基于DM6446的视频编解码实现与优化
摘要
视频会议系统可以帮助企业提高生产效率,即时把握市场动向,提高团队协作效率,因而得到广泛应用。H.264因其压缩率高、编码质量高和网络适应性好而逐渐成为视频会议系统编解码器的标准。达芬奇(DaVinci)平台具有ARM+DSP双核系统及专门的视频处理子系统,在数字视频处理方而较有优势,因此成为视频会议系统开发硬件平台的理想选择。
首先简要介绍了H.264标准及其关键技术,并对H.264的三大开源代码进行测试分析,然后对于达芬奇平台DM6446(包括硬件平台和软件平台)做了详细的介绍。
对x264编码器进行了裁剪,保留了支持基本档次的代码,并成功地移植到CCSv3.3。通过编译器选项优化,C代码优化(去除相关性,内联函数使用,软件流水技术)、CACHE优化和线性汇编改编等方法对编码器进行了优化,并达到了如下优化效果：qcif格式的视频文件编码速度达到87帧/s,cif格式的文件达到18帧/s。对解码器也进行了类似的移植和优化,优化后解码器的速度得到大大的提高, qcif格式的视频文件解码速度为495帧/s,cif格式的文件解码速度为125帧/s。整个系统编解码的能力满足实时传输和播放的要求。
解析了达芬奇开发套件的xDM的标准,介绍了封装的具体过程,并将封装成xDM格式的编码器成功集成在算法服务上。
分析了视频采集的具体过程,对视频驱动V4L2进行了详细的介绍,实现了视频的成功采集,完成了ARM端的应用程序开发,并最终完成完整的视频会议系统。
综上所述,本课题实现了一个完整的视频会议系统,为将来进一步提高系统性能和完善系统功能打下了基础。
Video conferencing system can help enterprises improve production efficiency, grasp the market trends immediately, and improve the efficiency of team collaboration, therefore, it has been widely used. Because of its high compression rate, high quality of coding and good adaptability to network, H.264has gradually become the standard for video conferencing codec. Equipped with ARM+DSP dual-core system and a dedicated video processing subsystem, DaVinci platform has more advantages in digital video processing, and thus has become an ideal choice for developing hardware platform of video conferencing system.
H.264video standard and its critical technologies are introduced in this article. Three main open source H.264encodecs/decodecs are tested and analyzed. TI's DM6446hardware and software platforms are introduced in detail.
X264code is tailored to the requirement of base profile of H.264and successfully ported to CCSv3.3. Code optimization is done via changing compiler options, C code optimization (removing the correlation,inlining function,pipelining technique), CACHE optimization, linear assembly language rewriting, etc.. After the optimization procedure, qcif format videos could be encoded at87fps and cif videos18fps. Decoder is also ported and optimized and decoding speed is greatly enhanced: qcif videos could be decoded at495fps and cif videos125fps. The system's encoding/decoding capability satisfies the requirements of real time transfer and playback.
The xDM standards of the DaVinci Development Kit and its encapsulation process are described, and then the codec is successfully encapsulated into xDM format and integrated into codec server.
The specific process of video capture is described at first, and then V4L2is introduced in in detail. Video capturing is realized, the ARM-side application development is completed, and eventually a complete video conferencing system is achieved.
In conclusion, a complete video conferencing system is completed. The foundation is laid, but further improvement of system performance is expected.
首先简要介绍了H.264标准及其关键技术,并对H.264的三大开源代码进行测试分析,然后对于达芬奇平台DM6446(包括硬件平台和软件平台)做了详细的介绍。
对x264编码器进行了裁剪,保留了支持基本档次的代码,并成功地移植到CCSv3.3。通过编译器选项优化,C代码优化(去除相关性,内联函数使用,软件流水技术)、CACHE优化和线性汇编改编等方法对编码器进行了优化,并达到了如下优化效果：qcif格式的视频文件编码速度达到87帧/s,cif格式的文件达到18帧/s。对解码器也进行了类似的移植和优化,优化后解码器的速度得到大大的提高, qcif格式的视频文件解码速度为495帧/s,cif格式的文件解码速度为125帧/s。整个系统编解码的能力满足实时传输和播放的要求。
解析了达芬奇开发套件的xDM的标准,介绍了封装的具体过程,并将封装成xDM格式的编码器成功集成在算法服务上。
分析了视频采集的具体过程,对视频驱动V4L2进行了详细的介绍,实现了视频的成功采集,完成了ARM端的应用程序开发,并最终完成完整的视频会议系统。
综上所述,本课题实现了一个完整的视频会议系统,为将来进一步提高系统性能和完善系统功能打下了基础。
Video conferencing system can help enterprises improve production efficiency, grasp the market trends immediately, and improve the efficiency of team collaboration, therefore, it has been widely used. Because of its high compression rate, high quality of coding and good adaptability to network, H.264has gradually become the standard for video conferencing codec. Equipped with ARM+DSP dual-core system and a dedicated video processing subsystem, DaVinci platform has more advantages in digital video processing, and thus has become an ideal choice for developing hardware platform of video conferencing system.
H.264video standard and its critical technologies are introduced in this article. Three main open source H.264encodecs/decodecs are tested and analyzed. TI's DM6446hardware and software platforms are introduced in detail.
X264code is tailored to the requirement of base profile of H.264and successfully ported to CCSv3.3. Code optimization is done via changing compiler options, C code optimization (removing the correlation,inlining function,pipelining technique), CACHE optimization, linear assembly language rewriting, etc.. After the optimization procedure, qcif format videos could be encoded at87fps and cif videos18fps. Decoder is also ported and optimized and decoding speed is greatly enhanced: qcif videos could be decoded at495fps and cif videos125fps. The system's encoding/decoding capability satisfies the requirements of real time transfer and playback.
The xDM standards of the DaVinci Development Kit and its encapsulation process are described, and then the codec is successfully encapsulated into xDM format and integrated into codec server.
The specific process of video capture is described at first, and then V4L2is introduced in in detail. Video capturing is realized, the ARM-side application development is completed, and eventually a complete video conferencing system is achieved.
In conclusion, a complete video conferencing system is completed. The foundation is laid, but further improvement of system performance is expected.
引文
[1].刘仕翔.基于DM642 DSP的x264编码器研究及实现[D].成都：西南交通大学硕士学位论文,2010.
[2].何勋.基于S3C2440的H.264软编解码器移植及优化[D].成都：电子科技大学硕士学位论文,2010.
[3].宋扬.基于TMS320DM642 DSP的H.264编码器优化与设计[D].南京：南京理工大学硕士学位论文,2007.
[4].毕厚杰.新一代视频压缩编码标准——H.264/AVC[M].北京：人民邮电出版社2004.
[5].张映伟.DM642上的X264编码器实时实现[D].太原：太原理工大学硕士学位论文,2010.
[6].李辉武.基于ARM9的H.264解码器的优化与实现[D].广东：广东工业大学硕士学位论文,2009.
[7].唐玲娜.H.264视频解码优化及DSP实现[D].成都：电子科技大学硕士学位论文,2009.
[8].白玉洁.基于TMS320DM6446的H.264编码器的设计与优化[J].今日电子,2011(1)：第50-52+55页.
[9].罗了平.基于TI达芬奇平台H.264编码器的研究与实现[D].成都：西华大学硕士学位论文,2010.
[10]. Texas Instruments. tms320dm6446.2007
[11]. Texas Instruments. RUIVA-DM6446-0012 User Guide 20110516.2007
[12]. Texas Instruments. sprue66f-TMS320DM6446 DVEVM v2.0 Getting Started Guide.2007
[13]. Texas Instruments. UserGuide.2007
[14].卢宁,柯熙政,贾贞.基于达芬奇技术的H.264视频编码器的研究与实现[J].内蒙古大学学报(自然科学版),2010(4)：第410-416页.
[15].张伟华.基于DaVinci技术的H.264实现与研究[D].北京邮电大学,2010.
[16]. Texas Instruments. CodecEngine_ServerIntegrator_UserGuide.2007
[17]. Texas Instruments. spru424c.2007
[18]. Texas Instruments. spruec8.2007
[19].成嘉,张文雄,李善劲.基于达芬奇技术的H.264视频编码器的实现[J].电视技术,2007(12)第34-36页.
[20].刘冲,刘岚.H.264标准中CABAC算法和CAVLC算法的比较与研究[J].中国科技论文在线,www.paper.edu.cn
[21].刘聪.视频编码中的隔行编码技术研究[J].咸宁学院学报,2008(3)：第67-70页.
[22].程鸿芳,赵森严.基于TMS320DM6446的H.264视频编码算法的移植与优化[J].安庆师范学院学报(自然科学版),2010(1)：第30-34+38页.
[23].oskycar. ccs3.3中建立-编译-调试工程及常见错误讲解[EB].2010：第13页.
[24].陈寅.H.264帧间编码算法研究与DSP移植[D].武汉：武汉理工大学硕士学位论文,2009.
[25].黄孝建,张颖.基于达芬奇DM6446处理器的H.264视频编码算法优化[J].数字技术与应用,2011(11)：第141-144页.
[26]. Texas Instruments. TMS320C6000 Programmer's Guide.2007
[27]. Texas Instruments. C6x programmer's guide(198d).2007
[28].李方慧,王飞,何佩琨.TMS320C6000系列DSPs原理与应用(第二版)[M].北京：电子工业出版社,2003
[29].基于DSP的软件流水优化研究[EB].
[30].Qian. M.软件流水[EB].
[31]. Wand. E.04-DaVinci DSP端代码优化策略[EB].2010.
[32]. Texas Instruments. TMS320C6000 DSP Cache User's Guide.2007
[33].Hhx. DSP_C6000软件优化一线性汇编[EB].2005.
[34]. Texas Instruments. MSP430系列汇编指令集.2007
[35].周德卿.CCS集成环境下C与汇编的混合编程实现.微电子技术[J],2002(2)：第23-26+36页.
[36].鲍协浩DaVinci平台上H.264视频编码算法移植优化与xDM封装[D].北京：北京邮电大学,2011.
[37]. Texas Instruments. CodecEngine_AlgorithmCreator_UserGuide.2007
[38]. Texas Instruments. CodecEngine_ExampleFeatures_UsersGuide.2007
[39]. Texas Instruments. spru360e.2007
[40]. Texas Instruments. Bios_Tconf_User_Guide.2007
[41].Nsq.4-2_Pthreads提供的基本线[EB],2009.
[42].张辉.基于V4L2视频采集缓存机制应用与实现[J].现代电子技术,2010(20)：第54-56页.
[43].张龙,李霁野.基于DM642的视频采集与传输系统设计[J].微计算机信息,2011(6)：第61-62+131页.
[44].赵满意.视频采集与编码技术在视频监控终端中的应用[J].计算机工程与科学,2011(1)：第122-126页.
[45]. TCP_IP协议栈在嵌入式系统中的实现[EB].
[2].何勋.基于S3C2440的H.264软编解码器移植及优化[D].成都：电子科技大学硕士学位论文,2010.
[3].宋扬.基于TMS320DM642 DSP的H.264编码器优化与设计[D].南京：南京理工大学硕士学位论文,2007.
[4].毕厚杰.新一代视频压缩编码标准——H.264/AVC[M].北京：人民邮电出版社2004.
[5].张映伟.DM642上的X264编码器实时实现[D].太原：太原理工大学硕士学位论文,2010.
[6].李辉武.基于ARM9的H.264解码器的优化与实现[D].广东：广东工业大学硕士学位论文,2009.
[7].唐玲娜.H.264视频解码优化及DSP实现[D].成都：电子科技大学硕士学位论文,2009.
[8].白玉洁.基于TMS320DM6446的H.264编码器的设计与优化[J].今日电子,2011(1)：第50-52+55页.
[9].罗了平.基于TI达芬奇平台H.264编码器的研究与实现[D].成都：西华大学硕士学位论文,2010.
[10]. Texas Instruments. tms320dm6446.2007
[11]. Texas Instruments. RUIVA-DM6446-0012 User Guide 20110516.2007
[12]. Texas Instruments. sprue66f-TMS320DM6446 DVEVM v2.0 Getting Started Guide.2007
[13]. Texas Instruments. UserGuide.2007
[14].卢宁,柯熙政,贾贞.基于达芬奇技术的H.264视频编码器的研究与实现[J].内蒙古大学学报(自然科学版),2010(4)：第410-416页.
[15].张伟华.基于DaVinci技术的H.264实现与研究[D].北京邮电大学,2010.
[16]. Texas Instruments. CodecEngine_ServerIntegrator_UserGuide.2007
[17]. Texas Instruments. spru424c.2007
[18]. Texas Instruments. spruec8.2007
[19].成嘉,张文雄,李善劲.基于达芬奇技术的H.264视频编码器的实现[J].电视技术,2007(12)第34-36页.
[20].刘冲,刘岚.H.264标准中CABAC算法和CAVLC算法的比较与研究[J].中国科技论文在线,www.paper.edu.cn
[21].刘聪.视频编码中的隔行编码技术研究[J].咸宁学院学报,2008(3)：第67-70页.
[22].程鸿芳,赵森严.基于TMS320DM6446的H.264视频编码算法的移植与优化[J].安庆师范学院学报(自然科学版),2010(1)：第30-34+38页.
[23].oskycar. ccs3.3中建立-编译-调试工程及常见错误讲解[EB].2010：第13页.
[24].陈寅.H.264帧间编码算法研究与DSP移植[D].武汉：武汉理工大学硕士学位论文,2009.
[25].黄孝建,张颖.基于达芬奇DM6446处理器的H.264视频编码算法优化[J].数字技术与应用,2011(11)：第141-144页.
[26]. Texas Instruments. TMS320C6000 Programmer's Guide.2007
[27]. Texas Instruments. C6x programmer's guide(198d).2007
[28].李方慧,王飞,何佩琨.TMS320C6000系列DSPs原理与应用(第二版)[M].北京：电子工业出版社,2003
[29].基于DSP的软件流水优化研究[EB].
[30].Qian. M.软件流水[EB].
[31]. Wand. E.04-DaVinci DSP端代码优化策略[EB].2010.
[32]. Texas Instruments. TMS320C6000 DSP Cache User's Guide.2007
[33].Hhx. DSP_C6000软件优化一线性汇编[EB].2005.
[34]. Texas Instruments. MSP430系列汇编指令集.2007
[35].周德卿.CCS集成环境下C与汇编的混合编程实现.微电子技术[J],2002(2)：第23-26+36页.
[36].鲍协浩DaVinci平台上H.264视频编码算法移植优化与xDM封装[D].北京：北京邮电大学,2011.
[37]. Texas Instruments. CodecEngine_AlgorithmCreator_UserGuide.2007
[38]. Texas Instruments. CodecEngine_ExampleFeatures_UsersGuide.2007
[39]. Texas Instruments. spru360e.2007
[40]. Texas Instruments. Bios_Tconf_User_Guide.2007
[41].Nsq.4-2_Pthreads提供的基本线[EB],2009.
[42].张辉.基于V4L2视频采集缓存机制应用与实现[J].现代电子技术,2010(20)：第54-56页.
[43].张龙,李霁野.基于DM642的视频采集与传输系统设计[J].微计算机信息,2011(6)：第61-62+131页.
[44].赵满意.视频采集与编码技术在视频监控终端中的应用[J].计算机工程与科学,2011(1)：第122-126页.
[45]. TCP_IP协议栈在嵌入式系统中的实现[EB].