Eye Research Institute, San Francisco
Models of shape processing generally infer the interior properties and depth structure from the 1D boundary curves. This approach will completely fail for objects presented via a random-dot stereogram, which is specifically designed to eliminate all such boundary information and provide the depth structure only through binocular disparity cues. Nevertheless, objects and their surface structure can readily be perceived by interpolation across the disparity cues in random-dot stereograms. The random-dot stereogram paradigm thus
reveals that human visual processing is fully capable of 3D object perception without prior boundary identification under sparse sampling conditions. Objects by their nature are 3D and have to be understood in 3D in order to be effectively perceived and manipulated from many angles. Sparse sampling is important because it is a typical feature of the visual environment, with many objects in the natural world represented only by sparse samples separated b! y intervening spaces across their surfaces. Thus, perceiving the 3D surface shape requires interpolation of the depth structure across spaces devoid of depth cues, and a sparse sampling paradigm is necessary for testing the domain of interpolation of object structure.
Our data reveal that depth interpolation in human vision is limited to the generic domain of depth structure (and is not possible within the domains of the individual cues). Since object perception and manipulation in sparse-sampled conditions depends on surface interpolation in depth, the 3D representation is the natural operating domain of object processing. A variety of approaches to the computation of 3D shape from the available visual information will be considered.