Abstract
Terrestrial laser scanning campaigns provide an important means to document the 3D structure of historical sites. Unfortunately, the process of converting the 3D point clouds acquired by the laser scanner into a coherent and accurate 3D model has many stages and is not generally automated. In particular, the initial cleaning stage of the pipeline—in which undesired scene points are deleted—remains largely manual and is usually labour intensive. In this article, we introduce a semi-automated cleaning approach that incrementally trains a random forest (RF) classifier on an initial keep/discard point labelling generated by the user when cleaning the first scan(s). The classifier is then used to predict the labelling of the next scan in the sequence. Before this classification is presented to the user, a denoising post-process, based on the 2D range map representation of the laser scan, is applied. This significantly reduces small isolated point clusters that the user would otherwise have to fix. The user then selects the remaining incorrectly labelled points and these are weighted, based on a confidence estimate, and fed back into the classifier to retrain it for the next scan. Our experiments, across 8 scanning campaigns, show that when the scan campaign is coherent, i.e., it does not contain widely disparate or contradictory data, the classifier yields a keep/discard labelling that typically ranges between 95% and 99%. This is somewhat surprising, given that the data in each class can represent many object types, such as a tree, person, wall, and so on, and that no further effort beyond the point labeling of keep/discard is required of the user. We conducted an informal timing experiment over a 15-scan campaign, which compared the processing time required by our software, without user interaction (point label correction) time, against the time taken by an expert user to completely clean all scans. The expert user required 95mins to complete all cleaning. The average time required by the expert to clean a single scan was 6.3mins. Even with current unoptimized code, our system was able to generate keep/discard labels for all scans, with 98% (average) accuracy, in 75mins. This leaves as much as 20mins for the user input required to relabel the 2% of mispredicted points across the set of scans before the full system time would match the expert’s cleaning time.
Supplemental Material
Available for Download
Supplemental movie, appendix, image and software files for, Semi-automated Cleaning of Laser Scanning Campaigns with Machine Learning
- Gustavo E. A. P. A. Batista, Ronaldo C. Prati, and Maria Carolina Monard. 2004. A study of the behavior of several methods for balancing machine learning training data. ACM SIGKDD Explor. Newslett. 6, 1 (2004), 20--29. Google ScholarDigital Library
- Mark de Berg, Otfried Cheong, Marc van Kreveld, and Mark Overmars. 2008. Computational Geometry: Algorithms and Applications (3rd ed.). Springer-Verlag TELOS, Santa Clara, CA. Google ScholarDigital Library
- Christopher M. Bishop. 2006. Pattern Recognition and Machine Learning. Springer-Verlag New York, Inc., Secaucus, NJ.Google Scholar
- Yuri Boykov, Olga Veksler, and Ramin Zabih. 2001. Fast approximate energy minimization via graph cuts. IEEE Trans. Patt. Anal. Machine Intell. 23, 11 (2001), 1222--1239. Google ScholarDigital Library
- Leo Breiman. 2001. Random forests. Machine Learn. 45, 1 (2001), 5--32. Google ScholarDigital Library
- L. C. Chen, G. Papandreou, I. Kokkinos, K. Murphy, and A. L. Yuille. 2018. DeepLab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs. IEEE Trans. Patt. Anal. Machine Intell. 40, 4 (2018), 834--838.Google ScholarCross Ref
- Paolo Cignoni, Marco Callieri, Massimiliano Corsini, Matteo Dellepiane, Fabio Ganovelli, and Guido Ranzuglia. 2008. MeshLab: An open-source mesh processing tool. In Proceedings of the 6th Eurographics Italian Chapter Conference. 129--136. http://vcg.isti.cnr.it/Publications/2008/CCCDGR08.Google Scholar
- David Doria and Richard J. Radke. 2012. Filling large holes in lidar data by inpainting depth gradients. In Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops (CVPRW’12). IEEE, 65--72.Google Scholar
- J. Elseberg, D. Borrmann, and A. Nüchter. 2011. Full wave analysis in 3D laser scans for vegetation detection in urban environments. In Proceedings of the 23rd International Symposium on Information, Communication and Automation Technologies. 1--7.Google Scholar
- Timo Hackel, Nikolay Savinov, Lubor Ladicky, Jan D. Wegner, Konrad Schindler, and Marc Pollefeys. 2017. Semantic3D. net: A new large-scale point cloud classification benchmark. Retrieved from arXiv preprint arXiv:1704.03847.Google Scholar
- Timo Hackel, Jan D. Wegner, and Konrad Schindler. 2016. Fast semantic segmentation of 3D point clouds with strongly varying density. ISPRS Ann. Photogram., Remote Sens. Spatial Inform. Sci. 3, 3 (2016).Google Scholar
- Jungong Han, Ling Shao, Dong Xu, and Jamie Shotton. 2013. Enhanced computer vision with Microsoft Kinect sensor: A review. IEEE Trans. Cyber. 43, 5 (2013), 1318--1334.Google ScholarCross Ref
- Alexander Hermans, Georgios Floros, and Bastian Leibe. 2014. Dense 3D semantic mapping of indoor scenes from RGB-D images. In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA’14). IEEE, 2631--2638.Google ScholarCross Ref
- Jing Huang and Suya You. 2016. Point cloud labeling using 3D convolutional neural network. In Proceedings of the 23rd International Conference on Pattern Recognition (ICPR’16). IEEE, 2670--2675.Google ScholarCross Ref
- Juha Hyyppä, Anttoni Jaakkola, Yuwei Chen, and Antero Kukko. 2013. Unconventional lidar mapping from air, terrestrial and mobile. In Proceedings of the Photogrammetric Week Conference. 205--214.Google Scholar
- Byung-soo Kim, Pushmeet Kohli, and Silvio Savarese. 2013. 3D scene understanding by Voxel-CRF. In Proceedings of the IEEE International Conference on Computer Vision (ICCV’13). IEEE, 1425--1432.Google Scholar
- J. Kisztner, J. Jelínek, T. Daněk, and J. Růžička. 2016. 3D documentation of outcrop by laser scanner—Filtration of vegetation. Perspect. Sci. 7 (2016), 161--165.Google ScholarCross Ref
- Hema S. Koppula, Abhishek Anand, Thorsten Joachims, and Ashutosh Saxena. 2011. Semantic labeling of 3D point clouds for indoor scenes. In Proceedings of the Conference on Advances in Neural Information Processing Systems. 244--252. Google ScholarDigital Library
- Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton. 2012. Imagenet classification with deep convolutional neural networks. In Proceedings of the Conference on Advances in Neural Information Processing Systems. 1097--1105. Google ScholarDigital Library
- Rickert Mulder and Patrick Marais. 2016. Accelerating point cloud cleaning. In Proceedings of the 14th Eurographics Workshop on Graphics and Cultural Heritage (GCH’16). Eurographics Association, 211--214. Google ScholarDigital Library
- Daniel Munoz, Nicolas Vandapel, and Martial Hebert. 2008. Directional associative Markov network for 3-D point cloud classification. In Proceedings of the International Symposium on 3D Data Processing, Visualization and Transmission.Google Scholar
- Guan Pang and Ulrich Neumann. 2013. Training-based object recognition in cluttered 3D point clouds. In Proceedings of the International Conference on 3D Vision (3DV’13). IEEE, 87--94.Google ScholarDigital Library
- Charles R. Qi, Hao Su, Kaichun Mo, and Leonidas J. Guibas. 2017. Pointnet: Deep learning on point sets for 3D classification and segmentation. Proc. Comput. Vis. Patt. Recog., IEEE 1, 2 (2017), 4.Google Scholar
- Heinz Rüther, Christoph Held, Roshan Bhurtha, Ralph Schroeder, and Stephen Wessels. 2012. From point cloud to textured model, the Zamani laser scanning pipeline in heritage documentation. South African J. Geomat. 1, 1 (2012), 44--59.Google Scholar
- Amir Saffari, Christian Leistner, Jakob Santner, Martin Godec, and Horst Bischof. 2009. On-line random forests. In Proceedings of the 12th IEEE International Conference on Computer Vision Workshops (ICCV’09). IEEE, 1393--1400.Google ScholarCross Ref
- Richard Socher, Brody Huval, Bharath Bath, Christopher D. Manning, and Andrew Y. Ng. 2012. Convolutional-recursive deep learning for 3D object classification. In Adv. Neural Inform. Proc. Syst. 656--664. Google ScholarDigital Library
- Rudolph Triebel, Kristian Kersting, and Wolfram Burgard. 2006. Robust 3D scan point classification using associative Markov networks. In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA’06). IEEE, 2603--2608.Google ScholarCross Ref
- Julien Valentin, Vibhav Vineet, Ming-Ming Cheng, David Kim, Jamie Shotton, Pushmeet Kohli, Matthias Niessner, Antonio Criminisi, Shahram Izadi, and Philip Torr. 2015. SemanticPaint: Interactive 3D labeling and learning at your fingertips. ACM Trans. Graph. 34, 5, Article 154 (Nov. 2015), 17 pages. Google ScholarDigital Library
- Martin Weinmann, Boris Jutzi, Stefan Hinz, and Clément Mallet. 2015. Semantic point cloud interpretation based on optimal neighborhoods, relevant features and efficient classifiers. ISPRS J. Photogram. Remote Sens. 105 (2015), 286--304.Google ScholarCross Ref
- Daniel Wolf, Johann Prankl, and Markus Vincze. 2015. Fast semantic segmentation of 3D point clouds using a dense CRF with learned parameters. In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA’15). IEEE, 4867--4873.Google ScholarCross Ref
- Jingyu Yang, Ziqiao Gan, Kun Li, and Chunping Hou. 2015. Graph-based segmentation for RGB-D data using 3-D geometry enhanced superpixels. IEEE Trans. Cyber. 45, 5 (2015), 927--940.Google ScholarCross Ref
- Na-Eun Yang, Yong-Gon Kim, and Rae-Hong Park. 2012. Depth hole filling using the depth distribution of neighboring regions of depth holes in the Kinect sensor. In Proceedings of the IEEE International Conference on Signal Processing, Communication and Computing (ICSPCC’12). IEEE, 658--661.Google ScholarCross Ref
- Zhi-Hua Zhou and Ji Feng. 2017. Deep forest: Towards an alternative to deep neural networks. Retrieved from arXiv preprint arXiv:1702.08835.Google Scholar
Index Terms
- Semi-automated Cleaning of Laser Scanning Campaigns with Machine Learning
Recommendations
Transductive Multilabel Learning via Label Set Propagation
The problem of multilabel classification has attracted great interest in the last decade, where each instance can be assigned with a set of multiple class labels simultaneously. It has a wide variety of real-world applications, e.g., automatic image ...
SPL-LDP: a label distribution propagation method for semi-supervised partial label learning
AbstractPartial label learning learns from examples represented by a single instance while associated with multiple candidate labels, among which only one valid label resides. However, in real-world applications, collecting candidate label sets for all ...
Comments