Abstract: A new phase-field model for segmenting images with intensity inhomogeneity and noise is proposed based on the local binary fitting (LBF) method. In our approach, an artificially constructed cubic function of the phase variable is creatively shown to represent the two different regions of images. We use the stabilized method to solve the Allen-Cahn type equation and prove the scheme having the monotonic-decaying and maximum-bound-principle properties. Ample numerical cases are offered to illustrate the effectiveness of our algorithm.
