A deformable model based on level sets for image segmentation

This paper presents a new level set-based image segmentation method. First, a Gabor filter is used to suppress of noise in the extracted regions of interest and guide the motion of the evolving contour detection. Second, Green's theorem is used to develop a region-based energy function, combined with diffusion-based smoothing, to separate low contrast regions. Results are presented for it's application to a variety of real and synthetic images, particularly those exhibiting texture properties. The results indicate the method is more effective than traditional region-based level set methods in coping with intensity inhomogeneities, noisy and textured images.