Skip to Main content Skip to Navigation

Pré-traitement de grosses bases de données pour la visualisation interactive

Xavier Décoret 1
1 iMAGIS - Models, Algorithms and Geometry for Computer Generated Image Graphics
GRAVIR - IMAG - Laboratoire d'informatique GRAphique, VIsion et Robotique de Grenoble, Inria Grenoble - Rhône-Alpes
Abstract : To navigate in real-time in a virtual environnement, one must be able to generate rapidly new views of the 3D model describing this environment. In this thesis, we got interested in pre-computing results that could later be used to speed-up the rendering of those views. Two directions have been explored. In a first part we pre-compute, for each possible location, a list of objects that are hidden by other objects and thus need not to be displayed. This pre-computation relies on a so-called "shrinking theorem": if from a view point a shrunk object is hidden by other objects which are also shrunk, then the object is certainly hidden from any neighbouring view point. The first contribution of this thesis is the demonstration of this theorem which formalizes and extends a previously known result. The second contribution is a robust algorithm to compute such shrinkings on any model, given it is watertight (with a well-defined exterior and interior). Once we have determined what is visible, we have to display it in the most efficient manner. A classical solution consists in having, for each object , different representations of different complexity. Depending on the distance between object ant the viewer, the most adequat simplification is displayed. The difficulty is then to build such level of details from an initial model. The third contribution of this thesis is a new representation, called billboard cloud which achieves extreme simplification while keeping a very good visual quality. A complex object is represented by a small number of textured polygons with transparency, whose interleaving allows to render both shape and appearance of the object. An automatic algorithm is described, to compute such a cloud from an initial polygonal model.
Document type :
Complete list of metadata

Cited literature [95 references]  Display  Hide  Download
Contributor : Team Artis <>
Submitted on : Friday, October 22, 2010 - 5:17:32 PM
Last modification on : Monday, December 28, 2020 - 3:44:02 PM
Long-term archiving on: : Sunday, January 23, 2011 - 3:03:19 AM


  • HAL Id : tel-00528890, version 1




Xavier Décoret. Pré-traitement de grosses bases de données pour la visualisation interactive. Interface homme-machine [cs.HC]. Université Joseph-Fourier - Grenoble I, 2002. Français. ⟨tel-00528890⟩



Record views


Files downloads