Spatial Error Concealment Based on Line Segmentation and Adaptive Edge Points Matching

A novel spatial error concealment algorithm with low-complexity is proposed in this paper for block losses in image and/or video communications. It is based on one dimensional (1-D) line segmentation and adaptive edge points matching, which can be divided into three steps. 1) A 1-D line growing based method is applied to obtain the edge points from the neighboring pixels of the lost areas. 2) The edge points will be adaptively matched in pairs and then connected to recover the lost edge information. 3) For each pixel in the lost areas, the neighboring intact pixels for interpolation will be searched by an up-down scanning procedure, after which the lost pixel will be interpolated using them. It has been shown that the proposed algorithm can recover the lost edge information efficiently. Compared with the existing algorithms, our proposed algorithm can achieve superior objective and subjective quality with even lower complexity.