Fast Video Restoration Method based on Segmentation
Video restoration is quite a challenging problem in image processing. Beacause of the large quantity of data to be processed, it is very time-consuming. In this paper we proposed a convenient method to deal with the problem. Firstlv.The input video is segmented into background and moving foreground.Secondly, the background and foreground are repaired respectively by using the spatialtemporal coherence and periodic movement.For the background, we completed it along both the time and space axis.The former can make full use of the redundant information in the video and the latter can use belief propagation (BP) to propagate known information to unknown regions.By searching the best match along the time axis.the completion of the foreground can be achieved. So we can preserve motion coherence better and avoid flick and shake at the same time.
video restoration moving foreground spatialtemporal information
Minzhi Zheng Shu Liu
Computer Science College Sichuan University Chengdu, China Dept. of Aliababa Clound Computing Hangzhou, China
国际会议
成都
英文
371-374
2011-01-14（万方平台首次上网日期，不代表论文的发表时间）