题  目: Variational 3D Mesh Segmentation

报告人: Juyong Zhang （张举勇）       Swiss Federal Institute of Technology in Lausanne

时  间：3月8日，周四，下午3:30－5:00

地  点：东区管研楼1218

Abstract: 3D Mesh segmentation refers to the process of extracting subparts or components from an existing mesh or decomposing a mesh model into multiple meaningful components. Mesh segmentation is of great practical importance in many mesh related processing and applications. The challenge with 3D mesh segmentation lies in the fact that the segmentation algorithm is expected to extract meaningful components or decompose a mesh into meaningful parts which are consistent with user intention, geometric mesh attributes and human shape perception. In this talk, I will present our work on how to use variational methods to formulate and solve interactive/automatic mesh segmentation problem. Flexibility and robustness are quite important for interactive mesh segmentation problem. However, they are heavily overlooked in the existing interactive methods. We put forward to use our proposed constrained random walks model and fast geodesic curvature flow model to achieve these requirements. To solve automatic mesh decomposition problem, we propose to extend Mumford-Shah model from 2D image to 3D mesh. It simultaneously handles segmentation and boundary smoothing, which are usually two separate processes in most previous work, and its efficiency is achieved by a new fast primal-dual numerical solver on mesh. Extensive experimental results demonstrate that our methods outperform existing same type algorithms.

Bio: Juyong Zhang received the B.S. degree in computer science and engineering from the University of Science and Technology of China, and the Ph.D. degree in computer engineering from NTU (Nanyang Technological University), Singapore, in 2011. From June 2010 to July 2010, he was a visiting student in Massachusetts Institute of Technology. From 2011, he worked as an exchanging Ph.D student and is a Postdoctoral Research Fellow in computer science of EPFL (Swiss Federal Institute of Technology in Lausanne), Lausanne, Switzerland. His research interests include computer graphics, geometry processing, image processing, architecture geometry, as well as numerical PDE. He served on the program committees of ICME 2012.

主办单位：365英国上市官网    国家数学与交叉科学中心合肥分中心

欢迎感兴趣的师生参加！