An Empirical Study on Fuzzy Image Clustering with Various Clustering Validity Indexes
Chih‐Hung Wu, Liwen Chen, Liwei Lu · 2012
An important issue in clustering analysis is to determine the number of clusters, which is usually approved by domain experts or evaluated by clustering validity indexes. This paper presents a new clustering validity index, WLI, that considers the median effects of image clustering using the fuzzy c-means (FCM) algorithm. the performance of WLI is compared with existing indexes including PC, PE, CHI, DBI, XBI, FSI, SCI, CSI, PCAES, and PBMF. Six images from various application domains, including synthetic, remote sensing, and CT-scan images, are tested and the results are analyzed and presented. the experimental results show that WLI has better performance on FCM-based image segmentation.