An Optimized Method for Automatic Registration of Remotely Sensed Images Based on Structural Image Matching

This paper introduces an optimized method for automatic tie points collection and spatial optimization, thereby realizing automatic registration of images. The method contains two crucial technologies, one is the image matching utilizing structural image matching, which can obtain a number of tie points, the other is spatial distribution optimization of tie points using cluster analysis, which can make the points distributed evenly and appropriate in number and accuracy. In the process of matching, an evaluation function based on the joint probability of a possible match and an informed tree search method are applied to find the best match. In order to shorten the search time, several strategies have been developed and integrated into the search method. When optimizing the distribution of tie points, the uniform index has been proposed by the introduction of the concept of monopolized circle. Experimental results indicate that this method optimized in time and space can achieve a high level of automation in image registration.