Remote Sensing Image Registration Based on Dynamic Threshold Calculation Strategy and Multiple-Feature Distance Fusion

Abstract

Remote sensing image registration is widely used in civilian and military applications such as target recognition, environmental transformation monitoring, and military damage assessment. The severe outliers in the extraction of feature points caused by nonrigid transformation and viewpoint changes in the process of capturing remote sensing images increase the difficulty of registration. Therefore, we present a remote sensing image registration method based on dynamic threshold calculation strategy (DTCS) and multiple-feature distance fusion. The main idea of our approach is to maximize inliers while ensuring the optimal correspondence. First, DTCS gradually screens reliable inliers to reduce the negative effect of outliers over the iterations. The multiple-feature distance fusion Gaussian mixture model is then introduced to compensate for the defect of a single feature, and DTCS acting as the prior probability combines with the deterministic annealing to achieve the optimal mapping from local to global scale. Moreover, structure constraint based on local applying force is added into the global constraint to control the alignment of feature points more accurately in the overlapping area, so as to guide the subsequent image transformation. Extensive experiments show that our method performs better in most cases compared with nine state-of-the-art methods.