计算机应用研究2012，Vol.29Issue(3)：1127-1129,1145,4.DOI:10.3969/j.issn.1001-3695.2012.03.089
自动分割图像新方法研究
Research of new automatic image segmentation method
摘要
Abstract
This paper applied nonparametric Dirichlet process mixtures model algorithm to segment images automatically, which could obtain the segmentation class numbers automatically during the process without initialization. The number of clusters as an input constant was substituted by a random variable with a control parameter, which specified a level of cluster resolution by adjusting the parameter. The algorithm was used to segment noisy natural images and magnetic resonance images with biasing field. The experiment results show that the algorithm has anti-noise performance and can restrain the biasing field effect of MR images well. The Dice similarity coefficients are all above 90% , which show that the proposed method is robust and accurate.关键词
图像分割/非参数/Dirichlet过程混合/聚类Key words
image segmentation/ nonparametric/ Dirichlet process mixtures/ clustering分类
信息技术与安全科学引用本文复制引用
卢易苏,陈武凡..自动分割图像新方法研究[J].计算机应用研究,2012,29(3):1127-1129,1145,4.基金项目
国家"973"重点基础研究发展计划资助项目(2010CB732505) （2010CB732505）
国家自然科学基金重点资助项目(30730036) （30730036）
