A Split and Merge Based Plane Segmentation Method with Octree Data Structure for Unorganized Point Cloud Data

In recent years, point cloud data have been used frequently in robotic applications. It is considered that the segmentation of that data is the first step in solving several problems such as scene and object classification. In this study, it is aimed to the segmentation of planar surfaces such as ceiling, floor, and walls via unorganized point cloud data captured with 3D laser scanner in indoor environments. To achieve that, firstly, the unorganized point cloud data is converted to organized point cloud data with octree data structure and the relation between neighbor points is defined. Then, a split and merge based segmentation method, which uses that relation, is proposed. The Fukuoka indoor laser dataset with different noise levels obtained with a 3D laser scanner was used to test the efficiency of the proposed methods. The proposed method was compared with the RANSAC and region growing methods in terms of segmentation accuracy and robustness against the noise.