Quantitative stereoscopic depth without binocular correspondence

Abstract

WHAT features in a stereogram define the disparities that lead to stereoscopic depth? The usual answer is that luminance-defined edges from the two eyes are matched and produce depth perception^1,2. But parts of an object may be occluded by other objects and absent from one eye's view. It was suggested that unpaired monocular elements might signal occlusion in depth, and the qualitative perception of depth associated with unmatched elements has been shown to be consistent with the geometry of occlusion^3–5. We designed a stereogram that simulates a particular in front of a large black rectangle pasted on a white background. The position of the occluder is adjusted so that its left edge obscures the left-hand edge of the black rectangle in the right eye view and its right edge obscures the right-hand edge of the black rectangle in the left eye view. We report here that quantitative stereopsis can be seen from this stereogram, even though there are no binocular corresponding luminance edges to match.