CAJ | 학술논문

针对模糊C均值聚类分割算法无法获得复杂图像的细节信息问题，提出一种相对熵模糊 C均值聚类分割算法。该算法利用划分隶属度构造相对熵，对传统模糊C均值聚类进行正则化约束，将其作为正则化因子添加到传统 FCM的目标函数，得到新的聚类目标函数。通过拉格朗日乘子最优化推导，得到新的隶属度和聚类中心的迭代更新表达式。实验结果表明，该算法对于图像与背景灰度相近的复杂图像可以清晰的分割出图像的轮廓，也比FCM方法获得更多的图像细节信息。
침대모호C균치취류분할산법무법획득복잡도상적세절신식문제，제출일충상대적모호 C균치취류분할산법。해산법이용화분대속도구조상대적，대전통모호C균치취류진행정칙화약속，장기작위정칙화인자첨가도전통 FCM적목표함수，득도신적취류목표함수。통과랍격랑일승자최우화추도，득도신적대속도화취류중심적질대경신표체식。실험결과표명，해산법대우도상여배경회도상근적복잡도상가이청석적분할출도상적륜곽，야비FCM방법획득경다적도상세절신식。
A segmentation method based on relative entropy fuzzy C-means clustering is proposed in this paper to tackle at the problem that the traditional fuzzy C-means clustering (FCM) segmentation method cannot get the details of a complex image.By using the membership to get the relative entropy,the traditional fuzzy C-means clustering is made to have regularization constraint,and then it is added as a positive function to the traditional FCM obj ective function to get a new obj ective function.By using optimal derivation of Lagrange multiplier,a new expression of memberships and cluster centers is obtained.Experimental results show that the algorithm is more suitable for complex image to acquire more details.