Visual coding and the phase structure of natural scenes

Although it is now well known that natural images display consistent statis tical properties which distinguish them from random luminance distributions , this ecological approach to vision has so far concentrated on those secon d-order image statistics which are quantified by image power spectra, and i t appears to be the image phase spectra which carry the majority of the ima ge-intrinsic information. The present work describes how conventional nth-o rder statistics can be modified so that they are sensitive to image phase s tructure only. The modified measures are applied to an ensemble of natural images, and the results show that natural images do have consistent higher- order statistical properties which distinguish them from random-phase image s with the same power spectra. An interpretation of this finding in terms o f higher-order spectra suggests that these consistent properties arise from the ubiquity of edge structures in natural images, and raises the possibil ity that the properties of ideal relative-phase-sensitive mechanisms could be determined directly from analyses of the higher-order structure of natur al scenes.