Depth interpolation, structure-from-motion for chromatic and luminance stimuli, color appearance under chromatic adaptation
My thesis consists of three psychophysical studies on visual perception. First, we investigate the interpolation of stereoscopic depth given only sparse disparity information. Interpolation occurs if the area to be interpolated is ambiguous, as in the case of a uniform rectangle with different disparities at the two edges. Interpolated depth is larger for a high mean contrast between rectangle and background than for a low mean contrast. Relative to a linear interpolation between the edges, a larger difference in edge disparity results in poorer depth interpolation. Rivalrous low spatial frequency content is resistant to interpolation; rivalrous high spatial frequencies do not interfere with depth interpolation.
Second, we ask whether luminance information is necessary to infer structure from motion. Structure-from-motion performance for luminance defined stimuli is superior to performance for chromatic stimuli when stimuli are expressed in terms of their length in cone contrast space. When the psychometric function for 3D shape identification performance is scaled in terms of contrast threshold for psychophysical tasks prior to the computation of structure-from-motion, performance with chromatic stimuli is still inferior to performance with luminance-defined stimuli. Performance in the structure-from-motion task increases with purely chromatic contrast. This result is at variance with the extreme view that chromatic information cannot be used by the motion system.
Third, the linearity of asymmetric color matches was tested under conditions of steady-state chromatic adaptation to lights of equal luminance differing either in L-M or in S cone excitation from the standard equal-energy achromatic adapting light. We asked whether the changes in color appearance induced by the transition from a grey to a chromatic adapting light, can be described by a linear transformation in the three-dimensional color space. Adapting lights differing in S cone excitation only from the standard adapting light produced nonlinearities. For adapting lights varying along the L-M axis from the standard adapting light we did not find any nonlinearities. Adaptational effects were not additive. Color appearance changes due to transition from grey to an adapting light differing in L-M as well as in S cone excitation could not be predicted by adding the color changes induced by adapting lights differing either along the L-M or along the S cone axis.