2018/01/15 by Rémi Cura, Cura, Remi, Julien Perret +3
Earth and Planetary Sciences · Environmental Science · Physics and Astronomy · #3D Surveying and Cultural Heritage #Advanced Optical Sensing Technologies #Computational Geometry (cs.CG) #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Remote Sensing and LiDAR Applications
paper · pdf · doi:10.48550/arxiv.1801.05038
openalex publication_date 2018/01/15 · openalex created_date 2023/04/19 · openalex updated_date 2026/07/28
Lidar datasets are becoming more and more common. They are appreciated for\ntheir precise 3D nature, and have a wide range of applications, such as surface\nreconstruction, object detection, visualisation, etc. For all this\napplications, having additional semantic information per point has potential of\nincreasing the quality and the efficiency of the application. In the last\ndecade the use of Machine Learning and more specifically classification methods\nhave proved to be successful to create this semantic information. In this\nparadigm, the goal is to classify points into a set of given classes (for\ninstance tree, building, ground, other). Some of these methods use descriptors\n(also called feature) of a point to learn and predict its class. Designing the\ndescriptors is then the heart of these methods. Descriptors can be based on\npoints geometry and attributes, use contextual information, etc. Furthermore,\ndescriptors can be used by humans for easier visual understanding and sometimes\nfiltering. In this work we propose a new simple geometric descriptor that gives\ninformation about the implicit local dimensionality of the point cloud at\nvarious scale. For instance a tree seen from afar is more volumetric in nature\n(3D), yet locally each leaves is rather planar (2D). To do so we build an\noctree centred on the point to consider, and compare the variation of the\noccupancy of the cells across the levels of the octree. We compare this\ndescriptor with the state of the art dimensionality descriptor and show its\ninterest. We further test the descriptor for classification within the Point\nCloud Server, and demonstrate efficiency and correctness results.\n