基于马氏距离的FCM图像分割算法

Mahalanobis distance-based FCM image segmentation algorithm

基于模糊C均值聚类的图像分割是应用较为广泛的方法之一,但大多数模糊C均值聚类方法都是基于欧式距离,且存在运算时间过长等问题。提出了一种基于Mahalanobis距离的模糊C均值聚类图像分割算法。实验分析表明,提出的算法在保证分割质量的前提下,能较快提高分割速度。实验结果表明了该方法的有效性。

Fuzzy C-Means（FCM） clustering is one of well-known unsupervised clustering techniques,which has been widely used in automated image segmentation. However, most of fuzzy partition clustering algorithms are based on Euclidean distance function which can only be used to detect spherical structural clusters,and have disadvantages in runtime. This paper presents a Mahalanobis distance-based fuzzy C-means clustering image segmentation algorithm. Experiments show that the proposed method improves the segmentation runtime on the basis of segmentation qualities. Experimental results show that the proposed method is effective.