Automatic Texture Segmentation Based on k-means Clustering and Co-occurrence Features

This work presents a method for automatic texture segmentation based on k-means clustering technique and cooccurrence texture features. A set of features was extracted from 256 gray-level co-occurrence information. These features were used to segment image regions regarding the textural homogeneity of its areas. As the process of calculating co-occurrence information demands the majority of computational time required, we propose a new methodology based on a gray-level co-occurrence indexed list (GLCIL) for fast element access, highly optimizing this step in the algorithm. Besides that, we compare the efficiency of the proposed method against other well known algorithms. The experiments show that GLCIL is the most efficient method in terms of computational time. Additionally, traditional Brodatz textures and other literature examples were tested to evaluate the appropriateness and robustness of the method.