Purchase this article with an account.
Leah Land, Igor Juricevic, Arnold Wilkins, Michael Webster; Visual discomfort and natural image statistics. Journal of Vision 2009;9(8):1046. doi: 10.1167/9.8.1046.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Images with excessive energy at medium spatial frequencies (Fernandez & Wilkins Perception 2007) or that have high color contrast or no luminance contrast (Wilkins et al ECVP 2008), appear uncomfortable or aversive and can induce headaches in hypersensitive observers. Such stimuli are uncharacteristic of natural images, and we therefore examined whether ratings of visual discomfort generally increase with deviations from the spatial and chromatic properties of natural scenes. Full-color 14° images were generated from noise or random overlapping rectangles (Mondrians). Slopes of the amplitude spectra for luminance or chromatic contrast were varied independently to create image sets ranging from strongly blurred to sharpened in luminance or color relative to a “natural” 1/f spectrum. Subjects viewed the images on a monitor for 10 sec each and rated both discomfort and artistic merit on a 7-point scale. Perceived blur was dominated by the luminance slopes, with discomfort rated lowest for the 1/f spectra for both the filtered noise and Mondrians, which were ranked similarly. In comparison, spatial variations in color had only weak effects. In a second set of Mondrians, focus and luminance contrast were fixed while color was varied along axes at 45° intervals in the LM vs. S chromatic plane (for images with the same mean gray but different hue angles), or was confined to one pole of each axis (for images that varied in a single average hue). Discomfort ratings were lowest for a blue-yellow axis (−45°), a direction that is again typical of natural outdoor scenes. Notably these ratings of discomfort were not related to judgments of artistic merit. Thus for both spatial and chromatic content the least aversive images corresponded to characteristic properties of the natural visual environment, and may reflect a normalization of visual coding to the natural world.
This PDF is available to Subscribers Only