Correspondence-less Image Alignment using a Geometric Framework

In this paper we describe a framework for aligning images without using explicit feature correspondences. We show how certain geometric properties of image contours are related to the parameters of the geometric transformation between the images. For a given transformation model, we show how the transformation parameters can be recovered using simple statistical distributions of these geometric properties. The use of these statistical descriptions eliminates the need for establishing explicit feature correspondence. Further, the proposed method is robust to problems of occlusion, clutter and errors in low-level processing. We demonstrate the eeectiveness of our method on various real images.