This paper introduces a robust normal vector estimator for point cloud data. It can handle sharp features as well as smooth areas. Our method is based on the inclusion of a robust estimator into a Principal Component Analysis in the neighborhood of the studied point which can detect and reject outliers automatically during the estimation. A projection process ensures robustness against noise. Two automatic initializations are computed leading to independent optimizations making the algorithm robust to neighborhood anisotropy around sharp features. An evaluation has been carried out in which the algorithm is compared to state-of-the-art methods. The results show that it is more robust against a low and/or a non-uniform sampling, a high noise level and outliers. Moreover, our algorithm is fast relatively to existing methods handling sharp features. The code and data sets will be available online.
Acquiring 3D data with LiDAR systems involves scanning multiple scenes from different points of view. In actual systems, the ICP algorithm (Iterative Closest Point) is commonly used to register the acquired point clouds together to form a unique one. However, this method faces local minima issues and often needs a coarse initial alignment to converge to the optimum. This paper develops a new method for registration adapted to indoor environments and based on structure priors of such scenes. Our method works without odometric data or physical targets. The rotation and translation of the rigid transformation are computed separately, using, respectively, the Gaussian image of the point clouds and a correlation of histograms. To evaluate our algorithm on challenging registration cases, two datasets were acquired and are available for comparison with other methods online. The evaluation of our algorithm on four datasets against six existing methods shows that the proposed method is more robust against sampling and scene complexity. Moreover, the time performances enable a real-time implementation.Data Set: https://liris.cnrs.fr/3d-registration/ Data Set License: ODC Attribute License
Abstract. This paper deals with 3D modeling of building interiors from point clouds captured by a 3D LiDAR scanner. Indeed, currently, the building reconstruction processes remain mostly manual. While LiDAR data have some specific properties which make the reconstruction challenging (anisotropy, noise, clutters, etc.), the automatic methods of the state-of-the-art rely on numerous construction hypotheses which yield 3D models relatively far from initial data. The choice has been done to propose a new modeling method closer to point cloud data, reconstructing only scanned areas of each scene and excluding occluded regions. According to this objective, our approach reconstructs LiDAR scans individually using connected polygons. This modeling relies on a joint processing of an image created from the 2D LiDAR angular sampling and the 3D point cloud associated to one scan. Results are evaluated on synthetic and real data to demonstrate the efficiency as well as the technical strength of the proposed method.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.