Segment matching using a neural network approach

We propose a new approach to solve the correspondence problem for a set of segments extracted from a pair of stereo images. The problem is first formulated as an optimization task where a cost function, which represents the constraints on the solution, is to be minimized. The optimization problem is then performed by a two-dimensional Hopfield neural network. The network uses several local constraints such as correlation and compatibility measures between segments of a pair of stereo images. Finally we show numerous results obtained with this approach.