Error concealment via best matching block selection strategy

In this paper, we propose a new error concealment method based on best matching block selection strategy, which is implemented under a Bayesian probabilistic framework. Best selected image block is forwarded to a block-based bilateral filter for restoring missing pixels, while operates in block wise manner. During the Bayesian framework, a non-local spatial correlation is introduced in block selection. Meanwhile, a Markov Random Field Prior is built as an image prior model. Finally, experimental results show favorable results of proposed algorithm against traditional methods.