Adaptive Neighborhood Size Method for Normal Vector Estimation Based on Multi-scale Point Cloud
To mitigate the impact of neighborhood size on the accuracy of normal vector estimation derived from principal component anal-ysis(PCA)in 3D point clouds,we propose an adaptive neighborhood size normal vector estimation method based on multi-scale point cloud data.This method determines the optimal neighborhood size by analyzing the local covariance matrix of 3D point clouds,construct-ing a feature entropy function,and minimizing this entropy function according to the principle of minimum entropy while incorporating collinearity assessments among neighboring points within the point cloud.Experimental results obtained from both simulated and real-world measured point clouds demonstrate that our proposed approach effectively addresses issues related to unreasonable neighborhood size selection in PCA methods and significantly enhances normal vector estimation accuracy across varying scales of point clouds.
Laser point cloudPCA normal vector estimationPoint cloud normal vectorPoint cloud collinearity
万方数据
维普
