Outdoor Illegal Construction Identification Algorithm Based on 3D Point Cloud Segmentation

作     者：Lu An Baolong Guo 

作者机构：Institute of Intelligent Control and Image Engineering Xidian University Xi'an China 

出 版 物：《IOP Conference Series: Materials Science and Engineering》 

年 卷 期：2018年第322卷第5期

摘      要：Recently, various illegal constructions occur significantly in our surroundings, which seriously restrict the orderly development of urban modernization. The 3D point cloud data technology is used to identify the illegal buildings, which could address the problem above effectively. This paper proposes an outdoor illegal construction identification algorithm based on 3D point cloud segmentation. Initially, in order to save memory space and reduce processing time, a lossless point cloud compression method based on minimum spanning tree is proposed. Then, a ground point removing method based on the multi-scale filtering is introduced to increase accuracy. Finally, building clusters on the ground can be obtained using a region growing method, as a result, the illegal construction can be marked. The effectiveness of the proposed algorithm is verified using a publicly data set collected from the International Society for Photogrammetry and Remote Sensing (ISPRS).