講題：Shell Clustering: An Introduction and Recent Developments
演講人：王才沛 教授（交通大學 資訊工程學系）
時間：2007年9月20日(星期四) 下午02：00 ∼ 04：00
地點：推廣教育大樓3樓9329多媒體電腦教室
茶會：下午01：40資科所辦公室（9213）
Abstract
Shell clustering represents a set of prototype-based clustering algorithms that use shell-like prototypes, instead of points, for the detection of particular shapes in the data. The main advantage of such algorithms is for efficient detection of object of particular shapes in 2-D or 3-D data sets without exhaustive search. This property is particular useful for image analysis and computer vision applications. I will describe both earlier examples of shell clustering algorithms geared toward particular geometrical shapes, followed by our more recent study on applying shell clustering to the detection of more generic shapes.