基于多视点云拼接算法研究

针对目前点云自动匹配效率低，拼接精度还有待提高，提出了一种自动匹配标志点的拼接算法。首先根据标志点之间的空间特征不变性，引入一种动态距离矩阵来记录搜索标志点间的距离，通过循环迭代比对动态距离矩阵来匹配标志点；然后采用最小二乘法求解坐标变换矩阵进行多视点云拼接。通过实验对比，可以发现本文的点云拼接算法明显优越于传统的方法，最后引入拼接精度进行验证。多视点云拼接实验表明，该方法方便、快捷、实用，拼接精度达0.0179mm，适合在工业生产中使用。

Research on Clouds Registration Algorithm Based on Multi-view Point

Aiming at the low automatic matching efficiency of point clouds, the stitching precision remains to be im-proved, in this paper, a stitching algorithm of automatic matching marked point is put forward. Firstly, according to the space invariant feature among marked point, a dynamic distance matrix is introduced to record the distance of marked points that are searched, marked points are matched by using cyclic iterative method to compare dynamic dis-tance matrix. Finally,the multi-view point clouds are stitched by using the method of least squares to solve the coordi-nate transformation matrix.Through the experimental comparison,the point cloud registration algorithm is obviously supe-rior to the traditional method, finally stitching precision is verified. Multi-view point clouds stitching experiments show that the method is convenient, fast and practical and suits for industrial production, and the stitching precision of 0.0179mm can be achieved.