点云数据分块是模型反求过程中的重要环节,分割优劣影响模型重建的效率和精度。微型复杂曲面零件由多个微小图形并列、交叉组合而成,特征点精简、图元识别难度大,是数据分割中的难点。根据模型造型特点,分离带状特征点的下边界点作为拟合特征线的真实特征点;由每个图元端点的邻近关系和端点附近特征点的排列趋势识别属于同一图形的图元;利用以边界为约束的区域生长算法和三角形叉积的算法分割同一曲面的点云。实验结果表明:该方法克服了现有方法处理微型复杂曲面点云时出现的过分分割和分割不足的问题,为高质量的模型重建提供了基础。

The segmentation of point cloud data is an important link in the process of model reversal. The quality of segmentation affects the efficiency and accuracy of model reconstruction. The parts with complex micro-surface are composed of several small graphics side by side and cross-combined. It is difficult to simplify feature points and identify elements, which is a difficulty in point cloud data segmentation. According to the modeling characteristics of the model, the lower boundary points of the banded feature points are separated as the real feature points of the fitting feature line, and the elements belonging to the same graph are identified by the proximity of the end points of each element and the arrangement trend of the feature points near the end points. The regional growth algorithm with boundary constraints and the triangle cross product algorithm are used to segment the point clouds on the same surface. The experimental results show that this method can overcome the problems of excessive segmentation and insufficient segmentation when dealing with complex micro-surface point clouds, which lays a foundation for high-quality model reconstruction.