基于ARM926EJ的H.264解码器设计与优化
摘要
近年来随着多媒体应用的领域逐渐扩展,人们在对新的先进编码技术进行理论研究的同时,也加大了实用化的力度。H.264是新一代的视频标准,比以往的视频标准在相同的信噪比下的压缩效果有明显的提升,并适合于在网络中传播,但算法复杂度成倍增加。计算复杂度的升高,使得H.264视频压缩标准很难在现有的条件下应用到实际中去,因此,对H.264的实现和优化是很必要、很有意义的一项研究工作。同时嵌入式系统,尤其是便携式的手持终端设备,需要找到一个性能和功耗的平衡点,而ARM处理器的出现正适应了这一需求。
本论文首先在详细研究H.264 Baseline Profile视频解码的算法原理和标准的基础上,测试并分析总结出未经优化的软件在ARM9EJ处理器上的性能瓶颈和优化思路;然后结合ARM9EJ系列处理器架构的特点,对程序的整体结构及各个功能模块展开优化,尽可能地提高处理器效率;最后在实际的硬件平台MX21(ARM926FA内核)平台上实现一个解码算法、并测试优化结果。测试表明,解码核心和回放系统经过优化以后,各种运动复杂度的QCIF视频序列基本实现了视频解码的需要,达到了所需的优化要求。
现阶段的工作成果为今后的实际应用做了很好的铺垫,并且本论文所采用的研究方法对于其它平台上的应用也具有指导意义。
With the extension in the field of multimedia applications in recent years, people begin to pay more attention to the field of how to apply advanced compression technologies to practical applications while doing researches of new thoeries. As a new video standard, H.264 raises the compression rate obviously than those used video standards. And it is more suitable for transmitting in the network, but its computation complexity is multiplied. For the increased computational complexity, it is difficult for H.264 video compression standard to be widly used to pratical applications in the current conditions. Therefore, the realization and optimization of the H.264 is very necessary and significant. As embedded systems, especially the portable handset terminals, they usually try to keep balance between performance and power consumption. ARM processors can fairly meet this requirement.
On the basis of understanding the theories and standards of H.264 Video Baseline Profile, this thesis analyses the bottleneck on the non-optimized decoder software on ARM9EJ platform and summarizes the aspects that need optimizing. H.264 video decoder software is optimized to improve the efficiency of the CPU according to the architecture of ARM processor. The optimization work is mainly focusing on the overall structure and various functional modues. In the last part, the implementation of a simple H.264 video playback system on MX21(ARM926EJ core) platform is discussed. The test results show that the QCIFs with various motion complexities meet the basic needs after the decoder core and the playback system being fully optimized.
The work done at this stage is a good preparation for the future practical application, and the methods used in this thesis are instructive for the applications on other platforms as well.
本论文首先在详细研究H.264 Baseline Profile视频解码的算法原理和标准的基础上,测试并分析总结出未经优化的软件在ARM9EJ处理器上的性能瓶颈和优化思路;然后结合ARM9EJ系列处理器架构的特点,对程序的整体结构及各个功能模块展开优化,尽可能地提高处理器效率;最后在实际的硬件平台MX21(ARM926FA内核)平台上实现一个解码算法、并测试优化结果。测试表明,解码核心和回放系统经过优化以后,各种运动复杂度的QCIF视频序列基本实现了视频解码的需要,达到了所需的优化要求。
现阶段的工作成果为今后的实际应用做了很好的铺垫,并且本论文所采用的研究方法对于其它平台上的应用也具有指导意义。
With the extension in the field of multimedia applications in recent years, people begin to pay more attention to the field of how to apply advanced compression technologies to practical applications while doing researches of new thoeries. As a new video standard, H.264 raises the compression rate obviously than those used video standards. And it is more suitable for transmitting in the network, but its computation complexity is multiplied. For the increased computational complexity, it is difficult for H.264 video compression standard to be widly used to pratical applications in the current conditions. Therefore, the realization and optimization of the H.264 is very necessary and significant. As embedded systems, especially the portable handset terminals, they usually try to keep balance between performance and power consumption. ARM processors can fairly meet this requirement.
On the basis of understanding the theories and standards of H.264 Video Baseline Profile, this thesis analyses the bottleneck on the non-optimized decoder software on ARM9EJ platform and summarizes the aspects that need optimizing. H.264 video decoder software is optimized to improve the efficiency of the CPU according to the architecture of ARM processor. The optimization work is mainly focusing on the overall structure and various functional modues. In the last part, the implementation of a simple H.264 video playback system on MX21(ARM926EJ core) platform is discussed. The test results show that the QCIFs with various motion complexities meet the basic needs after the decoder core and the playback system being fully optimized.
The work done at this stage is a good preparation for the future practical application, and the methods used in this thesis are instructive for the applications on other platforms as well.
引文
[1]马小虎,张明敏,严华明:多媒体数据压缩标准及实现,北京,清华大学出版社,1996
[2]姚栋,虞露.H.264指数哥伦布码解码部件的硬件设计和实现.电视技术2004,11,14-16.
[3]毛讯,姚庆栋,虞露.VSP中的变字长解码器设计研究.浙江大学学报(工学版),2001(11);583-587
[4]ISO/IEC 14496-2:2001(Coding of Audio-visual Objects-Part 2:Visual).MPEG N4530,2001
[5]Fernando Pereira,Touradj Ebrahimi:The MPEG-4 Book,Prentice Hall PTR,2002
[6]lain E.G.Richardson:Video Codec Design,John Wiley & Sons,2002
[7]KeithJack:Video Demystified,Third Edition,LLH Technology,2001
[8]Andrew N.Sloss,Dominic Symes,Chris Wright著,沈建华译:ARM嵌入式系统开发——软件设计与优化,北京,北京航空航天大学出版社,2005
[9]张颖,张济林.H264中CAVLC解码的分组优化.信号处理.2005年,第21卷第1期.17-21.
[10]朱冬冬,戴琼海.H.264解码器中CAVLC码表查找算法的改进.电视技术.2004,1,26-27
[11]蒋汉琪,朱秀昌.H.264编码器中插值运算和整数变换的优化.南京邮电学院学报.2005,第3期,41-44.
[12]周宁兆,宋彬.H.264中帧内编码算法及其快速实现研究[J].中国有线电视,2004年,24期:9-13.
[13]熊赞,于鸿洋.H264实时软件解码器的实现.中国有线电视.2005(6),562-565
[14]王延美,林争辉等.H264在手机电视中的应用.电视技术.2004(11),24-26
[15]]鹿宝生,称启美.H.264高哇能视频编码器的Dsp实现.2005(12),2824-2826
[16]陈维安,李典等.H.264软件解码器的优化.数据采集于处理.2005(4),493-498
[17]董海燕,张其善.H.264中多参考帧预测技术的优化.计算机应用.2005(6),10-11
[18]刘丽娜,邹雪妹.基于hitel指令集的H.264编码器优化.中国有线电视.2005(24),2406-2409.
[19]召区振,郑世宝,杨宇红基于SOC平台的H.264解码器IP核设计.电视技术.2006f3),2-27.
[20]ARM Ltd.:ARM920T Technical Reference Manual,ARM DDI0151C,2000
[21]ARM Ltd.:Performance of the ARM9TDMI~(TM) and ARM9E-STM cores compared to the ARM7TDMI~(TM) core,2000
[22]朱海涛,戚文芽.基于TMS320DM642的H.264视频解码器设计.电视技术.2005(4),25-28.
[23]王继山,李挥.一种快速去块滤波器结构.微电子与计算机.2005(11),14-16
[24]雷钊,薛少丽,梁篙.H264编/解码器中去块效应滤波系统的设计.电视技术.2005(10),18-19
[25]王正宁,彭强,诸昌铃.几种快速运动搜索算法在H.264中的实现与分析。计算机应用,2004.24(9):91-93
[26]ARM Ltd:ARM Architecture Reference Manual,2000
[27]Intel Corporation:A Fast Precise Implementation of 8×8 Discrete Cosine Transform Using the Streaming SIMD Extentions and MMX~(TM) Instructions,1999
[28]P.Pakdeepaiboonpol,S.Kittitornkun:Energy Optimization for Mobile MPEG-4Video Decoder,Mobile Technology,Application and Systems,2005 2~(nd)International Conference
[29]Eduardo Asbun and Chiouguey Chen:On the Implementation of MPEG-4 Motion Compensation Using the TMS320C62x,Texas Instruments,Inc.1999
[30]Steve Furber:ARM System-on-Chip Architecture,2~(nd) Edition,Addison Wesley Longman Limited,2000
[31]Zhiyong Xu,Rui Min,Yiming Hu:An Analysis of Cache Performance of Multimedia Applications,IEEE Transactions on Computers,VOL.53,No.1,2004
[32]ARM Ltd.:ARM Applications Library—Programmer's Guide,ARM DUI 0081B,2001
[2]姚栋,虞露.H.264指数哥伦布码解码部件的硬件设计和实现.电视技术2004,11,14-16.
[3]毛讯,姚庆栋,虞露.VSP中的变字长解码器设计研究.浙江大学学报(工学版),2001(11);583-587
[4]ISO/IEC 14496-2:2001(Coding of Audio-visual Objects-Part 2:Visual).MPEG N4530,2001
[5]Fernando Pereira,Touradj Ebrahimi:The MPEG-4 Book,Prentice Hall PTR,2002
[6]lain E.G.Richardson:Video Codec Design,John Wiley & Sons,2002
[7]KeithJack:Video Demystified,Third Edition,LLH Technology,2001
[8]Andrew N.Sloss,Dominic Symes,Chris Wright著,沈建华译:ARM嵌入式系统开发——软件设计与优化,北京,北京航空航天大学出版社,2005
[9]张颖,张济林.H264中CAVLC解码的分组优化.信号处理.2005年,第21卷第1期.17-21.
[10]朱冬冬,戴琼海.H.264解码器中CAVLC码表查找算法的改进.电视技术.2004,1,26-27
[11]蒋汉琪,朱秀昌.H.264编码器中插值运算和整数变换的优化.南京邮电学院学报.2005,第3期,41-44.
[12]周宁兆,宋彬.H.264中帧内编码算法及其快速实现研究[J].中国有线电视,2004年,24期:9-13.
[13]熊赞,于鸿洋.H264实时软件解码器的实现.中国有线电视.2005(6),562-565
[14]王延美,林争辉等.H264在手机电视中的应用.电视技术.2004(11),24-26
[15]]鹿宝生,称启美.H.264高哇能视频编码器的Dsp实现.2005(12),2824-2826
[16]陈维安,李典等.H.264软件解码器的优化.数据采集于处理.2005(4),493-498
[17]董海燕,张其善.H.264中多参考帧预测技术的优化.计算机应用.2005(6),10-11
[18]刘丽娜,邹雪妹.基于hitel指令集的H.264编码器优化.中国有线电视.2005(24),2406-2409.
[19]召区振,郑世宝,杨宇红基于SOC平台的H.264解码器IP核设计.电视技术.2006f3),2-27.
[20]ARM Ltd.:ARM920T Technical Reference Manual,ARM DDI0151C,2000
[21]ARM Ltd.:Performance of the ARM9TDMI~(TM) and ARM9E-STM cores compared to the ARM7TDMI~(TM) core,2000
[22]朱海涛,戚文芽.基于TMS320DM642的H.264视频解码器设计.电视技术.2005(4),25-28.
[23]王继山,李挥.一种快速去块滤波器结构.微电子与计算机.2005(11),14-16
[24]雷钊,薛少丽,梁篙.H264编/解码器中去块效应滤波系统的设计.电视技术.2005(10),18-19
[25]王正宁,彭强,诸昌铃.几种快速运动搜索算法在H.264中的实现与分析。计算机应用,2004.24(9):91-93
[26]ARM Ltd:ARM Architecture Reference Manual,2000
[27]Intel Corporation:A Fast Precise Implementation of 8×8 Discrete Cosine Transform Using the Streaming SIMD Extentions and MMX~(TM) Instructions,1999
[28]P.Pakdeepaiboonpol,S.Kittitornkun:Energy Optimization for Mobile MPEG-4Video Decoder,Mobile Technology,Application and Systems,2005 2~(nd)International Conference
[29]Eduardo Asbun and Chiouguey Chen:On the Implementation of MPEG-4 Motion Compensation Using the TMS320C62x,Texas Instruments,Inc.1999
[30]Steve Furber:ARM System-on-Chip Architecture,2~(nd) Edition,Addison Wesley Longman Limited,2000
[31]Zhiyong Xu,Rui Min,Yiming Hu:An Analysis of Cache Performance of Multimedia Applications,IEEE Transactions on Computers,VOL.53,No.1,2004
[32]ARM Ltd.:ARM Applications Library—Programmer's Guide,ARM DUI 0081B,2001