Regularized Level Set for Inhomogeneity Segmentation
Guoqi Liu, Haifeng Li · Frontiers in artificial intelligence and applications · 2016
Active contour model based on level set method is a popular method for image segmentation. However, intensity inhomogeneity universally exists in images and it greatly influences image segmentation. Local binary fitting model (LBF) is a effective method to cope with inhomogeneous intensity. However, the energy function of LBF is non-convex and it costs much computational cost. Otherwise, LBF could not preserve the weak edges. The non-convexity always causes the contour suffer from local minimum, and the computational cost is large. In order to cope with these shortcomings, we introduce a regularized minimization for improved LBF model. In proposed model, the edge information is integrated into the energy functional. The energy functional of improved LBF model is convex, and the local minimum is avoid. Furthermore, some fast optimal method can be utilized. In this paper, the regularized method is utilized to make the contour converge to minimization. Experimental results confirm that proposed method attains a similar segmentation effect with the LBF but costs less computation times.