一种基于分水岭变换和模糊C均值聚类的彩色图像分割算法

针对分水岭算法对在图像分割中容易产生过分割,提出了一种基于分水岭变换和模糊C均值(FCM)聚类算法的彩色图像分割算法。该算法先对图像进行分水岭分割,再对分水岭产生的过分割进行聚类合并。在合并过程中采用区间差异度和区域面积来确定模糊C均值聚类个数。该算法的优点是解决了分水岭变换算法的过分割问题的同时解决了模糊C均值聚类算法的初始值以及聚类中心难以确定的问题。实验结果表明,该算法可以准确地分割出目标并应用到自动分割系统中。

Based on Watershed Transform and Fuzzy C Means Clustering Algorithm for Color Image Segmentation

For the watershed algorithm is very sensitive to weak edges and noise, thus in the watershed algorithm is applied to image segmentation prone to over-segmentation, In this paper, a new color image segmentation algorithm basing on watershed transformation and a fuzzy C mean (FCM) clustering algorithm is proposed. In this algorithm, watershed segmentation is carried on image, and then clusters and mergers the over-segmentation generated by the watershed, during the merging process using the degree of sector differences and the region area to determine the cluster number of FCM. The advantage of this algorithm is to solve over-segmentation problems caused by watershed transformation algorithm and meanwhile solve problems which are difficult to determine the initial value and cluster centers of FCM algorithm. Experimental results show that the algorithm can segment the target accurately and apply to automatic segmentation system.