Temporal Shape Error Concealment for Video Objects

YU Ye,XIE Xu-dong,LU Jian-hua,ZHENG Jun-li,CHEN Chang-wen,于烨,谢旭东,陆建华,郑君里,陈长文
2008-01-01
Abstract:A novel temporal shape error concealment technique is proposed, which can be used in the context of object-based video coding schemes. In order to reduce the effect of the shape variations of a video object, the curvature scale space (CSS) technique is adopted to extract features, and then these features are used for boundary matching between the current frame and the previous frame. Because the temporal, spatial and statistical video contour information are all considered, the proposed method can find the optimal matching, which is used to replace the damaged contours. The simulation results show that the proposed algorithm achieves better subjective, objective qualities and higher efficiency than those previously developed methods.
What problem does this paper attempt to address?