A ROI Features Registration Method for SAR Images

The method of SAR image registration is a key point of SAR image processing. It has a great influence on the fusion and mosaicking of SAR image. The SAR image registration method based on region often has such defects as area extraction difficulty, great influence by speckle, etc. As a result an automated registration method for SAR images based on ROI features is presented in this paper. In this method, the ROI features in SAR images were extracted as matching units, feature matching was carried out by combining the row\column matching likelihood coefficients and the minimum distance measure of invariant moment together. As all the correct matching features of ROI were gained, the least squares method was employed to estimate parameter of affine transformation with the centroid of ROI. The real SAR images are used to demonstrate its performance. Compared with some existing methods，the proposed method has very good adaptability and accuracy of registration.