ZHENG De jiong, LU Keqing

(School of Mechanical Engineering, Hangzhou Dianzi University, Hangzhou 310018, China)

 

Abstract: Aiming at the quick picking problem of massive point cloud data in reverse engineering, the basic picking method of 3D graphics currently was researched, the basic process of point cloud picking and the key problem of quick picking was analyzed, a quick picking algorithm for 3D point cloud based on adaptive octree was proposed. After the picking polygon was given by the user on screen, firstly, the adaptive octree division based on the distribution density of point cloud data was made, then the projection of the octree node was made and the octree projected polygon on the screen was formed, the rectangular bounding box of the picking polygon was established, then the intersection detection between octree projected polygon and rectangular bounding box of the picking polygon was executed, the point cloud of the octree node not intersected with the rectangular bounding box was removed， thereby the point cloud picking judgment was reduced and the picking efficiency was improved. Finally, a picking test under different point cloud distribution density was conducted. The results indicate that the greater the density of point cloud distribution, the picking time is relatively shorter, the algorithm has a high picking speed and accuracy.

Key words: picking; 3D point cloud;octree; adaptive