Fuzzy C-Means Clustering Based Principal Component Averaging Fusion

Image fusion is a method of imparting all relevant and complementary image details into a single composite image extracted from images of same source or various sources. This paper proposes a fusion method based on segmented regions of source images which are obtained by a fuzzy C-Means clustering algorithm. Robust clustering is exhibited by Fuzzy C-means algorithm by assigning fuzzy membership values for given data towards various clusters. Principal components are evaluated for the clustered regions of source images and average of all the principal components is evaluated to get fused result as a linear combination of input images. Proposed fusion algorithm is experimented on the source images of same modality and the superiority of this fusion method is demonstrated over other methods by average quality index, mean structural similarity index and average mutual information.