Unsupervised segmentation of textured color images using fuzzy homogeneity decision

This paper proposes a fuzzy-based unsupervised segmentation of textured color images. L*a*b* color space is used to represent color features and statistical geometrical features (SGF) are adopted as texture descriptors. Homogeneity decision is used to make a fusion of texture features and color features with fuzzy-rule theory. Hierarchical segmentation based on the fuzzy homogeneity decision is performed in four processes: hierarchical splitting, local agglomerative merging, global agglomerative merging and pixelwise classification. Experiments on segmentation of some color texture mosaics and color natural images are presented to verify the effectiveness of the proposed segmentation approach.