Effective visual masking techniques in JPEG200

This paper introduces a very low complexity visual masking algorithm for the JPEG2000 image compression standard and evaluates its impact on the visual image quality by means of the multi-scale SSIM index. The algorithm derives suitable weighting masks indirectly from a statistical model of the wavelet data which is defined from the second moment and the average absolute amplitude of the data. If combined with an a priori rate allocation algorithm, the computation of the visual masking weights has almost no overhead at all.