Robust Optical and SAR Image Registration Based on Phase Congruency Scale Space

Abstract

Automatic optical and synthetic aperture radar (SAR) image registration is still a challenging task due to significant non-linear radiation distortions (NRD) between such images and the speckle noise in the SAR images. This letter addresses this problem by proposing a novel descriptor named the Phase Space Feature Descriptor (PSFD), which is based on the Phase Congruency Scale Space. The proposed PSFD aims to capture the structural similarity between optical and SAR images. In the definition of the proposed similarity metric, we first extend the phase congruency model in different frequencies to generate the Phase Congruency Scale Space. The PSFD is robust against nonlinear radiation distortions because Phase Congruency Scale Space uses different frequencies of log Gabor wavelet information to filter out noise while extracting information. In the first experiment, compared to other methods, PSFD can locate the correct match point precisely in different scenes.In the second matching, the results show that PSFD is robust against NRD and the speckle noise of the SAR images, outperforming the other proposed algorithm.