MPEG-4视频编解码DCT算法的实现研究
摘要
多媒体通信技术的迅速发展，要求视频通信能在保证质量的同时具有更低的码率。近年来MPEG-4以其低码率、基于对象的交互性等优点在多媒体领域表现出巨大的应用潜力，虽然标准中的部分相关技术还处于开发阶段，但基于简单框架(SP，Simple Profile)的MPEG-4视频压缩标准兼顾了传输码率与视频质量的要求，能在低码率、高压缩比的情况下获得较高的图象质量，因此在现代多媒体通信系统中SP视频框架成为MPEG-4应用的主轴。
尤其是随着信息电子和移动通信设备对视频应用的需求增长，低码率视频编解码技术的大量应用以硬件实现为主，尤其是更多的应用到嵌入式设备中，因此本论文以远程网络视频监控系统的开发为背景，重点研究MPEG-4简单框架视频编解码过程的嵌入式平台实现。
本文在研究MPEG-4视频关键技术及其特点的基础上，详细分析了MPEG-4 SP视频编解码算法的原理，用C语言程序实现了该算法并在PC机上对程序正确性及其视频效果进行测试，然后在linux操作系统的支持下，基于以ARCA2处理器为核心的嵌入式硬件平台，实现了编解码算法，同时从算法角度针对运动估计和DCT模块提出优化措施，尽可能提高算法的硬件实现速度。系统通过现场视频的测试，解码恢复后的视频利用Framebuffer技术显示，效果较好，同时达到低码率、视频恢复质量高、编码速度快的要求，在多媒体通信系统中有一定的应用价值。最后从芯片设计角度研究了DCT变换编码的ASIC实现，并提出相应的改进方案，在芯片的面积、速度、数据吞吐量上都达到较好的性能，为多媒体处理芯片的开发奠定了基础。
全文共分六章：第一章简单介绍了图象压缩的基础理论；第二章对MPEG-4标准视频部分的关键技术及特点做了全面的阐述；第三章详细分析了SP视频框架的算法原理；第四章基于对编解码方案的分析，完成算法的硬件实现，并提出优化措施；第五章针对DCT模块研究算法的ASIC实现，并作出相应的优化改进；第六章简要总结了本论文完成的工作。
With the rapid development of multimedia communication technology, low rate and high quality video coding schemes are attracting more attention. In recent years MPEG-4 is showing great values in multimedia application because of its low-rate and object-based interchangeability. Especially simple visual profile of MPEG-4 can efficiently solve the trade-off between bitrates and video quality with relatively simple tools and algorithms, therefore, it dominates MPEG-4 applications among multimedia communication system.
To meet the requirements of video application, MPEG-4 SP will be more implemented in embedded equipment. This paper attaches more importance to the embedded implementation and optimization of video coding and decoding in order to complete the development of remote video monitoring system.
In this paper, after studying the core technologies and characteristics of MPEG-4 visual part, the principles of MPEG-4 SP are analysed in detail. The C language implementation for these algorithms has been done in order to evaluate the quality of reconstructed video and to validate the correctness of the program.. Then by the embedded development platform based on ARCA2 processor, hardware implementation of algorithms is established. Meanwhile, many optimizations on the two key parts----Motion Estimation and DCT are presented in order to improve the speed of video coding and decoding. The experimental results indicate that the above hardware schemes can meet many actural requirements and have some value in multimedia system. At last, ASIC implementation of DCT is studied and many optimizations or improvements measures are presented, so a solid basement is formed for IC designing of multimedia processing.
The thesis is organized as following:
Chapter 1 briefly introduces basic theory about video coding and decoding. Chapter 2 analyzes characteristics and main technologys of MPEG-4 visual part in detail. Chapter 3 discusses the principles of MPEG-4 SP algorithm. Chapter 4 presents an hardware implementation scheme on the embedded development platform composed of ARCA2 processor. Chapter 5 completes the optimization and implementation of ASIC on DCT. Chapter 6 gives a brief conclusion of the thesis.
尤其是随着信息电子和移动通信设备对视频应用的需求增长，低码率视频编解码技术的大量应用以硬件实现为主，尤其是更多的应用到嵌入式设备中，因此本论文以远程网络视频监控系统的开发为背景，重点研究MPEG-4简单框架视频编解码过程的嵌入式平台实现。
本文在研究MPEG-4视频关键技术及其特点的基础上，详细分析了MPEG-4 SP视频编解码算法的原理，用C语言程序实现了该算法并在PC机上对程序正确性及其视频效果进行测试，然后在linux操作系统的支持下，基于以ARCA2处理器为核心的嵌入式硬件平台，实现了编解码算法，同时从算法角度针对运动估计和DCT模块提出优化措施，尽可能提高算法的硬件实现速度。系统通过现场视频的测试，解码恢复后的视频利用Framebuffer技术显示，效果较好，同时达到低码率、视频恢复质量高、编码速度快的要求，在多媒体通信系统中有一定的应用价值。最后从芯片设计角度研究了DCT变换编码的ASIC实现，并提出相应的改进方案，在芯片的面积、速度、数据吞吐量上都达到较好的性能，为多媒体处理芯片的开发奠定了基础。
全文共分六章：第一章简单介绍了图象压缩的基础理论；第二章对MPEG-4标准视频部分的关键技术及特点做了全面的阐述；第三章详细分析了SP视频框架的算法原理；第四章基于对编解码方案的分析，完成算法的硬件实现，并提出优化措施；第五章针对DCT模块研究算法的ASIC实现，并作出相应的优化改进；第六章简要总结了本论文完成的工作。
With the rapid development of multimedia communication technology, low rate and high quality video coding schemes are attracting more attention. In recent years MPEG-4 is showing great values in multimedia application because of its low-rate and object-based interchangeability. Especially simple visual profile of MPEG-4 can efficiently solve the trade-off between bitrates and video quality with relatively simple tools and algorithms, therefore, it dominates MPEG-4 applications among multimedia communication system.
To meet the requirements of video application, MPEG-4 SP will be more implemented in embedded equipment. This paper attaches more importance to the embedded implementation and optimization of video coding and decoding in order to complete the development of remote video monitoring system.
In this paper, after studying the core technologies and characteristics of MPEG-4 visual part, the principles of MPEG-4 SP are analysed in detail. The C language implementation for these algorithms has been done in order to evaluate the quality of reconstructed video and to validate the correctness of the program.. Then by the embedded development platform based on ARCA2 processor, hardware implementation of algorithms is established. Meanwhile, many optimizations on the two key parts----Motion Estimation and DCT are presented in order to improve the speed of video coding and decoding. The experimental results indicate that the above hardware schemes can meet many actural requirements and have some value in multimedia system. At last, ASIC implementation of DCT is studied and many optimizations or improvements measures are presented, so a solid basement is formed for IC designing of multimedia processing.
The thesis is organized as following:
Chapter 1 briefly introduces basic theory about video coding and decoding. Chapter 2 analyzes characteristics and main technologys of MPEG-4 visual part in detail. Chapter 3 discusses the principles of MPEG-4 SP algorithm. Chapter 4 presents an hardware implementation scheme on the embedded development platform composed of ARCA2 processor. Chapter 5 completes the optimization and implementation of ASIC on DCT. Chapter 6 gives a brief conclusion of the thesis.
引文
[1] 郑志航，惠新标等．数字电视原理与应用．北京：中国广播电视出版社，2001．
[2] 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——MPEG-4及其校验模型．北京：科学出版社，2000．
[3] 黎洪松等．数字视频技术及其应用．北京：清华大学出版社，1997．
[4] ISO/IEC14496-2. Coding of audio-visual objects——Part 2: Visual Amendmentl: Visualextensions (WG11N3056). Jan.,2000.
[5] 张宗念，余英林，马义德．MPEG-4的视频“档次”及其“等级”技术．电视技术．第209期．1999年11月．
[6] 陈禾，毛志刚，叶以正等．DCT快速算法及其VLSI实现．信号处理．第14卷(增刊)．Vol．14．1998年12月．
[7] M. Vetterli, H.J. Nussbaumer. Simple FFTandDCT Algorithms with Reduced Number of Operations. Signal Processing. Vol. 6. pp. 803-816. Aug.,1984.
[8] W.H. Chen, C.H. Smith, S. C. Fralick. A fast computational algorithm for the discrete cosine transform. IEEE Trans. Commun... vol. COM-25. pp. 1004-1009. Sept., 1977.
[9] C. Loeffler, A. Ligtcnberg, G.S. Moschytz. Practical fast 1-D DCT algorithms with 11 multiplications. Proc. IEEE ICASSP 89, pp. 988-991,1989.
[10] 涂亚明．基于块匹配的快速运动估计算法研究及应用．硕士学位论文．北京航空航天大学．2001年2月．
[11] 骆立俊，邹采荣等．视频编码中的块运动估计算法．电视技术．第12期，1997．
[12] ISO/IEC14496. MPEG-4 Video Verification Model version 14.0 (WG11N2932). Melboume. Oct.,1999.
[13] K.K. Ma, P.I. Hosur. Performance Report of Motion Vector Field Adaptive Search Technique (MVFAST) . MPEG2000/m5851. March, 2000.
[14] A.M. Tourapis, O.C. Au, M.L. Liou, G. Shen. Fast Block-Matching Motion Estimation using Predictive Motion Vector Field Adaptive Search Technique (PMVFAST). MPEG2000/m5866. March, 2000.
[15] A.M. Tourapis, O.C. Au, M. L. Liou. An Advanced Zonal Block Based Algorithm for Motion Estimation. Proc. IEEE International Conference on Image Processing. Vol. 2. pp. 610-614. Oct.,1999.
[16] ARCA Technology Corporration. Area2 CPU Core Reference Manual. Jan, 2003.
[17] ARCA Technology Corporration. GT2000 Area General Terminal Series SOC: Development Board Manual. Jan, 2003.
[18] C. Loeffler, A. Ligtenberg, G.S. Moschytz. Algorithm——Architecture Mapping for Custom DSP Chips. Proceeding IEEE International Symposium on Circuits and Systems. ISCAS-88. pp. 1953-1956. Helsinki. May, 1988.
[19] 洪波．基于对象的视频编码中纹理编码与运动估计的研究．博士学位论文．上海交通大学．2000年5月．
[20] J.Y. Tham, S. Ranganath, M. Ranganth. A Novel Unrestricted Center-Biased Diamond Search Algorithm for Block Motion Estimation. IEEE Trans. CASVT, Vol. 8, No. 4. August, 1998.
[21] A.M. Tourapis, G. Shen, M.L. Liou. A New Predictive Diamond Search Algorithm for Block Based Motion Estimation. Proc. of Visual Commu. And Image Processing 2000. Perth, Australia. June, 2000.
[22] http://www.linuxdoc.org/HOWTO/Framebuffer-HOWTO.html.
[23] Uwe Meyer-Baese著．刘凌，胡永生译．数字信号处理的FPGA实现．北京：清华大学出版社．2003年1月．
[24] Keshab K．Parhi(美)．VLSI数字信号处理系统设计与实现．北京：机械工业出版社．2003年10月．
[25] Ahmed Shams, Wendi Pan, Archana Chidanandan, Magdy A. Bayoumi. A Low Power High Performance Distributed DCT Architecture. Proc. IEEE Computer Society Annual Symposium on VLSI. Mar.,2002.
[26] 许琪，沈绪榜，钱刚等．LSRISC 32位浮点阵列乘法器的设计．微电子学与计算机．2001年第4期．
[27] Ahmed Shams, Magdy Bayoumi. A 108Gbps, 1.5Ghz 1D-DCT Architecture.
[28] A. Aggoun, I. Jalloh. Two-dimensional DCT/IDCT architecture. IEEE proc. Computer Digital Technology. Vol. 150, No. 1. Jan.,2003.
[29] STMicroelectronic. Mono and Colour Digital Video CMOS Image Sensors (VV5500 &VV6500) ,Customer Datasheet, Rev A. July, 2001.
[30] 刘振宇．二维条码识读器硬件平台的实现研究．硕士学位论文．厦门大学电子工程系． 2003年8月．
[31] Texas Instruments. TMS320C6000 EMIF to External FIFO Interface. May, 1999.
[32] 任丽香，马淑芬，李方慧．TMS320C6000系列DSPs的原理与应用．北京：电子工业出版社．2000年7月．
[33] Texas Instruments. TMS320C6000 Peripherals Reference Guide. Feb, 2001.
[34] Tommy Olsen. An improved image processing chain for mobile terminals. Graduate Thesis. Agder University, College Faculty of Engineering. May, 2002.
[35] Yao Wang，Jorn Ostermann著．候正信，杨喜等译．视频处理与通信．北京：电子工业出版社．2003年6月．
[2] 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——MPEG-4及其校验模型．北京：科学出版社，2000．
[3] 黎洪松等．数字视频技术及其应用．北京：清华大学出版社，1997．
[4] ISO/IEC14496-2. Coding of audio-visual objects——Part 2: Visual Amendmentl: Visualextensions (WG11N3056). Jan.,2000.
[5] 张宗念，余英林，马义德．MPEG-4的视频“档次”及其“等级”技术．电视技术．第209期．1999年11月．
[6] 陈禾，毛志刚，叶以正等．DCT快速算法及其VLSI实现．信号处理．第14卷(增刊)．Vol．14．1998年12月．
[7] M. Vetterli, H.J. Nussbaumer. Simple FFTandDCT Algorithms with Reduced Number of Operations. Signal Processing. Vol. 6. pp. 803-816. Aug.,1984.
[8] W.H. Chen, C.H. Smith, S. C. Fralick. A fast computational algorithm for the discrete cosine transform. IEEE Trans. Commun... vol. COM-25. pp. 1004-1009. Sept., 1977.
[9] C. Loeffler, A. Ligtcnberg, G.S. Moschytz. Practical fast 1-D DCT algorithms with 11 multiplications. Proc. IEEE ICASSP 89, pp. 988-991,1989.
[10] 涂亚明．基于块匹配的快速运动估计算法研究及应用．硕士学位论文．北京航空航天大学．2001年2月．
[11] 骆立俊，邹采荣等．视频编码中的块运动估计算法．电视技术．第12期，1997．
[12] ISO/IEC14496. MPEG-4 Video Verification Model version 14.0 (WG11N2932). Melboume. Oct.,1999.
[13] K.K. Ma, P.I. Hosur. Performance Report of Motion Vector Field Adaptive Search Technique (MVFAST) . MPEG2000/m5851. March, 2000.
[14] A.M. Tourapis, O.C. Au, M.L. Liou, G. Shen. Fast Block-Matching Motion Estimation using Predictive Motion Vector Field Adaptive Search Technique (PMVFAST). MPEG2000/m5866. March, 2000.
[15] A.M. Tourapis, O.C. Au, M. L. Liou. An Advanced Zonal Block Based Algorithm for Motion Estimation. Proc. IEEE International Conference on Image Processing. Vol. 2. pp. 610-614. Oct.,1999.
[16] ARCA Technology Corporration. Area2 CPU Core Reference Manual. Jan, 2003.
[17] ARCA Technology Corporration. GT2000 Area General Terminal Series SOC: Development Board Manual. Jan, 2003.
[18] C. Loeffler, A. Ligtenberg, G.S. Moschytz. Algorithm——Architecture Mapping for Custom DSP Chips. Proceeding IEEE International Symposium on Circuits and Systems. ISCAS-88. pp. 1953-1956. Helsinki. May, 1988.
[19] 洪波．基于对象的视频编码中纹理编码与运动估计的研究．博士学位论文．上海交通大学．2000年5月．
[20] J.Y. Tham, S. Ranganath, M. Ranganth. A Novel Unrestricted Center-Biased Diamond Search Algorithm for Block Motion Estimation. IEEE Trans. CASVT, Vol. 8, No. 4. August, 1998.
[21] A.M. Tourapis, G. Shen, M.L. Liou. A New Predictive Diamond Search Algorithm for Block Based Motion Estimation. Proc. of Visual Commu. And Image Processing 2000. Perth, Australia. June, 2000.
[22] http://www.linuxdoc.org/HOWTO/Framebuffer-HOWTO.html.
[23] Uwe Meyer-Baese著．刘凌，胡永生译．数字信号处理的FPGA实现．北京：清华大学出版社．2003年1月．
[24] Keshab K．Parhi(美)．VLSI数字信号处理系统设计与实现．北京：机械工业出版社．2003年10月．
[25] Ahmed Shams, Wendi Pan, Archana Chidanandan, Magdy A. Bayoumi. A Low Power High Performance Distributed DCT Architecture. Proc. IEEE Computer Society Annual Symposium on VLSI. Mar.,2002.
[26] 许琪，沈绪榜，钱刚等．LSRISC 32位浮点阵列乘法器的设计．微电子学与计算机．2001年第4期．
[27] Ahmed Shams, Magdy Bayoumi. A 108Gbps, 1.5Ghz 1D-DCT Architecture.
[28] A. Aggoun, I. Jalloh. Two-dimensional DCT/IDCT architecture. IEEE proc. Computer Digital Technology. Vol. 150, No. 1. Jan.,2003.
[29] STMicroelectronic. Mono and Colour Digital Video CMOS Image Sensors (VV5500 &VV6500) ,Customer Datasheet, Rev A. July, 2001.
[30] 刘振宇．二维条码识读器硬件平台的实现研究．硕士学位论文．厦门大学电子工程系． 2003年8月．
[31] Texas Instruments. TMS320C6000 EMIF to External FIFO Interface. May, 1999.
[32] 任丽香，马淑芬，李方慧．TMS320C6000系列DSPs的原理与应用．北京：电子工业出版社．2000年7月．
[33] Texas Instruments. TMS320C6000 Peripherals Reference Guide. Feb, 2001.
[34] Tommy Olsen. An improved image processing chain for mobile terminals. Graduate Thesis. Agder University, College Faculty of Engineering. May, 2002.
[35] Yao Wang，Jorn Ostermann著．候正信，杨喜等译．视频处理与通信．北京：电子工业出版社．2003年6月．