Automatic multiband SAR image registration using sparse-based despeckling and Affine Scale Invariant transfrom

A new method of automatic multiband SAR image registration based on Affine Scale Invariant and sparse-based despeckling is proposed. It first employs the sparse representation-based despeckling method to erase affection of speckle noise on image registration. Second, Normalized gradient feature space is designed to cope with affection of different grey value brought multiband SAR. Affine scale invariant transformation (Affine SIFT) is adopted to extract invariant feature and match them; RANSAC model is adopted to exclude the mismatched points. The image transformation model and parameter estimation using the rule of least root mean square error (RMSE) are discussed. The method is accurate, robust, and fast and has been tested on a lot of images. Experimental results using real multiband SAR images demonstrate the proposed method can achieve impressive results.