基于曲率的点云数据配准算法

为了实现不同视角下测得的数据的多视定位，提出一种点云数据配准算法。该算法针对近邻内的点，采用二次曲面逼近的方法来求得每个点的曲率，并根据曲率的Hausdorff距离来寻找有效点集，建立名义上的对应关系，最后用四元组法来求得坐标变换，把数据统一到一个坐标系下。该算法利用曲率的性质准确判断对应点集，解决了任意多视点云的拼合问题，试验结果验证了其有效性和精度。

Curvature-based registration algorithm of point clouds data

A 3-D measuring data registration algorithm was proposed in order to locate the different-view-measured cloud data without clear corresponding relation. In the algorithm, a local parabolic surface was fitted to each neighbor point. The curvature of each point was estimated according to the point and its neighbor points, using Hausdorff distance as measurement criterion. Thus a nominal correspondence was created between the control point set selected from the different positions in the overlapped view area. Then the rotation matrix and translation vector were deduced by quaternion method. The corresponding point sets were judged according to the characteristics of curvature, and the problem of combining arbitrary multi view point clouds was solved. Experimental results show the accurate and robust performance of the proposed algorithm.