Clustering Hierarchical Fuzzy Colors and its Application on Image segmentation

The purpose of this research is to analyze the lowlevel features of images as a representation of semantic concept. This paper proposes a flexible approach for clustering features of images and mapping the low-level image features to the highlevel concept recognition. Owing to the uncertainty of image features for human’s recognition, we use the approach of fuzzy colors clustering to analyze image features based on fuzzy entropy. The proposed approach first analyzes color features using HLS color space to determine the best number of fuzzy colors and cluster the colors into a hierarchical concept for human’s recognition. We also apply the clustered fuzzy colors to segment out meaningful regions from an image automatically. In this paper, the application of fuzzy colors on segmentation is demonstrated and compared with other methods. The experimental results show that the proposed method can extract meaningful regions from images as effectively as human visual perception.