Journal of South China University of Technology(Natural Science Edition)

• Electronics, Communication & Automation Technology • Previous Articles     Next Articles

Iterative Correlation Noise Refinement for Unidirectional Distributed Video Coding

WANG Jianpeng1,2 SONG Juan3 LIU Huan3   

  1. 1. Department of Mathematics and Physics,Changzhou Campus of Hohai University,Changzhou 213022,Jiangsu,China; 2. School of Artificial Intelligence,Xidian University,Xi'an 710071,Shaanxi,China; 3. School of Computer Science and Technology,Xidian University,Xi'an 710071,Shaanxi,China
  • Received:2018-11-15 Online:2019-04-25 Published:2019-03-01
  • Contact: 宋娟(1984-),女,博士,副教授,主要从事深度学习、图像处理等研究. E-mail:songjuan@mail.xidian.edu.cn
  • About author:王建鹏(1979-),男,在职博士生,讲师,主要从事神经网络、图像处理等研究. E-mail:wangjp1879@163. com
  • Supported by:
    Supported by the National Natural Science Foundation of China(61401324)

Abstract: An iterative correlation noise refinement (CNR) method was proposed for unidirectional distributed video coding (UDVC). In the iterative decoding process,the correlation noise was refined by using the previously reconstructed coefficients to improve the accuracy of correlation noise modeling. During the refinement,the correla- tion noise residuals were classified according to the decoding reliability and weighted refined respectively in order to avoid misleading refinement caused by wrongly decoded coefficients. Experimental results show that the reconstruc- tion quality is greatly improved after CNR. The average peak signal-to-noise ratio (PSNR) of reconstructed WZ frames from different video sequences is improved by 0. 32 ~ 0. 13 dB. Compared with UDVC without CNR,the average PSNR of the proposed UDVC with CNR is improved by 0. 21dB.

Key words: unidirectional distributed video coding, iterative decoding, correlation noise, noise refinement, cla- ssification

CLC Number: