Image decomposition model combined with sparse representation and total variation

Abstract

In this paper, we propose a new decomposition model combined with sparse representation and total variation (SRTV), which allows us to separate cartoon and texture components from an image. The SRTV model naturally fits into the framework of separation and produces separated layers, meanwhile, denoising and inpainting process appears as the byproducts. Therefore, the new approach incorporates separation, denoising, and inpainting as a unified framework. We demonstrate the performance of the new approach through several examples.