Unsupervised Texture Segmentation Using Permutation Entropy and Grey-level Feature

A new method based on permutation entropy and grey level feature is provided in this paper. Permutation entropy is a new complexity measure for time series based on comparison of neighbouring values. The definition applies to describe the texture feature of image. The new complexity measure feature combines with the grey-scale mean and grey-scale deviation, construct multi-dimension feature vector. Then, apply the fuzzy c-means algorithm as the classifier to cluster the feature vectors, get the texture segmentation results. Experiments show that the method is particularly useful in the presence of dynamical or observational noise and the advantages of the method are its simplicity, extremely fast calculation, its robustness