Motion deblurring in human vision.

Abstract

Under normal viewing conditions we are little conscious of blur in moving objects, despite the persistence of vision. Moving objects look more blurred in brief than in long exposures, suggesting an active mechanism for suppressing motion blur. To see whether blur suppression would improve visual discrimination of objects, we measured blur discrimination thresholds for moving Gaussian-blurred edges and bars. The observer's task was to decide which of two moving stimuli, presented successively, was the more blurred. It is known that for stationary objects the just-noticeable difference in blur increases with baseline blur; therefore, if motion increases blur, it would be expected to increase the just-noticeable difference in blur. An active deblurring mechanism, on the other hand, would be expected to counteract the detrimental effects of motion blur on discrimination performance. We found, however, that motion increased thresholds for blur discrimination, both for brief (40 ms) and for longer (150 ms) exposures. We conclude that motion deblurring is a subjective effect, which does not enhance visual discrimination performance. Moving objects appear sharp, not because of some special mechanism that removes blur, but because the visual system is unable to perform the discrimination necessary to decide whether the moving object is really sharp or not.