Efficient Feature Correspondence for Image Registration

Feature correspondence is an important step in image registration. Chamfer matching is a process of establishing the feature correspondence of an object (subimage) in an image where both the subimage and the image are binary. Chamfer matching establishes correspondence based on low level features. It is the process of locating the template within the image by shifting the template within the image and at each shift position determining the sum of distances of closest object points in the template and the image. The smaller the sum, the closer object points in the template and the image. Correspondence is achieved by calculating the minimum distance out of all the translations which is computationally expensive. The problem is further aggravated if reference image is of much greater dimensions as compared to template image. Since the reference image consists of objects and background, calculating the minimum distance for all the pixel locations becomes time consuming. In this paper a method to decrease the computation time of existing technique is presented, to make existing schemes suitable for real time registration. Key-Words: Image registration, Feature correspondence, Distance transformations, Chamfer matching, Edge matching, Reference image, Template image.