Deep learning architecture for building extraction using LiDAR point clouds

doi:10.13474/j.cnki.11-2246.2021.379

Abstract

Abstract: Aiming at the problem of applying deep neural network algorithm to LiDAR point cloud for large-scale building extraction, PointCNN and PointNet++ models are selected for modification and comparison in this paper. For PointCNN, the parameters are adjusted to make it more suitable for large scenes. For PointNet++, in order to add more features and speed up the training efficiency of network model in large scenes, a K-means layer is added after the sampling layer. Finally, through training and verification on the test data set, it is found that the deep learning methods can well solve the disordered characteristics of point cloud and make better use of the spatial correlation between points. The accuracy of the improved models is more than 96% and they are also better than the original models in time consumption and extraction effect.

Key words: PointNet++, PointCNN, LiDAR, point clouds, building, K-means

CLC Number:

P237

HU Chuanwen, LU Shijie, YANG Wenjing, ZHU Xiaoyong. Deep learning architecture for building extraction using LiDAR point clouds[J]. Bulletin of Surveying and Mapping, 2021, 0(12): 88-93.

References

[1] LAI Xudong, YANG Jingru, LI Yongxu, et al. A building extraction approach based on the fusion of LiDAR point cloud and elevation map texture features[J]. Remote Sensing, 2019, 11(14):1636.
[2] NÆSSET E. Practical large-scale forest stand inventory using a small-footprint airborne scanning laser[J]. Scandinavian Journal of Forest Research, 2004, 19(2):164-179.
[3] REUTEBUCH S E, ANDERSEN H E, MCGAUGHEY R J. Light detection and ranging (LIDAR):an emerging tool for multiple resource inventory[J]. Journal of Forestry, 2005, 103(6):286-292.
[4] KULAWARDHANA R W, POPESCU S C, FEAGIN R A. Fusion of lidar and multispectral data to quantifysalt marsh carbon stocks[J]. Remote Sensing of Environment, 2014, 154:345-357.
[5] VO A V, TRUONG-HONG L, LAEFER D F, et al. Octree-based region growing for point cloud segmentation[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2015, 104:88-100.
[6] LACH S R, KEREKES J P. Robust extraction of exterior building boundaries from topographic lidar data[C]//IGARSS 2008-2008 IEEE International Geoscience and Remote Sensing Symposium. Boston, MA:IEEE, 2008.
[7] RONNEBERGER O, FISCHER P, BROX T. U-net:convolu-tional networks for biomedical image segmentation[C]//Medical Image Computing and Computer-Assisted Intervention-MICCAI 2015.[S.l.]:Springer, Cham,2015.
[8] HE Kaiming, ZHANG Xiangyu, REN Shaoqing, et al. Deep residual learning for image recognition[J]. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016:770-778.
[9] CHARLES R Q, HAO Su, MO Kaichun, et al. PointNet:deep learning on point sets for 3D classification and segmentation[C]//IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Honolulu, HI:IEEE, 2017:77-85.
[10] QI C R, YI L, SU H, et al. PointNet++:deep hierarchical feature learning on point sets in a metric space[EB/OL].(2017-06-07)[2021-08-09]. https://arxiv.org/abs/1706.02413.
[11] BOULCH A, GUERRY J, LE SAUX B, et al. SnapNet:3D point cloud semantic labeling with 2D deep segmentation networks[J]. Computers & Graphics, 2018, 71:189-198.
[12] LI YANGYAN, BU RUI, SUN MINGCHAO, et al. Point CNN[EB/OL]. (2018-11-01)[2020-08-09].https://arxiv.org/abs/1801.07791.
[13] AXELSSON P. DEM generation from laser scanner data using adaptive TIN models[J]. International Archives of Photogrammetry and Remote Sensing, 2000, 23(B4):110-117.
[14] LONG J, SHELHAMER E, DARRELL T. Fully convolu-tional networks for semantic segmentation[J]. 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015:3431-3440.
[15] ROTTENSTEINER F, SOHN G, GERKE M, et al. Results of the ISPRS benchmark on urban object detection and 3D building reconstruction[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2014, 93:256-271.
[16] HOFFMAN R, JAIN A K. Segmentation and classifica-tion of range images[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1987, PAMI-9(5):608-620.
[17] SHAN Jie, SAMPATH A. Building extraction from LiDAR point clouds based on clustering techniques[M]//Topographic Laser Ranging and Scanning.[S.l.]:CRC Press, 2008:421-444.
[18] FILIN S, PFEIFER N. Segmentation of airborne laser scanning data using a slope adaptive neighborhood[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2006, 60(2):71-80.
[19] XU Y S, TUTTAS S, HOEGNER L, et al. Geometric primitive extraction from point clouds of construction sites using VGS[J]. IEEE Geoscience and Remote Sensing Letters, 2017, 14(3):424-428.
[20] DONG Z, YANG B S, HU P B, et al. An efficient global energy optimization approach for robust 3D plane segmentation of point clouds[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2018, 137:112-133.