Sequential error concealment via canonical correlation analysis

In this paper, we propose a new error concealment algorithm for video/image based on canonical correlation analysis (CCA). Motivated by the Intra prediction in H.264/AVC, it is reasonable to assume that there is a strong spatial correlation relationship between the lost regions and their known adjacent regions. Based on the above idea, we use CCA to estimate a correlation projection matrix which utilizes the loss of macro block adjacent spatial information, then we use the projection matrix and the adjacent region to estimate missing pixel area. In addition, in order to use the spatial information efficiently, we apply the neighbor-embedding-type weight into the aforementioned CCA model. Experimental results demonstrate that the proposed method improves the subjective and objective image quality to a large extent in comparison with other existing methods.