A pyramidal stereovision algorithm based on contour chain points

Abstract

We are interested in matching stereoscopic images involving both natural objects (vegetation, sky, reliefs,…) and man made objects (buildings, roads, vehicles,…). In this context we have developed a pyramidal stereovision algorithm based on” contour chain points.” The matching process is performed at different steps corresponding to the different resolutions. The nature of the primitives allows the algorithm to deal with rich and complex scenes. Goods results are obtained for extremely fast computing time.