FCM algorithm besed on Normalized Mahalanobis distances in image clustering

Author

Yih, Jeng-Ming

Author_Institution

Dept. of Math. Educ., Nat. Taichung Univ., Taichung, Taiwan

Volume

5

fYear

2010

fDate

11-14 July 2010

Firstpage

2724

Lastpage

2729

Abstract

The popular fuzzy c-means algorithm (FCM) based on Euclidean distance function converges to a local minimum of the objective function, which can only be used to detect spherical structural clusters. Gustafson-Kessel(GK) clustering algorithm was developed to detect non-spherical structural clusters. However, GK clustering algorithm needs added constraint of fuzzy covariance matrix, In this paper, an improved Fuzzy C-Means algorithm based on a Normalized Mahalanobis distance (FCM-NM) by taking a new threshold value and a new convergent process is proposed The experimental results of two real data sets in image classification show that our proposed new algorithm has the better performance.

Keywords

covariance matrices; fuzzy set theory; image classification; pattern clustering; Euclidean distance function; FCM algorithm; FCM-NM; Gustafson-Kessel clustering algorithm; convergent process; fuzzy c-means algorithm; fuzzy covariance matrix; image classification; image clustering; normalized Mahalanobis distance; spherical structural cluster; Accuracy; Algorithm design and analysis; Classification algorithms; Clustering algorithms; Covariance matrix; Equations; Machine learning algorithms; FCM; FCM-NM algorithm; GK-algorithm;

fLanguage

English

Publisher

ieee

Conference_Titel

Machine Learning and Cybernetics (ICMLC), 2010 International Conference on

Conference_Location

Qingdao

Print_ISBN

978-1-4244-6526-2

Type

conf

DOI

10.1109/ICMLC.2010.5580475

Filename

5580475