Point-Pattern Matching Under Perspective Transformation
基于p2-不变量的透视变换下的点模式匹配方法

Point-pattern matching is an important problem in the fields of computer vision and pattern recognition. Its main task is to pair up the points in two images of a same scene when there is a geometric transformation relating the two images. Once the correspondences between the two sets of points are set up, we can recognize objects and locate their poses in many optical sensor applications. However it is well known that the imaging geometry of a TV camera is a nonlinear perspective transformation. To simplify the difficult solving process, many researchers approximate the perspective-transformation relation with an affine transformation. Obviously it will increase the possibility of fault matchings. In this paper, a new algorithm is proposed to solve the problem of matching two point sets with the different cardinality under a perspective transformation without simplifying the original perspective-transformation relation. Supposing two sets of three points have been matched as a whole beforehand, based on projective coordinates andp2-invariant theory, a new concept named generalized distance is introduced. Theoretical analysis and simulation results show that the new algorithm is fast and effective.