Parametric probability models for lossless coding of natural images

Parametric modelling based entropy coding is an alternative technique to the non-parametric, contex-based techniques that are widely used in current image coding algorithms. Compared to the non-par ametric techniques, the proposed parametric technique is more tractable mathematically and is easier to understand and manuplate. In this paper, we examine a number of issues related with the application of this technique to image coding. We discuss the strong and weak points of several probability distributions that can be used in our algorithm. We then present a working framework that can perform close to the state-of-the-art lossless image compression techniques.