Point cloud segmentation of 3D scattered parts sampled by RealSense

Abstract

Segmenting 3D objects from a set of disordered point cloud is always one challenge work in Bin-pi cking systems. In this paper, an easy and valid algorithm is proposed to solve this problem. The point data set obtained from a low-cost depth camera, RealSense, can be quickly filtered to be a much clean compact one, significantly saving the computing resources. The filtered points set is next divided into a number of sub-patch fragments, and all sub-patches will be finally joined independently to be a set of larger ones by two criteria, i.e., extended convexity criterion and angle criterion, each of which corresponds to one working piece. The validity of the proposed algorithm has been verified by testing cases, which have shown its potential applications in robot grabbing systems in practice.