A Texture-Based Energy for Active Contour Image Segmentation

This paper presents a two-dimensional deformable model-based image segmentation method that integrates texture feature analysis into the model evolution process. Typically, the deformable models use edge and intensity-based features as the influencing image forces. Incorporation of the image texture information can increase the methods effectiveness and application possibilities. The algorithm generates a set of texture feature maps and selects the features that are best suited for the currently segmented region. Then, it incorporates them into the image energies that control the deformation process. Currently, the method uses the Grey Level Co-occurrence Matrix (GLCM) texture features, calculated using hardware acceleration. The preliminary experimental results, compared with outcomes obtained using standard energies, show a clearly visible improvement of the segmentation on images with various texture patterns.