基于DSP的AVS实时解码器的研究与实现
摘要
AVS是我国自主研发的音视频标准,其中第二部分:视频(AVS1-P2)已获批成为国家标准。它有性能高,计算复杂度低,专利授权费用低等优点。但目前依然没有成熟的AVS编解码器解决方案,不利于AVS向新一代数字音视频产业的应用,因此AVS1-P2解码器实时播放的研究与实现,就成为AVS产业化的重要内容。论文研究的目的就是在DSP平台上对AVS1-P2解码器进行解码速度的优化,提高输出帧率,达到实时播放的要求。
首先,研究了AVS1-P2的主要编解码算法和DSP平台TMS320DM642的特点以及常用的DSP优化方法,为解码器的优化打下基础。
解码器解码速度的提高,关键是各核心解码模块的效率提高,因此论文的重点内容是针对解码器各功能模块的特点,提出核心模块专门的优化方案,以使解码器的解码速度提高。主要研究了以下模块:
亚象素插值模块,简化了四分之一样本插值算法,提出了共享样本复用的方法,在此基础上从采用DMA并行预取参考样本提高存取效率和采用内联指令提高计算效率两个方面来提高插值模块的效率。对该模块的优化使解码速度提高了7帧/秒,是各模块种优化效果最好的。
帧内预测、重建、环路滤波三模块,建立了三模块共用存储系统,以减少三个模块间的冗余存取操作,提高三模块的运行效率。为此打破了传统环路滤波的滤波方式,提出了一种新的环路滤波模式,创新的调整了环路滤波的滤波顺序以及解码器的结构。
哥伦布码解析模块,提出了按字(32位)读取比特流的方法,并简化了哥伦布码前缀和后缀解析的方法。
另外,对AVS解码程序还普遍采用了如下优化策略:Cache优化、数据传输管理(采用DMA)、内联指令、排循环流水等。
通过验证实现证明,采用以上优化方案后,AVS1-P2解码器可以达到实时解码播放的要求。
AVS (Audio and Video Coding Standard),the second part of which has been authorized, is the standard of video and audio developed by our country independently.It has the superiority of simple computational complexity and low patent licensing fees.At present,the application of AVS in the field of Video and Audio domain has encountered obstacle ,because of short of feasible encoder and decoder solutions, so the research on AVS1-P2 has become necessary to solve the problem. The purpose of the thesis is to improve the speed of frame to meet the request of real-time video displaying.
At first,the algorithms of AVS1-P2 is studied,the characteristics of TMS320DM642 is analyzed the the methods of DSP to optimize Video are researched,establishing the foundation for optimizing decoder.
The speed of every modules is the key factor in the process of optimizing the decoder.The improvement of the speed of decoder So the keystone is to provide solutions for the kernel specially with the purpose of improving the frame rate of decoder.
Sub-pixel motion interpolation filter module reusing shareable samples not only simplifies the 1/4 sample interpolation algorithm,but also improves interpolation filter efficiency through using DMA parallel prefetching from the reference samples to improve access efficiency and using intrinstics to enhance counting efficiency.The optimization for the sub-pixel interpolation filter module has increased 7 frame/second for the decoder,which is best among all the modules.
The establishment of shared memory system between Intra- Prediction ,Reconstruction, and In-loop filter has reduced the redundancy between the three functional modules and improved operating frequency. Different from the tradition,a new methods of adjusting the sequence of In-loop filter and architecture of decoder has come out innovatly.
The parsing golomb module advising to read a word (32bit) but not a byte simplifies the method of parse prefix and suffix of golobm code.
In addition,the following mehods are also applied to the program of the decoder:Cache optimizing,data transmission managing (adopt DMA), using intrinstics,and optimizing software pipelining.
The result of the test proves that AVS1-P2 could meet the requirements of teal-time decoding through the optimizing methods above.
首先,研究了AVS1-P2的主要编解码算法和DSP平台TMS320DM642的特点以及常用的DSP优化方法,为解码器的优化打下基础。
解码器解码速度的提高,关键是各核心解码模块的效率提高,因此论文的重点内容是针对解码器各功能模块的特点,提出核心模块专门的优化方案,以使解码器的解码速度提高。主要研究了以下模块:
亚象素插值模块,简化了四分之一样本插值算法,提出了共享样本复用的方法,在此基础上从采用DMA并行预取参考样本提高存取效率和采用内联指令提高计算效率两个方面来提高插值模块的效率。对该模块的优化使解码速度提高了7帧/秒,是各模块种优化效果最好的。
帧内预测、重建、环路滤波三模块,建立了三模块共用存储系统,以减少三个模块间的冗余存取操作,提高三模块的运行效率。为此打破了传统环路滤波的滤波方式,提出了一种新的环路滤波模式,创新的调整了环路滤波的滤波顺序以及解码器的结构。
哥伦布码解析模块,提出了按字(32位)读取比特流的方法,并简化了哥伦布码前缀和后缀解析的方法。
另外,对AVS解码程序还普遍采用了如下优化策略:Cache优化、数据传输管理(采用DMA)、内联指令、排循环流水等。
通过验证实现证明,采用以上优化方案后,AVS1-P2解码器可以达到实时解码播放的要求。
AVS (Audio and Video Coding Standard),the second part of which has been authorized, is the standard of video and audio developed by our country independently.It has the superiority of simple computational complexity and low patent licensing fees.At present,the application of AVS in the field of Video and Audio domain has encountered obstacle ,because of short of feasible encoder and decoder solutions, so the research on AVS1-P2 has become necessary to solve the problem. The purpose of the thesis is to improve the speed of frame to meet the request of real-time video displaying.
At first,the algorithms of AVS1-P2 is studied,the characteristics of TMS320DM642 is analyzed the the methods of DSP to optimize Video are researched,establishing the foundation for optimizing decoder.
The speed of every modules is the key factor in the process of optimizing the decoder.The improvement of the speed of decoder So the keystone is to provide solutions for the kernel specially with the purpose of improving the frame rate of decoder.
Sub-pixel motion interpolation filter module reusing shareable samples not only simplifies the 1/4 sample interpolation algorithm,but also improves interpolation filter efficiency through using DMA parallel prefetching from the reference samples to improve access efficiency and using intrinstics to enhance counting efficiency.The optimization for the sub-pixel interpolation filter module has increased 7 frame/second for the decoder,which is best among all the modules.
The establishment of shared memory system between Intra- Prediction ,Reconstruction, and In-loop filter has reduced the redundancy between the three functional modules and improved operating frequency. Different from the tradition,a new methods of adjusting the sequence of In-loop filter and architecture of decoder has come out innovatly.
The parsing golomb module advising to read a word (32bit) but not a byte simplifies the method of parse prefix and suffix of golobm code.
In addition,the following mehods are also applied to the program of the decoder:Cache optimizing,data transmission managing (adopt DMA), using intrinstics,and optimizing software pipelining.
The result of the test proves that AVS1-P2 could meet the requirements of teal-time decoding through the optimizing methods above.
引文
[1] GB/T 20090.2- 2006 信息技术先进音视频编码, 第2部分: 视频[S]. 2006.
[2]信息技术 先进音视频编码 第七部分:移动视频 ,2004
[3]高文 ,黄铁军 .信源编码标准 AVS 及其在数字电视中的应用.电视技术,2003(11)4-6
[4]高文 等. AVS数字音视频编解码标准.中兴通讯技术,2006年6月第12卷第3期
[5]虞 露 等 .AVS 视频的技术特征. 电视技术 , 2004 年第 7 期(总第 277 期)
[6]ISO/IEC 13818-2-2000,Generic Coding of Moving Pictures and Associated Audio Information Part 2:Vdeo.2000
[7] ZHANG Nan, YIN Bao- cai, KONG De- hui, et al. Spatial prediction based intra- coding. Proceedings of the 2004 IEEE International Conference on Multimedia and Expo(ICME'2004), Vol 1. Jun. 27- 30, 2004, Taibei,China. New York, NY,USA:IEEE, 2004:97- 100.
[8]Draft ITU-T recommendation and final draft international standard of joint video sepefication (ITU-T Rec . H.264/ISO/IEC . 14496-10 AVC),JVT of ISO/IEC MPEG and ITV-T VCEC ,JVT-C050 ,2003
[9]ZHANG Nan, YIN Bao- cai, KONG De- hui, et al. Spatial prediction based intra- coding. Proceedings of the 2004 IEEE InternationalConference on Multimedia and Expo(ICME'2004), Vol 1. Jun. 27- 30, 2004, Taibei,China. New York, NY,USA:IEEE, 2004:97- 100.
[10] Peng Zhang, Debin Zhao, Siwei Ma, Yan Lu, Wen Gao . Multiple Modes Intra-Prediction in Intra Coding . IEEE InternationalConference on Multimedia and Expo(ICME'2004),2004
[11] JI Xiang- yang, ZGAO De- bin, GAO Wen, et al. New scaling technique for direct modecoding in B pictures .IEEE InternationalConference on Image Proces s ing (ICIP 2004).Oct 24- 27, 2004, Singapore . Piscataway, NJ ,USA:IEEE, 2004:469- 472.
[12] JI Xiang- yang, ZHAO De- bin, GAO Wen, et al. New Bi- prediction techniques for B pictures coding.Proceedings of the 2004 IEEE International Conference on Multimedia and Expo (ICME'2004), Vol 1. Jun.27- 30,2004, Taibei, China. New York, NY,USA:IEEE,2004: 101- 104.
[13] WANG Rong- gang, HUANG Chao, LI J in- tao, et al. Sub- pixel motion compensation interpolation filter in AVS . Proceedings of the 2004 IEEE International Conference on Multimedia and Expo(ICME'2004), Vol 1. Jun. 27- 30, 2004, Taibei, China. New York, NY,USA:IEEE, 2004:93- 96.
[14] 王荣刚 等.一种分像素运动补偿插值滤波方法及高效 VLSI 实现. 计算机学报,2005,第 28 卷 第 12 期,2052-2058
[15] MA Si- wei, GAO Wen, FAN Xiao- peng. Low complexity integer trans form and high definition coding.Proceedings of SPIE 49thAnnual Meeting,Vol 58. Aug 02- 06, 2004,Denver, CO, USA. Bellingham, WA,USA:SPIEPres s , 2002:547- 554.
[16] Zhang Ci-xun,Lou Jian ,Yu lu ,et al .The technique of pre-scaled integer transform. IEEE international Conference on Multimedia and Expo ,2004.
[17] WANG Qiang, ZHAO De- bin, MA Si- wei, et al. Context- based 2D- VLC for video coding.Proceedings of the 2004 IEEEInternational Conference on Multimedia and Expo (ICME'2004), Vol 1. Jun.27- 30, 2004,Taibei, China.New York, NY,USA:IEEE, 2004: 89- 92.
[18]陈光法,姚立敏, 虞 露 .AVS熵解码与dsp实现 .电视技术,2004年第10期,总第238 期
[19] Reeve H, Lim J S. Reduction of blocking effect s in image coding[J] . Opt Eng, 1984 , 23(1) :34 - 73.
[20] lain E. G. Richardson Video Codec Design Developing Image and Video Compression Systems Copyright 0 2002 by John Wiley & Sons Ltd,Baffins Lane, Chichester,West Sussex P019 IUD, England The Robert Gordon University, Aberdeen, UK
[21]George A. Triantafyllidis, Dimitrios Tzovaras, and Michael Gerassimos Strintzis,“Blocking Artifact Detection and Reduction in Compressed Data”, IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, OCTOBER 2002.VOL. 12, NO. 10: 877-890
[22]S.Minami and A.Zakhor. An optimization approach for removing blocking effects in transform coding〔J〕. IEEE Transaction on Circuits and Systems for Video Technology,Apr.1995Vol.5:74-82
[23]颜金尧 ,宋宜纯.低码率视频压缩技术的块效应消除方法〔J〕.通信学报,2000 , 4 .Vol.21:42～47
[24]Mei-Yin Shen, Jong Won Kim, and C.-C.Jay Kuo. Nonlinear filtering based fast compression artifact reduction technique for H.263. ITU-Telecommunications Standardization Sector STUDY GROUP 16 Video Coding Group(Question15),Document:Q15-F-13, Generated:27 Oct.1998.
[25]P.List, A.Joch, J.Lainema,Cz Bj}ntegaard, and M.Karczewicz, "Adaptive deblocking filter,”IEEE Transactions on Circuits and Systems for Video Technology, Vol.l3, N0.7,July 2003.
[26] 黄铁军 . 我视频编码国家标准 AVS 与国际标准 MPEG 的比较 . 互连网
[27] 国家广播电视产品质量监督检验中心. AVS视频编/解码方案图像质量主观评价试验报告. 2004年12月
[28] 国家广播电影电视总局广播电视规划院. AVS视频压缩质量主观评价(AVS参考软件5.2版)测试报告. 2005年9月
[29]梁凡 . AVS视频标准的技术特点 . 电视技术, 2005 第7期(总第277期)
[30] Texas Instruments .TMS320DM642 Technical Overview . SPRU615 - September 2002
[31]Taeksang Hwang, Wonyong Sung. Implementation of a Digital Copier Using TMS320C6414 VLIW DSP Processor. IEEE, IC ASSP, 2003
[32] Texas Instruments .TMS320C6000 CPU and Instruction Set Reference Guide. Oct. 2000
[33] Texas Instruments .TMS320C6000 DSP Cache User's Guide. May 2003
[34]丁刚. 从 TMS320C6000 EDMA 的结构来理解和优化其应用.无线电工程,2002 , 32(7) :28一 32.
[35] TI. TMS320C6000 C/C++ Optimizing Compiler User's Guide. 2002,10.
[36]李方慧,王飞,何佩馄.TMS320C6000 系列 DSPs 原理与应用. 第 2 版.电子工业出版社,2003 年 1 月,
[37].许小东,徐佩霞.基于 TM S320DM642 的视频解码系统优化数据采集与处理.2005 20(1): 91--95
[38] 董 斌,姜昱明.AVS 软件解码器的优化 .计算机工程与设计,2006,第 27 卷 第 4 期,618-620
[39] TI .TMS320C64x DSPTwo-Level Internal Memory Reference Guide .Texa Instruments Incorporated,August 2002
[40] TI .TMS320C6000 Programmer’s Guide Texa Instruments Incorporated.August 2002
[41]陈武凡. 小波分析及其在图像处理中的应用[M]. 北京: 科学出版社, 2002-04
[42] 徐士麟, 喻 莉等. AVS1.0中熵编码技术的研究 .电视技术 ,2005年第10期(总第280期)
[2]信息技术 先进音视频编码 第七部分:移动视频 ,2004
[3]高文 ,黄铁军 .信源编码标准 AVS 及其在数字电视中的应用.电视技术,2003(11)4-6
[4]高文 等. AVS数字音视频编解码标准.中兴通讯技术,2006年6月第12卷第3期
[5]虞 露 等 .AVS 视频的技术特征. 电视技术 , 2004 年第 7 期(总第 277 期)
[6]ISO/IEC 13818-2-2000,Generic Coding of Moving Pictures and Associated Audio Information Part 2:Vdeo.2000
[7] ZHANG Nan, YIN Bao- cai, KONG De- hui, et al. Spatial prediction based intra- coding. Proceedings of the 2004 IEEE International Conference on Multimedia and Expo(ICME'2004), Vol 1. Jun. 27- 30, 2004, Taibei,China. New York, NY,USA:IEEE, 2004:97- 100.
[8]Draft ITU-T recommendation and final draft international standard of joint video sepefication (ITU-T Rec . H.264/ISO/IEC . 14496-10 AVC),JVT of ISO/IEC MPEG and ITV-T VCEC ,JVT-C050 ,2003
[9]ZHANG Nan, YIN Bao- cai, KONG De- hui, et al. Spatial prediction based intra- coding. Proceedings of the 2004 IEEE InternationalConference on Multimedia and Expo(ICME'2004), Vol 1. Jun. 27- 30, 2004, Taibei,China. New York, NY,USA:IEEE, 2004:97- 100.
[10] Peng Zhang, Debin Zhao, Siwei Ma, Yan Lu, Wen Gao . Multiple Modes Intra-Prediction in Intra Coding . IEEE InternationalConference on Multimedia and Expo(ICME'2004),2004
[11] JI Xiang- yang, ZGAO De- bin, GAO Wen, et al. New scaling technique for direct modecoding in B pictures .IEEE InternationalConference on Image Proces s ing (ICIP 2004).Oct 24- 27, 2004, Singapore . Piscataway, NJ ,USA:IEEE, 2004:469- 472.
[12] JI Xiang- yang, ZHAO De- bin, GAO Wen, et al. New Bi- prediction techniques for B pictures coding.Proceedings of the 2004 IEEE International Conference on Multimedia and Expo (ICME'2004), Vol 1. Jun.27- 30,2004, Taibei, China. New York, NY,USA:IEEE,2004: 101- 104.
[13] WANG Rong- gang, HUANG Chao, LI J in- tao, et al. Sub- pixel motion compensation interpolation filter in AVS . Proceedings of the 2004 IEEE International Conference on Multimedia and Expo(ICME'2004), Vol 1. Jun. 27- 30, 2004, Taibei, China. New York, NY,USA:IEEE, 2004:93- 96.
[14] 王荣刚 等.一种分像素运动补偿插值滤波方法及高效 VLSI 实现. 计算机学报,2005,第 28 卷 第 12 期,2052-2058
[15] MA Si- wei, GAO Wen, FAN Xiao- peng. Low complexity integer trans form and high definition coding.Proceedings of SPIE 49thAnnual Meeting,Vol 58. Aug 02- 06, 2004,Denver, CO, USA. Bellingham, WA,USA:SPIEPres s , 2002:547- 554.
[16] Zhang Ci-xun,Lou Jian ,Yu lu ,et al .The technique of pre-scaled integer transform. IEEE international Conference on Multimedia and Expo ,2004.
[17] WANG Qiang, ZHAO De- bin, MA Si- wei, et al. Context- based 2D- VLC for video coding.Proceedings of the 2004 IEEEInternational Conference on Multimedia and Expo (ICME'2004), Vol 1. Jun.27- 30, 2004,Taibei, China.New York, NY,USA:IEEE, 2004: 89- 92.
[18]陈光法,姚立敏, 虞 露 .AVS熵解码与dsp实现 .电视技术,2004年第10期,总第238 期
[19] Reeve H, Lim J S. Reduction of blocking effect s in image coding[J] . Opt Eng, 1984 , 23(1) :34 - 73.
[20] lain E. G. Richardson Video Codec Design Developing Image and Video Compression Systems Copyright 0 2002 by John Wiley & Sons Ltd,Baffins Lane, Chichester,West Sussex P019 IUD, England The Robert Gordon University, Aberdeen, UK
[21]George A. Triantafyllidis, Dimitrios Tzovaras, and Michael Gerassimos Strintzis,“Blocking Artifact Detection and Reduction in Compressed Data”, IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, OCTOBER 2002.VOL. 12, NO. 10: 877-890
[22]S.Minami and A.Zakhor. An optimization approach for removing blocking effects in transform coding〔J〕. IEEE Transaction on Circuits and Systems for Video Technology,Apr.1995Vol.5:74-82
[23]颜金尧 ,宋宜纯.低码率视频压缩技术的块效应消除方法〔J〕.通信学报,2000 , 4 .Vol.21:42～47
[24]Mei-Yin Shen, Jong Won Kim, and C.-C.Jay Kuo. Nonlinear filtering based fast compression artifact reduction technique for H.263. ITU-Telecommunications Standardization Sector STUDY GROUP 16 Video Coding Group(Question15),Document:Q15-F-13, Generated:27 Oct.1998.
[25]P.List, A.Joch, J.Lainema,Cz Bj}ntegaard, and M.Karczewicz, "Adaptive deblocking filter,”IEEE Transactions on Circuits and Systems for Video Technology, Vol.l3, N0.7,July 2003.
[26] 黄铁军 . 我视频编码国家标准 AVS 与国际标准 MPEG 的比较 . 互连网
[27] 国家广播电视产品质量监督检验中心. AVS视频编/解码方案图像质量主观评价试验报告. 2004年12月
[28] 国家广播电影电视总局广播电视规划院. AVS视频压缩质量主观评价(AVS参考软件5.2版)测试报告. 2005年9月
[29]梁凡 . AVS视频标准的技术特点 . 电视技术, 2005 第7期(总第277期)
[30] Texas Instruments .TMS320DM642 Technical Overview . SPRU615 - September 2002
[31]Taeksang Hwang, Wonyong Sung. Implementation of a Digital Copier Using TMS320C6414 VLIW DSP Processor. IEEE, IC ASSP, 2003
[32] Texas Instruments .TMS320C6000 CPU and Instruction Set Reference Guide. Oct. 2000
[33] Texas Instruments .TMS320C6000 DSP Cache User's Guide. May 2003
[34]丁刚. 从 TMS320C6000 EDMA 的结构来理解和优化其应用.无线电工程,2002 , 32(7) :28一 32.
[35] TI. TMS320C6000 C/C++ Optimizing Compiler User's Guide. 2002,10.
[36]李方慧,王飞,何佩馄.TMS320C6000 系列 DSPs 原理与应用. 第 2 版.电子工业出版社,2003 年 1 月,
[37].许小东,徐佩霞.基于 TM S320DM642 的视频解码系统优化数据采集与处理.2005 20(1): 91--95
[38] 董 斌,姜昱明.AVS 软件解码器的优化 .计算机工程与设计,2006,第 27 卷 第 4 期,618-620
[39] TI .TMS320C64x DSPTwo-Level Internal Memory Reference Guide .Texa Instruments Incorporated,August 2002
[40] TI .TMS320C6000 Programmer’s Guide Texa Instruments Incorporated.August 2002
[41]陈武凡. 小波分析及其在图像处理中的应用[M]. 北京: 科学出版社, 2002-04
[42] 徐士麟, 喻 莉等. AVS1.0中熵编码技术的研究 .电视技术 ,2005年第10期(总第280期)
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |