3-D information derivation from a pair of binocular images

摘要

A new scheme using a pair of stereo images for 3-D information derivation about a scene is presented. First, we establish a triangulation geometry between two camera models and a point in 3-D scene. In this scheme the successful implementation of a matching algorithm of features in images is considered as the key step of stereovision analysis. After a brief discussion of feature extraction, representation and internal implementation, an efficient stereo feature matching algorithm is proposed. Finally, some results obtained by applying the matching algorithm to several pairs of stereo images are shown in the paper.