126/WBl-l Contrast Sensitivity Functions for Analysis and Simulation of Visual Perception

The response of visual filters has been used to simulate vision and to design image enhancement techniques for the visually impaired. For the simulations and enhancements to be useful, they must be based on appropriate measurements of the filters being studied. Most previous measurements of spatial contrast sensitivity functions (CSFs) have been based on stimuli that give maximal contrast sensitivity. We measured CSF using stimuli that are better suited for analyzing perception of images·. Use of the different types of CSF yields notably different simulations. Measurements of maximal threshold sensitivity may enable earlier detection of visual loss. Extended grating stimuli are also convenient for experiments where global Fourier analysis is used to interpret results. However, extended grating CSFs may be inappropriate for siinulation of vision in normals (Ginsburg, 1978) and low vision patients (Lundh et al., 1981; Peli et al., 1989) and for the design of image enhancement techniques as visual aid (peli & Peli, 1984; Lawton, 1988). ' Repeated, continuous cycles at any frequency are rarely seen in natural scene images. Only when periodic patterns (such as a picket fence) appear in the image does one find more than two consecutive high-contrast cycles anywhere within the images. Thus, the senSitivity added by spatial summation is unlikely to be representative of visual perception of images other than gratings. The use of localized stimuli matched to the impulse response of simple cortical cells is a natural way to obtain "valid" CSFs that minimize spatial summation. The temporal presentation of the gratings is usually a gradual fade-in and -outratherthan flash-on and -off, as the former presentation results in increased threshold senSitivity measure. In free viewing, saccadic eye movements result in abrupt changes of eye position. Thus, the local presentation of stimuli to receptive fields usually does not follow the gradual fade-in and -out presentation. Therefore, thresholds measured with abrupt presentation of grating patches better represent the perception of localized contrast in complex images. Detection thresholds are commonly measured using paradigms in which only the presence of the target is recorded, without need for the subject to perceive any aspect of form, such as orientation, profile, or contrast. This often produces the lowest threshold, but does not necessarily correspond to the information required for useful image interpretation. For this purpose the experimental task should include some form discrimination. In view of the above arguments, we designed our CSF measurements to be used for low vision simulation and image enhancement based on visual loss in the following way: Stimuli presented were patches of sinusoidal gratings in a Gaussian envelope (Gabor functions) of 1octave bandwidth, flashed on for I sec at either horizontal or vertical orientation, and abruptly removed. We compared the CSF obtained with these stimuli to the data obtained with fIXed size screen, narrower bandwidth (more cycles),andgradual temporal presentation. Simulations of the appearance of images to normal observers, using different types of CFSs measured, are used to demonstrate the effect of these differences.