Alpha-divergence maximization for statistical region based active contour segmentation with non-parametric PDF estimations
Abstract
In this article, a complete original framework for non supervised statistical region based active contour segmentation is proposed. More precisely, the method is based on the maximization of alphadivergences between non paramterically estimated probability density functions (PDF) of the inner and outer regions defined by the evolving curve. In this paper, we define the variational context associated to distance maximization in the particular case of alphadivergence and we also provide the complete derivation of the partial differential equation leading the segmentation. Results on synthetic data (corrupted with a high level of Gaussian and Poisonian noises) but also on clinical images (X-ray images) show that the proposed non supervised approach improves classical approach of that kind.