Robust Fuzzy Active Contour Model for Mixed Noise Image Segmentation

In this paper, a novel region based active contour model is proposed to segment images with mixed noise. The energy function of the proposed model is composed of global fuzzy energy term and local fuzzy energy term. The global term is constructed by using grayscale statistical characteristics of the global region, mainly used to improve evolution efficiency of the model and to avoid unsatisfactory results caused by inappropriate initial position. In order to improve model robustness to mixed noise, using local spatial and grayscale information to establish local term, which can adaptively adjust the influence of the neighborhood on the membership of the central pixel according to the intensity of the mixed noise in the local region. In addition, this model provides a fast convergence to the desired object edge by directly calculating the change of fuzzy energy in place of solving Euler–Lagrange equations corresponding to the energy function. Experiments are carried out with synthetic data sets and real data sets. The experimental results show that the proposed model has better performance than the currently commonly used active contour model.