基于联合估计的Wyner-Ziv系统边信息重建

摘要

In order to improve the quality of Side Information(SI), without significantly increase the complexity at the same time, a new SI generation algorithm jointly using full-pixel and half-pixel Motion Compensation Interpolation(MCI) is proposed.A quality decision factor namely SI confidence is proposed to indicate the quality of SI.According to the SI confidence whether overtakes the threshold, the algorithm adaptively switches between full-pixel MCI and half-pixel MCI.Experimental results indicate that the proposed algorithm can significantly reduce the decode complexity without causing efficiency loses when compared with traditional SI reconstruction based on sub-pixel method.%在不显著增加解码复杂度的前提下,为提高Wyner-Ziv视频编码系统的边信息(SI)重建质量,提出一种基于SI信度的整像素/半像素联合估计算法.该算法引入一个SI质量判定因子判断边信息的重建质量,根据SI置信度是否超过设定的判别阈值进行整像素和半像素边信息重建方法的自适应切换.实验结果表明,在保持甚至超过传统亚像素边信息重建算法性能的基础上,该算法明显降低了系统解码复杂度.