The Application of Mean Shift in Image Segmentation

ZHANG Ke,SUN Xiaowei,LI Yanjun,LIU Yan

DOI: https://doi.org/10.3969/j.issn.1673-9728.2006.01.145

2006-01-01

Abstract:A practical approach to image segmentation based statistics is presented in this paper.It adopts kernel function to analysis digital image that is called nonparametric cluster analysis.Firstly,a certain quantity of points is selected from the image randomly.Then the points are clustered by mean shift procedure.Finally,the regions,which have the same characteristic,are merged according to a certain principle.The experimental results show that mean shift proce- dure is an effective method to image segmentation.