The role of perceptual contrast non-linearities in image transform quantization

The conventional quantizer design based on average error minimization over a training set does not guarantee a good subjective behavior on individual images even if perceptual metrics are used. In this work a novel criterion for transform coder design is analyzed in de pth. Its aim is to bound the perceptual distortion in each individual quant ization according to a non-linear model of early human vision. A common com parison framework is presented to describe the qualitative behavior of the optimal quantizers under the proposed criterion and the conventional rate-d istortion based criterion. Several underlying metrics, with and without per ceptual non-linearities, are used with both criteria. Analytical results sh ow that the proposed design criterion gives rise to a JPEG-like quantizatio n if a simple linear metric is used. Experimental results show that signifi cant improvements over the perceptually weighted rate-distortion approach a re obtained if a more meaningful non-linear metric is used. (C) 2000 Elsevi er Science B.V. All rights reserved.