A Robust Level Set Image Segmentation Model Driven by Markov Random Field and Dual Regularization

This paper presents a robust level set model for image segmentation. In order to enhance the robustness of the proposed model against noise interference, we construct a special energy term based on Markov random field (MRF) theory which has strong local neighborhood modeling ability and embed it into the commonly used level set segmentation framework. In its numerical implementation, two efficient regularization strategies are introduced to ensure the stability of the level set evolution process, one is the adaptive regularization constraint that is directly applied to the zero level curve, its purpose is to reduce the influence of noise and ensure that the active contour does not pass through the weak object boundaries; the other is the reaction diffusion-based level set regularization constraint which is imposed on the whole level set function, under its constraints, the time-consuming level set re-initialization step that is widely used in the traditional level set methods (LSMs) is completely removed. The extensive and promising experimental results on wide variety of images demonstrate the excellent performance of the proposed method.