CCD noise filtering based on surface diffusion model combined with min/max curvature

Abstract

We propose to combine the min/max curvature to the surface diffusion model. The resulting technique is an automatic, extremely robust, computationally efficient, and a straightforward scheme. The flow under min/max curvature was presented by Malladi and Sethian (see Graphic. Models Image Processing, vol.58, no.2, p.127-141, 1996) in a differential equation in terms of the image intensity. If it is combined to the surface diffusion model then the new scheme automatically picks the termination criteria; continued application of the scheme produces no further change. In other words it has a convergence state. It has another important advantage over the MCD (mean curvature diffusion) and other diffusion models; removing notch noise on the side of a structure does not diffuse the shape of the structure.