High-Fidelity Image-Based Modeling and Rendering
Développer des logiciels perfectionnés de modélisation 3D à partir d'images, ce qui fera des appareils photos et des caméras grand public des capteurs 3D flexibles, précis et peu coûteux.
Il existe un besoin grandissant de "contenu" tri-dimensionnel (3D) dans le domaine des loisirs, de l'ingénierie et de la science. Concrètement, nous axerons notre recherche, menée en vision par ordinateur, sur la modélisation et le rendu 3D de haute précision à partir d'images, et nous proposerons des applications des technologies développées dans le cadre de ce projet à la post-production cinématographique, aux effets spéciaux et à la conservation de l'héritage culturel, par le biais de collaborations avec des partenaires extérieurs.
Objectif Général
There is an increasing need for three-dimensional (3D) “content" in entertainment, engineering, and scientific applications. We predict that, for most of these, today's specialized 3D sensors will eventually be replaced by ordinary, consumer-grade digital cameras equipped with advanced image-based modeling and analysis software. We propose core computer vision and computer graphics research that will enable the development of this software and its application to real-world problems.
Porteur
CERTIS
Partenaires
Domaine
Image, Son, Interactivité
AAP
ANR
Objectif Sectoriel / Contenus Technologiques
Concretely, we will focus on high-fidelity image-based modeling and 3D shape and appearance matching. We will focus on the following three problems:
1) Core image-based modeling. We propose a novel approach to calibrated multi-view, wide-baseline stereopsis, as a core technology for image-based modeling.
2) Physical and statistical shape models. We propose to construct realistic physical and statistical models of object shape and appearance and to integrate them as priors into the image-based modeling and rendering process.
3) Advanced image-based rendering. We focus on the efficient and realistic rendering of image-based models integrated inside classical geometrical scenes, including lighting, shadowing and re-lighting effects.
Philosophie / Démarche / Enjeux
We believe that the emergence of consumer-grade digital cameras that match (and may soon exceed) the resolution and image quality of argentic ones is a tremendous opportunity to develop advanced image-based modeling and analysis software that will turn these cameras into flexible, affordable and accurate 3D sensors, and allow the widespread acquisition and use of 3D content in engineering and scientific applications. We propose computer vision and computer graphics research that will enable the development of this software.
