Inpainting for satellite imagery using thin plate spline radial basis function neural networks in shearlet domain

Abstract

Remotely sensed images are occasionally subject to the problem of dead pixels. Removing or filling dead pixels is known as image inpainting. In this paper, thin plate spline radial basis function based inpainting in the Shearlet domain is presented to solve this problem. The proposed algorithm takes the advantage of a Discrete Shearlet Transform (DST) which has a good performance with textured and non textured images. Additionally, the algorithm works well for multichannel remote sensing data. The algorithm is tested with the QuickBird Satellite images for inpainting. The test results and the computed Peak Signal to Noise Ratio (PSNR) values demonstrate the feasibility of the use of Radial Basis Function (RBF) method for inpainting of remotely sensed images.