Alignment using distributions of local geometric properties

We describe a framework for aligning images without needing to establish ex plicit feature correspondences. We assume that the geometry between the two images can be adequately described by an affine transformation and develop a framework that uses the statistical distribution of geometric properties of image contours to estimate the relevant transformation parameters. The estimates obtained using the proposed method are robust to illumination con ditions, sensor characteristics, etc., since image contours are relatively invariant to these changes. Moreover, the distributional nature of our meth od alleviates some of the common problems due to contour fragmentation, occ lusion, clutter, etc. We provide empirical evidence of the accuracy and rob ustness of our algorithm. Finally, we demonstrate our method on both real a nd synthetic images, including multisensor image pairs.