Digital color image processing and psychophysics within the framework of a human visual model

A three-dimensional homomorphic model of human color vision based on neurophysiological and psychophysical evidence is presented. This model permits the quantitative definition of perceptually important parameters such as brightness. saturation, huo and strength. By modelling neural interaction in the human visual system as three linear filters operating on perceptual quantities, this model accounts for the automatic gain control properties of the eye and for brightness and color contrast effects. In relation to color contrast effects, a psychophysical experiment was performed. It utilized a high quality color television monitor driven by a general purpose digital computer. This experiment, based on the cancellation by human subjects of simultaneous color contrast illusions, allowed the measurement of the low spatial frequency part of the frequency responses of the filters operating on the two chromatic channels of the human visual system. The experiment is described and its results are discussed. Next, the model is shown to provide a suitable framework in which to perform digital images processing tasks. First, applications to color image enhancement are presented and discussed in relation to photographic masking techniques and to the handling of digital color images. Second, application of the model to the definition of a distortion measure between color images (in the sense of Shannon's rate-distortion theory), meaningful in terms of human evaluation, is shown. Mathematical norms in the "perceptual" space defined by the model are used to evaluate quantitatively the amount of subjective distortion present in artificially distorted color presented. Results of a coding experiment yielding digital color images coded at an average bit rate of 1 bit/pixel are shown. Finally conclusions are drawn about the implications of this research from the standpoints of psychophysics and of digital image processing.