A SAR Image Registration Method Based on Pixel Migration of Edge-Point Feature

In this letter, a synthetic aperture radar (SAR) image registration method based on edge-point features is presented to improve precision, robustness, and applicability of SAR image registration. First, an affine transformation model is decomposed into six parameters with explicit geometric meanings. Next, based on the strength and direction features of the edge point, a matching similarity criterion and a joint similarity metric, i.e., the square summation joint feature, are constructed. Then, the parameters of the transformation model between SAR images are solved with a modified genetic algorithm that is able to get a global optimal solution of the metric. Finally, the performance of the proposed method is validated with two SAR image registration experiments.