Research on Improvement of Point Cloud Segmentation Method Used for Flexible Packaging Bag Recognition

Abstract

Industrial robots, in the flexible plant bagged material identification and grasping work occupies an important position, for the traditional region growth algorithm is prone to under-segmentation problems when processing 3D images, put forward a K-mean clustering fusion region growth point cloud segmentation algorithm, in this study, the region growth algorithm is applied for the initial segmentation and clustering of point clouds, by carrying out the segmentation of the target clustering set of the situation of the determination, automatic selection of algorithms. After multiple iterations of the region growing and clustering iterations, the target point cloud data can be successfully segmented. The segmentation and identification of single-layer bagged materials in the stacking position is realized, and the feasibility of the algorithm is verified through point cloud segmentation experiments under different experimental conditions.