Adaptive Weight Texture Segmentation Based on Criterion of Separability
基于可分离性判据的自适应加权纹理图像分割

In this paper, we propose an adaptive weighted method based on feature separability. We use co-occurrence matrix to extract features , then select better features according to criterion of separable, and give different feature adaptive weight to let them more efficient in clustering. Finally, experimental results show that the segmentation has less faults than before.