Abstract

This work aims at the study and implementation of occlusion removal. Occlusion is a block which disturbs eyesight. In order to get a better clarity we remove the block. It involves removing the block and filling the gap left behind within the image. There are so many varieties of algorithms. First, segment the image into structure and texture image. The sparse technique is used for decomposition of the image and reconstruction of the structure image. Texture image is reconstructed based on anisotropic diffusion. Finally, the structure and texture image is recombined and we get an image where occlusion is filled. This paper includes texture synthesis, exemplar based method and hybrid method for occlusion removal, sparse method, PDE method. The results are found better since the patches selected are from the source region, and also regularization is carried out during diffusion. This paper aims at the study of various types of inpainting techniques, and how efficiently they work.