A new two-view stereo matching method and its application

The authors describe a new two-view stereo image matching method which is quite effective for a kind of object called block objects. The matching technique uses local geometric constraints among edges, corners and the two-view stereo images to resolve matching ambiguity. The authors describe its application in object recognition and localization. Some experimental results are given to illustrate the effect of applying it to the two-view stereo vision system. Compared with other matching methods, the proposed matching technique is simple, effective, rapid and has a high corrective match rate