关键词: 图像通信, 空域差错掩盖, 突发块丢失, 信号延拓

Abstract:

A block-coded image over error-prone networks may result in burst block loss, which drastically degrades the quality of the received image. The spatial error concealment techniques are accepted to alleviate the degradation caused by the missing blocks. In this paper, we propose a spatial-frequency selective extrapolation (SFSE) algorithm that introduces the prior knowledge into the sparse modeling to select the required support area with suitable base functions and optimal expansion coefficients by a distance-adaptive method, so as to conceal the damaged blocks in burst error situations. Experimental results demonstrate that the proposed SFSE algorithm obtains a gain 0.5 dB higher on average than that by the second-best KMMSE algorithm, and effectively improves the subjective quality of concealed image with appropriate complexity.

Key words: simage communications, spatial error concealment, burst block loss, signal extrapolation