改进蚁群算法的三维激光点云聚类方法

为了提高点云聚类方法的效率和精度，本文提出了一种蚁群优化投影寻踪算法。试验采用机载LiDAR点云数据，通过构建蚁群算法中信息素系数更新的对数反正切函数模型来减少所需的信息素更新系数的迭代次数，不断优化的投影寻踪方向投影，提高寻找最佳投影方向的连续空间的效率，实现了树木和建筑的分割。试验使用人工方法对比评估树木和建筑物的位置和数量的准确性。

3D Point Cloud Clustering Based on Improved Ant Colony Algorithm

In order to improve the efficiency and accuracy of the existing point cloud clustering method, an ant colony optimization projection pursuit is proposed. This method provides better clustering results with respect to airborne LiDAR point clouds. The paper constructs the log and arctan function model to reduce the number of iterations required to update pheromone coefficients in ant colony algorithms,continuously optimize projection on the projection pursuit direction, and improve the efficiency of searching for the best projection direction in continuous spaces, which realizes point cloud clustering for trees and buildings through the ant colony optimization projection.The accuracy of the location and the number of the trees and the building are evaluated using manual methods.