Cell average image transform algorithms with exact error control

摘要

In this paper we present non-separable two-dimensional multi-resolution algorithms based on Harten's cell average framework for multi-resolution which guarantee a priori prescribed quality in the reconstructed image, hence being suitable for applications where quality control is most important, yet performing economically in terms of storage and speed of computation. Moreover, after applying the algorithm the exact error between the original and the decoded images measured in the L (2) discrete norm is known without being necessary to decode the image.