Similarity evaluation algorithm for concealing frame loss error in multi-view video transmission

Abstract

An efficient error concealment algorithm for reconstructing the loss frame in multi-view video sequences at the decoder is presented. The proposed method has a capacity of adopting the appropriate error concealment mode to recover the loss frame according to its reference frames similarity level. There are three modes in the method. For a high similarity level, the mode using temporal related frame copy (FC) to recover the lost frame is employed. When the similarity level is low, the spatially related frame is copied. Adapting to the medium similarity condition, both the similar MBs in temporal frames and the matching MBs in spatial frames are used to reconstruct the loss frame. And block matching principle is employed for filling empty blocks. Experimental results have shown that the proposed multi-view video error concealment method can effectively reconstruct the lost frames and has more attractive performance than previous algorithms.