基于进化聚类的图像分割方法

Method for image segmentation based on evolving clustering

介绍了一个与模糊C均值FCM算法等效的图像颜色分割的方法。首先利用进化聚类对图像中的像素依据其RGB的值进行进化聚类划分,对划分后的各个类的类中心用遗传算法进行优化,然后再对图像中像素进行归类划分,使其满足各类中元素具有较高的相似度,而不同类中的元素相似度差别较大的目标,并与FCM算法进行了实验对比,结果表明经人工评价该算法与模糊C均值FCM算法等效。

A new method that is equivalent to C-means clustering in image color clustering segmentation is introduced. Firstly, pixels ofthe image are segmented based on the value of RGB by evolving clustering. Secondly the center of clustering that had been segmented is optimized by genetic algorithm. Finally, pixels of the image are segmented again, which making pixels of the same cluster high similarity rather than the different cluster. The results of the two methods are equivalent in image color clustering segmentation, which proved by experiment.