Skip to Main content Skip to Navigation
Conference papers

Study of Volumetric Methods for Face Reconstruction

Gang Zeng Sylvain Paris 1 Maxime Lhuillier 2 Long Quan 3
1 ARTIS - Acquisition, representation and transformations for image synthesis
GRAVIR - IMAG - Laboratoire d'informatique GRAphique, VIsion et Robotique de Grenoble, Inria Grenoble - Rhône-Alpes, CNRS - Centre National de la Recherche Scientifique : FR71
Abstract : This paper presents an early study about a reconstruction method that extends previous space carving methods to handle all characteristics of human faces: complex non-Lambertian materials, untextured areas, highly detailed geometry, etc. This face study is seen as a preliminary step to a more general framework for non-Lambertian reconstruction. We therefore avoid specific techniques like parameterized face models. Since we expose our early studies, we mainly present the related existing work and discuss the pros and cons of each approach. We aim at discerning the strength and weaknesses of the classical tools in order to adapt and improve them to handle non-Lambertian materials while overcoming their limitations. We show our first results which are promising and validate our global approach. Throughout the discussion, we raise several questions that identify complex issues related to our goal. We provide some hints that pave the way for a better understanding of the global problem. We are confident in that further studies will lead to significant improvements over existing methods.
Document type :
Conference papers
Complete list of metadatas
Contributor : Team Evasion <>
Submitted on : Thursday, October 14, 2010 - 10:58:09 AM
Last modification on : Monday, December 28, 2020 - 3:44:01 PM
Long-term archiving on: : Saturday, January 15, 2011 - 2:29:14 AM


Files produced by the author(s)


  • HAL Id : inria-00510186, version 1


Gang Zeng, Sylvain Paris, Maxime Lhuillier, Long Quan. Study of Volumetric Methods for Face Reconstruction. Proceedings of IEEE Intelligent Automation Conference, 2003, Hong Kong, China. ⟨inria-00510186⟩



Record views


Files downloads