针对FCM彩色图像分割算法需要给定聚类数且受初始值影响较大的缺点,提出一种自适应FCM彩色图像分割算法.该方法首先引入主分量变换,通过直方图分析,在RGB彩色空间上自动确定聚类数目,其次引入2D直方图,分析对角线投影1D直方图,并采用最大最小距离法选取初始聚类中心.利用该算法对各种彩色图像进行分割实验,实验结果表明:该算法能够有效且自适应地分割彩色图像.
In view of the FCM, need to a given number of clustering color image segmentation algorithm and greatly influenced by the initial value of faults,puts forward an adaptive color image segmentation FCM algorithm. The method first introduced by princi- pal component transform, histogram analysis,automatically determine the number of clustering in the RGB color space,then the in- troduction of 2D histogram analysis of diagonal projection 1 D histogram, and using the maximum and mininmm distance method to select the initial cluster center. Using the algorithm to a variety of color image segmentation experiments, experimental results show that : the algorithm can effectively and adaptive segmentation of color images.