一种新的基于分水岭变换的聚类分析算法

New clustering algorithm based on watershed transform

提出了一种基于分水岭变换的聚类分析算法。该算法首先根据数据的密度信息把二维数据样本集转化成数字灰度图像,然后对该灰度图像进行分水岭变换,根据变换结果得到聚类结果。实验结果表明,该算法在准确度方面与传统的K-means算法相当,是一种完全无监督的聚类算法。

Based on watershed transform, a new clustering analysis algorithm was proposed： In this algorithm, the 2- dimension data set was converted to digital grey image according to its density at first. And watershed transform was performed on the grey image, and then the clustering results were got. The result indicates that the accuracy is slightly worse than that of K-means algorithm, but it has the capability of recognizing the number of clusters automatically.