A new approach for unsupervised segmentation

Texture is an important spatial feature useful for identifying objects or regions of interest in an image. This paper presents two fuzzy logic-based approaches for colour textures. They are fuzzy texture unit and colour fuzzy texture spectrum. A method is proposed for segmentation using colour fuzzy texture descriptor for colour texture images. This paper discusses two steps. Firstly, the presence of colour texture is locally as well as globally represented. Secondly, with the proposed global descriptor, segmentation operation is performed by SOMs algorithm. Compared with moment features, our method achieves good segmentation results on colour texture images, demonstrating the performance of our proposal.