Image Filtering with a Fourth Order Geometry-Driven Flow

We present a image filtering algorithm via a new fourth order geometry-driven flow. In geometry framework for image processing, a gray image is regarded as a surface in R³. Spirited by this idea, we try to construct a new energy functional based on the second fundamental form of a surface. With this motivation, a new fourth order geometry-driven flow for image filtering is obtained. The new model can adjust diffusion speed according to local structure of the image. Numerical experiment results on real image data are given to demonstrate the performance of the proposed method.