Impainting SONAR images by exemplar-based algorithm

Abstract

An essential problem in image processing is the recovery of lost or damaged data, which can be done by inpainting. In the recent years, a large variety of image inpainting algorithms have been proposed in the literature. This paper presents a comparison between two exemplar based image inpainting algorithms, in the case of SONAR images. A number of simulations on real SONAR images prove the effectiveness of these algorithms in reconstructing the affected parts of the images. The performance is analyzed in terms of the visual quality and comparing the Peak Signal to Noise Ratio.