STEREO CORRESPONDENCE USING THE HOPFIELD NEURAL-NETWORK OF A NEW ENERGY FUNCTION

This paper presents an approach using a Hopfield neural network to the stereo correspondence problem for extracting the 3D structure of a sc ene. The stereo correspondence problem can be defined in terms of find ing a disparity map that satisfies three competing constraints: simila rity, smoothness and uniqueness. In order to solve the stereo correspo ndence problem using a Hopfield neural network, these constraints are transformed into the form of an energy function, whose minimum value c orresponds to the best solution of the problem, on the Hopfield networ k. In the process of mapping the constraints into energy function, the energy functions are derived so that the network ensures Hopfield's c onvergence rule. Stereo correspondence then is carried out through the network evolving energy surface to find the minimum energy correspond ing to the solution of the problem. The examples for random-dot stereo grams and real images are shown in the experiment, illustrating how th e proposed network works.