针对多视点数据配准问题提出了一种基于几何匹配的定位方法.该方法是在最近点迭代匹配算法(Iterative Closest Point Algorithm)的基础上,对参与最佳旋转矩阵计算的对应点对进行了限制,因而扩大了算法的适用范围,同时又对用结构光方法重建获得的三维物体表面数据进行了加权处理,从而减少了曲面重建所产生的误差对配准精度的影响,并通过实验证明了该方法的有效性.

In this paper, we propose a new geometric registration algorithm which can register multi-view range data of 3D free-form object efficiently. This method is based on the Iterative Closest Point(ICP) algorithm, but adding a restriction to the corresponding points in estimation of optimal rotation matrix. Further more, we takes a membership of weighting value to points of 3D surface reconstructed using structure lighting to reduce the reconstruction error influence to the registration. The experiments demonstrate the feasibility of our approach.