Integration of Fractal and Grey-Level Features for Texture Segmentation

Fractal analyses have been recently applied successfully for texture segmentation. As for distinct multi-textured image with similar fractal dimensions (FD), efficient feature extraction based on FD is one of recent concerns. The novelty of our proposed method for texture segmentation lies in the integrating statistical features derived from directional FD and gray-level statistics in a given preset window. The proposed approach consists of three steps: firstly the fractal values at eight different directions are calculated; the statistical features on the directional FD are then combined with gray-level statistics. Finally we apply the fuzzy c-means algorithm as the classifier based on formed feature vectors to achieve good segmentation performance. The experiments taken from both textile and medical images show significantly effective of the proposed scheme.