Clustering Algorithm of Similarity Segmentation based on Point Sorting

We propose a clustering algorithm of similarity segmentation based on point sorting to improve the clustering performance. Taking full advantage of segmentation sorting of the clustering algorithm based on minimum spanning tree, the algorithm use a variety of methods for different situations to sort these cluster elements with their similarity and segment them where there are large changes in their similarity to obtain cluster results. In order to compare the performance of the method, we select some traditional cluster analysis methods like k-means, hierarchical clustering and density clustering with noise data, etc. In the experimental testing, we select three sets of two-dimensional artificial data sets and four sets of real data sets as test data. And three evaluation indexes are applied to measure the quality of clustering. The simulation results in test data show that this algorithm can improve the accuracy of the algorithm effectively and achieved good clustering performance.