Detection and Removal of Distorted Regions in Remote Sensing Images

Abstract: In sensor imaging, an existing open challenge problem is to restore a clear image from a sole remote sensor image. In our proposed system, we concentrate mainly on how to restore a motion distortion image which is caused due to satellite tracking. We have proposed a regularization based approach, which removes motion distortion from the sole image by carefully regularizing the sparsity of both the real image and the noise distorted image under a tight wavelet frame systems. Moreover, we have adapted a model of the Split Bregman method that effectively answers the occurrence of minimization issues in the system. The implementation of our proposed algorithm on both the original image and synthesized image proves that complex motion distortion is effectively removed from the satellite images and it does not require any information about the kernel in advance.

Keywords: image processing, satellite images, split bregman method, regularization etc.