三维扫描系统中的数据配准技术

Data registration in 3-D scanning systems

通过引入特征点和改进最近点迭代法,提出了一种在三维扫描系统中对三维点云数据进行配准的方法。该方法通过对特征点的提取,首先得到一组匹配点对,然后运用SVD矩阵分解算法求出转换参数R和T,进而以此作为最近点迭代法的初始值,并对最近点的求法和迭代截止条件作了改进,得到了很好的配准效果。该文论述了该方法的基本原理,并通过不同视觉下物体三维测量点云数据配准的应用实例证明了该方法的有效性。

A 3-D measuring data registration method for 3-D scanning system was developed based on feature points and iterative closet point (ICP) algorithm. The feature points are first identified and matched. Then, the rotation matrix R and the translation vector T are found using the singular value decomposition algorithm. Thus, initialized by the former result, the ICP algorithm then leads to a perfect registration after improving on finding closet distance and deciding when to terminate the iterations. The basic principle of this method is discussed, and a registration application example of 3-D measuring data taken from different viewpoints is carried out to show the perfect feasibility of this approach.