John P. Frisby

PMF: A Stereo Correspondence Algorithm Using a Disparity Gradient Limit

The advantages of solving the stereo correspondence problem by imposing a limit on the magnitude of allowable disparity gradients are examined. It is shown how the imposition of such a limit can provide a suitable balance between the twin requirements of disambiguating power and the ability to deal with a wide range of surfaces. Next, the design of a very simple stereo algorithm called PMF is described. In conjunction with certain other constraints used in many other stereo algorithms, PMF employs a limit on allowable disparity gradients of 1, a value that coincides with that reported for human stereoscopic vision. The excellent performance of PMF is illustrated on a series of natural and artificial stereograms. Finally, the differences between the theoretical justification for the use of disparity gradients for solving the stereo correspondence problems presented in the paper and others that exist in the stereo algorithm literature are discussed.

Psychophysical and computational studies towards a theory of human stereopsis

Psychophysical studies are described which pose a strong challenge to models of human stereopsis based on the processing of disparity information within independent spatial frequency tuned binocular channels. These studies support instead the proposal that the processes of human binocular combination integrally relate the extraction of disparity information with the construction of raw primal sketch assertions. This proposal implies global binocular combination rules using principles of figural continuity and cross-channel correspondence to disambiguate local matches found independently within spatial frequency channels. Exploratory small-scale computational experiments with stereo algorithms based on these rules are described and found to be successful in dealing with a variety of stereo inputs. The constraints presented by objects which are exploited by these algorithms are discussed.

Illusory Contours: Curious Cases of Simultaneous Brightness Contrast?

It is suggested that simultaneous brightness contrast mediated by lateral inhibition plays an important role in generating many illusory contours. These contours might reflect a further way in which lateral inhibition serves to clarify and sharpen the neural encoding of retinal images.

Stereopsis, vertical disparity and relief transformations.

The pattern of retinal binocular disparities acquired by a fixating visual system depends on both the depth structure of the scene and the viewing geometry. This paper treats the problem of interpreting the disparity pattern in terms of scene structure without relying on estimates of fixation position from eye movement control and proprioception mechanisms. We propose a sequential decomposition of this interpretation process into disparity correction, which is used to compute three-dimensional structure up to a relief transformation, and disparity normalization, which is used to resolve the relief ambiguity to obtain metric structure. We point out that the disparity normalization stage can often be omitted, since relief transformations preserve important properties such as depth ordering and coplanarity. Based on this framework we analyse three previously proposed computational models of disparity processing; the Mayhew and Longuet-Higgins model, the deformation model and the polar angle disparity model. We show how these models are related, and argue that none of them can account satisfactorily for available psychophysical data. We therefore propose an alternative model, regional disparity correction. Using this model we derive predictions for a number of experiments based on vertical disparity manipulations, and compare them to available experimental data. The paper is concluded with a summary and a discussion of the possible architectures and mechanisms underling stereopsis in the human visual system.

The computation of binocular edges.

A computational model is described which effects the binocular combination of monocular edge information. The distinctive features of the model are: (i) it identifies edge locations in each monocular field by searching for zero crossings in nonorientated centre-surround convolution profiles; (ii) it selects amongst all possible binocular point-for-point combinations of edge locations only those which satisfy a (quasi-)collinear figural grouping rule; (iii) it presents a concept of the orientated and spatial-frequency-tuned channel as a nonlinear grouping operator. The success of the model is demonstrated both on a stereo pair of a natural scene and on a random-dot stereogram.

The relationship between apparent depth and disparity in rivalrous-texture stereograms.

A series of experiments is reported on rivalrous-texture stereograms composed of narrow-band-filtered random noise. Experiment 1 found that the apparent depth—disparity function for such stereograms was different from that observed with similar but nonrivalrous stimuli. In particular, rivalrous divergent disparities produced the same depth as rivalrous zero disparity and this latter disparity itself produced a significant degree of protruding (i.e. ‘convergent’) depth in a certain type of rivalrous-texture stereogram. Free inspection was permitted and disparities were in the range 16 min convergent to 16 min divergent. Experiment 2 found no convincing evidence for reliable qualitative depth discriminations from tachistoscopic presentations of rivalrous-texture stereograms, using a forced-choice task requiring a discrimination between 16 min convergent and 16 min divergent conditions. This task was solved easily for equivalent nonrivalrous stimuli. Experiment 3 measured a hitherto unreported binocular depth effect, termed ‘paradepth’, which is produced by presenting a target in one field only. This effect appears to be a genuine binocular depth effect and not just the result of an ordinary monocular masking depth cue. The size of the depth effect was found to be a function of the width of the target. The overall conclusion derived from the series of experiments is that rivalrous-texture stereograms are complex stimuli capable of yielding curious and unexpected depth effects which are not readily explained in detail within any existing theoretical framework.

Contrast sensitivity function for stereopsis.

Contrast thresholds for stereopsis from narrow-band-filtered random-dot stereograms were compared with contrast thresholds for simple detection of similar narrow-band noise. Centre frequencies of filters were in the range 2·5–15 cycles deg−1. It was found that the contrast sensitivity function for stereopsis is similar in shape to that for detection, suggesting that as far as contrast requirements are concerned the mechanisms of global stereopsis do not show a bias in sensitivity to any particular spatial frequency but instead require a constant level of suprathreshold contrast regardless of spatial frequency.

Depth cue reliance in surgeons and medical students

Background: Depth perception is reduced in endoscopic surgery, although little is known about the effect this has on surgical performance. Methods: To assess the role of depth cues, 45 subjects completed tests of depth cue reliance. Surgical skill was assessed using the Minimally Invasive Surgical Trainer-Virtual Reality, a previously validated laparoscopic simulator. Results: We could demonstrate no difference in cue reliance for three depth cues — namely stereo, texture, and outline — between surgeons and medical students. Greater dominance on stereo for medical students was a positive finding and a negative finding for the surgeons when correlated with surgical performance. Conclusions: We suggest that surgeons learn to adapt to the nonstereo environment in MIS, and this is the first study to show evidence of this phenomenon. This difference in stereo reliance is a reflection of the experience that surgeons have with laparoscopy compared with medical students, who have none.

Evidence for good recovery of lengths of real objects seen with natural stereo viewing.

Six experiments are described in which good performance of the task of matching the lengths of two stationary real objects, gnarled wooden sticks, under a variety of binocular viewing conditions, including variations in viewing distances was demonstrated. Relatively poor matching performance was observed when the sticks were viewed monocularly in oscillatory motion, or monocularly and stationary. The results suggest that stereo can support good representations of metric scene structure when length judgments of natural objects are required under (quasi-)natural viewing. The implications of these results for theories of structure from stereo and structure from motion are discussed.

Matching geometrical descriptions in three-space

Abstract A matching strategy for combining two or more three-space descriptions, obtained here from edge-based binocular stereo, of a scene is discussed. The scheme combines features of a number of recent model-matching algorithms with heuristics aimed to reduce the space of potential rigid transformations that relate scene descriptions.