A perceptual error measure for sampled and interpolated complex colour images

Error measures are needed for optimizing display parameters. Most of the existing error measures are inappropriate since they correlate poorly with the perceived impairment of the reproduced images. The reason is that these measures, often distance measures, do not or do not sufficiently incorporate properties of the human visual system. We therefore present a perceptual error measure that uses the distance in a perceptual space spanned by the sampling and interpolation artifacts periodic structure and blur, to predict the impairment of sampled and interpolated images. It is shown how the sensorial strengths of periodic structure and blur that are along the axes of this perceptual space can be determined from the image. In addition, we demonstrate that after the perceptual error measure is calibrated to a particular observer, it can predict experimentally determined subjective image quality of sampled and interpolated colour images, and may thereby be used to optimize display parameters.