TWI324484B - Method for reducing calculation complexity of video compression standard - Google Patents

Description

1324484 IX. Description of the Invention: [Technical Field] The present invention relates to an algorithm for reducing computational complexity in a video compression standard, and more particularly to a reduction calculation based on spatial correlation in a video compression standard The algorithm of complexity. [Prior Art] H.264/AVC is the Video Coding Experts Group (VCEG) and the International Organization for Standardization (ISO)/International Electrotechnical Commission (IEC) in the International Telecommunication Union Telecommunication Standardization Group (ITU-T) The video compression standard developed by the Joint Video Team (JVT), which is composed of Moving Picture Experts Group (MPEG). Among them, H.264 is based on ITU-T's H.26x series, and AVC is the name of ISO/IEC MPEG. The first draft of this standard, the Final Draft, was completed in May 2003. H.; 264/AVC's initial main goal is to provide higher compression ratios than previous related technologies (such as MPEG-2 or H.263) under approximate video quality. At the same time, there is no need to add a lot of complexity. The codec tool makes the hardware difficult to implement. Another goal is adaptability, that is, the video compression standard can be used in a wide range (including high bit rate and low bit rate video, and different video resolution), and can be used on various networks and systems. Work (such as multimedia mobile phone systems or high definition TV). H.264/AVC internal coding is divided into several important parts: Intra Prediction, conversion and quantization (DCT/Q/IQ/IDCT), context 6 1324484 Adaptive variable length coding (Context Adaptive Variable Length)

Coding, CAVLC), Mode Decision, etc. For I-Frame coding, the internal prediction and mode decision are about 70% of the internal code. This is because internal predictions must produce 13 different brightness predictions and must also produce 4 different chroma predictions. In order to calculate the minimum distortion (Rate-Distortion Cost), each predicted value must be subtracted from the original image, and then subjected to a 2-dimensional Hadamard conversion to accumulate the converted coefficients to determine the optimal prediction mode. Reducing the amount of computation in the prediction mode will greatly improve the overall system performance, while reducing the prediction mode will also affect the quality of the image. The present invention proposes an effective algorithm, which can reduce a large amount of computational complexity only under the condition of slight image distortion through an accurate prediction mode. SUMMARY OF THE INVENTION To solve the above and other problems 'and to achieve the technical advantages claimed by the present invention, the present invention provides an algorithm for fast encoding in a video compression standard. Therefore, the object of the present invention is to provide a video compression standard. An algorithm for predicting the process of the Macroblock prediction process to reduce the amount of Luminance calculations due to video compression per unit time. Another object of the present invention is to provide an acceleration color in a video compression standard. Chroma distortion prediction process algorithm to reduce the single 7

Claims (1)

1324484 X. Patent application scope: 1. An algorithm for reducing the computational complexity in the video compression standard, comprising the following steps: Performing an internal 4x4 macroblock search algorithm to accelerate the luminance prediction process of the macroblock' The 4x4 macroblock search algorithm includes a state-related search mode, which includes the following search conditions: If one of the upper and the left blocks of an image is not available, only mode 2 is performed. (DC mode) calculation; if only the upper block is available, perform mode 〇, mode 2, mode 3 and mode 7 calculation; if only the left block is available, mode 丨, mode 2 and mode 8 calculation; and if both the upper block and the left block are available, a half domain search mode and a context search mode are performed. 2. The algorithm for reducing the computational complexity in the video compression standard as described in claim 1 of the patent scope, the half domain search mode includes only the execution mode when the upper block is calculated in mode 2 (DC mode) 〇, mode i, mode 2, mode 3 and mode 4. 3. The algorithm for reducing the computational complexity in the video compression standard as described in claim 1, wherein the half-domain search mode is included in the left block 15 when the mode 2 (DC material) is used for mode 2 Mode 3 and Mode 4. Execute only mode 0, mode ▲ 4. The reduced video > 1 standardization command described in item 1 of the towel 4 patent fan®: the algorithm for calculating the complexity, the half domain search mode is included in the upper block and the left side (four). (DC mode) when calculating, 〇 'mode i, mode 2, mode 3 and mode 4β, ▲ 5. The algorithm for reducing the complexity of the juice compression standard in the video compression standard as described in the patent application _ , The contextual correlation search mode includes when the upper block and the left block are not calculated by the mode 2 (DC mode), and the spatial orientation similarity prediction between the upper block and the left block is calculated. Mode / / _ _ patent scope of the first to reduce the computational complexity of the video compression algorithm, including the implementation - internal 16x16 macro block search algorithm to accelerate the brightness prediction process of the macro block, The internal 16x16 macroblock search algorithm includes calculating the absolute conversion difference value (SATD) of the 16 macroblocks of the image to obtain the distortion of each macroblock, and calculating the absolute conversion. The sum of differences (SATD) includes the following steps: Get each of the macros 16 conversion remainders of the block (Transf() med Residues); Hadamard conversion for each macro block; and ^month) 3⁄4 correction replacement page absolute value of the output value after the cumulative conversion to calculate 16 macroblocks The sum of the absolute conversion differences (SATD). 7. The algorithm for reducing the computational complexity in the video compression standard as described in claim 1 of the patent application, further comprising performing a chroma search algorithm to accelerate the chroma prediction process of the macro block, the chroma search algorithm The method includes calculating the distortion of the -U macroblock and the distortion of the -v macroblock, and then adding the two to represent the distortion of the internal chroma macroblock; wherein, calculating the U macroblock The distortion is calculated only by the distortion of one of the top leftmost blocks of the image to represent the distortion of the macroblock; and the distortion of the V-mega block is calculated only for the shadow | top left The distortion of one of the blocks represents the degree of distortion of the V macroblock. 8. The algorithm for reducing computational complexity in video compression standards as described in clause 7. of the patent scope, wherein the ^^ macroblock contains four prediction modes. 9. The algorithm for reducing the computational complexity in the video compression standard according to claim 7 of the patent scope, wherein the 1] macroblock comprises 4 4χ4 blocks 0 10· as described in claim 7 The algorithm for reducing the computational complexity in the video compression standard 'where the V macro block contains 4 pre- 17 1324484 test modes. 11. The algorithm for reducing computational complexity in a video compression standard as described in claim 7 wherein the V-major block comprises four 4x4 blocks. 18 1324484 November ^^Correct this state The correlation block is not available in the upper block. The left block is not available. Mode 2 Mode 0, 2, 3, 7 Get 1, 2, 8 half-domain search and context correlation Search 12 13 Half Domain Search Table Upper Block (Available) Mode 2 (DC) Other Prediction Mode Left Block (Available) Mode 2 (DC) 0,1,2, 3,4 0,1,2, 3,4 Other prediction modes 0, 1, 2, 3, 4 Context-sensitive search 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 01234 246537 2465 24650 24651 2465703 246581 CO 2037 23705 237168 01234 237465 237540 237614 2370 | 23781 CNJ 2037 01234 01234 01234 01234 01234 01234 01234 01234 01234 ▼-Η 2037 2057168 2168 01234 237168 216845 2168540 21684 2168703 2168 CD 2037 2057 2057168 01234 20573 205746 20574 2057614 20573 205781 Not available &C<l oo 03 iH OO CNJ OO Cs3 OO oa OO CNJ rH OO oa i-H OO CNJ r~-J oo 03 PQ Not available cz> 03 CO LO CO Bu OO 凾Z-rich 1324484 31 32 4x4 blk 0 4x4 blk 1 4x4 blk 2 4x4 blk 3 4x4 blk 4 4x4 blk 5 4x4 blk 6 4x4 blk 7 4x4 blk 8 4x4 blk 9 4x4 blk 10 4x4 blk 11 4x4 blk 12 4x4 blk 13 4x4 blk 14 4x4 blk 15 trO trl tr2 tr3 tr4 tr5 tr6 tr7 tr8 Tr9 trlO trll trl2 trl3 trl4 trl5 4x4 block Luma 16x16 macro block 33 15 SATD 4X4 blk Σ tr 34 15 COST, 16MB =4X4 blk=0 Y^SATD 4X4 blk 1324484 41 4x4 blkO 4x4 blkl 4x4 blk2 4x4 blk3 U 8x8 macroblock 42 4x4 blkO 4x4 blkl 4x4 blk2 4x4 blk3 V 8x8 macro block 43 COST Chroma COST, + COSTv SATD + SATD 4x4 blkO 4x4 blkO