基于立体成像几何特性的动态有限搜索匹配法

提出了一种基于立体成像几何特性的动态有限搜索匹配法。该法根据被测物体与摄像机标定参照物在空间上的相互关系及被测物体形态，确定被测物体特征点视差匹配大致范围。利用立体成像几何特性，采用动态有限搜索法，以求出的视差为参考，根据左(右)图像中特征点与所得匹配对应点的水平视差值，确定在右(左)图像中的搜索方向和搜索范围，采用灰度区域相关计算，确定下一个匹配特征点，并依此类推。由于特征点灰度不稳定，且随视点的不同而有很大差异，为能得到正确匹配，需结合几何相似性约束。通过对一已知三维坐标标准件的计算机仿真实验，证实了该方法的有效性。

Dynamic Limited-Search Matching Algorithm Based on Geometric Properties of Stereo Imaging

A dynamic limited search matching algorithm based on geometric properties of stereo imaging is developed in this paper. The approximate scope of disparity for matching feature points on the measurement object is ascertained according to the shape of measured object and correlated location spatially between the measured object and target of camera calibration. Based on the geometric properties of stereo imaging, the disparity is obtained by dynamic limited search method. After the disparity computed, the searching orientation and scope of disparity in right (left) image can be determined by way of the disparity difference horizontally between the feature point in left (or right) image and the computed matching point, so the next disparity can be confirmed by calculating correlation value of gray in some region. The rest may be deduced by analogy. In order to gained the correct correspondence, the restriction of geometric comparability should be taken into account due to the instability of gray and much difference from viewpoints also. It is proved that the algorithm proposed in the paper is feasible by computer imitated experiment on standard piece with given 3D coordinates.