|
Abstract : |
This publication can be retrieved by anonymous ftp to publications.ai.mit.edu. The pathname for this publication is: ai-publications/1994/AIM-1463.ps.Z In order to recognize an object in an image, we must determine the best-fit transformation which maps an object model into the image. In this paper, we first show that for features from coplanar surfaces which undergo linear transformations in space, there exists a class of transformations that yield projections invariant to the surface motions up to rotations in the image field. To use this property, we propose a new alignment approach to object recognition based on centroid alignment of corresponding feature groups built on these invariant projections of planar surfaces. This method uses only a single pair of 2D model and data pictures. Experimental results show that the proposed method can tolerate considerable errors in extracting features from images and can tolerate perturbations from coplanarity, as well as cases involving occlusions. As part of the method, we also present an operator for finding planar surfaces of an object using two model views and show its effectiveness by empirical results., |
