Continuous Level Set Algorithm Based on Improved CV Model for Magnetic Resonance Breast Image Segmentation
Wang Fangme · Xi'an Jiaotong Daxue xuebao · 2014
A continuous level set algorithm based on improved Chan-Vese(CV)model is proposed to segment magnetic resonance(MR)breast images with weak edges,large information,and low signal-to-noise ratio.B-spline basis functions are used to express the traditional discrete level set function as a continuous level set function,and the re-initialization of the level set function is replaced by the solution of variational problem in B-spline space.Anα-CV model is constructed to form energy function by introducing a shifted Heaviside function. Experimental results and comparison with the traditional CV model method show that the proposed algorithm is able to avoid local minimum,to improve segmentation accuracy,and to suppress noise effectively with a reduction in the number of algorithm iterations by magnitude of 101.