Segmentation by a Local and Global Fuzzy Gaussian Distribution Energy Minimization of an Active Contour Model

We propose a novel region-based active contour model, which incorporates the image local and global information into a fuzzy energy function, providing a robust and accurate segmentation while accounting for intensity inhomogeneity. In this model, image intensities are assumed to have a Gaussian distribution with different means and variances. While the local information contributes to dealing with intensity inhomogeneity, the global information allows improving the performance of the model in the case of very noisy or blurred images. Another interesting property is that the energy function of the proposed model is convex with respect to the variable used to determine the contour. This makes the accuracy of the result invariant with respect to the position of the initial contour and more suitable for an automatic segmentation. Moreover, the energy function of the proposed model is minimized in a computationally efficient way by calculating the fuzzy energy alterations directly. Experiments are carried out to validate the capabilities of the proposed model. Comparisons are provided with ground truth and other methods in the field to underline the superiority of our method in terms of accuracy.