Abstract: This paper proposes a 3D measured data registration algorithm.In the algorithm firstly，a triangular pyramid is constructed to each neighborhood.And then it computes the angle of the side edges and the middle axis of the triangular pyramid and constructs the geometric features based on those angles of the triangular pyramid.Thirdly the effective correspondent point pairs are extracted from the geometric features.Lastly the rotation and translation matrix are deduced by quaternion method.Experimental results show the accurate and robust performance of the proposed algorithm.

Key words: 3D measured data, registration, geometric features

摘要： 提出一种基于几何特征的三维数据配准算法。该算法针对点云中各点p_i的k邻近点Nbhd(p_i)构造三棱锥体，将三棱锥体各侧棱p_iv_j（j=1,2,...,）和其中轴线p_io的夹角记作θⁱ_j，所有夹角按照右手系来依次形成夹角序列（θⁱ₁θⁱ_2...）作为三棱锥的几何特征。通过比较三棱锥体的几何特征来确定有效点对。算法实现时，首先对初始数据通过抽取有效点对，建立名义上的对应关系，然后采用四元组法求得坐标变换的旋转和平移矩阵，实现数据配准。

关键词: 三维测量数据, 配准, 几何特征