In this paper, a new and robust approach to mesh segmentation is presented. There are various algorithms which deliver satisfying results on clean 3D models. However, many reverse-engineering applications in computer vision such as 3D reconstruction produce extremely noisy or even incomplete data. The presented segmentation algorithm copes with this challenge by a robust semi-global clustering scheme and a cost-function that is based on a probabilistic model. Vision based reconstruction methods are able to generate colored meshes and it is shown, how the vertex color can be used as a supportive feature. A probabilistic framework allows the algorithm to be easily extended by other user defined features. The segmentation scheme is a local iterative optimization embedded in a hierarchical clustering technique. The presented method has been successfully tested on various real world examples. © 2011 Springer-Verlag.
CITATION STYLE
Wekel, T., & Hellwich, O. (2011). A robust approach to multi-feature based mesh segmentation using adaptive density estimation. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 6854 LNCS, pp. 244–252). https://doi.org/10.1007/978-3-642-23672-3_30
Mendeley helps you to discover research relevant for your work.