CN1290340C - predictive decoding method - Google Patents

Abstract

The invention provides a predictive decoding method for decoding a video picture to generate a plurality of predictors corresponding to a plurality of blocks of the video picture, the predictive decoding method comprising: storing a plurality of first vertical predictors of a first block to a storage column of a first memory device, and storing a plurality of first horizontal predictors of a second block to a storage row of the first memory device; performing a prediction operation to generate a plurality of target vertical predictors and a plurality of target horizontal predictors for a first target block according to the plurality of first vertical predictors and the plurality of first horizontal predictors; and updating the storage column and the storage row of the first storage device respectively by using the target vertical predictors and the target horizontal predictors respectively.

Description

predictive decoding method

technical field

The present invention relates to a predictive decoding method using memory access, in particular to a method for using at least one storage device to access multiple predictors in a video frame and reducing the usage of the storage device.

Background technique

In 1988, MPEG (Motion Picture Experts Group, animation expert group) was established. MPEG is a working group of ISO (International Standards Organization). This working group has established and promoted some standard compression formats for digital video and audio. These compression formats are now widely used in the production of digital content products around the world. Since its establishment in 1988, MPEG has announced several important standards. Among the audio and video file formats, the dynamic image compression standards MPEG-2 and MPEG-4 are among the best, and the two also have many similarities in the operation process of adding and decoding. Please refer to FIG. 1 for a partial decoding procedure of the dynamic image compression standard MPEG-4. FIG. 1 is a functional block diagram of some devices in an MPEG-4 decoder (decoder) 10 . The decoder 10 of Fig. 1 comprises a variable length decoding unit (variable length decoder) 12, an inverse scanning unit 14 (inverse scanner), an inverse DC/AC prediction module (inverse DC/AC prediction module) 16, An inverse quantization unit (inverse quantization unit) 18 , an inverse discrete cosine transformer (inverse DCT) 20 , and a motion compensation unit (motion compensation unit) 22 . Please see Fig. 1, after the compressed video (video) data passes through the variable length decoding unit 12, after being processed by a variable length decoding program, a basic decoding operation can be performed on the video data transmitted from the encoding end Next, an anti-scan program is carried out through the anti-scanning unit 14, and the video data of the one-dimensional matrix format (one-dimensional array) is restored to the video data of the two-dimensional matrix format (two-dimensional array), and the anti-DC/AC A detailed description of the prediction module 16 will be provided later. The aforementioned video data in the two-dimensional matrix format will be restored to the original discrete cosine transform coefficient (DCT coefficient) in the frequency domain after the dequantization unit 18 executes the dequantization process. Next, the generated DCT coefficients are then input to the inverse discrete cosine converter 20 to generate a number of non-zero-valued outputs to convert the signal of the image in the frequency domain into a signal in the space (spatial domain) to convert The signal is restored to the data stream before compression. Finally, the motion compensation unit 22 performs a motion compensation process to integrate the received data streams and generate the final decoded output picture according to the information of the reference picture and the motion vector.

Please note that the relevant functions of the inverse DC/AC prediction module 16 not mentioned in the above brief description are actually one of the most important technical features of the MPEG-4 decoder 10 . Please refer to FIG. 2 . FIG. 2 is a schematic diagram of the operation of a conventional inverse DC/AC prediction module 16 . FIG. 2 inherits the structure of FIG. 1, and illustrates that the video signal in the two-dimensional matrix format generated by the inverse scanning unit 14 will then perform a prediction operation (prediction operation) through the inverse DC/AC prediction module 16, and further in the two-dimensional The pixel coefficient (pixel coefficient) of the entire video frame is decoded spatially. During the operation of the inverse DC/AC prediction module 16, the entire video frame is divided into a plurality of blocks (blocks) with a size of 8*8 pixels, and the pixel coefficients of the new block to be decoded next, Then, according to some pixel coefficients of blocks that have been decoded before, the prediction selection unit 24 is sent to the inverse DC/AC prediction module 16, and the differential value calculated by the variable-length decoding unit 12 ( difierential value), the combination of the two is completed. In order to clearly illustrate the operation of the inverse DC/AC prediction module 16 and the prediction selection unit 24 in FIG. 2 on a video frame, please refer to FIG. 3 first. FIG. 3 is a schematic diagram of performing the above prediction operation in a macro-block 32 on a video frame 30 . In the dynamic image compression standard MPEG, the most basic processing unit is a block with a size of 8*8 pixels, and a macroblock 32 has a size of 16*16 pixels and includes four blocks. Since the dynamic image compression standard MPEG adopts the color representation format of luminance and chrominance, the macroblock 32 in this embodiment actually refers to a luminance macro-block. Thus, if the horizontal and vertical chrominance data are sampled at half the sampling frequency of luma, the original 16x16 pixels are sampled and the sampled luma and chrominance data are encoded to produce a luma macro block (Y) and two chrominance blocks (Cr and Cb), wherein the luminance macroblock contains luminance information corresponding to the 16×16 pixels, and the two chrominance blocks contain information corresponding to the Chroma information for half of a 16x16 pixel.

Please continue to refer to FIG. 3 , the macroblock 32 includes a first block B, a second block C, a third block A, and a fourth block X to be decoded. Before decoding the fourth block X, a plurality of predictors (predictors) are respectively defined in adjacent blocks (the first block B, the second block C, and the third block A), According to the spatial configuration in each block, the plurality of predictors can be further divided into two types: DC coefficient (DC coefficient) and AC (AC coefficient) coefficient. As shown in Figure 3, the first block B, the second block C, and the third block A each contain a DC coefficient, which are respectively DC _B , DC _C , and DC _A (the cells filled with double oblique lines in the figure ), and for the fourth block X, in this prediction operation, the second block C also contains multiple (seven) AC coefficients: AC _C (expressed in a two-dimensional matrix format: AC _C [ 0][n], n is an integer from 1 to 7); similarly, the third block A contains seven AC coefficients: AC _A (expressed in a two-dimensional matrix format: AC _A [n][0], n is an integer from 1 to 7). Since some of the pixel coefficients in the fourth block X have two sources of predictors: the second block C located above or the third block A located on the left, in order to determine whether the pixel coefficients of the fourth block X are Obtained from the predictors in the second block C or the third block A, a motion vector must be determined through the following judgment formula: (|DC _A -DC _B |<|DC _B -DC _C |) (Equation 1) If Equation 1 is established, then the prediction selection unit 24 in FIG. 2 will judge that some of the pixel coefficients in the fourth block X come from the second block C, and determine a first motion vector (DCAC direction vector ) M1, on the contrary, if formula 1 is not established, it is judged that part of the pixel coefficients in the fourth block X come from the third block A, and a second motion vector M2 is determined. After determining the source and motion vector of the predictor, the inverse DC/AC prediction module 16 in FIG. 2 adds a DC differential value to the DC coefficient of the source block, and sets it as the DC coefficient of the fourth block X, that is, The first pixel coefficient in the fourth block X shown in FIG. 3 . Next, the prediction selection unit 24 in FIG. 2 will send the AC coefficient of the predictor of the source block to the inverse DC/AC prediction module 16, and add an AC differential value to the AC coefficient to set the AC of the fourth block X coefficient, and the AC coefficient is located in a first storage row or a first storage column in the fourth block X. For example _, if the source is the upper second block _C , the inverse DC/AC prediction module 16 in FIG. Calculate the DC differential value and the AC differential value, and set the resulting value as the pixel coefficient of a first storage column (33) in the fourth block X, and then decode the entire fourth block X. Similarly, if the source is the third block A on the left, DC _A and AC _A of the third block A will be respectively added with the DC difference value and AC difference calculated by the variable-length decoding unit 12 Set the obtained result value as the pixel coefficient of a first storage row (35) in the fourth block X, and then decode the entire fourth block X.

It can be seen from the above that the compression standard of MPGE-4 obtains some pixel coefficients in the block to be decoded according to the predictors of the spatially adjacent and decoded blocks. For the entire video frame 30 , after each block is sequentially processed by the above prediction operation, the pixel coefficients of the entire video frame 30 can be generated. In addition, in the process of decoding, since it is necessary to continuously determine the predictors for the blocks to be decoded, the system (decoder 10 shown in FIG. 1 and FIG. The storage device is used for storing multiple predictors. In order to deduce the complete prediction operation procedure, the following example of FIG. 4 is used to describe the situation of decoding a macroblock 42, and also observe the number of predictors that need to be stored during the process. Please refer to FIG. 4 , which is a schematic diagram of an embodiment of the known technology. The definition of the embodiment in FIG. 3 is slightly different. In this embodiment, the blocks to be decoded include a first block X, a second block Y, a third block X', and a fourth block Y', and these four blocks are merged into one macroblock 42 . Surrounding the four blocks are respectively a reference block (reference block) B, a first adjacent block A (adjacent block), a second adjacent block C, a third adjacent block C', and a first adjacent block Four adjacent blocks A'. At this point, please refer back to the architecture in Figure 2 at the same time. Inheriting the feature of using the upper and left adjacent blocks in the embodiment of FIG. 3 to obtain the data of the block to be decoded, when processing the first block X, it is necessary to provide the reference block B located at the upper left One predictor (DC _B ), eight predictors (one DC _C and seven AC _C [0][1-7]) of the first adjacent block A above it, and the second above it For the eight predictors (one DC _C and seven AC _C [1-7][0]) of the adjacent block C, the prediction selection unit 24 in FIG. 2 judges the correctness by DC _B , DC _C and DC _A After the source of the predictor, the remaining AC coefficients are sent to the inverse DC/AC prediction module 16 in Figure 2, and combined with the differential value sent by the variable length decoding unit 12, the pixels of the first block X are obtained coefficient. That is, in order to complete the decoding of the first block X, a storage device capable of storing 17 predictors (17=1+8+8) needs to be provided in the system. After completing the encoding of the first block X, before processing the second block Y, since the first block X and the third adjacent block C' are immediately adjacent to the second block Y, the first block needs to be The leftmost row of block X is set as the predictor, continue to use DC _C of the second adjacent block C, DC _X and AC _X [1-7][0] of the first block X, and the third adjacent block C 'DC _C' and AC _C' [0][1-7] to determine some pixel coefficients of the second block Y. Similarly, DC A of the first adjacent block _A , DC _X and AC _X [0][1-7] of the first block X, and DC _A' and AC A' of the fourth adjacent block _A' [ 1-7][0] can determine some pixel coefficients of the third block X'. DC _X of the first block X, DC _Y and AC _Y [0][1-7] of the second block Y, and DC _X' and AC _X' [1-7][ of the third block X' 0] can determine some pixel coefficients of the fourth block Y'.

For each macroblock 42 (including one (16×16 pixel size) luma macroblock 32 described in FIG. 3 , one (8×8 pixel size) Cb chrominance block, and one (8×8 pixel size) Pixel-sized (Cr chrominance block), the above-mentioned prediction operation and decoding procedure can be summarized in FIG. 5, and FIG. 5 is a flow chart of an embodiment of a known technology:

Step 100: start;

Step 101: When processing any block in the macroblock (such as the first block X in FIG. 4 ), determine whether there are predictors that can perform prediction operations. (For example, for the first block X in FIG. 4, the required predictors include the predictor DC _B of the block, the predictors DC _A and AC _A [1-7][0 of the first adjacent block A ], the predictor DC _C and AC _C [0][1-7] of the second adjacent block C), if there is, then proceed to step 103, if not, then proceed to step 102;

Step 102: Set the predictor required by the block to be decoded in a predetermined manner, such as setting the required DC coefficient to a fixed value, and simultaneously setting the required AC coefficient to 0, and the preset is completed Carry out step 104 afterwards;

Step 103: Judging and determining the source of the predictor and a related motion vector to obtain the predictor required by the block to be decoded. In actual implementation, the required predictor can be read from the above-mentioned storage device used to store the predictor in the system, and step 104 is performed;

Step 104: Add a DC coefficient in multiple predictors to a DC differential value calculated by the variable-length decoding unit to generate a DC coefficient of the block to be decoded; and then add DC coefficients in the multiple predictors The AC coefficients of the block to be decoded are generated by adding the AC differential value calculated by the variable-length decoding unit to the AC coefficients, and placed in the first storage row or in the block to be decoded. first storage line;

Step 105: use a counter (counter) to add one to the value (integer) of the counter memory;

Step 106: Determine whether the value of the counter memory is greater than 4, if not, go back to step 101, continue to process another block in this macroblock, if the value of the counter memory is greater than 4, it means that the macroblock will be The four blocks of are all processed, then proceed to step 107;

Step 107: Execute prediction operation and decoding procedure on a Cb chrominance block of 8×8 pixel size and a Cr chrominance block of 8×8 pixel size;

Step 108: End the prediction operation and decoding procedure in this macroblock, and skip to the next macroblock.

In addition to the above-mentioned known technologies that have been partly regulated in the dynamic image compression standard MPEG-4, many related methods and structures are disclosed in US Patent No. 6,005,622 proposed by Haskell et al., "Video coder providing implicit or explicit prediction for image coding and intra coding of video". Please note that, as described in the embodiment of FIG. 4, during the prediction operation and decoding process, since it is necessary to continuously determine the predictor for the block to be decoded, at least one storage device must be configured in the system for storage Multiple predictors are used. According to the general calculation of the content described in the embodiment of FIG. 4, macroscopically speaking, in any block of each macroblock, the pixel coefficients located in the leftmost row and the uppermost column of the block will be determined as predictors. In order to decode a macroblock, a storage device capable of storing the capacity equivalent to 17*6 predictors must be provided in the system (the calculation method of 6 of the aforementioned 17*6 is: four Y blocks, one Cb block and a Cr block). Continuing in this way, if it is desired to decode all the pixel coefficients in an entire video frame in two-dimensional space, the number of predictors required to be accessed is considerable (taking a video frame with a size of 720×480 pixels as an example , roughly 30*45*17*4 predictors will be determined). This also means that the capacity of the storage device set in the system must be able to store such a huge number of predictors. Just, the consideration about the capacity of storage device is not important in the known technology (such as known patent US Patent No.6,005,622) that mainly completes the decoding program with software (Software) operation, yet, in today's industry circle all want Under the trend of integrating related video codecs/decoders with one hardware unit, it becomes unfeasible to install such a large-capacity storage device in an on-chip manner in the system.

Contents of the invention

Therefore, the main purpose of the present invention is to provide a method for accessing multiple predictors in a video frame with less storage space, so as to solve the above-mentioned problems.

The present invention provides a predictive decoding method for decoding a video picture to generate multiple predictors corresponding to multiple blocks of the video picture. The predictive decoding The method includes: (a) storing a plurality of first vertical predictors of a first block into a storage column of a first storage device, and storing a plurality of vertical predictors of a second block The first horizontal predictor (horizontal predictor) is stored in a storage row (storing row) of the first storage device; (b) perform a prediction operation (prediction operation) to base on the plurality of first vertical predictors and the plurality of A first horizontal predictor is used to generate a plurality of target vertical predictors and a plurality of target horizontal predictors of a first target block, wherein the first target block is adjacent to the first and the second blocks, and the first target block is adjacent to the first and the second blocks, and the The first block and the first target block are located in the same row; and (c) using the plurality of target vertical predictors to update the storage row of the first storage device, and using the plurality of target horizontal predictors to update updating the storage rank of the first storage device.

The present invention also provides a method for storing a plurality of predictors of a macro-block comprising a first storage device and a second storage device. A block, a second block, a third block, and a fourth block, the method includes: (a) according to a first adjacent block (adjacent block) and a second adjacent block to Generate a plurality of predictors of the first block; (b) after performing step (a), store the plurality of predictors of the first block in the first storage device; (c) perform step ( b) afterwards, generate a plurality of predictors of the second block according to a third adjacent block and the first block; (d) after performing step (c), generate the plurality of predictors of the second block The predictor is stored in the first storage device; (e) after step (d), a plurality of predictors of the third block are generated according to a fourth adjacent block or the first block; (f) After performing step (e), storing the multiple predictors of the third block in the first storage device and the second storage device; (g) after performing step (f), according to the second block and the third block to generate a plurality of predictors for the fourth block; and (h) after performing step (g), storing the plurality of predictors for the fourth block in the first The storage device and the second storage device.

Description of drawings

Fig. 1 is a functional block diagram of some devices in an MPEG-4 decoder.

FIG. 2 is a schematic diagram of the operation of a conventional inverse DC/AC prediction module.

FIG. 3 is a schematic diagram of performing predictive decoding in a macroblock on a video frame.

Fig. 4 is a schematic diagram of an embodiment of the known technology.

FIG. 5 is a flowchart of a method embodiment of a known technology.

FIG. 6 is a spatial configuration diagram of a video frame.

Fig. 7 is a schematic diagram of an embodiment of the present invention.

FIG. 8 is a flowchart of a detailed method embodiment of the present invention.

FIG. 9 is a flowchart of related operations of adding a second storage device in the embodiment shown in FIG. 8 .

FIG. 10 is a list of content changes of storage rows and storage columns in the storage device of FIG. 7 .

Description of reference symbols

10 Decoder 12 Variable length decoding unit

14 Anti-scanning unit 16 Anti-DC/AC prediction module

18 Inverse Quantization Unit 20 Inverse Discrete Cosine Converter

22 Motion Compensation Unit 24 Prediction Selection Unit

30, 50 Video picture 32, 42, 52 Macro blocks

33 The first storage row 35 The first storage row

51 Macroblock Column 53 Cb Chroma Block

55 Cr Chroma Block 56 Storage Device

58 Second storage device

Detailed ways

In the predictive decoding method of the present invention, the video data bitstream (video bitstream) includes a plurality of video frames (picture), and each of the video frames can be a dynamic image compression standard MPEG (moving picture expert group, A frame (frame), a top field (top field) or a bottom field (bottom field) defined by the specification of MPEG for short. A video frame can be divided into multiple macro-blocks 52 , and each macro-block 52 with a size of 16*16 pixels can be regarded as a processing unit. Please refer to FIG. 6 . FIG. 6 is a spatial configuration diagram of a video frame 50 . The video image 50 includes a plurality of macro-block rows (macro-blockrow) 51, each macro-block row 51 includes a plurality of macro-blocks 52, and each macro-block 52 includes four 8*8 pixel-sized block. Taking a video frame 50 with a size of 720×480 pixels as an example, it includes 30 macroblock rows 51 , and each macroblock row 51 includes 45 macroblocks 52 . In addition, an arrow MR1 is also shown in FIG. 6 , which represents the processing order of the macroblocks 52 in the present invention (from left to right) to process each macroblock 52 in a macroblock column 51 sequentially. Inheriting the basic structure of FIG. 4, the macroblock 52 shown in this embodiment can refer to a luminance macroblock (luminancemacro-block), and if the horizontal and vertical chrominance data are at a sampling frequency equivalent to luminance Half of the sampling frequency sampling, then two chrominance blocks (respectively Cr and Cb) 53 and 55 corresponding to a macroblock 52 with a size of 8×8 pixels will be processed first, and then will be decoded A macroblock 52 with a size of 16×16 pixels.

Please refer to FIG. 7 . FIG. 7 is a schematic diagram of an embodiment of the present invention, showing how the present invention utilizes a storage device for predictor access to reduce the required capacity of the related storage device. Similar to the definition in the known embodiment of FIG. 4, the macroblock 52 to be decoded in this embodiment includes a first block X", a second block Y", and a third block X". ', and a fourth block Y"'. Surrounding the four blocks are a reference block (reference block) B', a first adjacent block A' (adiacent block), a second adjacent block C', a third adjacent block C", and a fourth adjacent block A". Please note that in FIG. 7, a storage device (memory device) 56 is shown around the macroblock 52 to be decoded, and the storage device 56 includes a storage row 56C (storing column) and a storage row 56R (storing column). row) is used to store the predictors generated during the predictive decoding process. For convenience of description, each storage unit (memory cell) in the storage device 56, such as each storage row 56C and storage row 56R in FIG. 7 A block can be used to store a predictor. According to the predictive decoding rules of the motion image compression standard MPEG-4, when processing the block to be decoded, the predictors of its adjacent blocks are used to obtain the data of the block to be decoded. Please refer to FIG. 7, when processing the first block X", the system needs to provide a predictor (DC _B' ) of the reference block B' located on the upper left of it, and the first adjacent block A' located to the left of it. The eight predictors of (DC _A' and AC _A' [1-7][0]), and the eight predictors of the second adjacent block C' above it (DC _C' and AC _C' [0 ][1-7]).

In the present invention, we divide the predictors into three categories according to their spatial positions: vertical predictors, horizontal predictors, and diagonal predictors. For example, the eight predictors (including AC _A' [1-7][0] and DC _A' ) located in the same row in the first adjacent block A' are all regarded as vertical predictors, and the second adjacent block C' The eight predictors (including AC _C' [0][1-7] and DC _C' ) located in the same column are all regarded as horizontal predictors, and the predictor DC _{B' located in the lower left corner of the reference block B'} is treated as a pair of diagonal predictors. In this embodiment, the upper-left predictor in each block is defined as a pair of corner predictors required by another block. For example, the predictor DC _{B' of the reference block B'} is A diagonal predictor of the first block X" and a predictor DC _{C' of a second adjacent block C'} is a diagonal predictor of the second block Y"; in other words, a A horizontal predictor can also be considered as a diagonal predictor of another block, and/or a vertical predictor of one block can also be considered as a diagonal predictor of another block. The eight memory cells storing column 56R (defined as 56R[0-7]) will store the eight horizontal predictors of the second contiguous block C', while the eight memory cells storing row 56C (defined as 56C[0-7] 7]) will store eight vertical predictors of the first adjacent block A', and the storage device 56 has another storage unit 56D reserved for storing the diagonal predictor DC B' of the reference block B _' .

Please continue to refer to Figure 7. After the system determines the correct predictor source based on DC _B' , DC _C' and DC _A' , it combines the remaining AC coefficients with the relevant differential values (about the operation of combining with the differential values The program can refer to the known embodiment of FIG. 2 ) to obtain the pixel coefficient (pixelcoefficient) of the first block X ". Then, in order to complete the predictive decoding of the remaining blocks in the macroblock 52, the first block X " The pixel coefficients in the leftmost row and the uppermost column are set as predictors and stored in the storage device 56 . In this embodiment, the newly determined (horizontal) predictors DC _X" and AC _X" [0][1-7] of the first block X" will replace the original storage row 56R[0-7] The eight horizontal predictors of the second adjacent block C' are stored in the storage column 56R[0-7]. Similarly, the newly determined (vertical) predictor DC _X of the first block X"_" and AC _X" [1-7][0] will replace the eight vertical predictors of the first contiguous block A' originally stored in row 56C[0-7] and store them in row 56C[0 -7] in. In this way, by using the permutation method in the storage device 56, there is no need to waste extra space for storing the predictors. After the decoding of the first block X" is completed, the second block Y" is ready to be processed. Since the first block X" and the third adjacent block C" are adjacent to the second block Y", it is first necessary to determine the correct source of the predictor based on DC _C' , DC _X" and DC _C" , and then use DC _X” and AC _X” [1-7][0] of the first block X” and DC _C” and AC _C” [0][1-7] of the third adjacent block C” Part of the pixel coefficients of the predictors of the second block Y″ are determined. Afterwards, the newly determined (horizontal) predictors DC _Y" and AC _Y" [0][1-7] of the second block Y" will replace the third one previously stored in the storage row 56R[8-15]. Eight horizontal predictors of adjacent block C" are stored in row 56R[8-15]. The newly determined (vertical) predictors DC _Y" and AC _Y" [1-7][0] of the second block Y" will be replaced and stored in the storage line 56C[0-7] in the previous operation. The DC _X" and AC _X" [1-7][0] of the first block X" are stored in the storage row 56C[0-7].

Similarly, DC A' in the first adjacent block A _' , DC X" and AC _{X " [0][1-7] in the first block X"} and DC A in the fourth adjacent block A _" After _" and AC _A" [1-7][0] determine some pixel coefficients of the third block X"', the predictor DC _X"' and the predictor AC of the third block X"' will be determined _X"' [0][1-7] and predictor AC _X"' [1-7][0], and then the newly determined predictor DC _X"' and AC _X"' [1-7][ 0] will replace the eight vertical predictors of the fourth adjacent block A" originally stored in row 56C[8-15] and be stored in row 56C[8-15]. Then, DC X" of the first block X _" , DC Y" and AC _Y" of the second block Y _" [0][1-7], and DC _{X"' of the third block X"'} and AC _X"' [1-7][0] will determine some pixel coefficients of the fourth block Y"', and then determine the predictor DC _Y"' and predictor of the fourth block Y"' AC _Y”’ [0][1-7] and predictor AC _Y”’ [1-7][0], and then, the newly determined predictor DC _Y”’ and AC _Y”’ [1-7 ][0] will replace the DC _X "' and AC _X"' [1-7][0] of the third block X"' stored in row 56C[8-15] in the previous operation, And store it in storage line 56C[8-15].

In this embodiment, the vertical predictor, the horizontal predictor and the diagonal predictor are stored in the storage row 56C, the storage column 56R and an additional storage unit 56D of the storage device 56, and the replacement method is used in the storage device 56 to save storage device 56 space. In actual implementation, the storage device 56 may be a processing register, or may be a common register if hardware performance permits. In this embodiment, a storage device 56 is used to access multiple predictors in a video frame 50, and the method for performing predictive decoding can be summarized in the following steps, and please refer to FIG. 8 , which is a detailed method of the present invention. The flow chart of embodiment:

Step 200: start;

Step 201: When processing a first block X" in a macroblock 52, refer to a plurality of adjacent blocks to generate a plurality of predictors (including a plurality of horizontal predictors, a plurality of vertical predictors, and a pair of diagonal predictors), and then proceed to step 202; please refer to FIG. There is a reference block B' located on the upper left of the first block X", a first adjacent block A' located on the left of the first block X", and a reference block A' located above the first block X". The second adjacent block C';

Step 202: Store the plurality of vertical predictors in row 56C[0-7], store the plurality of horizontal predictors in row 56R[0-7], and store a pair of diagonal predictors in the aforementioned The storage unit 56D then proceeds to step 203; and the plurality of horizontal predictors of the newly determined first block X" will replace the second adjacent block C originally stored in the storage row 56R[0-7] ', the newly determined vertical predictors of the first block X" will replace those of the second adjacent block C' originally stored in the storage row 56C[0-7] The plurality of vertical predictors, and the diagonal predictor DC _C ' required by the second block Y" will replace the diagonal predictor DC _B required by the first block X" originally stored in the storage unit 56D _' ;

Step 203: When processing a second block Y" in the macroblock 52, refer to a plurality of adjacent blocks to generate multiple predictors (including multiple horizontal predictors, multiple predictors) required by the second block Y " vertical predictors, and a pair of diagonal predictors), then go to step 204. Please refer to FIG. 7, wherein the second block Y" is located at the upper right of the macro block 52, and the aforesaid plurality of adjacent blocks include the second adjacent block C', the second adjacent block located at the upper left of the second block Y", A first block X" located to the left of the second block Y" and a third adjacent block C" located above the second block Y".

Step 204: Store the plurality of vertical predictors (8 in total) of the second block Y" in the storage row 56C[0-7], and store the plurality of horizontal predictors (8 in total) in the storage row 56R [8-15], and store a pair of corner predictors in the storage unit 56D, and then proceed to step 205. The newly determined horizontal predictor of the second block Y" will replace the original storage row 56R The plurality of horizontal predictors of the third adjacent block C" in [8-15], the plurality of vertical predictors of the newly determined second block Y" will replace the original storage in step 202 to the storage The plurality of vertical predictors (8 in total) of the first block X" in row 56C[0-7], and the diagonal predictor DC _{A' required by the third block X"'} will replace the original The diagonal predictor DC _C' required for the second block Y" stored in the storage unit 56D;

Step 205: When processing a third block X"' in the macroblock 52, refer to a plurality of adjacent blocks to generate multiple predictors (including multiple vertical predictors) required by the third block X"' , and a pair of diagonal predictors), then go to step 206. Please refer to FIG. 7, wherein the third block X"' is located at the lower left of the macro block 52, and the plurality of adjacent blocks include a first adjacent block A' located at the upper left of the third block X"' , the fourth adjacent block A" located on the left of the third block X"', and the first block X" located above the third block X"';

Step 206 : Store the plurality of vertical predictors in storage row 56C[8-15], and store a pair of diagonal predictors in storage unit 56D, then proceed to step 207 . The plurality of vertical predictors of the newly determined third block X"' will replace the plurality of vertical predictors of the fourth adjacent block A" originally stored in the storage row 56C[8-15], The diagonal predictor DC _X " required by the fourth block Y"' will replace the diagonal predictor DC _A ' required by the third block X"' originally stored in the storage unit 56D;

Step 207: When processing a fourth block Y"' in the macroblock 52, refer to a plurality of adjacent blocks to generate multiple predictors (including multiple vertical predictors) required by the fourth block Y"' , and a pair of diagonal predictors), then proceed to step 208; please refer to FIG. The first block X" on the upper left of the block Y"', the third block X"' on the left of the fourth block Y"', and the second block on the upper left of the fourth block Y"' block Y";

Step 208 : Store the plurality of vertical predictors in the storage row 56C[8-15], and store a pair of diagonal predictors in the storage unit 56D, then go to step 209 . The plurality of vertical predictors of the newly determined fourth block Y"' will replace the plurality of vertical predictors of the third block X"' originally stored in the storage row 56C[8-15], The diagonal predictor DC _C required for a first block of the next macroblock will replace the diagonal predictor DC _X required for the fourth block Y"' originally stored in the storage unit 56D;

Step 209: Perform prediction and decoding procedures on a Cb chrominance block of 8×8 pixels and a Cr chrominance block of 8×8 pixels; and

Step 210: End the prediction operation and decoding procedure in this macroblock 52, and then process the next macroblock.

Please refer to the flow chart in FIG. 8 and refer back to the spatial configuration diagram in FIG. 6 , you can clearly see the default sequence (as shown by the arrow MR2 ) when processing each macroblock 52 in the embodiment of the present invention. Therefore, when processing each macroblock 52, four blocks are processed sequentially according to a zigzag direction, and after processing this (16×16 pixel size) luminance macroblock 52, then process the other The corresponding two chrominance blocks (respectively Cr and Cb) with a size of 8×8 pixels. After processing a macroblock 52, continue to process the next macroblock 52 in the same macroblock row, and process the macroblocks in the macroblock row sequentially from left to right (as shown by arrow MR1). 52 per macroblock. As can be seen from the above, the processing sequence of macroblocks is carried out in a macroblock column 51 from left to right, so when the system finishes processing a macroblock 52, it does not need to immediately process the macroblock located in the just processed Another macroblock 52 below the macroblock 52, therefore, there is no need to immediately associate the horizontal predictors (DC _X and AC _X [0][1-7]) of the third block X"' with the fourth block The horizontal predictors (DC _Y and AC _Y [0][1-7]) of Y"' are stored in the storage device 56. This is also because the above-mentioned multiple horizontal predictors are located in the macro area that has just been processed. The processing of a plurality of macroblocks 52 below the block 52 is relatively irrelevant and helpful, and itself is still waiting to be processed, so this is why in the steps of the flow chart of FIG. The newly determined horizontal predictor of the third block X"' replaces the horizontal predictor of the first block X" originally stored in the storage row 56R[0-7] and/or "the newly determined horizontal predictor of the first block X" The horizontal predictor of the fourth block Y"' replaces the horizontal predictor " of the second block Y" originally stored in the storage row 56R[8-15] and other operation steps. In another embodiment, in addition to the aforementioned storage device 56, a second storage device 58 is also planned in Fig. 9, and the second storage device 58 can be a dynamic random access memory (DRAM), a static random access memory Memory (SRAM), or a register to store the horizontal predictor of the third block (lower left) and the horizontal predictor of the fourth block (lower right) in each macroblock 52, After all the macroblocks 52 in the entire macroblock row are processed, the predictive element stored in the second storage device 58 can be used to perform predictive decoding on the macroblocks 52 in the next macroblock row. See also Fig. 9, Fig. 9 is the flow chart of the relevant operation of adding the second storage device 58 again in the embodiment shown in Fig. 8:

Step 211: In step 206 of FIG. 8 , store a plurality of horizontal predictors of the third block X"' in the second storage device 58;

Step 212: In step 208 of FIG. 8 , store a plurality of horizontal predictors of the fourth block Y"' in the second storage device 58 .

According to the embodiment disclosed in FIG. 7 , we can store the respective contents in the storage row 56C[0-7], the storage row 56C[8-15], the storage row 56R[0-7] and the storage row 56R[8-15] The replaced cases are tabulated in Figure 10. In the aforementioned prediction operation and decoding process, since the present invention divides the storage device into a storage row and a storage column, so as to respectively access and replace multiple vertical predictors, multiple horizontal predictors and diagonal predictors stored therein Therefore, even if it is necessary to continuously determine the predictors for the macroblocks to be decoded, the present invention can completely achieve the function of predictive decoding with a storage space far less than the number of actually generated predictors. There is no need to use a large amount of storage space to separately store all generated predictors as in the prior art. In addition, compared with the known embodiment, when a macroblock is predictively decoded, the present invention only needs a storage device with 33 storage units (the calculation method of 33 storage units is as shown in FIG. 7 The embodiment is an example: storage row 56C[0-7], storage row 56C[8-15], storage row 56R[0-7] and storage row 56R[8-15] consume 4*8=32 (storage unit) ), plus 1 storage unit required to store a diagonal predictor), and even if the capacity of the above-mentioned second storage device is calculated, a macroblock only needs 16 additional storage units.

In the present invention, the required space of the storage device 56 required in the system is greatly reduced. Taking a whole (such as 720×480 pixel size) video picture as an example, the capacity of the saved storage space is even more surprising. , In this way, not only the use space capacity requirement of the storage device 56 implemented with the operation buffer is reduced, but the second storage device 58 implemented with the dynamic random access memory also only needs to be occupied with a small capacity to be able to The storage device 56 (and the second storage device 58) shown in FIG. 7 can be integrated into the hardware system in an on-chip method, so as to reduce costs and accelerate integration of related components with a single chip. Trends in all hardware components of video codecs (video CODECs).

The above descriptions are only preferred embodiments of the present invention, and all equivalent changes and modifications made according to the claims of the present invention shall fall within the scope of the present invention.

Claims (13)

1. A predictive decoding method for decoding a video frame to generate a plurality of predictors corresponding to a plurality of blocks of the video frame, the predictive decoding method comprising: (a) storing a plurality of first vertical predictors of a first block in a storage row of a first memory device, and storing a plurality of first horizontal predictors of a second block in the first storage device a memory bank of the device; (b) performing a prediction operation to generate a plurality of target vertical predictors and a plurality of target horizontal predictors of a first target block according to the plurality of first vertical predictors and the plurality of first horizontal predictors, wherein the first target block is adjacent to the first and second blocks, and the first block and the first target block are located in the same row; and (c) updating the storage row of the first storage device with the plurality of target vertical predictors, and updating the storage column of the first storage device with the plurality of target horizontal predictors. 2. The predictive decoding method according to claim 1, wherein the first and second blocks and the first target block are located in different macroblocks, and each macroblock includes a plurality of blocks , the first target block is not located in the lowest column of a corresponding macroblock. 3. The predictive decoding method according to claim 2, wherein step (a) further comprises storing a pair of diagonal predictors corresponding to the first target block into a storage unit of the first storage device, step (b) generating the plurality of target horizontal predictors and the plurality of target vertical predictors according to the diagonal predictor, the plurality of first vertical predictors, and the plurality of first horizontal predictors, and step (c) It also includes updating the storage unit of the first storage device with a pair of diagonal predictors corresponding to a second target block, wherein the second target block is processed later than the first target block, and the second target block is processed later than the first target block. The first and the second target block are located in the same macroblock. 4. The predictive decoding method as claimed in claim 1, wherein the first block and the first target block are located in a macroblock comprising a plurality of blocks, and the second block is located in another macroblock comprising In a macroblock with multiple blocks, the first target block is not located in the lowest column of the corresponding macroblock, and the plurality of first vertical predictors are stored in a first part of the storage row, The plurality of first horizontal predictors are stored in a first portion of the storage row, and step (c) uses the plurality of target vertical predictors to update the first portion of the storage row, and uses the plurality of target vertical predictors to update the first portion of the storage row. target level predictor to update the first portion of the row. 5. The predictive decoding method as claimed in claim 4, wherein step (a) further comprises storing a pair of diagonal predictors corresponding to the first target block into a storage unit of the first storage device, step (b) generating the plurality of target horizontal predictors and the plurality of target vertical predictors according to the diagonal predictor, the plurality of first vertical predictors, and the plurality of first horizontal predictors, and step (c) It also includes updating the storage unit of the first storage device with a pair of diagonal predictors corresponding to a second target block, wherein the second target block is processed later than the first target block, and the second target block is processed later than the first target block. The first and the second target block are located in the same macroblock. 6. The predictive decoding method as claimed in claim 1, wherein the first block is located in a macroblock comprising a plurality of blocks, and the second block and the first target block are located in another In a macroblock comprising a plurality of blocks, the target block is located at the lowest column of the corresponding macroblock, and step (c) uses the plurality of target vertical predictors to update the storage row without updating The storage column. 7. The predictive decoding method as claimed in claim 6, wherein step (a) further comprises storing a pair of diagonal predictors corresponding to the first target block into a storage unit of the first storage device, step (b) generating the plurality of target horizontal predictors and the plurality of target vertical predictors according to the diagonal predictor, the plurality of first vertical predictors, and the plurality of first horizontal predictors, and step (c) It also includes updating the storage unit of the first storage device with a pair of diagonal predictors corresponding to a second target block, wherein the second target block is processed later than the first target block, and the second target block is processed later than the first target block. The first and the second target block are located in the same macroblock. 8. The predictive decoding method as claimed in claim 6, further comprising storing the plurality of target level predictors in a second storage device. 9. The predictive decoding method as claimed in claim 1, wherein the first, the second block and the first target block are located in a macroblock comprising a plurality of blocks, the first target area The block is located at the lowest row of the macroblock, and step (c) updates the storage row with the plurality of target vertical predictors without updating the storage row. 10. The predictive decoding method as claimed in claim 9, wherein step (a) further comprises storing a pair of diagonal predictors corresponding to the first target block into a storage unit of the first storage device, step (b) generating the plurality of target horizontal predictors and the plurality of target vertical predictors according to the diagonal predictor, the plurality of first vertical predictors, and the plurality of first horizontal predictors, and step (c) Also includes updating the storage unit of the first storage device with a pair of diagonal predictors corresponding to a second target block, wherein the second target block is processed later than the first target block, and The first and second target blocks are not located in the same macroblock. 11. The predictive decoding method as claimed in claim 9, further comprising storing the plurality of target level predictors in a second storage device. 12. The predictive decoding method as claimed in claim 1, wherein the video frame conforms to the specification of a motion image compression standard MPEG. 13. The predictive decoding method according to claim 12, wherein the plurality of vertical predictors of a block are located at the leftmost row of the block, and the plurality of horizontal predictors of the block are located at the top of the block row, the plurality of horizontal predictors and the plurality of vertical predictors of the block respectively include a DC coefficient and a plurality of AC coefficients.

Images