SUBJECTIVE IMAGE FIDELITY METRIC BASED ON BIT ALLOCATION OF THE HUMANVISUAL-SYSTEM IN THE DCT DOMAIN

Until now, subjective image distortion measures have partially used di verse empirical facts concerning human perception: non-linear percepti on of luminance, masking of the impairments by a highly textured surro und, linear filtering by the threshold contrast frequency response of the visual system, and non-linear post-filtering amplitude corrections in the frequency domain. In this work, we develop a frequency and con trast dependent metric in the DCT domain using a fully non-linear and suprathreshold contrast perception model: the Information Allocation F unction (IAF) of the visual system. It is derived from experimental da ta about frequency and contrast incremental thresholds and it is consi stent with the reported noise adaptation of the visual system frequenc y response. Exhaustive psychophysical comparison with the results of o ther subjective metrics confirms that our model deals with a wider ran ge of distortions more accurately than previously reported metrics. Th e developed metric can, therefore, be incorporated in the design of co mpression algorithms as a closer approximation of human assessment of image quality. (C) 1997 Elsevier Science B.V.