时间：7月11日（星期五）上午10：00
地点：浙江大学紫金港校区蒙民伟楼402房间
报告人：鞠立力　教授
主持人：刘新国　教授
报告题目：Centroidal Voronoi Tessellation Based Methods for Image Segmentation
Abstract: Centroidal Voronoi tesssellations (CVTs) are special Voronoi tessellations whose generators are also the centers of mass (centroids) of the Voronoi regions with respect to a given density function and CVT-based methodologies have been proven to be very useful in many diverse applications in science and engineering. In this talk we will review recent advances of image segmentation methods based on  CVT and its variants. In the context of image processing, the classic CVT model reduces to the well-known k-means algorithm that minimizes the classic clustering (similarity) energy. This model lacks a regularization term and thus often produces too many details. The edge-weighted CVT (EWCVT) model beautifully combines the global color/intensity information and the physical information by adding a edge-based regularity term in the classic CVT model. It works very well in general but sometimes has difficulties in segmenting images with strong background inhomogeneity. The local variation and edge-weighted CVT (LVEWCVT) model improves the EWCVT model on this issue by replacing the classic clustering energy by the local variation energy. We will illustrate these interesting segmentation methods through extensive examples.
Bio：鞠立力，理学博士，教授。主要从事数值分析及科学计算方面的研究工作，已主持由美国国家科学基金会（NSF）, 美国能源部（DOE）等联邦机构资助的多项科研项目。目前担任数值分析领域国际顶级学术期刊（SIAM Journal on Numerical Analysis）的编委。
1995年毕业于武汉大学数学系数学专业，获理学学士。
1998年于中国科学院计算数学及科学工程计算研究所获得理学硕士学位
2002年于美国爱荷华州立大学(Iowa State University)获理学博士学位。
2002年-2004年明尼苏达大学(University of Minnesota)数学与应用研究所博士后。
2004年进入美国南卡莱罗纳大学(University of South Carolina) 工作, 历任数学系助理教授，副教授，及正教授。
2012年起兼任北京计算科学研究中心应用数学实验室客座教授。