Centre de documentation
scientifique

Geometric approximation of structured scenes from images / Muxingzi Li (2021)  

Public Titre : Geometric approximation of structured scenes from images Type de document : Thèse/HDR Auteurs : Muxingzi Li, Auteur ; Renaud Marlet, Directeur de la recherche Editeur : Nice : Université Côte d'Azur Année de publication : 2021 Importance : 122 p. Format : 21 x 30 cm Note générale : bibliographie Thèse de Doctorat résentée en vue de l’obtention du grade de docteur en Informatique de l’Université Côte d’Azur Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] approximation [Termes IGN] chaîne de traitement [Termes IGN] détection d'objet [Termes IGN] extraction automatique [Termes IGN] maillage [Termes IGN] modélisation 3D [Termes IGN] primitive géométrique [Termes IGN] recalage de données localisées [Termes IGN] reconstruction d'image [Termes IGN] scène urbaine [Termes IGN] segmentation d'image [Termes IGN] semis de points [Termes IGN] superposition de données [Termes IGN] vectorisation [Termes IGN] vision par ordinateur Index. décimale : THESE Thèses et HDR Résumé : (auteur) Geometric approximation of urban objects with compact and accurate representation is a challenging problem that concerns both computer vision and computer graphics communities. Existing literature mainly focuses on reconstruction from high-quality point clouds obtained by laser scanning which are too costly for many practical scenarios. This motivates the investigation into the problem of geometric approximation from low-budget image data. Dense reconstruction from a collection of images is made possible by recent advances in multi-view stereo techniques, yet the resulting point cloud is often far from perfect for generating a compact model. In particular, our goal is to describe the captured scene with a compact and accurate representation. In this thesis, we propose two generic algorithms which address different aspects of image-based geometric approximation. First, we present an algorithm for extracting and vectorizing objects in images with polygons. Second, we present a global registration algorithm from multi-modal geometric data, typically 3D point clouds and surface meshes. Both approaches exploit detection of linear geometric primitives to approximate either 2D silhouettes with polygons consisting of line segments, or 3D point sets with a collection of planar shapes. The proposed algorithms could be used sequentially to form a pipeline for geometric approximation of an urban object from a set of image data, consisting of an overhead shot for coarse model extraction and multi-view stereo data for point cloud generation. We demonstrate the robustness and scalability of our methods for structured scenes and objects, as well as applicative potential for free-form objects. Note de contenu : 1- Introduction 2- Literature review 3- Polygonal image segmentation 4- 3D registration of multi-modal geometry 5- Application to floor modeling 6- Conclusion and perspectives Numéro de notice : 28627 Affiliation des auteurs : non IGN Thématique : IMAGERIE/INFORMATIQUE Nature : Thèse française Note de thèse : Thèse de Doctorat : Informatique : Côte d'Azur : 2021 Organisme de stage : INRIA DOI : sans En ligne : https://tel.hal.science/tel-03388295v2/ Format de la ressource électronique : URL Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=99557