As one of the main components of the urban interior, buildings play a pivotal role in the construction of smart cities. In order to segment the structural characteristics contained in building facades more accurately and efficiently and intelligently, this paper constructs a small building facade dataset based on on-board LiDAR point clouds, and improves it based on the original PointNet++ network by adding a multi-headed attention mechanism module in model feature extraction; in model training, an extraction function is introduced after the training function for unlabelled In terms of model training, an extraction function is introduced after the training function to extract features from the unlabelled building façade data, and the extraction results are finally given labels and saved to the training data set, so as to achieve the effect of enhancing the training sample data, and by improving the model structure and model training method, the effective extraction of building façade features is achieved. The final accuracy of the proposed method is 88.2% and the average intersection-to-merge ratio (mIoU) is 77.1%, which is 1% and 3.3% higher than the original PointNet++ network respectively.

Junli Yang, Yongqiang Li, Xiang Pan, Jing Zang. Research on Building Facade Feature Information Extraction. Academic Journal of Computing & Information Science (2023), Vol. 6, Issue 7: 72-83. https://doi.org/10.25236/AJCIS.2023.060711.

[1] Yang Qingke, Li Yongqiang, Liu Cong, et al. Study on Window Boundary Extraction Method of Introducing Dynamic Ellipse Based on Vehicle LiDAR[J]. Geography and Geo-Information Science, 2019, 35(02):61-67.  

[2] Wang Y, Ma Y, Zhu A, et al. Accurate facade feature extraction method for buildings from three-dimensional point cloud data considering structural information [J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2018, 139: 146-153. 

[3] Arachchige N H, Perera S N, Maas H G. Automatic processing of mobile laser scanner point clouds for building facade detection [J]. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci, 2012, 39: 187-192. 

[4] Li Hongmei, Zhang Chunchao, Zhang Xia, Luo Zhu. Feature extraction of building point clouds based on Morse theory [J]. Bulletin of Surveying and Mapping, 2020(05):31-35+42. 

[5] Fan Huilong, Li Yongqiang, Yang Qingke, et al. Building façade window model construction of vehicle point cloud data [J]. Bulletin of Surveying and Mapping, 2020, No. 518(05):95-100. DOI: 10. 13474/ j. cnki. 11-2246. 2020. 0153. 

[6] Wu B, Wan A, Yue X, et al. Squeezeseg: Convolutional neural nets with recurrent crf for real-time road-object segmentation from 3d lidar point cloud[C]//2018 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2018: 1887-1893. 

[7] Iandola F N, Han S, Moskewicz M W, et al. SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and< 0. 5 MB model size[J]. arXiv preprint arXiv:1602. 07360, 2016. 

[8] Wu B, Zhou X, Zhao S, et al. Squeezesegv2: Improved model structure and unsupervised domain adaptation for road-object segmentation from a lidar point cloud[C]//2019 International Conference on Robotics and Automation (ICRA). IEEE, 2019: 4376-4382. 

[9] Imad M, Doukhi O, Lee D J. Transfer learning based semantic segmentation for 3d object detection from point cloud [J]. Sensors, 2021, 21(12): 3964. 

[10] Maturana D, Scherer S. Voxnet: A 3d convolutional neural network for real-time object recognition [C]//2015 IEEE/RSJ international conference on intelligent robots and systems (IROS). IEEE, 2015: 922-928. 

[11] Verdoja F, Thomas D, Sugimoto A. Fast 3D point cloud segmentation using supervoxels with geometry and color for 3D scene understanding[C]//2017 IEEE International Conference on Multimedia and Expo (ICME). IEEE, 2017: 1285-1290. 

[12] Li Y, Pirk S, Su H, et al. Fpnn: Field probing neural networks for 3d data [J]. Advances in neural information processing systems, 2016, 29. 

[13] Tchapmi L, Choy C, Armeni I, et al. Segcloud: Semantic segmentation of 3d point clouds[C]//2017 international conference on 3D vision (3DV). IEEE, 2017: 537-547. 

[14] Le T, Duan Y. Pointgrid: A deep network for 3d shape understanding[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 9204-9214. 

[15] Su H, Maji S, Kalogerakis E, et al. Multi-view convolutional neural networks for 3d shape recognition[C]//Proceedings of the IEEE international conference on computer vision. 2015: 945-953. 

[16] Jiang J, Bao D, Chen Z, et al. MLVCNN: Multi-loop-view convolutional neural network for 3D shape retrieval [C]//Proceedings of the AAAI Conference on Artificial Intelligence. 2019, 33(01): 8513-8520. 

[17] Tatarchenko M, Park J, Koltun V, et al. Tangent convolutions for dense prediction in 3d [C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2018: 3887-3896. 

[18] Qi C R, Su H, Mo K, et al. PointNet: Deep learning on point sets for 3d classification and segmentation[C]. IEEE, 2017, 652-660. 

[19] Qi C R, Yi L, Su H, et al. PointNet++: Deep hierarchical feature learning on point sets in a metric space [J]. Advances in neural information processing systems, 2017, 30. 

[20] Lai Ming, Zhao Jiankang, Liu Chuanqi, et al. Semantic Segmentation of LiDAR Point Cloud Based on CAFF-PointNet [J]. Laser & Optoelectronics Progress, 2021, 58(20):488-497. 

[21] Hu Chuanwen, Lu Shijie, Yang Wenjing, et al. Deep learning architecture for building extraction using LiDAR point clouds[J]. Bulletin of Surveying and Mapping, 2021, No. 537(12):88-93. 

[22] Dai Mofan, Xing Shuai, Xu Qing, et al. Semantic segmentation of airborne LiDAR point cloud based on multi-feature fusion and geometric convolution[J]. Journal of Image and Graphics, 2022, 27(02):574-585. 

[23] Meng Congtang, Zhao Yindi, Han Wenquan, et al. RandLA-Net-based detection of urban building change using airborne LiDAR point clouds[J]. Remote Sensing for Natural Resources, 2022, 34(04):113-121. 

[24] Wu Jun, Cui Yue, Zhao Xuemei, et al. SSA-PointNet++:A Space Self-Attention CNN for the Semantic Segmentation of 3D Point Cloud[J]. Journal of Computer-Aided Design & Computer, 2022, 34(03):437-448. 

[25] Jiang M, Wu Y, Zhao T, et al. Pointsift: A sift-like network module for 3d point cloud semantic segmentation [J]. arXiv preprint arXiv:1807. 00652, 2018. 

[26] Liang Zhenhua, Wang Feng. Attention weighted feature aggregation PointNet network for part segmentation [J/OL]. Application Research of Computers: 1-8 [2023-01-27]. 

[27] Qian G, Hammoud H, Li G, et al. Assanet: An anisotropic separable set abstraction for efficient point cloud representation learning [J]. Advances in Neural Information Processing Systems, 2021, 34: 28119-28130. 

[28] Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[J]. Advances in neural information processing systems, 2017, 30. 

[29] Yang Qingke, Li Yongqiang, Li Lixue, et al. Building façade modeling combining Vehicle-borne LiDAR data with airborne point cloud data[J]. Science of Surveying and Mapping, 2019, 44(02):94-101.