损伤零件点云模型配准的ICP算法

针对损伤零件的传统点云模型配准过程中存在着运算效率低与损伤部位损伤量确定误差较大等问题,提出一种基于原始ICP算法的改进算法。考虑模型因为损伤而引起的特征与表面形貌的改变,利用法矢夹角进行点云数据的精简,保留模型主要特征,再利用对应点的曲率约束与距离约束设定阈值剔除损伤区域点云,保证对应点之间快速准确地配准。最后,运用Matlab实现改进算法,并利用损伤的轮机叶片点云数据的配准验证该算法的有效性。

ICP algorithm for registration of point cloud model of damaged parts

In the traditional point cloud model registration process for damaged parts,there are some problems such as low computational efficiency and large error in the damage of damage parts. This paper proposes an improved algorithm based on the original ICP algorithm.Considering the changes of the features and surface topography caused by the damage of the model,the point cloud data is reduced by the normal vector angle,and the main features of the model are retained. Then the curvature constraint and the distance constraint of the corresponding points are used to set the threshold to eliminate the damage point cloud. To ensure fast and accurate registration between the corresponding points.Finally,the improved algorithm is implemented by using Matlab,and the effectiveness of the algorithm is verified by the registration of the damaged blade point cloud data.