Binocular disparity (the difference in the location of a target's image on the retina) is the most effective cue to depth perception. Stereoscopic 3D displays use this cue to cause sensation of depth in viewers. However, some stereoscopic 3D presentation protocols are prone to distortion in perceived depth. One such protocol is temporal interlacing. Temporal interlacing shows the left eye's view at one time and the right eye's at another. This introduces a temporal delay between the two eyes' views, which can be interpreted as spatial disparity by the visual system.
The figure below illustrates how this happens. In this example, a temporal interlacing display is showing a dot moving with zero disparity (left panel). The display shows the left-eye image first, and then the right-eye image after a temporal delay. With respect to the left-eye image at t=0, there can be more than one disparity pair, depending on which right-eye image is used for disparity calculation. The right panel plots the calculated binocular disparity as a function of the time delay between the left-eye image and a selected right-eye image (the black points). It is natural to assume that the calculated disparity is more certain if the temporal difference is smaller (the blue curve). The overall estimation is the weighted sum of the disparity values, where the certainty curve serves as a weighting coefficient.
Dolby uses wavelength multiplexing to achieve a temporal interlacing stereoscopic 3D presentation. Not surprisingly, it is effectively the same as normal temporal interlacing: the red, green, and blue channels for the left-eye view are shown during one subframe, and the RGB channels for the right-eye view during the other subframe. As we described in the depth distortion section, temporal interlacing techniques are prone to distortion in perceived depth. Noting that the visual system mainly uses luminance information to calculate binocular disparity, we proposed a novel stereoscopic 3D presentation technique that enhances depth perception. See the figure below. Instead of showing all three color-channels from one eye's view during each subframe, we proposed showing green from one eye's view and red and blue from the other eye's view during one subframe and vice versa in the other. This distributes the luminance more evenly at any time point and may enhance depth perception. Through a psychophysical experiment, we verified that this innovation significantly reduces the depth distortion present in the temporal interlacing S3D method.
Collaborator: Paul Johnson [home], Martin Banks