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

2017 International Symposium on Application of Materials Science and Energy Materials

材料科学;能源学

An, Lu^1 ; Guo, Baolong^1

Institute of Intelligent Control and Image Engineering, Xidian University, Xi'an, China^1

关键词: 3D point cloud;    Compression methods;    Identification algorithms;    International society;    Minimum spanning trees;    Multi-scale filtering;    Processing time;    Region growing methods;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/322/5/052013/pdf
DOI  :  10.1088/1757-899X/322/5/052013

学科分类：材料科学（综合）

来源: IOP

【 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).