BAND-LIMITED CONTRAST IN NATURAL IMAGES EXPLAINS THE DETECTABILITY OFCHANGES IN THE AMPLITUDE SPECTRA

The psychophysical task of discriminating changes in the slopes of the amplitude spectra of complex images has been used in the past to test ,whether the human visual system might be optimised for coding the spa tial structure in natural images (e.g. Knill et al., 1990; Tadmor & To lhurst, 1994). We have reported that the dependency of these discrimin ation thresholds on the reference slope has the same overall general f orm, regardless of the particular digitised photographs that are used for generating the stimuli, The actual discrimination thresholds, howe ver, differ markedly in magnitude for stimuli that are derived from di fferent digitised photographs, Here, we describe a model that aims at explaining this diversity of threshold magnitudes: we suppose that the observer is detecting small changes in image contrast estimated,withi n limited spatial-frequency bands of about 1 octave bandwidth. This lo cal-contrast analysis reveals that contrast changes in only one freque ncy band are of comparable magnitudes to the changes that observers ne ed for detecting differences in the Michelson contrast of simple sinus oidal gratings, The success of this band-limited contrast model is fur ther shown in experiments where the slopes of the amplitude spectra of stimuli were changed only within restricted frequency bands, We show that when the slope is changed outside the limited frequency band impl icated by the contrast model, the observer's thresholds are greatly el evated, Thresholds remain unchanged when slope changes are made within the implicated band, We also find that the exact bandwidth of the con trast operator is not critical, provided that it is in the range of ab out 0.6-1.5, which is the characteristic bandwidth range of V1 neurons . (C) 1997 Elsevier Science Ltd.