Variational Image Segmentation with Constraints

Abstract

The research of Huizhu Pan addresses the problem of image segmentation with constraints though designing and solving various variational models. A novel constraint term is designed for the use of landmarks in image segmentation. Two region-based segmentation models were proposed where the segmentation contour passes through landmark points. A more stable and memory efficient solution to the self-repelling snakes model, a variational model with the topology preservation constraint, was also designed.