RELAXATION BY HOPFIELD NETWORK IN STEREO IMAGE MATCHING

This paper outlines a relaxation approach using the Hopfield neural ne twork for solving the global stereovision matching problem. The primit ives used are edge segments. The similarity, smoothness and uniqueness constraints are transformed into the form of an energy function whose minimum value corresponds to the best solution of the problem. We com bine two methods: (a) optimization/relaxation((1)) and (b) relaxation merit((2)) with the above three constraints mapped in an energy functi on. The main contribution is made (1) by applying a learning strategy in the similarity constraint and (2) by introducing specific condition s to overcome the violation of the smoothness constraint and to avoid the serious problem arising from the required fixation of a disparity limit. So, we improve the stereovision matching process. A better perf ormance of the proposed method is illustrated with a comparative analy sis against a classical relaxation method. (C) 1998 Pattern Recognitio n Society. Published by Elsevier Science Ltd. All rights reserved.