Abstract
It is well established that monocular regions arising from occlusion of one object by another contribute to stereoscopic depth perception. However, the exact role of monocular occlusions in 3D scene perception remains unclear. One possibility is that monocular occlusions define object boundaries or discontinuities in depth. This is an attractive possibility, but to date it has not been tested empirically. Here we describe a series of experiments that directly test this hypothesis. Our novel stereoscopic stimulus consists of a foreground rectangular region set against a random-dot background positioned at zero disparity. One side of the foreground region is filled with a random-dot texture shifted towards the observer in apparent depth. The remaining area of the foreground is blank and carries no disparity information. In several experiments, we vary the presence or absence and the width of occluded areas at the border of the central blank area and the background texture. Our data show that the presence of occluded elements on the boundary of the blank area dramatically influences the perceived shape of the foreground region. If there are no occluded elements, the foreground appears to contain a depth step, as the blank area lies at the depth of the zero disparity border. When occluded elements are added, the blank region is seen vividly at the same depth as the texture, so that the foreground is perceived as a single opaque planar surface. We show that the depth perceived via occlusion is not due to the presence of binocular disparity at the boundary, and that it is qualitative, not quantitative in nature. Taken together, our experiments provide strong support for the hypothesis that monocular occlusion zones are important signals for the presence and location of depth discontinuities.
This work was supported by NSERC grants to L.W. and R.A.