Robust error concealment algorithm using iterative weighted boundary matching criterion

This paper proposes an iterative error concealment algorithm based on boundary matching criterion. For variable length coded data, a transmission error in a codeword affects not only the underlying codeword but also subsequent codewords. The objective of this approach is to minimize the total boundary error of whole lost macroblocks in one inter frame. In the proposed error concealment algorithm, extended boundary marching algorithm (BMA) is simply used in the first recovery stage. The recovered frame is used for following stage iteratively. However, to avoid the dependent recovery problem, uncertainty parameter is imposed on the boundary matching criterion as the condition that the current boundary matching information is correctly received one or recovered one. This error concealment stage iterates until the total sum of boundary errors in one frame is the minimum. The results show that the proposed iterative error concealment algorithm has advantages compared to conventional BMAs in a view of computation and image quality. The proposed iterative weighted BMA is much superior to other BMAs in both PSNR value and subjective image quality especially in the circumstance that succeeding two or more slices are lost together.