Strategic Selections of Temporal and Spatial Error Concealment Methods using Cooperative Game Structure

Abstract

In the error-prone transmission environment, the quality of a video sequence may be damaged because of noise or packet loss. Moreover, since more and more real time video stream applications are needed, high compression rates of video streams have been reached by some video coding standards such as H.264 or MPEG4. It will make the video stream more vulnerable against stream error, and the error will also propagate to the successive frames. In order not to increase the complexity of the encoder, error concealment techniques were proposed, and it is a post-processing concept on the decoder. It utilizes the temporal or spatial information to recover the damaged data, and more and more temporal, spatial or even hybrid error concealment methods were proposed over the years. We propose two temporal error concealment methods, which use bilateral filter as a tool to reconstruct the lost macro-block. Bilateral filter is mainly used to deal with de-noising problem. Therefore, we apply the de-noising algorithm to tackle our error concealment problem. Afterward, we propose a cooperative game theoretic error concealment structure to adaptively select the proper algorithms to reconstruct the error macro-block. By applying the cooperative game theory, we can effectively take advantages of both temporal and spatial information to do error concealment. Therefore, the result using our method can achieve better visual quality compared with using single error concealment method. Finally, because the blocking effect may occur after replacing the missing macro-block with the concealed macro-block; therefore, we also introduce a de-blocking scheme to detect and remove the blocking effect for better image quality