Mesh compression from geometry

Mesh compression from geometry
Thomas Lewiner

PhD Thesis (december 2005)
directed by Jean--Daniel Boissonnat


Images invaded most of contemporary publications and communications. This expansion has accelerated with the development of efficient schemes dedicated to image compression. Nowadays, the image creation process relies on multidimensional objects generated from computer aided design, physical simulations, data representation or optimization problem solutions. This variety of sources motivates the design of compression schemes adapted to specific class of models. This work introduces two compression schemes for geometrical models. The first one encodes level sets in any dimension, in a direct or progressive manner, with both state of the art compression ratios for low dimensions. The second one encodes meshes of any dimension and topology, even non--pure or non-manifold models, embedded in arbitrary space. The compression ratios for surfaces are comparable with famous mesh compression methods such as the Edgebreaker.


PDF paper (5.4 MB)
PPT presentation (42.5 MB)
Mesh compression from geometry


    author = {Thomas Lewiner},
    title = {Mesh compression from geometry},
    year = {2005},
    month = {december},
    school = {Géométrica Project, INRIA--Sophia Antipolis, delivered by Université Paris VI},
    note = {Advised by Jean--Daniel Boissonnat},
    url = {\url{}}

Last modifications on July 3rd, 2013