Compressed Image Quality Evaluation using Power Law Models

This paper presents a compressed image quality measure based on the properties of Zipf law, which is a power-law model adapted from linguistic analysis. It describes the frequency distribution of image patterns, and can be used to put into evidence image details which are affected by compression. Two image quality evaluation measures based on Zipf law have been proposed and tested, one designed to be used on JPEG images and the other on JPEG2000 images. These measures have also been used with the tabu search algorithm for automatic compression optimization