Abstract
Abstract: :
Purpose: Recent evidence suggests that oblique phase disparities play an important role in the perception of stereoscopic depth. Here we investigate how the orientation and degree of inter-ocular matching of spatial frequency components in the images presented to the two eyes affect the threshold for stereoscopic depth. Methods: Observers (N=2) fused a pair of random-dot (RD) images on the 2 sides of a monitor, presented to the 2 eyes using pairs of orthogonal polarizers. Each eye’s image consisted of a 1-deg inner square of RDs centered in a 3.3-deg outer square of RDs. Identical outer squares in the 2 images provided a reference plane for disparities that were introduced in the inner square. The outline of the two inner squares remained fixed at zero disparity and coherent position disparities were produced within the 2 inner squares by manipulating each image’s Fourier phase spectrum. The amplitude spectrum of each image was identical. Stereothresholds were measured using the method of constant stimuli in 17 conditions that varied according to the axis and power of the cylindrical blur in each eye. The conditions were grouped based on whether the axis of cylindrical blur was the same (parallel condition) or orthogonal (orthogonal condition) in the two eyes. The blur axes used were 45, 90 (vertical), 135, and 180 (horizontal) o[rientation]deg, for powers of 2 and 3D. From trial to trial, the disparity in the inner square varied randomly according to the method of constant stimuli and stereothresholds were defined as the inverse slope (50% to 84%) of the resulting psychometric function. Results: The average stereothreshold without blur was 10.2+/-2.8 arc-sec. In the parallel condition, stereothresholds (for 3D of blur) were elevated most when the axis was 90 odeg (65.1+/-18.6 arc-sec) and least when the axis was 180 odeg (25.6+/-7.9 arc-sec). Stereothresholds were considerably higher in the orthogonal condition, but were similar for 90/180 odeg (206+/-133.1 arc-sec) and 45/135 odeg (221.3+/-195.7 arc-sec) axis combinations. Although stereothresholds were lower, a similar pattern of results was obtained with 2D of blur. Conclusions: The results indicate that the stereovision system prefers images that include spatial frequency components with similar orientations in the two eyes. The results also indicate that oblique disparities are highly likely to increase the signal-to-noise ratio of the stereovision system.
Keywords: depth • spatial vision • astigmatism