We address the problem of matching two pictures of a scene taken from two separate viewpoints and potentially sharing only a small overlapping part. While the image deformation is non-rigid, it underlies the geometric constraint that at least one part of the scene is rigid. We formulate the problem as a search problem in the Cartesian product of all possible correspondences. In this space, candidate matches vote for geometry hypotheses with a vote weight depending on local image similarity. The voting process can be written as a Radon transform and the question is how it can be computed efficiently. We prove that for the geometry of two views, the Radon transform can be written as a correlation in the product of two spheres with the product of two rotations as an acting group. The talk will be concluded by applications both in localization and 3d retrieval. Biography: Kostas Daniilidis is Associate Professor of Computer and Information Science at the University of Pennsylvania. His research interests include visual space and motion applications in immersive environment.