Improvement on SAR Image Matching Algorithm Based on SURF

Abstract

The matching between SAR images is characterized by the rich texture of images and massive amount of computation,as well as the requirement of uniformly distribution for the matched points.After analysis the deficiencies of several common matching approaches,we propose a new algorithm for SAR image matching based on SURF,which can achieve sub-pixel accuracy.We then employ bidirectional search strategy and RANSAC algorithm to improve the accuracy and speed of searching correct matching points.Finally,we testify the validity of the method through affine model and quadratic polynomial model.