Image Clustering Using Intuitionistic Fuzzy C Means with Partial Supervision

In this paper, we propose a new image clusteringalgorithm, referred to as Intuitionistic Fuzzy C Means withPartial Supervision (IFCMPS). In the proposed method someprior knowledge as the class labels of some data objects in theintuitionistic fuzzy set theory are used. In the intuitionistic fuzzyC means setting an extra uncertainty parameter which is thehesitation degree that arises while defining the membershipfunction is applied that leads the cluster centers converge to adesirable location than the cluster centers obtained using justfuzzy or possibilistic C means algorithm. Also, in the proposedobjective function that considers the cluster scatter and hesitationentropy of data objects another component for supervisedinformation is proposed. Simulations on real life image databasesprove that the proposed method significantly outperformsstandard fuzzy C means as well as intuitionistic fuzzy C meansclustering methods.

Keywords:

image clustering , intuitionistic fuzzy sets , partial supervision , intuitionistic fuzzy-c mean