Block loss recovery techniques for image and video communications

Block loss recovery technique is a widely used method on most coding system when error happens during transmission. We discuss the effect of transmission error and corresponding block loss and recovery. To be suitable for the restoration on most coding standards, we suggest a reliable spatial interpolation block recovery algorithm improving restored image quality. The proposed procedure consists of image conditioning, a convex hull forming, and the method of alternating projections. In order to achieve the best restoration quality, a convex hull and convex sets are formed according to local image structure. Surrounding vector and recovery vector are introduced for interpolation reference and missing points. We have evaluated the proposed algorithm with other existing methods in severe block losses. We discuss an interpolation method in the case of consecutive block losses. The experimental results show the proposed spatial interpolation scheme provides good restored image quality, and outperforms other methods. We also present a temporal block loss restoration technique including motion vector recovery and spatial motion compensation. Proposed motion vector recovery utilizes local image structure of a missing blocks and adaptively imposes weights to the distortion metric to be suitable to the local condition. Proposed method also utilizes propagation and restoration error. The weights of the distortion metric for the propagation error are adaptively changed according to the error state. For the fast implementation, an evolutionary algorithm based fast search method is presented. Experimental results shows the proposed algorithm outperforms existing schemes. We apply the proposed algorithm to the image coding and the packet priority setting problem for the inter-active solution of image transmissions. The presented interpolation based coding and packet priority setting method is compatible with the widely accepted image and video coding standards such as JPEG, H.261, H.263+, H.26L, MPEG-1, MPEG-2, and MPEG-4.