Visual interpolation of surfaces defined through motion

Abstract

The ability of human visual system to interpolate surfaces when the structure of the objects was defined through motion was investigated in a series of experiments. The type of the surface, the position of an area devoid of dots (the gap) and the type of gap edges was varied. The local orientation of the interpolated surfaces was estimated by a method of adjustment. The results show that the interpolated surfaces possessed less depth than the simulated ones. The type of the surface and the position of the gap had a significant effect on the performance. The worst results were obtained for a surface with orientation discontinuity (dihedral angle) where not only the shape of the interpolated surface differed significantly from the simulated one but the variability of the estimates was largest as well. The type of the gap edges did not affect the performance. The results are discussed in relation to the algorithms of surface interpolation.