC component 3
309 AC component 4
310 Block division example 2
601 Example of stream structure of this configuration 602 Block type information stream 603 Motion vector information stream 604 DC component information stream 605 Low frequency component information stream 606 High frequency component information stream 607 Example of conventional stream structure 1101 Encoding target signal storage unit 1102 Block division unit 1103 Orthogonal transform / quantization unit 1104 Coefficient group storage unit 1105 Division position determination unit 1106 Coefficient division unit 1107 Coefficient first half entropy coding unit 1108 Coefficient latter half part entropy coding unit 1109 Division position coding unit 1110 Multiplexing unit 1201 Stream buffer 1202 Demultiplexing unit 1203 Orthogonal coefficient first half entropy decoding unit 1204 Orthogonal transform coefficient latter half entropy decoding unit 1205 Orthogonal transform coefficient governing unit 1206 Inverse quantization / inverse orthogonal transformation Part 1207 decoded signal buffer 1208 division position decoding section

Claims (20)

Means for storing a signal to be encoded;
Means for dividing the encoding target signal into encoding unit blocks;
Means for orthogonally transforming and quantizing the signal of the coding unit block;
Means for storing quantized orthogonal transform coefficients of one or more coding unit blocks;
Means for determining a division position for dividing and entropy-encoding the quantized orthogonal transform coefficient from the stored quantized orthogonal transform coefficient of the one or more coding unit blocks;
Means for dividing a quantized orthogonal transform coefficient of a coding unit block into a plurality of regions at a division position;
Means for independently entropy encoding a plurality of divided quantized orthogonal transform coefficients. The means for determining the division position is:
For the quantized orthogonal transform coefficients of all blocks stored in the means for storing the quantized orthogonal transform coefficients,
The image coding according to claim 1, wherein the division position is determined by comparing the absolute value sum of all the coefficient levels of one coding unit block with the absolute value sum of a part of the coefficient levels in the entropy coding order. apparatus. The means for determining the division position is:
For the quantized orthogonal transform coefficients of all blocks stored in the means for storing the quantized orthogonal transform coefficients,
The image coding apparatus according to claim 1, wherein the division position is determined by comparing the number of all non-zero coefficients of one coding unit block with the number of non-zero coefficients of a part in the entropy coding order. Of the means for independently entropy-encoding the plurality of divided quantized orthogonal transform coefficients, the means for entropy-encoding the quantized orthogonal transform coefficient that is not the first in the entropy coding order is the quantized orthogonal transform coefficient of the target portion. The image according to any one of claims 1 to 3, wherein whether or not all are 0 is determined, and if all are 0, information indicating that all are 0 is encoded, and the quantized orthogonal transform coefficient is not encoded. Encoding device. Means for storing a decoding target stream in which one block of orthogonal transform coefficients is divided and multiplexed;
Demultiplexing means for separating the decoding target stream into two or more orthogonal transform coefficient streams and a division position stream;
A division position decoding unit that decodes the division position stream and creates division positions of orthogonal transform coefficients of one block;
Quantized orthogonal transform coefficient entropy decoding means for independently decoding the orthogonal transform coefficient stream composed of the two or more,
Means for governing two or more quantized orthogonal transform coefficients subjected to entropy decoding based on the division position, and generating one block of quantized orthogonal transform coefficients;
An image decoding apparatus comprising: an inverse quantization / inverse transform unit that inversely quantizes and performs inverse transform on a governed quantized orthogonal transform coefficient to generate a decoded signal. Among the quantized orthogonal transform coefficient entropy decoding means, the means for entropy decoding the quantized orthogonal transform coefficient that is not the first in the entropy decoding order is a partial non-zero coefficient whether or not the quantized orthogonal transform coefficients of the target part are all zero 6. The information is decoded, and when the quantized orthogonal transform coefficients are all 0, the quantized orthogonal transform coefficients of the target part are all set to 0, and the coefficient level is entropy-decoded only when it is not. Image decoding device. Storing a signal to be encoded;
Dividing the encoding target signal into encoding unit blocks;
Orthogonally transform and quantize the signal of the coding unit block;
Storing quantized orthogonal transform coefficients of one or more coding unit blocks;
Determining a division position for dividing the quantized orthogonal transform coefficient and entropy coding from the stored quantized orthogonal transform coefficient of the one or more coding unit blocks;
Dividing the quantized orthogonal transform coefficient of the coding unit block into a plurality of regions at the division position;
And a step of independently entropy-encoding the plurality of divided quantized orthogonal transform coefficients. Storing a decoding target stream in which one block of orthogonal transform coefficients is divided and multiplexed;
A demultiplexing step for separating the decoding target stream into two or more orthogonal transform coefficient streams and a divided position stream;
Decoding the split position stream to create a split position for one block of orthogonal transform coefficients;
A quantized orthogonal transform coefficient entropy decoding step for independently decoding the orthogonal transform coefficient stream composed of the two or more;
Governing two or more quantized orthogonal transform coefficients subjected to entropy decoding based on the division position, and generating a block of quantized orthogonal transform coefficients;
An image decoding method comprising: dequantizing and transforming the governed quantized orthogonal transform coefficient to generate a decoded signal. Block dividing means for dividing the encoding target image into encoding units;
A reference frame buffer for inter-screen prediction;
Means for storing intra-screen reference encoding information for intra-screen prediction;
Means for irreversibly predictive encoding a block to be encoded by the reference frame buffer and the intra-screen reference encoding information;
Means for dividing the encoded information created by the predictive encoding means for each syntax element;
Lossless encoding means for individually lossless arithmetic coding each syntax element divided by the dividing means for each syntax element;
An image encoding device comprising: a multiplexing unit that multiplexes the lossless encoding information of each syntax element that has been losslessly encoded by the lossless encoding unit. The lossless encoding means includes block type lossless encoding means for losslessly encoding a block type indicating whether a prediction method of an encoding target block is intra prediction or inter prediction. Image encoding device. The image encoding device according to claim 9 or 10, wherein the lossless encoding means includes motion vector lossless encoding means for losslessly encoding motion vector information of a block to be encoded. The image encoding device according to any one of claims 9 to 11, wherein the lossless encoding means includes quantization orthogonal transform coefficient lossless encoding means for losslessly encoding the quantized orthogonal transform coefficient information of the encoding target block. The image coding apparatus according to claim 12, wherein the quantized orthogonal transform coefficient lossless encoding means includes a plurality of frequency component lossless encoding means. A block division step for dividing the encoding target image into encoding units;
Storing intra-screen reference encoding information for intra-screen prediction;
A step of performing irreversible predictive encoding of a block to be encoded using a reference frame buffer for inter-screen prediction and intra-screen reference encoding information;
Dividing the encoding information created in the predictive encoding step into syntax elements;
A lossless encoding step for individually lossless arithmetic encoding each syntax element divided in the division step for each syntax element;
A multiplexing step of multiplexing lossless encoding information of each syntax element that has been losslessly encoded in the lossless encoding step. Means for dividing the stream to be decoded according to syntax elements;
Lossless decoding means for individually performing lossless arithmetic decoding of each of the lossless encoding information of each syntax element divided in the means for dividing the syntax element;
Means for storing reversible decoding information in block units created by the lossless decoding means;
A reference frame buffer for inter-screen prediction;
Means for storing intra-screen reference decoding information for intra-screen prediction;
An image decoding apparatus comprising: the reference frame buffer; and means for predictively decoding a decoding target block from intra-screen reference decoding information. The image decoding apparatus according to claim 15, wherein the lossless decoding means includes block type lossless decoding means for losslessly decoding a block type indicating whether a prediction method of a decoding target block is intra prediction or inter prediction. The image decoding device according to claim 15 or 16, wherein the lossless decoding means includes motion vector lossless decoding means for losslessly decoding motion vector information of a decoding target block. The image decoding device according to any one of claims 15 to 17, wherein the lossless decoding means includes a quantized orthogonal transform coefficient lossless decoding means for losslessly decoding the quantized orthogonal transform coefficient information of a decoding target block. The image decoding device according to claim 18, wherein the quantized orthogonal transform coefficient lossless decoding means includes a plurality of frequency component lossless decoding means. Dividing the stream to be decoded according to syntax elements;
A lossless decoding step of individually performing lossless arithmetic decoding of each lossless encoding information of each syntax element divided in the division step according to the syntax element;
Storing the block-based lossless decoding information created in the lossless decoding step;
Storing intra-screen reference decoding information for intra-screen prediction;
An image decoding method comprising: a reference frame buffer for performing inter-screen prediction; and a step of predictively decoding a decoding target block from intra-screen reference decoding information.

Images