Fuzzy texture unsupervised classification approach using conditional local variance model

This paper presents a fuzzy-based unsupervised segmentation of textured images. We treated texture features and color features separately. Fuzzy-based homogeneity decision is used to measure the similarity of regions by making a soft fusion of texture features and color features. Texture model, the Multiplicative Autoregressive Random Field with Conditional variance (MARC), uses a neighborhood-dependent conditional variance.