Object Removal and Sparsity Based Restoration in Images

Abstract

Object Removal is the research area in the field of image processing whose goal is to remove objects or restore the damaged regions in an undetectable manner. Filling the region of missing information from a signal using the neighboring information and then reforming the signal is the basic work of inpainting algorithms. Image is composed two components texture and structure. Applying object removal or inpainting algorithms alone on an image doesn’t produce an effective result. Because some of the algorithm concentrate on textural part and some others concentrate on structural part. This conclusion gives rise to the concept of decomposing the image to texture and structure and thus completing the work in two stages. In the first stage we do a structure completion and then it is followed by texture completion. Structure completion is done by inpainting the missing region using sparse reconstruction method. The second stage is texture completion and is performed on the structure intact image. It is done using an exemplar based inpainting technique so that the texture gets filled in an effective manner.