Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

L’utilisation des nuages de points s’est accrue au cours des dernières années. Afin d’exploiter correctement ces données, la segmentation et la classification semblent essentielles et font dès lors l’objet de nombreuses recherches. La plupart de ces recherches se concentrent uniquement sur le post-traitement des données. Il semble toutefois intéressant de considérer l’intégration d’informations sémantiques lors de l’acquisition pour faciliter ces traitements. C’est le principe de la géocodification, principalement utilisée par les géomètres lors de leurs levés.
Ce travail envisage donc l’utilisation d’une géocodification lors de l’acquisition d’un nuage de points avec l’objectif d’améliorer le processus de segmentation/classification d’un environnement bâti grâce aux informations récoltées.
Deux méthodologies différentes ont été développées à cet effet. La première se rapproche d’une géocodification traditionnelle puisqu’elle consiste à combiner les informations spatiales et sémantiques issues d’un levé topographique avec un nuage de points afin de procéder à sa classification. La seconde méthode, plus innovante, se base sur le positionnement de cibles sur les éléments d’intérêts afin de guider le processus de segmentation/classification. Les différentes étapes nécessaires à la réalisation de ces deux méthodes sont détaillées dans ce travail, depuis l’établissement des géocodifications jusqu’à l’obtention des nuages de points classifiés, en passant par l’acquisition et les différents prétraitements à appliquer. 
Les résultats obtenus pour les deux méthodes sont ensuite présentés, validés et comparés entre eux. Il en ressort que les deux méthodes proposées sont plus rapides mais un peu moins précises qu’une segmentation manuelle avec un F1-score moyen compris entre 0,82 et 0,93 pour le nuage étudié. La méthode avec levé topographique présente des résultats de meilleure qualité que la méthode des cibles mais elle nécessite un temps d’acquisition plus long. Ce travail se clôture sur la mise en évidence des limites rencontrées pour les deux méthodes. Des perspectives d’amélioration et de développement sont également présentées. The use of point clouds has increased in recent years. In order to exploit these data correctly, segmentation and classification appear essential and are therefore the subject of numerous researches. Most of these researches only focus on post-processing data. However, it seems interesting to consider the integration of semantic information during the data acquisition to facilitate the treatments. This is the principle of feature codes, mainly used by land surveyors during their surveys. 
Therefore, this work considers the use of feature codes during a point cloud acquisition with the aim of improving the segmentation/classification process of a built environment through the collected information.
Two different methodologies have been developed for this purpose. The first one is similar to traditional feature codes since it consists in combining spatial and semantic information from a topographical survey with a point cloud in order to proceed with its classification. The second method, more innovative, is based on the positioning of targets on the elements of interest to guide the segmentation/classification process. The different steps necessary to achieve these two methods are detailed in this work, from the establishment of the feature codes list to the point clouds classification, passing by the acquisition and the pre-processing.
The obtained results for both methods are then presented, validated and compared. It shows that the two proposed methods are faster but slightly less accurate than a manual segmentation with an average F1-score between 0.82 and 0.93 for the point cloud studied. The topographic survey method presents better results than the target method but it requires a longer acquisition time. This work concludes by highlighting the limitations encountered for the two methods. Improvement and development perspectives are also presented.

Géocodification --- Segmentation --- Classification --- Scanner laser terrestre --- Nuage de points --- Point cloud --- Physique, chimie, mathématiques & sciences de la terre > Sciences de la terre & géographie physique

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Mobile mapping is applied widely in society, for example, in asset management, fleet management, construction planning, road safety, and maintenance optimization. Yet, further advances in these technologies are called for. Advances can be radical, such as changes to the prevailing paradigms in mobile mapping, or incremental, such as the state-of-the-art mobile mapping methods. With current multi-sensor systems in mobile mapping, laser-scanned data are often registered in point clouds with the aid of global navigation satellite system (GNSS) positioning or simultaneous localization and mapping (SLAM) techniques and then labeled and colored with the aid of machine learning methods and digital camera data. These multi-sensor platforms are beginning to undergo further advancements via the addition of multi-spectral and other sensors and via the development of machine learning techniques used in processing this multi-modal data. Embedded systems and minimalistic system designs are also attracting attention, from both academic and commercial perspectives.This book contains the accepted publications of the Special Issue 'Advances in Mobile Mapping Technologies' of the Remote Sensing journal. It consists of works introducing a new mobile mapping dataset (‘Paris CARLA 3D’), system calibration studies, SLAM topics, and multiple deep learning works for asset detection. We, the Guest Editors, Ville Lehtola from University of Twente, Netherlands, Andreas Nüchter from University of Würzburg, Germany, and François Goulette from Mines Paris- PSL University, France, wish to thank all the authors who contributed to this collection.

Technology: general issues --- History of engineering & technology --- LiDAR --- RetinaNet --- inception --- Mobile Laser Scanning --- point clouds --- data fusion --- Lidar --- point cloud density --- point cloud coverage --- mobile mapping systems --- 3D simulation --- Pandar64 --- Ouster OS-1-64 --- mobile laser scanning --- lever arm --- boresight angles --- plane-based calibration field --- configuration analysis --- accuracy --- controllability --- evaluation --- control points --- TLS reference point clouds --- visual–inertial odometry --- Helmert variance component estimation --- line feature matching method --- correlation coefficient --- point and line features --- mobile mapping --- manhole cover --- point cloud --- F-CNN --- transfer learning --- CAM localization --- loop closure detection --- visual SLAM --- semantic topology graph --- graph matching --- CNN features --- deep learning --- view planning --- imaging network design --- building 3D modelling --- path planning --- V-SLAM --- real-time --- guidance --- embedded-systems --- 3D surveying --- exposure control --- photogrammetry --- parking statistics --- vehicle detection --- robot operating system --- 3D camera --- RGB-D --- performance evaluation --- convolutional neural networks --- smart city --- georeferencing --- MSS --- IEKF --- DSIEKF --- geometrical constraints --- 6-DoF --- DTM --- 3D city model --- dataset --- laser scanning --- 3D mapping --- synthetic --- outdoor --- semantic --- scene completion

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Mobile mapping is applied widely in society, for example, in asset management, fleet management, construction planning, road safety, and maintenance optimization. Yet, further advances in these technologies are called for. Advances can be radical, such as changes to the prevailing paradigms in mobile mapping, or incremental, such as the state-of-the-art mobile mapping methods. With current multi-sensor systems in mobile mapping, laser-scanned data are often registered in point clouds with the aid of global navigation satellite system (GNSS) positioning or simultaneous localization and mapping (SLAM) techniques and then labeled and colored with the aid of machine learning methods and digital camera data. These multi-sensor platforms are beginning to undergo further advancements via the addition of multi-spectral and other sensors and via the development of machine learning techniques used in processing this multi-modal data. Embedded systems and minimalistic system designs are also attracting attention, from both academic and commercial perspectives.This book contains the accepted publications of the Special Issue 'Advances in Mobile Mapping Technologies' of the Remote Sensing journal. It consists of works introducing a new mobile mapping dataset (‘Paris CARLA 3D’), system calibration studies, SLAM topics, and multiple deep learning works for asset detection. We, the Guest Editors, Ville Lehtola from University of Twente, Netherlands, Andreas Nüchter from University of Würzburg, Germany, and François Goulette from Mines Paris- PSL University, France, wish to thank all the authors who contributed to this collection.

Technology: general issues --- History of engineering & technology --- LiDAR --- RetinaNet --- inception --- Mobile Laser Scanning --- point clouds --- data fusion --- Lidar --- point cloud density --- point cloud coverage --- mobile mapping systems --- 3D simulation --- Pandar64 --- Ouster OS-1-64 --- mobile laser scanning --- lever arm --- boresight angles --- plane-based calibration field --- configuration analysis --- accuracy --- controllability --- evaluation --- control points --- TLS reference point clouds --- visual–inertial odometry --- Helmert variance component estimation --- line feature matching method --- correlation coefficient --- point and line features --- mobile mapping --- manhole cover --- point cloud --- F-CNN --- transfer learning --- CAM localization --- loop closure detection --- visual SLAM --- semantic topology graph --- graph matching --- CNN features --- deep learning --- view planning --- imaging network design --- building 3D modelling --- path planning --- V-SLAM --- real-time --- guidance --- embedded-systems --- 3D surveying --- exposure control --- photogrammetry --- parking statistics --- vehicle detection --- robot operating system --- 3D camera --- RGB-D --- performance evaluation --- convolutional neural networks --- smart city --- georeferencing --- MSS --- IEKF --- DSIEKF --- geometrical constraints --- 6-DoF --- DTM --- 3D city model --- dataset --- laser scanning --- 3D mapping --- synthetic --- outdoor --- semantic --- scene completion

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

The Special Issue collects papers showing the progress made in key areas of photogrammetry and remote sensing such as modern and/or forthcoming sensors, improvements in data processing strategies and assessment of their reliability, application of innovations as proof of the contribution offered in the observation of the natural and built environment with better understanding of phenomena at required spatial scale.

VHR tri-stereo satellite imagery --- digital elevation model --- isolated objects --- dense image matching --- change detection --- natural disasters --- deep learning --- threshold selection --- optical flow estimation --- Structure from Motion (SfM) --- 3D reconstruction --- noise estimation --- point clouds --- roughness --- surface reconstruction --- mesh model --- visibility constraints --- volumetric methods --- dense point cloud --- multiple view stereo (MVS) --- dense image matching (DIM) --- photogrammetry --- computer vision --- Copernicus --- Sentinel-1 --- Sentinel-2 --- InSAR --- damage proxy map --- Beirut --- Lebanon --- explosion --- radiometric calibration --- modeling --- geometric error --- high-precision calibration --- n/a --- preprocessing --- enhancement --- point cloud --- image processing --- image histogram --- UAV --- camera calibration --- GNSS-assisted block orientation --- dome effect --- Monte Carlo simulation --- soil moisture content --- artificial neural network --- sample optimization --- synthetic aperture radar --- optical remote sensing image

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

The Special Issue collects papers showing the progress made in key areas of photogrammetry and remote sensing such as modern and/or forthcoming sensors, improvements in data processing strategies and assessment of their reliability, application of innovations as proof of the contribution offered in the observation of the natural and built environment with better understanding of phenomena at required spatial scale.

Technology: general issues --- History of engineering & technology --- VHR tri-stereo satellite imagery --- digital elevation model --- isolated objects --- dense image matching --- change detection --- natural disasters --- deep learning --- threshold selection --- optical flow estimation --- Structure from Motion (SfM) --- 3D reconstruction --- noise estimation --- point clouds --- roughness --- surface reconstruction --- mesh model --- visibility constraints --- volumetric methods --- dense point cloud --- multiple view stereo (MVS) --- dense image matching (DIM) --- photogrammetry --- computer vision --- Copernicus --- Sentinel-1 --- Sentinel-2 --- InSAR --- damage proxy map --- Beirut --- Lebanon --- explosion --- radiometric calibration --- modeling --- geometric error --- high-precision calibration --- preprocessing --- enhancement --- point cloud --- image processing --- image histogram --- UAV --- camera calibration --- GNSS-assisted block orientation --- dome effect --- Monte Carlo simulation --- soil moisture content --- artificial neural network --- sample optimization --- synthetic aperture radar --- optical remote sensing image