Application of Rapid Ambiguity-avoiding Feature Matching in Multi-view Stereo Vision Inspection System

摘要

One of the important tasks of stereo vision measurement is to search the matching same points between pictures. The matching method applying epipolar slope is one of the significant means not difficult to find the corresponding points. But a new problem is brought by this method likely and there are several feature points in one image, which match with one point in another image. That is the ambiguity problem of macthing. Based on the theory of epipolar slope matching, this paper proposes a new method of rapid ambiguity-avoiding feature matching which can be used to solve the problem of ambiguity in matching effectively aiming at the characteristic of the inspection system of multi-view stereo vision. The mathematical model to compute the epipolar slope is also put forward in this paper. The rapid ambiguity-avoiding feature matching method can resolve the ambiguity problem in matching very quickly and efficiently, that has been proved by experiments.