Abstract
In real-world settings, objects and backgrounds have varied textures. The spatial frequency (SF) of a texture is likely an important cue for depth perception. Here, we used bandpass-filtered noise patterns to study the interactions between the texture of targets and backgrounds with different SF combinations in a virtual reality setting. The experiment was programmed using Three.js, a Javascript library to render three-dimensional graphics in a web browser. Wearing an HTC Vive headset that showed separate images to each eye, subjects (n=3) viewed the stimuli, which consisted of 3 tiles placed at a virtual distance of 3m and a background surface 6m away from the observer. Each tile subtended an angle of 3°x3°, with a center-to-center distance of 4.7° between adjacent tiles. In each trial, the targets appeared for 1sec, with one tile positioned closer to the observer, scaled to compensate for monocular cues. The task was to indicate the closer tile with a keypress. The threshold distance from the reference tiles was determined using a staircase procedure and converted to disparity sensitivity. Both the target and background were tested with different combinations of bandpass-filtered noise patterns (0.3, 0.6, 1, 3, and 6 cpd) and also a uniform gray background. With a gray background, reliable thresholds were measured for all target textures except 6cpd. With a textured background, thresholds varied from a virtual distance of 11cm (equivalent disparity ~3 arcmins) to 86cm (~45 arcmins), depending on the condition. The subjects had difficulty discriminating the target from the background when both were 6cpd, which neared the screen resolution limit. We found that the disparity sensitivity has a bandpass shape, peaking at 1 to 3cpd for both target and background. Thus the textures of both targets and background are important for extracting depth information.