Summary: Transductive Segmentation of Textured Meshes
Anne-Laure Chauve, Jean-Philippe Pons, Jean-Yves Audibert, and
IMAGINE, ENPC/CSTB/LIGM, Universit´e Paris-Est, France
Abstract. This paper addresses the problem of segmenting a textured
mesh into objects or object classes, consistently with user-supplied seeds.
We view this task as transductive learning and use the flexibility of
kernel-based weights to incorporate a various number of diverse features.
Our method combines a Laplacian graph regularizer that enforces spa-
tial coherence in label propagation and an SVM classifier that ensures
dissemination of the seeds characteristics. Our interactive framework al-
lows to easily specify classes seeds with sketches drawn on the mesh and
potentially refine the segmentation. We obtain qualitatively good seg-
mentations on several architectural scenes and show the applicability of
our method to outliers removing.
The generalization of digital cameras, the increase in computational power
brought by graphical processors and the recent progress in multi-view recon-
struction algorithms allow to create numerous and costless textured 3D models
from digital photographs. In this work, we address the problem of segmenting