Adaptive MB size selecting based video error concealment algorithm

Abstract

At present, most commonly used techniques in video encoding standards are based on motion compensation, discrete cosine and variable length encoding techniques. These techniques have the advance of improving the compression ratio, but at the same time, the compressed data stream will be very sensitive to transmission errors. In the error-prone network, the frame transmission error, which caused by the amount of video data and the noise of transmission channel, will result in a decline in the quality of received video, and affect its viewing effect. Error concealment techniques have the ability to keep the video quality as much as possible in the conditions of not changing the structure of video stream in the case of transmission errors taking place. To reduce the impact of errors on video quality, an adaptive macro-block (MB) size selection error concealment algorithm that based on edge matching rule is proposed in this paper. By means of using the adjacent macro-block information and the edge matching principle to adaptively select MB size, and based on the distribution and the edge direction of the video frame, made use of the spatial correlation of the image to recover the motion vector of corrupted macro-block and keep the edge feature at the same time. The experimental results demonstrate that the proposed algorithm is superior to common used concealment algorithm of H.264 in both subjective and objective quality.