报 告 人:Wei Zhu教授 (University of Alabama)
报告题目:Augmented Lagrangian method for an Euler's elastica based segmentation model that prefers convex contours
报告时间:5月18日下午4:00-5:30
报告地点:静远楼1506学术报告厅
Abstract: In this talk, we will consider a novel Euler's elastica based image segmentation modeland discuss a new augmented Lagrangian method for this model. The interesting feature of this model lies in its preference of convex segmentation contour.However, due to the high order and non-differential term, it is often very challenging to minimize the associated functional. In this work, we propose using augmented Lagrangian method to tackle this problem. We design a new Lagrangian functional that deals with the mean curvature term differently from those existing ones. The new treatment reduces the number of Lagrange multipliers, and more importantly, it helps represent the curvature more effectively and faithfully. Numerical experiments will be presented to validate the efficiency of the proposed augmented Lagrangian method and to demonstrate some new features of this new segmentation model, such as shape driven and data driven properties. (This is a joint work with Xue-Cheng Tai and Egil Bae.)
Wei Zhu教授个人简介:
08/2008— present Assosiate Professor
Department of Mathematics, University of Alabama
08/2004— 08/2008 Research Scientist/Courant Instructor
Courant Institute of Mathematical Sciences, NYU
06/2004 Ph.D, University of California,Los Angeles
07/1999 M.S, Peking University, P.R.China
07/1994 B.S, Tsinghua University, P.R.China