A multiphase level set formulation for image segmentation using a MRF-based nonsymmetric Student’s-t mixture model

This paper proposes a new level set energy function framework in which the Markov random field-based nonsymmetric Student’s-t mixture model (SMM) is incorporated for labelling static images. This framework provides a general strategy by taking the best components of the Bayesian theory and level set technique. Therefore, the proposed segmentation method can bring the topology shape constraints to a statistical finite mixture model. An advantage of this method is that it can overcome the weakness of the conventional level set formulation by filtering out the outliers and stopping at the boundary points. Another feature is that the local relationship among neighbouring pixels is introduced into the prior probability so that the proposed framework is more robust against noise. The method is mainly implemented by modelling the probability density function of the observed data using nonsymmetric SMM. The proposed model has a simplified structure, which effectively reduces the computational complexity. Finally, numerical experiments on various synthetic, real-world images are conducted.