分布式视频编码边信息研究
摘要
随着3G通信技术的飞速发展,出现了大量的低能耗的视频摄取设备,由于这些设备的内存和电池能量有限,他们要求低复杂度的编码算法,这对传统的高复杂度的编码算法提出了挑战。
分布式视频编码(Distributed Video Coding, DVC)是一种新型的视频编码方式,由于将计算量大的运动估计去相关算法从编码端移动到了解码端,实现了低复杂编码算法,这样的特点使其非常适用于无线视频通信,因此,DVC逐渐成为视频编码领域的研究热点。传统视频编码成功的一个重要因素就是编码端采用了运动估计算法,相比之下,现有的DVC方案将运动估计转移到了解码端,利用解码端的恢复帧来进行运动估计以产生边信息,然而,不准确的恢复帧造成了运动估计的不准确,这将引起边信息的性能下降,最终造成DVC性能的下降。因此,为了提高DVC的率失真性能,本文将研究如何提高运动估计性能,以获得较准确的边信息。
本文首先在DVC理论基础之上,研究了传统的几种运动估计方法,重点讨论了基于块匹配运动估计的几种快速搜索算法的算法复杂度和率失真性能;其次研究了现有的几种边信息产生方法,重点介绍了经典的运动补偿时域内插产生边信息的方法和最新的基于最大后验概率的期望最大化算法(Expectation Maximization,EM),并做实验比较了每种方法的性能,得出了EM算法产生的边信息的率失真性能最高。由于现有的EM算法有学习运动矢量时间比较长的缺陷,在此基础上,提出了将菱形搜索与EM算法相结合的技术方案,实验表明,在保持较高的率失真性能情况下,缩短了学习时间。通过研究EM算法,发现在运动矢量概率模型的学习过程中,初始的概率模型起到一个很关键的作用,因此,我们采用了基于概率自适应的EM算法,实验结果表明,采用概率自适应的EM算法在保持同样的率失真性能下,比先前的EM算法缩短了学习时间。
With the rapid development of 3-Generation communication, a lot of low power video devices appear. Due to the limit storage and battery power, these devices require low-complexity encoding, which poses challenge to traditional video coding algorithms.
Recently, distributed video coding (DVC) is a new video encoding, due to the calculation of large correlation algorithm of motion estimation move from the encoder to the decoder to achieve a low-complexity encoding algorithm, this feature makes it reality for wireless video communications. Thus it has been the research hotspot in the field of video coding. Traditional video coding owes one success to the use of motion estimation algorithm in encoder, the existing motion estimation DVC framework will be transferred to the decoding end, which using the decoded recovery frame in the decoder to estimate the motion to generate the side information. However, the inaccurate recovery frame will cause the inaccurate motion estimation, this will result in degradation of the performance of side information and the DVC ultimately. Therefore, in order to improve the rate distortion performance of DVC, the paper will examine how to improve the performance of motion estimation in order to obtain more accurate side information.
First,the article based on DVC theory introduces several studies of the traditional motion estimation methods, focusing on discuss block matching motion estimation based on several fast searching algorithm complexity and rate distortion performance. Second, it researches several of the existing methods for generating the side information, focusing on the classical motion compensation interpolation in time domain method of generating side information and the latest expectation maximization (EM) algorithm based on the maximum a posteriori probability, and do Experimental Comparison of the performance of each method, As can be seen, the side information that EM algorithm generates obtains the highest rate-distortion performance. As EM algorithm learning motion vectors needs a long time in this basis, the technical program combines the proposed diamond search to the EM algorithm, and experimental results show that it can shorten the learning time under maintaining a high rate distortion performance of the case. By studying the algorithm, we find that the initial probability model plays a key role in the motion vectors probability learning process. Therefore, the adaptive probability-based EM algorithm is presented. Experimental results show that compared with previous EM algorithm, the algorithm we presented reduces the learning time under maintaining the same rate-distortion performance of the case.
分布式视频编码(Distributed Video Coding, DVC)是一种新型的视频编码方式,由于将计算量大的运动估计去相关算法从编码端移动到了解码端,实现了低复杂编码算法,这样的特点使其非常适用于无线视频通信,因此,DVC逐渐成为视频编码领域的研究热点。传统视频编码成功的一个重要因素就是编码端采用了运动估计算法,相比之下,现有的DVC方案将运动估计转移到了解码端,利用解码端的恢复帧来进行运动估计以产生边信息,然而,不准确的恢复帧造成了运动估计的不准确,这将引起边信息的性能下降,最终造成DVC性能的下降。因此,为了提高DVC的率失真性能,本文将研究如何提高运动估计性能,以获得较准确的边信息。
本文首先在DVC理论基础之上,研究了传统的几种运动估计方法,重点讨论了基于块匹配运动估计的几种快速搜索算法的算法复杂度和率失真性能;其次研究了现有的几种边信息产生方法,重点介绍了经典的运动补偿时域内插产生边信息的方法和最新的基于最大后验概率的期望最大化算法(Expectation Maximization,EM),并做实验比较了每种方法的性能,得出了EM算法产生的边信息的率失真性能最高。由于现有的EM算法有学习运动矢量时间比较长的缺陷,在此基础上,提出了将菱形搜索与EM算法相结合的技术方案,实验表明,在保持较高的率失真性能情况下,缩短了学习时间。通过研究EM算法,发现在运动矢量概率模型的学习过程中,初始的概率模型起到一个很关键的作用,因此,我们采用了基于概率自适应的EM算法,实验结果表明,采用概率自适应的EM算法在保持同样的率失真性能下,比先前的EM算法缩短了学习时间。
With the rapid development of 3-Generation communication, a lot of low power video devices appear. Due to the limit storage and battery power, these devices require low-complexity encoding, which poses challenge to traditional video coding algorithms.
Recently, distributed video coding (DVC) is a new video encoding, due to the calculation of large correlation algorithm of motion estimation move from the encoder to the decoder to achieve a low-complexity encoding algorithm, this feature makes it reality for wireless video communications. Thus it has been the research hotspot in the field of video coding. Traditional video coding owes one success to the use of motion estimation algorithm in encoder, the existing motion estimation DVC framework will be transferred to the decoding end, which using the decoded recovery frame in the decoder to estimate the motion to generate the side information. However, the inaccurate recovery frame will cause the inaccurate motion estimation, this will result in degradation of the performance of side information and the DVC ultimately. Therefore, in order to improve the rate distortion performance of DVC, the paper will examine how to improve the performance of motion estimation in order to obtain more accurate side information.
First,the article based on DVC theory introduces several studies of the traditional motion estimation methods, focusing on discuss block matching motion estimation based on several fast searching algorithm complexity and rate distortion performance. Second, it researches several of the existing methods for generating the side information, focusing on the classical motion compensation interpolation in time domain method of generating side information and the latest expectation maximization (EM) algorithm based on the maximum a posteriori probability, and do Experimental Comparison of the performance of each method, As can be seen, the side information that EM algorithm generates obtains the highest rate-distortion performance. As EM algorithm learning motion vectors needs a long time in this basis, the technical program combines the proposed diamond search to the EM algorithm, and experimental results show that it can shorten the learning time under maintaining a high rate distortion performance of the case. By studying the algorithm, we find that the initial probability model plays a key role in the motion vectors probability learning process. Therefore, the adaptive probability-based EM algorithm is presented. Experimental results show that compared with previous EM algorithm, the algorithm we presented reduces the learning time under maintaining the same rate-distortion performance of the case.
引文
[1] D. Slepian and J. K. Wolf. Noiseless coding of correlated information sources.IEEE Transaction on Information Theory, Vol.19, July 1973: 471-480.
[2] B. Griod, A. Aaron, S. Rane.Distirbuted video coding, proceedings of the IEEE, Vol,93,No. 1, January 2005, pp.71-83.
[3]陈爽文.分布式视频编码[J].中国传媒大学, 2007.
[4]干宗良,朱秀昌.分布式视频编码技术的研究现状及其展望[J].信号处理,2007, 23(1).
[5]王安红.分布式视频编码研究[D].北京:北京交通大学, 2009.
[6] R. G. Gallager.Low density parity check codes, IEEE Trans. Inform.Theory, vol. IT-8, Jan. 1962, pp. 21–28.
[7]周伟.低密度奇偶校验码译码研究及其应用[D].北京:北京邮电大学, 2007.
[8] Aaron, Rui Zhang and B. Griod. Wyner-Ziv coding of motion video, presented at the Asilomar Conf. Signals and Systems, Pacific Grove, CA, 2002.
[9] Aaron, E. Setton, and B. Girod. Towards practical Wyner-Ziv coding of video, in Proc. IEEE International Conference on Image Processing, Barcelona, Spain, Sept.
[10] Aaron, S.Rane, E.Setton and B.Griod. Transform-domain Wyner-Ziv codec for video[C], In Proceedings of Visual Communications and Image Processing, 2004.
[11]郑义.基于优化TCQ的分布式视频编码[D].太原:太原科技大学, 2010.
[12] Horn B, Schunch B. Determining optical flow[J].Artificial Intelligence,1981,17: 185-203.
[13]黎洪松.数字视频处理[M].北京:北京邮电大学出版社, 2006.
[14]干宗良.基于视频压缩的分布式编码技术研究[D].南京:南京邮电大学, 2007.
[15]许磊.基于块匹配的序列图像运动估计算法研究[D].山东:山东大学,2007.
[16] Aroh Barjatya. Block Matching Algorithm for Motion Estimation. IEEE Transactions Evolution Computation, 2004. 8(3): p. 225-239.
[17] A. Wyner and J. Ziv.The rate-distortion function for source coding with side information at the decoder. IEEE Transaction on Information Theory,Vol. 22, No. 1, January 1976: 1-10.
[18]史萍,罗坤.分布式视频编码中边信息的产生[J].电视技术, 2010, 4(11).
[19]曾宪科.分布式视频编码中边信息的研究[D].成都:电子科技大学,2009.
[20]张婷,尹明. Wyner-Ziv视频编码中边信息估计研究[J].计算机工程与应用,2010, 46(34).
[21]刘艳红.分布式视频编码中基于块的运动补偿插值边信息估计算法研究[D].西安:西安电子科技大学, 2010.
[22] A.Aaron, S. Rane, B. Griod. Wyner-Ziv video coding with hash-based motion compensation at the receiver[C], In: Proceedings of IEEE Int. Conf. Image Processing, Singapore, 2004.
[23] D. Varodayan, A. Mavlankar, M. Flierl, B. Girod. Distributed coding of random dot stereograms with unsupervised learning of disparity, in: Proc. IEEE Internat.Workshop Multimedia Signal Processing, Victoria.
[24] D. Chen, D. Varodayan, M. Flierl, B. Girod. Distributed stereo image coding with improved disparity and noise estimation, in: Proc. IEEE Internat. Conf. Acoustic, Speech and Signal Processing, Las Vegas, NV, 2008
[25] D. Varodayan, Y.-C. Lin, A. Mavlankar, M. Flierl,B. Girod. Wyner-Ziv coding of stereo images with unsupervised learning of disparity, in: Proc. Picture Coding Symp., Lisbon, Portugal, 2007.
[26] D. Chen, D. Varodayan, M. Flierl, B. Girod. Wyner-Ziv coding of multiview images with unsupervised learning of disparity and Gray code, in: Proc. IEEE Internat. Conf. Image Processing, San Diego, CA, 2008, submitted.
[27] D. Varodayan, A. Mavlankar, M. Flierl, B. Girod. Distributed grayscale stereo image coding with unsupervised learning of disparity, in: Proc. IEEE Data Compression Conf., Snowbird, UT, 2007.
[28] A. Dempster, N. Laird, D. Rubin. Maximum likelihood from incomplete data via the EM algorithm, J.Royal Stat. Soc., Series B 39 (1) (1977) 1–38.
[29] F. R. Kschischang, B. J. Frey, H.-A. Loeliger. Factor graphs and the sum-product algorithm, IEEE Trans. Inform. Theory 47 (2) (2001) 498–519.
[30] A. Liveris, Z. Xiong, C. Georghiades. Compression of binary sources with side information at the decoder using LDPC codes, IEEE Commun. Lett. 6 (10) (2002) 440–442.
[31] D. Varodayan, A. Aaron, B. Girod. Rate-adaptive distributed source coding using low-density parity-check codes, in: Proc. Asilomar Conf. on Signals, Syst., Comput., Pacific Grove, CA, 2005.
[32] D. Varodayan, A. Aaron, B. Girod. Rate-adaptive codes for distributed source coding, EURASIP Signal Processing J. 86 (11) (2006) 3123–3130.
[33] D. Chen, D. Varodayan. Unsupervised Learning of Motion for Distributed Video Coding (2008).
[34] D.Varodayan, D. Chen, M.Flierl and B.Girod, Wyner-Ziv Coding of Video with Unsupervised Motion Vector Learning. Signal Processing: Image Communication, 2008. 23(5): p. 369-378.
[35] Young Min Kim, Stephanie Kwan, Karen Zhu. Distributed Video Coding with Unsupervised Learning of Motion Estimation.
[2] B. Griod, A. Aaron, S. Rane.Distirbuted video coding, proceedings of the IEEE, Vol,93,No. 1, January 2005, pp.71-83.
[3]陈爽文.分布式视频编码[J].中国传媒大学, 2007.
[4]干宗良,朱秀昌.分布式视频编码技术的研究现状及其展望[J].信号处理,2007, 23(1).
[5]王安红.分布式视频编码研究[D].北京:北京交通大学, 2009.
[6] R. G. Gallager.Low density parity check codes, IEEE Trans. Inform.Theory, vol. IT-8, Jan. 1962, pp. 21–28.
[7]周伟.低密度奇偶校验码译码研究及其应用[D].北京:北京邮电大学, 2007.
[8] Aaron, Rui Zhang and B. Griod. Wyner-Ziv coding of motion video, presented at the Asilomar Conf. Signals and Systems, Pacific Grove, CA, 2002.
[9] Aaron, E. Setton, and B. Girod. Towards practical Wyner-Ziv coding of video, in Proc. IEEE International Conference on Image Processing, Barcelona, Spain, Sept.
[10] Aaron, S.Rane, E.Setton and B.Griod. Transform-domain Wyner-Ziv codec for video[C], In Proceedings of Visual Communications and Image Processing, 2004.
[11]郑义.基于优化TCQ的分布式视频编码[D].太原:太原科技大学, 2010.
[12] Horn B, Schunch B. Determining optical flow[J].Artificial Intelligence,1981,17: 185-203.
[13]黎洪松.数字视频处理[M].北京:北京邮电大学出版社, 2006.
[14]干宗良.基于视频压缩的分布式编码技术研究[D].南京:南京邮电大学, 2007.
[15]许磊.基于块匹配的序列图像运动估计算法研究[D].山东:山东大学,2007.
[16] Aroh Barjatya. Block Matching Algorithm for Motion Estimation. IEEE Transactions Evolution Computation, 2004. 8(3): p. 225-239.
[17] A. Wyner and J. Ziv.The rate-distortion function for source coding with side information at the decoder. IEEE Transaction on Information Theory,Vol. 22, No. 1, January 1976: 1-10.
[18]史萍,罗坤.分布式视频编码中边信息的产生[J].电视技术, 2010, 4(11).
[19]曾宪科.分布式视频编码中边信息的研究[D].成都:电子科技大学,2009.
[20]张婷,尹明. Wyner-Ziv视频编码中边信息估计研究[J].计算机工程与应用,2010, 46(34).
[21]刘艳红.分布式视频编码中基于块的运动补偿插值边信息估计算法研究[D].西安:西安电子科技大学, 2010.
[22] A.Aaron, S. Rane, B. Griod. Wyner-Ziv video coding with hash-based motion compensation at the receiver[C], In: Proceedings of IEEE Int. Conf. Image Processing, Singapore, 2004.
[23] D. Varodayan, A. Mavlankar, M. Flierl, B. Girod. Distributed coding of random dot stereograms with unsupervised learning of disparity, in: Proc. IEEE Internat.Workshop Multimedia Signal Processing, Victoria.
[24] D. Chen, D. Varodayan, M. Flierl, B. Girod. Distributed stereo image coding with improved disparity and noise estimation, in: Proc. IEEE Internat. Conf. Acoustic, Speech and Signal Processing, Las Vegas, NV, 2008
[25] D. Varodayan, Y.-C. Lin, A. Mavlankar, M. Flierl,B. Girod. Wyner-Ziv coding of stereo images with unsupervised learning of disparity, in: Proc. Picture Coding Symp., Lisbon, Portugal, 2007.
[26] D. Chen, D. Varodayan, M. Flierl, B. Girod. Wyner-Ziv coding of multiview images with unsupervised learning of disparity and Gray code, in: Proc. IEEE Internat. Conf. Image Processing, San Diego, CA, 2008, submitted.
[27] D. Varodayan, A. Mavlankar, M. Flierl, B. Girod. Distributed grayscale stereo image coding with unsupervised learning of disparity, in: Proc. IEEE Data Compression Conf., Snowbird, UT, 2007.
[28] A. Dempster, N. Laird, D. Rubin. Maximum likelihood from incomplete data via the EM algorithm, J.Royal Stat. Soc., Series B 39 (1) (1977) 1–38.
[29] F. R. Kschischang, B. J. Frey, H.-A. Loeliger. Factor graphs and the sum-product algorithm, IEEE Trans. Inform. Theory 47 (2) (2001) 498–519.
[30] A. Liveris, Z. Xiong, C. Georghiades. Compression of binary sources with side information at the decoder using LDPC codes, IEEE Commun. Lett. 6 (10) (2002) 440–442.
[31] D. Varodayan, A. Aaron, B. Girod. Rate-adaptive distributed source coding using low-density parity-check codes, in: Proc. Asilomar Conf. on Signals, Syst., Comput., Pacific Grove, CA, 2005.
[32] D. Varodayan, A. Aaron, B. Girod. Rate-adaptive codes for distributed source coding, EURASIP Signal Processing J. 86 (11) (2006) 3123–3130.
[33] D. Chen, D. Varodayan. Unsupervised Learning of Motion for Distributed Video Coding (2008).
[34] D.Varodayan, D. Chen, M.Flierl and B.Girod, Wyner-Ziv Coding of Video with Unsupervised Motion Vector Learning. Signal Processing: Image Communication, 2008. 23(5): p. 369-378.
[35] Young Min Kim, Stephanie Kwan, Karen Zhu. Distributed Video Coding with Unsupervised Learning of Motion Estimation.