CAJ | 학술논문

为改进传统的模糊C均值聚类（FCM）算法应用于图像分割时计算代价大、性能依赖于初始聚类个数和聚类中心、分割过程中易陷入局部极值的问题，提出一种基于均值漂移和模糊C均值聚类的图像分割算法。首先，利用优化的均值漂移算法对原始图像进行分割，分割后形成带权的分割图像并且得到聚类数目和聚类中心；然后，以带权分割图像为输入图像同时把聚类数和聚类中心引入加权FCM算法进行分割；最后，对分割结果进行形态学优化和二值化处理以提升分割效果。实验表明，该方法相对于传统的模糊C均值聚类算法有更好的图像分割效果和分割效率，且分割效果与人类视觉感知具有更高的一致性。
위개진전통적모호C균치취류（FCM）산법응용우도상분할시계산대개대、성능의뢰우초시취류개수화취류중심、분할과정중역함입국부겁치적문제，제출일충기우균치표이화모호C균치취류적도상분할산법。수선，이용우화적균치표이산법대원시도상진행분할，분할후형성대권적분할도상병차득도취류수목화취류중심；연후，이대권분할도상위수입도상동시파취류수화취류중심인입가권FCM산법진행분할；최후，대분할결과진행형태학우화화이치화처리이제승분할효과。실험표명，해방법상대우전통적모호C균치취류산법유경호적도상분할효과화분할효솔，차분할효과여인류시각감지구유경고적일치성。
To improve the problem of traditional fuzzy c-means clustering (FCM) algorithm that when applied to image segmentation , it has big computational cost , its performance depends on the initial clustering number and clustering centre , and it is easy to fall into local ex-tremum in segmentation process , an image segmentation algorithm based on mean shift and fuzzy c-means clustering is proposed .First, the algorithm uses the optimised mean shift algorithm to segment the original image , after the segmentation there forms the image with the right , and the clustering number and clustering centre are obtained as well .Then, the algorithm chooses the image with the right as the input image , and introduces the clustering number and clustering centre into the weighted FCM algorithm for segmentation .Finally, the algorithm applies morphologic optimisation and binarisation to the segmentation result to improve the segmentation effect .Experimental results show that , com-pared with traditional fuzzy c-means clustering method , the proposed algorithm has better segmentation effect and efficiency , and the segmen-tation effect has a higher consistency with human visual perception .