2D patterns for 3D surface matching

Surface matching is the process that compares the shape of two surfaces. To facilitate surface matching, we have developed a 2-D representation for comparing 3-D surfaces of arbitrary shape. Using a cylindrical mapping, a descriptive pose-invariant 2-D pattern is created for every point on the surface of an object. By comparing patterns, point correspondences between different surfaces are established. When a sufficient number of point correspondences exist between two surfaces, the surfaces match. We show that the simplicity and generality of our representation makes it widely applicable to problems in 3D computer vision including object modeling, interior modeling and object recognition