Application of inhomogeneous diffusion to image and video coding

We describe the properties of algorithms based on methods of inhomogeneous diffusion and their application to image and video coding. Filtering by inhomogeneous diffusion has the desirable property of reducing noise and other artifacts while preserving the image structure, maintaining or improving image quality. As a preprocessor, this filtering removes non-perceptible noise, leading to improved compression. As a postprocessor, it suppresses coding artifacts, such as the block impairments of DCT-based coders. Maintaining the basic structure of JPEG, MPEG, and other coders, we describe the improvements possible with inhomogeneous diffusion preprocessing and postprocessing.