Catalogue en ligne IGN

Reconstruction of 3D building models from aerial images and maps / ldikó Süveg (2003)

Public

Titre :	Reconstruction of 3D building models from aerial images and maps
Type de document :	Thèse/HDR
Auteurs :	ldikó Süveg, Auteur
Editeur :	Delft : Netherlands Geodetic Commission NGC
Année de publication :	2003
Collection :	Netherlands Geodetic Commission Publications on Geodesy, ISSN 0165-1706 num. 53
Importance :	143 p.
Format :	17 x 24 cm
ISBN/ISSN/EAN :	978-90-6132-280-1
Note générale :	Bibliographie
Langues :	Anglais (eng)
Descripteur :	[Vedettes matières IGN] Photogrammétrie numérique [Termes IGN] arbre de décision [Termes IGN] base de données urbaines [Termes IGN] bibliothèque de formes [Termes IGN] contour [Termes IGN] fusion de données [Termes IGN] image aérienne [Termes IGN] modèle 3D de l'espace urbain [Termes IGN] modélisation 3D [Termes IGN] reconstruction 3D du bâti [Termes IGN] segmentation [Termes IGN] zone urbaine
Index. décimale :	33.30 Photogrammétrie numérique
Résumé :	(Auteur) The 3D reconstruction of buildings has numerous applications in areas that include urban planning, construction, environment, communication, transportation, energy and property management, tourism, and virtual tours of cities. In this thesis, the reconstruction of 3D building models from aerial images is addressed. The approach presented in this thesis integrates the aerial images analysis with information from a GIS database and domain knowledge. The problem of automatic 3D building reconstruction has been a central research topic in computer vision and image understanding communities as well as in digital photogrammetry for many years. A variety of approaches has been suggested for the reconstruction of buildings from aerial images. Despite considerable research effort, there is no complete system that can reliably perform autonomous 3D building reconstruction in a wide variety of scene domains. This is particularly true in complex urban areas containing buildings with different shapes and roof types as well as in complicated underlying terrain. Of course, some progress has been made, but there is room for improvement. This improvement can be achieved by fusing multiple data sources and some a priori information. In this project, largescale 2D GIS databases were used as additional information source. Combination of image data and map data turned out to improve the reliability of the reconstruction. Generic knowledge about the shape of the buildings is also incorporated in the system. Since most buildings can be described as an aggregation of simple building types, the knowledge about the problem domain can be represented in a building library containing simple building models. Therefore, a building library was defined containing the most common building primitives, such as flat roof, and different types of gable roofs. The building reconstruction process was formulated as a multilevel hypothesis generation and verification scheme and it was implemented as a search tree. A method that can localize the buildings in images using map information has been developed. Also, a method for generating building hypotheses corresponding to the primitives defined in the building library has been developed. This implies stereo matching of image features (corners, lines) which correspond to map primitives and fitting of the building hypotheses to images. A further contribution is the definition of a metric for evaluating the generated building hypotheses in order to select the one which best describes the image. The metric is based on the formulation of the mutual information between the building model and the images. Methods for the estimation of the mutual information from training samples were analyzed. This metric has been rigorously derived from information theory and does not require a priori information about the surface properties of the object and is robust with respect to variations of illumination. Also, no assumption about the shape of the objects are made. As result the method is quite general and may be used in a wide variety of applications. The produced approach is able to meet most of the requirements of an automatic 3D building reconstruction system. The developed system has been used in urban and suburban areas to reconstruct buildings and showed good results. Experiments were carried out on two data sets with different characteristics. The system was able to reconstruct more than 80% of the buildings and the accuracy of the reconstruction is good enough for mapping purposes.
Numéro de notice :	15043
Affiliation des auteurs :	non IGN
Thématique :	IMAGERIE
Nature :	Thèse étrangère
DOI :	sans
En ligne :	https://ncgeo.nl/downloads/53Suveg.pdf
Format de la ressource électronique :	URL
Permalink :	https://documentation.ensg.eu/index.php?lvl=notice_display&id=55049

Exemplaires(2)

Code-barres	Cote	Support	Localisation	Section	Disponibilité
15043-02	33.30	Livre	Centre de documentation	Photogrammétrie - Lasergrammétrie	Disponible
15043-01	33.30	Livre	Centre de documentation	Photogrammétrie - Lasergrammétrie	Disponible

IGN

Centre de documentation
scientifique

Accueil

Sélection de la langue

Adresse

Se connecter

Actualité

L'actu ! Horaires du CDOS

Informations pratiques

Exemplaires(2)

IGN / ENSG

L'IGN a pour vocation

Accès directs

2014 -2024 IGN

IGN

Centre de documentationscientifique

Accueil

Sélection de la langue

Adresse

Se connecter

Actualité

L'actu ! Horaires du CDOS

Informations pratiques

Exemplaires(2)

IGN / ENSG

L'IGN a pour vocation

Accès directs

2014 -2024 IGN

Centre de documentation
scientifique