Object boundary location by region and contour deformation

Many applications in image analysis need to distinguish an object from its surroundings. The `snake' method (Kass, 1987) is a well known method to solve such problems by locating the object boundary from an initial plan, but it has the limitation that the distance between the initial plan and the object boundary should be `very small'. The authors present a two-step method, which combines region and contour deformation, to locate the boundary of an object from a designated initial boundary plan. First, they propose a new deformable region model to represent an object, and use the model to locate the boundary of the object by region deformation. This step fills the gap between a fairly rough initial plan and the `snake' method. The resulting boundary is then used as the initial plan of a `snake' method to further refine the boundary. The results of the method are shown in simulated images and MRI images of brain tumour patients