摘要
如何确定分类数是模式识别和图像分割中常要面对的难题。文献犤1犦提出了一种基于阴阳机制确定最优分类数的模式聚类方法。该文在文献犤1犦的基础上,导出了一个新的用于评价灰度图像多阈值分割的评价标准,并结合遗传算法,提出了一种用于灰度图像的多阈值自动分割方案。试验表明,该法具有较强的鲁棒性、适用性、非监督性。
How to determine the number of clusters is an open problem that is faced by pattern recognition and image segmentation.The Bayesian YING-YANG learning theory and system proposed byprovides people the model selection criteria for selecting the number of clusters in the clustering analysis.Based onand the genetic algorithm,this paper proposes a novel scheme for gray image segmentation.Experimental results show the effectiveness of this method.
出处
《计算机工程与应用》
CSCD
北大核心
2003年第23期79-80,83,共3页
Computer Engineering and Applications
基金
部委科技预研基金
