Andrew Glennerster

The Effect of Display Size on Disparity Scaling from Differential Perspective and Vergence Cues

The present study compared the relative effectiveness of differential perspective and vergence angle manipulations in scaling depth from horizontal disparities. When differential perspective and vergence angle were manipulated together (to simulate a range of different viewing distances from 28 cm to infinity), approximately 35% of the scaling required for complete depth constancy was obtained. When manipulated separately the relative influence of each cue depended crucially on the size of the visual display. Differential perspective was only effective when the display size was sufficiently large (i.e., greater than 20 deg) whereas the influence of vergence angle, although evident at each display size, was greatest in the smaller displays. For each display size the independent effects of the two cues were approximately additive. Perceived size (and two-dimensional spacing of elements) was also affected by manipulations of differential perspective and vergence. These results confirm that both differential perspective and vergence are effective in scaling the perceived two-dimensional size of elements and the perceived depth from horizontal disparities. They also show that the effect of the two cues in combination is approximately equal to the sum of their individual effects.

The task-dependent use of binocular disparity and motion parallax information.

Binocular disparity and motion parallax are powerful cues to the relative depth between objects. However to recover absolute depth, either additional scaling parameters are required to calibrate the information provided by each cue, or it can be recovered through the combination of information from both cues (Richards, W. (1985). Structure from stereo and motion. Journal of the Optical Society of America, 2, 343-349). However, not all tasks necessarily require a full specification of the absolute depth structure of a scene and so psychophysical performance may vary depending on the amount of information available, and the degree to which absolute depth structure is required. The experiments reported here used three different tasks that varied in the type of geometric information required in order for them to be completed successfully. These included a depth nulling task, a depth-matching task, and an absolute depth judgement (shape) task. Real world stimuli were viewed (i) monocularly with head movements, (ii) binocularly and static, or (iii) binocularly with head movements. No effect of viewing condition was found whereas there was a large effect of task. Performance was accurate on the matching and nulling tasks and much less accurate on the shape task. The fact that the same perceptual distortions were not evident in all tasks suggests that the visual system can switch strategy according to the demands of the particular task. No evidence was found to suggest that the visual system could exploit the simultaneous presence of disparity and motion parallax.

Efficiency of locating centres of dot-clusters by human observers.

A spatial interval acuity task was used to determine the efficiency with which observers located the centroids of circular clusters of individually-small elements. A cluster was defined by its radius and by the number of elements within it. The observers task was to compare the horizontal distance between the centres of two such clusters to a standard distance, and to decide whether it was greater or smaller. The standard distance was made explicit by two markers on the display screen. As expected on statistical grounds, thresholds increased with the radius of the clusters, and decreased as a function of the number of elements. There were considerable individual differences in efficiency between observers, but the best observers achieved efficiencies close to 100% in many of the conditions. One anomalous observer showed virtually no decrease in thresholds as the number of elements increased, and thus her efficiencies declined to as little as 10%. The ability to combine positional information over spatially-separated elements to obtain an estimate of their central tendency appears in general to be high, and may underlie the perceptual biases revealed in certain geometric illusions, such as the Müller-Lyer.

Humans Ignore Motion and Stereo Cues in Favor of a Fictional Stable World

As we move through the world, our eyes acquire a sequence of images. The information from this sequence is sufficient to determine the structure of a three-dimensional scene, up to a scale factor determined by the distance that the eyes have moved. Previous evidence shows that the human visual system accounts for the distance the observer has walked and the separation of the eyes when judging the scale, shape, and distance of objects. However, in an immersive virtual-reality environment, observers failed to notice when a scene expanded or contracted, despite having consistent information about scale from both distance walked and binocular vision. This failure led to large errors in judging the size of objects. The pattern of errors cannot be explained by assuming a visual reconstruction of the scene with an incorrect estimate of interocular separation or distance walked. Instead, it is consistent with a Bayesian model of cue integration in which the efficacy of motion and disparity cues is greater at near viewing distances. Our results imply that observers are more willing to adjust their estimate of interocular separation or distance walked than to accept that the scene has changed in size.

Spatial calibration of an optical see-through head mounted display

We present here a method for calibrating an optical see-through head-mounted display (HMD) using techniques usually applied to camera calibration (photogrammetry). Using a camera placed inside the HMD to take pictures simultaneously of a tracked object and features in the HMD display, we could exploit established camera calibration techniques to recover both the intrinsic and extrinsic properties of the HMD (width, height, focal length, optic centre and principal ray of the display). Our method gives low re-projection errors and, unlike existing methods, involves no time-consuming and error-prone human measurements, nor any prior estimates about the HMD geometry.

dmax for stereopsis and motion in random dot displays

The upper displacement limit for motion was compared with the upper disparity limit for stereopsis using two-frame random dot kinematograms or briefly presented stereograms. dmax (the disparity/displacement at which subjects make 20% errors in a forced-choice paradigm) was found to be very similar for motion and stereo at all dot densities, and to fall with increasing dot density (0.006% or two dots to 50%) according to a power law (exponent -0.2). If dmax is limited by the spacing of false targets, this pattern of results suggests that the spatial primitives in the input to the correspondence process may be derived from multiple spatial scales. A model using MIRAGE centroids provides a good fit to the data.

Systematic distortions of perceptual stability investigated using immersive virtual reality

Using an immersive virtual reality system, we measured the ability of observers to detect the rotation of an object when its movement was yoked to the observers own translation. Most subjects had a large bias such that a static object appeared to rotate away from them as they moved. Thresholds for detecting target rotation were similar to those for an equivalent speed discrimination task carried out by static observers, suggesting that visual discrimination is the predominant limiting factor in detecting target rotation. Adding a stable visual reference frame almost eliminated the bias. Varying the viewing distance of the target had little effect, consistent with observers underestimating distance walked. However, accuracy of walking to a briefly presented visual target was high and not consistent with an underestimation of distance walked. We discuss implications for theories of a task-independent representation of visual space.

Bias and sensitivity of stereo judgements in the presence of a slanted reference plane

The perceived depth of features is known to be affected by the presence of a slanted reference plane. Mitchison and Westheimer ((1984). Vision Research, 24, 1063-1070) reported that two lines appear to be at the same depth when they lie in a plane approximately parallel with the reference plane. We measured the perceived depth of two lines presented in front of a regular grid of dots that was either fronto-parallel or slanted about a vertical axis. The effect of the slanted grid on perceived depth diminished as the grid was moved further in disparity from the lines. We also found that the slanted grid affected the sensitivity to differences in disparity. The minimum threshold for detecting changes in disparity is normally lowest in the fixation plane and rises systematically with increasing pedestal disparity. In the presence of a slanted reference plane, the minimum threshold is at or close to the plane of equal perceived depth and rises with increasing disparity from this plane.

Stereoacuity thresholds in the presence of a reference surface

With isolated binocular targets, the best depth discrimination is found in the fixation plane (Blakemore, C., Journal of Physiology 211 (1970) 599). More recent studies have suggested that stereoscopic thresholds are not always a simple function of absolute disparity, but depend on the relative disparities in the stimulus. Here, we explored the effects of relative disparity in more detail, taking particular care to control for the possibility that subjects might change their binocular eye position or exploit monocular information provided by additional reference cues. Subjects judged the depth of a vertical target line presented above a comparison line in a blank window within a fronto-parallel reference surface composed of randomly positioned dots. On individual trials, the reference surface was presented at one of three disparities (-10, 0 and +10 arc min). To control for changes in binocular eye position, exposure duration was 150 ms, and experimental conditions with different disparities of the reference surface and comparison line were randomly interleaved. To control for monocular cues, changes in threshold were determined with respect to a disparity noise condition that was in all ways identical to the reference plane condition, except that the disparities of the dots were randomly assigned between 10 and +10 arc min. Stereo-thresholds were lowered by a factor of about 2 when the surface was at the same depth as the comparison line. Thresholds were also lowered when the comparison disparity was close to the same depth as the reference surface, but were often raised when the comparison disparity had the opposite disparity sign. These results provide unequivocal evidence that the fundamental sensitivity of the disparity detecting system can be influenced by relative disparity cues that are not related to the task.

Cues to viewing distance for stereoscopic depth constancy.

A veridical estimate of viewing distance is required in order to determine the metric structure of objects from binocular stereopsis. One example of a judgment of metric structure, which we used in our experiment, is the apparently circular cylinder task (E B Johnston, 1991 Vision Research 31 1351 – 1360). Most studies report underconstancy in this task when the stimulus is defined purely by binocular disparities. We examined the effect of two factors on performance: (i) the richness of the cues to viewing distance (using either a naturalistic setting with many cues to viewing distance or a condition in which the room and the monitors were obscured from view), and (ii) the range of stimulus disparities (cylinder depths) presented during an experimental run. We tested both experienced subjects (who had performed the task many times before under full-cue conditions) and naive subjects. Depth constancy was reduced for the naïve subjects (from 62% to 46%) when the position of the monitors was obscured. Under similar conditions, the experienced subjects showed no reduction in constancy. In a second experiment, using a forced-choice method of constant stimuli, we found that depth constancy was reduced from 64% to 23% in naive subjects and from 77% to 55% in experienced subjects when the same set of images was presented at all viewing distances rather than using a set of stimulus disparities proportional to the correct setting. One possible explanation of these results is that, under reduced-cue conditions, the range of disparities presented is used by the visual system as a cue to viewing distance.