Transition Region Extraction and Segmentation Based on Image Fuzzy Entropy Neighborhood Unhomogeneity

A transition region extraction and segmentation method based on image fuzzy entropy neighborhood unhomogeneity(NU-TRES) is presented,which is characterized by its speed,ability to deal with noise and robustness.Obtaining a transition region located between the object and background,an optimal segmentation threshold is attained based on the transition region histogram.The membership degree which describes fuzzy division,fuzzy entropy and neighborhood unhomogeneity measurement are created.NU-TRES is not so sensitive to noise as the conventional transition region extraction methods,depending no more on Llow and Lhigh.Analysis and experimental results show that this new algorithm(NU-TRES) significantly outperforms the existing transition region extraction and segmentation methods.