摘要
最近几十年来,数字图像编解码技术的研究一直在不断前进,各种编解码技术不断成熟。尤其在最近几年中,各个领域涌现出一批新一代视频编码标准,结束了MPEG-2一统天下的局面。在数字电视,IPTV,视频会议,多媒体通信和网络监控等诸多领域中,视频编解码标准都起着核心的作用。
由中国音视频编码标准工作组组织开发的AVS-M视频编码标准在取得良好图像质量的同时,保证了较高的压缩率和较低的编码复杂度(理论复杂度为AVS视频编码标准的70%),它引入了一系列先进的视频编码技术,并且提供了差错恢复能力和网络适应性。对AVS-M编解码技术的研究在很多移动多媒体应用领域都具有实际价值。
高效快速的编码器实现对AVS-M标准的应用和推广具有极其重要的意义。本文首先对国际音视频编码标准的发展和视频编码的主要技术做了介绍,对AVS-M视频编解码标准的关键技术做了详细分析,最后在IA-32体系结构上实现对AVS-M参考编码器的优化。
对AVS-M参考编码器的优化工作包括对编码器流程的调整、数据结构的调整以及对关键代码的SIMD指令优化。其中,对编码器流程的调整的是对冗余计算的消除,包括帧内预测、运动估计、离散余弦变换、反离散余弦变换、量化、反量化和率失真优化等模块的流程的调整;数据结构的调整包括为减少存储冗余而对控制数据进行的调整和为了SIMD指令的优化而对运算数据结构的调整;关键代码的SIMD指令重写则是编码器加速的关键因素,本文对AVS-M编码器的运动搜索、DCT、IDCT、量化、反量化、环路滤波等关键模块均以纯汇编或内嵌汇编的形式进行了重写,均取得了不错的效果。优化后的参考编码器在图像质量和码率损失很小的情况下,运行速度获得了很大提高。
In the past tens of years, study on the digital video coding technology has never ceased. Various techniques in video coding and decoding become applicable. Especially in the latest several years, a lot of new video compression standards were established for different purposes, which finished the era of MPEG-2. Video coding standard plays a key role in many fields, such as digital TV, IPTV, video conference and multimedia communication, etc.
AVS-M video coding standard is developed by the Audio Video Coding Standard Working Group of China. AVS-M standard provides good trades-off between compression rate and computation complexity (about 70% of AVS video coding standard). It introduced a number of advanced video coding techniques, and offered error resilience and net adaptation ability. Research on AVS-M coding technology offers great importance in developing multi-media applications on mobile platform.
A fast implementation of encoder is of great importance for the practical use of AVS-M standard. This paper briefly introduced the development of the audio/video coding standards in the world and the major coding technologies used in these standards, and then studied the key technologies in the AVS-M video compression standard. Finally the AVS-M reference encoder was optimized on IA-32 architecture.
The optimization is based on the AVS-M reference encoder. The major optimization strategy includes adjustment of the program and data structure, and rewriting of the most time-consuming functions with Intel SIMD instructions. Adjustment of the program structure includes the following parts: intra prediction, motion estimation, DCT, IDCT, quantization, de-quantization and rate distortion optimization etc; Adjustment of the data structure includes the program control data and computation data structure. They were adjusted to avoid redundant data calculation and to use SIMD instructions. After rewritten these functions, the encoder is accelerated with no descendant of picture quality and bit-rate.
引文
1王新城.视频编码器的昨天和今天.电视技术. 1997,(9):26~31
2黄铁军,高文. AVS标准制定背景与知识产权状况.电视技术. 2005,(7):4~8
3 Claude. E. Shannon. A Mathematical Theory of Communication. Bell System Technical Journal, July, 1948,(27):379~423
4张晓芸,宋建社,郑永安.视频编码技术研究进展.计算机工程. 2004, 30(14):4~5
5 Yao Wang, Jorn Ostermann, Ya-Qin Zhang. Video Processing and Communication. Beijing: Tsinghua University Press,2003:217~223
6 H. G. Mussman, M. Heter, Jorn Ostermann. Object-oriented Analysis-synthesis Coding of Moving Objects. Signal Processing: Image communication, 1989, 1:117~138
7 Markus Kampmann, Jorn Ostermann. Automatic Adaptation of A Face Model in A Layered Coder with An Object-based Analysis-synthesis Layer and A Knowledge-based Layer. Signal Processing: Image Communication, 1997,9:201~220
8 D. Terzopoulos, K. Waters. Physically-based Facial Modeling, Analysis, and Animation. Journal of Visualization and Computer Animation, 1990,1:73~80
9张乐平,杨少华,吴乐南.基于霍夫曼编码的SAR浮点图像数据压缩.江西理工大学学报. 2007, 28(1):28~31
10 Memon N, Sayood K. Lossless image compression: A comparative study. In:Proc of SPIE, Vol 2814. San Jose, California: SPIE Optical Engineering Publisher, 1995. 8~20
11 ISO/IEC 11172(MPEG1 DIS). Coding of Moving Picture and Associated Audio for Digital Storage Media at Up to About 1.5Mbps. 1991, 11:12~15
12 T. Koya, K. Iinuma, A Hirano, Y. Iiyima, T. Ishi-guro. Motion-compensated Inter-frame Coding for Video Conferencing. In Proc. NTC81, New Orleans, LA, 1981, 11: C9.6.1~9.6.5
13 S. Kappagantula, K. R. Rao. Motion Compensated Inter-frame Image Prediction. IEEE trans. Commun, Sept 1985, COM-33:1011~1015
14 J. R. Jain, A. K. Jain. Displacement Measurement and Its Application in Inter-frame Image Coding. IEEE Trans. Commun., Dec 1981, COM-29:1799~1808
15 Ghanbari. The Cross-search Algorithm for Motion Estimation. IEEE Trans. Commun. July 1990, COM-38:950~953
16倪伟,郭宝龙,陈龙潭,冯宗哲. H.264变换编码和量化算法的研究.计算机工程与应用. 2004,(3):33~36
17张晓燕,谢珺堂. H.264的整数DCT变换编码与量化过程.军民两用技术与产品. 2005,(5): 40~45
18刘波,王荣国,洪雨,赵明慧,刘丽娟.各种视频编码标准简介.有线电视技术. 2006,(10):41~43
19 T. Wiegand, G. J. Sullivan, G. Bjontegaard, A. Lithia. Overview of the H.264/AVC Video Coding Standard. IEEE Trans. Circuits Syst. Video Technol, 2003,13: 560~576
20 Ralf Schafer, Thomas Wiegand, Heiko Schwarz. The Emerging h.264/AVC Standard. EBU Technical Review, 2003,1:30~35
21倪伟,郭宝龙,等.一种适用于H.264/AVC的自适应空域帧内预测算法.计算机学报, 2005,28(11):1853~1860
22 M. Chan, Y. Yu, A. Constantinides. Variable Size Block Matching Motion Compensation with Applications to Video Coding. IEEE Proceedings. 1990, 137(4):205~212
23 J Jain, A Jain. Displacement Measurement and Its Application in Inter-frame Image Coding. IEEE Trans. Commun. 1981,29:1799~1808
24 Pao I-Ming, Sun Ming-Ting. Modeling DCT Coefficients for Fast Video Encoding. IEEE Trans CSVT, 1999,9(4):608~616
25 Luo Ying,Ward R K.Removing the Blocking Artifacts of Block-based DCT Compressed Images.IEEE Trans Image Processing,2003, 12(7):838~842
26 Zeng B.Reduction of Blocking Effect in DCT-coded Images Using Zero-Masking Techniques. Signal Process. 1999,79(2):205~211
27 Kris Kaspersky著.谭明金译.代码优化:有效使用内存.电子工业出版社. 2004:292~313
28王文生,杨铭,崔慧娟,唐昆. H.264运动估计的自适应提前退出算法.清华大学学报(自然科学版). 2004,44(10):1427~1429
29 Yu Alice, Lee R. Flynn M. Early Detection of All-zero Coefficients in H.263. PCS’97. Berlin, Germany, Sep,1997:159~164
30周璇,谭径微,余松煜. H.263中预先判别全零系数的新方法.上海交通大学学报, 1998,32(9):107~109
31 B. J. Meng, C. A. Oscar, C. W. Wong, et al. Efficient Intra-prediction Algorithm in H·264[A].IEEE International Conference on Multimedia & Expo. Maryland, 2003,Ⅲ:837~840
32赵晓燕,李金良.一种用于H.264的快速帧内预测模式判别方法.计算机工程与设计. 2005,26(4):1031~1035
33王维东,姚庆栋,刘鹏.小数像素运动估计快速算法.通信学报. 2003, 24(4):128~132
34徐天成.基2FFT的二进制理论分析.南京理工大学学报. 2000,(2): 160~163
35方延龙,周军. H.264分像素内插的快速算法.计算机工程. 2006,32(1):218~223
36 Agner Fog. Optimizing Subroutines in Assembly Language—an Optimization Guide for X86 Platform. www.agner.org. Aug 2006: 11~22
37张睿,彭宇行. MPEG-4中运动估计模块的改进与优化.计算机与信息技术. 2006,(7):16~17
38 Bing Zeng, Renxiang Li, Ming L Liou. Optimization of Fast Motion Estimation Algorithm. IEEE Trans. Circuit Syst. Video Technol, 1997,7(6):833~844
39杨育红,徐烜,季晓勇.快速运动估计UMHexagonS算法的探讨与改进.计算机工程与应用. 2006,(11):52~54
40 Shan Zhu, Kai Kuang Ma. A New Diamond Search Algorithm for Fast Block Matching Motion Estimation. Proc of Int Conf Information, Communications and Signal Processing. 1997: 292~296
41 Tourapis Alexis M, Au Oscar C, Liou Ming L. Highly Efficient Predictive Zonal Algorithms for Fast Block-matching Motion Estimation. IEEE Trans CSVT, 2002,12(10):934~947
42严明,胡国荣. AVS视频标准中的分像素插值算法设计.中国传媒大学学报自然科学版. 2006,13(4):44~48
43孙小叶,刘济林,陈建乐.一种低码率下简单有效的环路滤波方法.浙江大学学报(工学版). 2005,39(7):932~934
44 Liu Shizhong, A C Bovik. Efficient DCT-Domain Blind Measurement and Reduction of Blocking Artifacts. IEEE Trans Circuits Syst Video Technol, 2002, 12(12):1139~1149