Centre de documentation
scientifique

A versatile and efficient data fusion methodology for heterogeneous airborne LiDAR and optical imagery data acquired under unconstrained conditions / Thanh Huy Nguyen (2020)  

Public Titre : A versatile and efficient data fusion methodology for heterogeneous airborne LiDAR and optical imagery data acquired under unconstrained conditions Type de document : Thèse/HDR Auteurs : Thanh Huy Nguyen, Auteur ; Jean-Marc Le Caillec, Directeur de thèse ; Sylvie Daniel, Directeur de thèse Editeur : Institut Mines-Télécom Atlantique IMT Atlantique Année de publication : 2020 Autre Editeur : Québec : Université Laval Importance : 173 p. Format : 21 x 30 cm Note générale : Bibliographie Thèse de Doctorat de l'Ecole Nationale Supérieure des Mines-Telecom Atlantique Bretagne Pays de la Loire-IMT Atlantique, Spécialité : Signal, Image, Vision Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Applications photogrammétriques [Termes IGN] détection du bâti [Termes IGN] données lidar [Termes IGN] données localisées 3D [Termes IGN] fusion de données [Termes IGN] image optique [Termes IGN] recalage de données localisées [Termes IGN] reconstruction 3D [Termes IGN] scène urbaine [Termes IGN] semis de points Index. décimale : THESE Thèses et HDR Résumé : (auteur) The necessity and importance of representing a scene in 3-D have been exemplified through numerous remote sensing applications, such as urban planning, disaster management, etc. In these applications, LiDAR and optical imagery data have been used extensively. A complementarity existing between airborne LiDAR and aerial/satellite optical imagery datasets motivates the fusion between them, allowing to represent the observed scenes in 3-D with a better precision and completeness. In recent years, automatic building footprint extraction in urban and residential scenes has become a subject of growing interest among the field of 3-D scene representation and reconstruction. With the rising availability of massive amount of data captured by different LiDAR and imagery sensors onboard airborne and spaceborne platforms, new opportunities arise to perform this task on a large scale. However, existing fusion methods generally consider either hybrid acquisition systems consisting of LiDAR and optical cameras rigidly fixed, or datasets acquired from the same platform at identical or very close dates, and having the same spatial resolution. They do not intend to cope with datasets collected from different platforms with different acquisition configuration at different moments, having different spatial resolutions and levels of detail. Such a context is referred to as unconstrained acquisition context. Furthermore, extracting buildings on a large scale is a complex task. Existing methods reported over the years have achieved relatively significant results by assuming building shapes, enforcing geometrical constraints, or limiting on specific urban areas. Such assumptions are no longer applicable when dealing with large-scale datasets. This research work is devoted to the development of a versatile coarse-to-fine registration method between airborne LiDAR and aerial/satellite optical imagery datasets collected in an unsconstrained acquisition context. It aims at overcoming the challenges associated with this context such as the spatial shift between the datasets, the differences of spatial resolution and level of detail, etc. In addition, this research work elaborates an efficient building footprint extraction method, providing a high accuracy level while being an unsupervised method dedicated to largescale applications. The proposed method, called Super-Resolution-based Snake Model (SRSM), consists in an adaptation of snake models—a conventional image segmentation technique—to operate on high-resolution LiDAR-based elevation images generated by a super-resolution process. It pertains the unconstrained data acquisition context, serving as a prime application example. Relevant results have been achieved when rigorously assessing the proposed methods, namely a highly desirable accuracy level compared to existing methods. Note de contenu : Introduction 1- State of the art 2- Coarse-to-fine Registration of Airborne LiDAR and Optical Imagery Data on Urban Scenes 3- Building Extraction Based on the Fusion of Airborne LiDAR and Optical Imagery Data 4- Conclusions and Perspectives Numéro de notice : 28327 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Thèse française Note de thèse : Thèse de Doctorat : Sciences Géomatiques : Mines-Télécom Atlantique : 2020 Organisme de stage : Lab-STICC DOI : sans En ligne : https://tel.archives-ouvertes.fr/tel-03123328/document Format de la ressource électronique : URL Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=98401