Fast Method for Locating Spheres in Large 3D Point Clouds in PCL

Abstract

Scanning spheres are used for a variety of reasons in large point clouds captured by large volume 3D scanners. A couple these reasons are: point cloud registration, automatic alignment, and creation of geometry and center points for further analysis. Locating the scanning spheres automatically in large point clouds can take quite a lot of computing time, and produce false positives or negatives. To reduce computation time and reduce false positives or negatives a novel approach of utilizing the Intensity values or RGB values was explored. This presentation will discuss the developed approach and quantify the improvements seen on various large clouds.