Abstract
The blur discrimination function provides a standard measure of sensitivity to image blur, but is typically one-sided (extending from focused to increasingly blurred pedestals). We examined the shape of the function for stimuli that were also over-sharpened relative to the level of physical focus. Edge blur and sharpness were varied symmetrically about the focus point by varying the slope of the amplitude spectrum. Discrimination of slope changes has been examined in a number of studies, but has typically been assessed with natural images or noise. We instead examined slope discriminations for grayscale checkerboard patterns, which more closely parallel the simple edges used to assess blur discrimination. Subjects judged which of two images shown above or below a central fixation point appeared more blurred, with the threshold slope difference estimated with a 2AFC staircase. For these simple edges the discrimination function shows a minimum shifted toward slightly blurry images (i.e. steeper slopes), reminiscent to the dipper function characteristic of Gaussian-blurred edges. The lack of a corresponding dip for sharper (shallower) slopes suggests that the "norm" of image focus does not reflect a special neutral point in blur sensitivity, despite being a well-defined subjectively. In further experiments we examined how blur discrimination was affected by adaptation to blurred or sharpened edges. These showed shifts in directions consistent with the changes in perceived blur (i.e. after adapting to more blurred edges, test edges appeared sharper and there was a shift in the discrimination function toward blurrier edges). This suggests that adaptation can produce short-term changes in some aspects of blur sensitivity, in addition to the large changes it produces in the appearance of blur. Supported by EY10834
Meeting abstract presented at VSS 2016