Reconstruction of lost image blocks in wireless transmission & compression using structure and texture filling-in

Abstract

In wireless scenarios, the image is transmitted over the wireless channel block by block. Due to severe fading, we may lose an entire block, even several consecutive blocks of an image. We aim to reconstruct the lost data using correlation between the lost block and its neighbors. The basic idea is to first automatically classify the block as textured or structured (containing edges), and then fill-in the missing block with information propagated from the surrounding pixels. If the lost block contained structure, it is reconstructed using an image inpainting algorithm, while texture synthesis is used for the textured blocks. We also combine this approach with JPEG compression itself, where the encoder voluntarily skips blocks, and these are reconstructed at the decoder in the same fashion as in the wireless scenario. The switch between the two schemes is done in a fully automatic fashion based on the surrounding available blocks. The performance of this method is tested for various images and combinations of lost blocks.