Stereo Resolution Research Articles

Human Stereopsis Is Not Limited by the Optics of the Well-Focused Eye

Human stereopsis, the perception of depth from differences in the two eyes' images, is very precise: image differences smaller than a single photoreceptor can be converted into a perceived difference in depth. To better understand what determines this precision, we examined how the eyes' optics affects stereo resolution. We did this by comparing performance with normal, well-focused optics and with optics improved by eliminating chromatic aberration and correcting higher-order aberrations. We first measured luminance contrast sensitivity in both eyes and showed that we had indeed improved optical quality significantly. We then measured stereo resolution in two ways: by finding the finest corrugation in depth that one can perceive, and by finding the smallest disparity one can perceive as different from zero. Our optical manipulation had no effect on stereo performance. We checked this by redoing the experiments at low contrast and again found no effect of improving optical quality. Thus, the resolution of human stereopsis is not limited by the optics of the well-focused eye. We discuss the implications of this remarkable finding.

Ocular Prevalence: Difference Between Crossed and Uncrossed Disparities of Stereo Objects

Background: Many observers perceive a horizontal offset between a stereoscopically viewed object and a reference object in the fixation plane, although both objects are presented along a single line that projects through the midpoint between both eyes. These observers weigh one eye more than the other when determining the visual directions of stereo objects. This unequal weighting has been called ocular prevalence. We considered that ocular prevalence might be asymmetric, that is, different for crossed and uncrossed stereo disparities. Methods: We examined 11 subjects with normal eyes and a stereo resolution of less than 100 sec arc. A vertical line was dichoptically presented for 100 ms at crossed or uncrossed stereo disparities of 1, 2, 3, 4, 5, 6, or 7 min arc; this stereo object was judged in relation to a reference line in the fixation plane. Results: Ten of the 11 subjects had ocular prevalence, but only 3 of the 10 showed a significant difference between crossed and uncrossed disparities. Conclusions: Ocular prevalence occurs frequently among observers with normal binocular vision, but only a minority of these observers show a difference between crossed and uncrossed disparities of the stereo object. It is unclear whether this difference bears clinical importance.

Commentary on high-performance computing and human vision II: Virtual reality and eyeball tracking

Visualization of scientific data that are not normally visible can be greatly enhanced by understanding human visual perception. Mars the Movie illustrated the use of stereo resolution for producing realistic vistas of the Mars surface. The real-time recording and analysis of eyeball-tracking data is another example of human vision research which benefits from high-performance computing. Currently the bottleneck is not in the hardware but in the available software to transform data efficiently.

Brief auditory stereograms

Auditory analogs of Julesz' random‐dot visual stereograms were created using 100‐msec tone bursts as dot analogs and interaural phase shift as the analog of visual angle disparity. Noise bursts consisting of 500 simultaneous tone components with random phase were presented to one ear. The frequencies of the components ranged from 10 to 5000 Hz by 10‐Hz steps. At the same time the other ear received an identical burst except that some components were phase shifted relative to the same frequency component in the other ear an amount corresponding to a 1‐msec delay in one ear (figure components) and all others were phase shifted according to a 1 msec delay in the other ear (ground components). Surprisingly poor stereo resolution was obtained; only very broad band figures could be reliably resolved.