TV Regularized Reweighted Joint Low-Rank and Sparse Decomposition for Pansharpening

Pansharpening is the fusion of high-resolution Panchromatic (PAN) and the corresponding lower resolution Multispectral (MS) imagery to create a single high-resolution color image. The proposed method improves the quality of pansharp-ened images. It utilizes total variation regularized reweighted joint low-rank and sparse decomposition with an assumption that multiple data have a common low-rank component. The proposed method decomposes each data into common low-rank component, specific low-rank component and specific sparse component. Details Injection (DI) type pansharpening is employed here. The spatial details to be injected are calculated as the linear combination of decomposed components of PAN image. To reduce unwanted details and to preserve important details such as edges, Total Variation (TV) regularization is employed. Also, an iterative reweighting scheme is utilized for enhancing low-rank and sparsity simultaneously. The experimental results conducted on several data sets reveals that the proposed method provide better results compared with the state-of-art methods.