Registration of Remote Sensing Images with Steerable Pyramid Transform and Robust SIFT Features

This paper proposes an automatic registration technique based on steerable pyramid transform and robust shift invariant feature transform (SIFT) features, which can deal with the large variations of scale, rotation and illumination of the images. First, the steerable pyramid transform is used to the two input images and the sub-band images along certain orientations are obtained at each layer. Then, a robust SIFT algorithm is developed to calculate the initial transformation parameters for registration of these obtained images. Finally, the final refined parameters determined by using gradual optimization are adopted to achieve the registration result. The effectiveness of the proposed method is demonstrated by the experimental results.