Text this: Saliency-based variational active contour model for image with intensity inhomogeneity.