中文摘要：

在视频图像的内容运动剧烈或者场景切换的情况下,解码图像出现的差错无法用时域上的信息加以恢复,利用当前解码帧内的空域相关性来进行差错掩盖就成为关键技术。本文从匹配特征矢量和块相似性统计指标出发,提出一种基于模糊聚类的空域差错掩盖的新方法。实验表明,与传统的块匹配和多方向插值掩盖方法相比,该方法改善了恢复图像的边缘细节,提高了重建图像的质量。

英文摘要：

As no temporal reference information can be obtained when the video content moves fast or switches to another scene, error concealment using spatial correlation becomes the key technology for reducing the degradation of reconstruct image. In this paper, we proposed a novel spatial error concealment algorithm based on fuzzy clustering. This algorithm gives a new selection of feature vector and similarity measurement. The simulation results show that compared with the classical algorithms based on block-matching principle or multi-directional interpolation, the proposed algorithm can improve the edge details and give a better quality of the recovered image.