An Improved Fast Global k-means Clustering Segmentation Algorithm

A new improved fast global k-means clustering segmentation is presented to solve the problem that k-means clustering algorithm for clustering results easily affected by noise and clustering centers randomly selected.the proposed method is divided into a series of sub-clustering problem to solve the problem that the k-means clustering original centers randomly selected clustering lead to an incorrect results.Sorting through the median,this method alleviate the problem easily affected by noise.Experimental results show that the proposed method improve the accuracy of the clustering segmentation.