Discrete Point Cloud Filtering And Searching Based On VGSO Algorithm
Fengjun Hu, Yanwei Zhao, Wanliang Wang, Xianping Huang · 2013
The massive point cloud data obtained through the computer vision is uneven in density together with a lot of noise and outliers, which will greatly reduce the point cloud search efficiency and affect the surface reconstruction. Based on that, this paper presents a filtering algorithm based on Voxel Grid Statistical Outlier (VGSO): Firstly, 3D voxel grid is created for the massive point cloud data approximating other points inside the voxel with the centroid of all points; then, the neighborhood of discrete points is analyzed statistically, calculating average distance of every point to its neighboring points and filtering the outliers outside the reference ranges of average distance from the data set; finally, the segmentation rules are improved according to the characteristics of KD-Tree. A large amount of experimental results show that this stable and reliable algorithm can compress and filter the point cloud data quickly and effectively. At the same time, it greatly accelerates the search speed.