Human sensitivity to phase perturbations in natural images: a statistical framework

Fourier phase information is important in determining the appearance of nat ural scenes, but the structure of natural-image phase spectra is highly com plex and difficult to relate directly to human perceptual processes. This p roblem is addressed by extending previous investigations of human visual se nsitivity to the randomisation and quantisation of Fourier phase in natural images. The salience of the image changes induced by these physical proces ses is shown to depend critically on the nature of the original phase spect rum of each image, and the processes of randomisation and quantisation are shown to be perceptually equivalent provided that they shift image phase co mponents by the same average amount. These results are explained by assumin g that the visual system is sensitive to those phase-domain image changes w hich also alter certain global higher-order image statistics. This assumpti on may be used to place constraints on the likely nature of cortical proces sing: mechanisms which correlate the outputs of a bank of relative-phase-se nsitive units are found to be consistent with the patterns of sensitivity r eported here.