基于DM642的MPEG-4视频服务器设计与优化
摘要
MPEG-4是国际运动图像专家组(MPEG)于1999年制定的新一代面向对象的多媒体压缩国际标准。MPEG-4引入了视频对象(VO)的概念,它不再是一个单纯的视频和音频编解码标准,它将内容和交互性作为核心,强调多媒体通信的交互性和灵活性,从而为多媒体数据压缩提供了一个更为广阔的平台。
本系统是一套拥有自主知识产权的视频监控服务系统,满足多个行业迅速发展的需要,例如在网络安全、视频会议、远程监控等领域中的应用。本系统采用的是TI公司推出的一款高性能可编程数字信号处理器TMS320DM642 DSP。在此基础上设计开发一套基于MPEG-4标准的视频服务器(亦称视频编码器),并对其进行优化,以达到实时性要求。
论文第一部分介绍了图像编码的技术原理与发展现状,第二部分着重介绍MPEG-4编码技术的框架结构与技术特点。我们讨论了MPEG-4包含的两种量化技术,并最终确立H.263量化方案作为我们编码器的量化方案。我们对比了MPEG-4的核心编码算法,并确定系统采用的编码方案为MPEG-4简单框架(SPSimple Profile)下的VOP(Video Object Plane)编码。论文第三部分对系统硬件平台的设计开发进行了深入探讨,其中对我们采用的主芯片DM642的硬件结构特点、指令系统、存储器结构、流水线结构等都进行了详细的描述,重点对系统硬件平台的原理与开发步骤进行了深入研究。论文第四部分提出了C64x平台下软件实现与优化方案,研究了DSP C64X系列系统的软件开发优化流程与VOP编码流程等,随后结合算法特点,总结提出了编码流程的三级优化方案:一级为算法优化,采用高效算法,提高模块效率;二级为程序优化,合理安排程序流程,提高DSP内存利用率;三级为汇编优化,将多次调用及以循环体为主的函数改写为汇编语言,利用汇编优化器进行优化。
通过我们的实验结果表明,经过优化后的视频服务系统,可以实现对于QCIF格式的视频源,达到每秒25帧的实时视频监控要求。
The project's original intention was to project the development of independent intellectual property rights of a video surveillance system to meet the various needs of the rapidly development, such as security, network video conferencing, intelligent transport, remote video surveillance, intelligent buildings, and other works . The system is based on TI's programmable launched a high-performance digital signal processor TMS320DM642 DSP. On this basis, designed and developed a set of standards-based MPEG-4 video services, using TMSDM642DSP hardware platform to realize the MPEG-4 video encoder and optimize them to achieve real-time requirements.
The international Motion Picture Experts Groups (MPEG) formally brought forward the new MPEG-4 Standard in 1999. MPEG-4 firstly introduced the concept of video object and contained object-based coding and model-based coding, and then supported content-based interactive and object-based scalable encoding. The standard paid mostly attention to interaction and flexibility in multimedia communication compared with former international standard of video compression, it is different from those old standards.So it provides a broader platform to multimedia data Compression.
Papers firstly introduced the image coding technology and the development of principles, focused on MPEG-4 encoding technology in the framework structure and technical characteristics, and compared its core coding algorithm to determine the coding scheme for MPEG-4 simple framework (SP Simple Profile) under the VOP (Video Object Plane) Coding. We study the hardware architecture of DM642, instruction system, memory architecture of DM642, pipeline and software explored flow. This paper take depth discussions on the design of hardware platform development and optimization. We concluded three optimization process under theC64Xplatform:1, algorithm optimization.Efficient use of algorithms to improve Module efficiency;2, procedures optimization. Reasonable procedures processes, improve utilization of DSP memory.3, compilation optimization.Rewritten the multiple calls and recycling-based function for the compilation language,and optimized it by the compilation Optimizer.
The result of the experiment shows that the optimized video system can transfer the video source at QCIF format with a speed of 25fps.
本系统是一套拥有自主知识产权的视频监控服务系统,满足多个行业迅速发展的需要,例如在网络安全、视频会议、远程监控等领域中的应用。本系统采用的是TI公司推出的一款高性能可编程数字信号处理器TMS320DM642 DSP。在此基础上设计开发一套基于MPEG-4标准的视频服务器(亦称视频编码器),并对其进行优化,以达到实时性要求。
论文第一部分介绍了图像编码的技术原理与发展现状,第二部分着重介绍MPEG-4编码技术的框架结构与技术特点。我们讨论了MPEG-4包含的两种量化技术,并最终确立H.263量化方案作为我们编码器的量化方案。我们对比了MPEG-4的核心编码算法,并确定系统采用的编码方案为MPEG-4简单框架(SPSimple Profile)下的VOP(Video Object Plane)编码。论文第三部分对系统硬件平台的设计开发进行了深入探讨,其中对我们采用的主芯片DM642的硬件结构特点、指令系统、存储器结构、流水线结构等都进行了详细的描述,重点对系统硬件平台的原理与开发步骤进行了深入研究。论文第四部分提出了C64x平台下软件实现与优化方案,研究了DSP C64X系列系统的软件开发优化流程与VOP编码流程等,随后结合算法特点,总结提出了编码流程的三级优化方案:一级为算法优化,采用高效算法,提高模块效率;二级为程序优化,合理安排程序流程,提高DSP内存利用率;三级为汇编优化,将多次调用及以循环体为主的函数改写为汇编语言,利用汇编优化器进行优化。
通过我们的实验结果表明,经过优化后的视频服务系统,可以实现对于QCIF格式的视频源,达到每秒25帧的实时视频监控要求。
The project's original intention was to project the development of independent intellectual property rights of a video surveillance system to meet the various needs of the rapidly development, such as security, network video conferencing, intelligent transport, remote video surveillance, intelligent buildings, and other works . The system is based on TI's programmable launched a high-performance digital signal processor TMS320DM642 DSP. On this basis, designed and developed a set of standards-based MPEG-4 video services, using TMSDM642DSP hardware platform to realize the MPEG-4 video encoder and optimize them to achieve real-time requirements.
The international Motion Picture Experts Groups (MPEG) formally brought forward the new MPEG-4 Standard in 1999. MPEG-4 firstly introduced the concept of video object and contained object-based coding and model-based coding, and then supported content-based interactive and object-based scalable encoding. The standard paid mostly attention to interaction and flexibility in multimedia communication compared with former international standard of video compression, it is different from those old standards.So it provides a broader platform to multimedia data Compression.
Papers firstly introduced the image coding technology and the development of principles, focused on MPEG-4 encoding technology in the framework structure and technical characteristics, and compared its core coding algorithm to determine the coding scheme for MPEG-4 simple framework (SP Simple Profile) under the VOP (Video Object Plane) Coding. We study the hardware architecture of DM642, instruction system, memory architecture of DM642, pipeline and software explored flow. This paper take depth discussions on the design of hardware platform development and optimization. We concluded three optimization process under theC64Xplatform:1, algorithm optimization.Efficient use of algorithms to improve Module efficiency;2, procedures optimization. Reasonable procedures processes, improve utilization of DSP memory.3, compilation optimization.Rewritten the multiple calls and recycling-based function for the compilation language,and optimized it by the compilation Optimizer.
The result of the experiment shows that the optimized video system can transfer the video source at QCIF format with a speed of 25fps.
引文
[1]陈传波,陈先级.数字图像处理.北京:机械工业出版社.2004.16-25
[2]胡国荣.数字视频压缩及其标准.北京广播学院出版社.1999.13-28
[3]沈兰荪,卓立,田栋,汪孔桥.视频编码与低速率传输.电子工业出版社.2001.17-26
[4]余兆明,李晓飞等.MPEG标准及其应用.北京邮电大学出版社.2002.32-45
[5]Olivier.Avaro.ITU-T Recommendation Video coding for low bit rate communication.Signal Processing:Image Communication Jan.Vol.15.2000.281-298.
[6]谭仕勇,戴琼海,陈慧蓉MPEG-4标准中的视频编码.广播与电视技术.2001.6:51-57
[7]夏天,陈磊,余兆明MPEG-4的基于内容的视频编码技术.电子工程师.1999.10:9-10
[8]田元,叶秀清,顾伟康.实时图像处理系统中的优化编程.电子技术应用.2000.10.43-46.
[9]王栋,张兆杨等.基于TMS320C6000 DSP的视频解码器设计及其硬件实现方案.电子技术应用.2002.2.78-80.
[10]章毓晋.图像处理与分析 清华大学出版社.1999.25-28
[11]Kenneth R.Castleman(美),朱志刚,林学訚,石定机译.数字图像处理.电子工业出版社.2002.31-35
[12]S.Zhu,K.Ma New Diamond Search Algorithm for Fast Blocking matching Motion Estimation IEEE.Trans.Image Processing.2000.Feb.vol.9.No.2.287-290.
[13]J.Y.Tham,S.Ranganath,M.Ranganath.A novel unrestricted center-biased diamond search algorithm for block motion estimation.IEEE Trans.on Circ.And Sys.vol.8.No.4.August 1998.369-377.
[14]Texas Instruments,Inc.TMS320DM642 Technical Overview.2002.9.156-163
[15]Texas Instruments,Inc.TMS320DM642 Evaluation Module Technical Reference.2003.35-37
[16]周霖等.DSP信号处理技术的应用[M].国防工业出版社.2004.67-71
[17]Texas Instruments.Inc.TMS320C64x Image/Video Processing Library Programmer Reference.2003.10.39-56
[18]Texas Instruments,Inc.MS320C6000 Peripherals.2001.230-258
[19]Texas Instntments.Inc.MS320C6x Peripherals Support Library User's Guide.2000.24-39
[20]钟玉琢,王琪,贺玉文.基于对象的多媒体数据压缩编码国际标准—MPEG及其校验模型.科学出版社.2000.10.66-68
[21]Texas Instruments,Inc.MS320C64X DSP Two-level Internal Memory Reference Guide 2002.1.25-31
[22]李芳慧,王飞,何佩琨等.TMS320C6000系列DsPS的原理与应用(第二版).电子工业出版社.2003.45-46
[23]Texas Instruments,Inc.Code Composer Studio Getting Started Guide 2001.88-97
[24]Ut-Va Koc,Ray Liu.DCT based Motion Estimation.IEEE Trans.on Image processing.Vol.7.No.7.Jul.1998.948-965.
[25]Texas Instruments,Inc.TMS320C6000 Optimizing C Compiler Tutorial.2002.9.396-415
[26]Texas Instruments,Inc.MS320C6000 Optimizing Compiler User Guide.2002.10.453-459
[27]Texas Instruments,Inc.External Programming of the TMS320C64x EDMA for Low Overhead Data Transfers.2004.7.115-123
[28]Texas Instruments,Inc.TMS320C6x Peripherals Support Library User's Guide 2000.267-288
[29]Olivier.Avaro.ITU-T Recommendation Video coding for low bit rate communication.Signal Processing.Image Communication Jan.Vol.15.2000.281-298.
[30]P.Pirsch.H.-J.Stolberg.VLSI implementation of image and video multimedia processing systems.IEEE Trans.Circuits Systems Video Techno.1999.11.878-891.
[31]Texas Instruments,Inc.External Programming of the TMS320C64x EDMA for Low Overhead Data Transfers.2004.7.546-588
[32]Texas Instruments,Inc.MS320C6x Peripherals Support Library User's Guide 2000.268-289
[2]胡国荣.数字视频压缩及其标准.北京广播学院出版社.1999.13-28
[3]沈兰荪,卓立,田栋,汪孔桥.视频编码与低速率传输.电子工业出版社.2001.17-26
[4]余兆明,李晓飞等.MPEG标准及其应用.北京邮电大学出版社.2002.32-45
[5]Olivier.Avaro.ITU-T Recommendation Video coding for low bit rate communication.Signal Processing:Image Communication Jan.Vol.15.2000.281-298.
[6]谭仕勇,戴琼海,陈慧蓉MPEG-4标准中的视频编码.广播与电视技术.2001.6:51-57
[7]夏天,陈磊,余兆明MPEG-4的基于内容的视频编码技术.电子工程师.1999.10:9-10
[8]田元,叶秀清,顾伟康.实时图像处理系统中的优化编程.电子技术应用.2000.10.43-46.
[9]王栋,张兆杨等.基于TMS320C6000 DSP的视频解码器设计及其硬件实现方案.电子技术应用.2002.2.78-80.
[10]章毓晋.图像处理与分析 清华大学出版社.1999.25-28
[11]Kenneth R.Castleman(美),朱志刚,林学訚,石定机译.数字图像处理.电子工业出版社.2002.31-35
[12]S.Zhu,K.Ma New Diamond Search Algorithm for Fast Blocking matching Motion Estimation IEEE.Trans.Image Processing.2000.Feb.vol.9.No.2.287-290.
[13]J.Y.Tham,S.Ranganath,M.Ranganath.A novel unrestricted center-biased diamond search algorithm for block motion estimation.IEEE Trans.on Circ.And Sys.vol.8.No.4.August 1998.369-377.
[14]Texas Instruments,Inc.TMS320DM642 Technical Overview.2002.9.156-163
[15]Texas Instruments,Inc.TMS320DM642 Evaluation Module Technical Reference.2003.35-37
[16]周霖等.DSP信号处理技术的应用[M].国防工业出版社.2004.67-71
[17]Texas Instruments.Inc.TMS320C64x Image/Video Processing Library Programmer Reference.2003.10.39-56
[18]Texas Instruments,Inc.MS320C6000 Peripherals.2001.230-258
[19]Texas Instntments.Inc.MS320C6x Peripherals Support Library User's Guide.2000.24-39
[20]钟玉琢,王琪,贺玉文.基于对象的多媒体数据压缩编码国际标准—MPEG及其校验模型.科学出版社.2000.10.66-68
[21]Texas Instruments,Inc.MS320C64X DSP Two-level Internal Memory Reference Guide 2002.1.25-31
[22]李芳慧,王飞,何佩琨等.TMS320C6000系列DsPS的原理与应用(第二版).电子工业出版社.2003.45-46
[23]Texas Instruments,Inc.Code Composer Studio Getting Started Guide 2001.88-97
[24]Ut-Va Koc,Ray Liu.DCT based Motion Estimation.IEEE Trans.on Image processing.Vol.7.No.7.Jul.1998.948-965.
[25]Texas Instruments,Inc.TMS320C6000 Optimizing C Compiler Tutorial.2002.9.396-415
[26]Texas Instruments,Inc.MS320C6000 Optimizing Compiler User Guide.2002.10.453-459
[27]Texas Instruments,Inc.External Programming of the TMS320C64x EDMA for Low Overhead Data Transfers.2004.7.115-123
[28]Texas Instruments,Inc.TMS320C6x Peripherals Support Library User's Guide 2000.267-288
[29]Olivier.Avaro.ITU-T Recommendation Video coding for low bit rate communication.Signal Processing.Image Communication Jan.Vol.15.2000.281-298.
[30]P.Pirsch.H.-J.Stolberg.VLSI implementation of image and video multimedia processing systems.IEEE Trans.Circuits Systems Video Techno.1999.11.878-891.
[31]Texas Instruments,Inc.External Programming of the TMS320C64x EDMA for Low Overhead Data Transfers.2004.7.546-588
[32]Texas Instruments,Inc.MS320C6x Peripherals Support Library User's Guide 2000.268-289