Qualitative depth from stereo, with applications

doi:10.1016/0734-189X(90)90138-L

Computer Vision, Graphics, and Image Processing

Volume 49, Issue 2, February 1990, Pages 222-241

Computer Vision, Gra...

https://doi.org/10.1016/0734-189X(90)90138-L Get rights and content

Abstract

Obtaining exact depth from binocular disparities is hard if camera calibration is needed. We will show that qualitative information can be obtained from stereo disparities with little computation and without prior knowledge (or computation) of camera parameters. First, we derive two expressions that order all matched points in the images by depth in two distinct ways from image coordinates only. Using one for tilt estimation and point separation (in depth) demonstrates some anomalies observed in psychophysical experiments, most notably the “induced size effect.” We apply the same approach to detect qualitative changes in the curvature of a contour on the surface of an object, with eitherx- ory-coordinate fixed. Second, we develop an algorithm to compute axes of zero-curvature from disparities alone. The algorithm is shown to be quite robust against violations of its basic assumptions for synthetic data with relatively large controlled deviations. It performs almost as well on real images, as demonstrated on an image of four cans at different orientations.

References (17)

ArditiA.
The dependence of the induced effect on orientation and a hypothesis concerning disparity computations in general
Vision Res.
(1982)
ArditiA. et al.
A simple explanation of the induced size effect
Vision Res.
(1981)
MitchisonG.J. et al.
The perception of depth in simple figures
Vision Res.
(1984)
PrintzH.
Finding the orientation of a cone or cylinder
Longuet-HigginsH.C. et al.
The interpretation of a moving retinal image
FoleyJ.M. et al.
Effects of voluntary eye movement and convergence on the binocular appreciation of depth
Percept. Psychophys.
(1972)
HornB.K.P.
Robot Vision
(1986)
Longuet-HigginsH.C.
A computer algorithm for reconstructing a scene from two projections
Nature
(1981)

There are more references available in the full text version of this article.

Cited by (29)

Fixation could simplify, not complicate, the interpretation of retinal flow
2001, Vision Research
The visual system must generate a reference frame to relate retinal images in spite of head and eye movements. We show how a reference frame for storing the visual direction and depth of points can be composed from the angles and changes in angles between pairs and triples of points. The representation has no unique origin in 3-D space nor a unique set of cardinal directions (basis vectors). We show how this relative representation could be built up over a series of fixations and for different directions of translation of the observer. Maintaining gaze on a point as the observer translates helps in building up this representation. In our model, retinal flow is divided into changes in eccentricity and changes in meridional angle. The latter, called ‘polar angle disparities’ for binocular viewing (Weinshall, 1990. Computer Vision Graphics and Image Processing, 49 222–241), can be used to recover the relief structure of the scene in a series of stages up to full Euclidean structure. We show how the direction of heading can be recovered by a similar series of stages.
An orientation anisotropy in the effects of scaling vertical disparities
2000, Vision Research
Gårding et al. (Vis Res 1995;35:703–722) proposed a two-stage theory of stereopsis. The first uses horizontal disparities for relief computations after they have been subjected to a process called disparity correction that utilises vertical disparities. The second stage, termed disparity normalisation, is concerned with computing metric representations from the output of stage one. It uses vertical disparities to a much lesser extent, if at all, for small field stimuli. We report two psychophysical experiments that tested whether human vision implements this two-stage theory. They tested the prediction that scaling vertical disparities to simulate different viewing distances to the fixation point should affect the perceived amplitudes of vertically but not horizontally oriented ridges. The first used elliptical half-cylinders and the ‘apparently circular cylinder' judgement task of Johnston (Vis Res 1991;31:1351–1360). The second experiment used parabolic ridges and the amplitude judgement task of Buckley and Frisby (Vis Res 1993;33:919–934). Both studies broadly confirmed the anisotropy prediction by finding that large scalings of vertical disparities simulating near distances had a strong effect on the perceived amplitudes of the vertically oriented stimuli but little effect on the horizontal ones. When distances >25 cm were simulated there were no significant differential effects and various methodological reasons are offered for this departure from expectations.
Depth object recovery using radial basis functions
1999, Optics Communications
The main task of robot vision systems is to recover the shape of the objects in a scene and use this information for recognition and classification. The fringe projection technique may be used to recover depth range data. To use this technique, we need to estimate the phase of the projected fringes (related to the object height) by using conventional phase demodulation techniques (phase shifting, phase locked loop, or spatial synchronous detection among others). The final step of the process is to recover the object surface using a calibration procedure. This calibration procedure uses a phase-to-depth conversion obtained from the experimental optical geometry. Set-up parameters to consider are the optical characteristics of the lenses used to project and acquire the fringe patterns, which are usually unknown. Some experimental factors arise when the object under analysis is located near the projector–camera system, such as the diverging fringe projection, the video camera perspective distortion and the use of crossed-optical-axes geometry. In this work, we present a neural network based on radial basis functions (RBF) to estimate actual depth data of the object from the recovered phase using the projected fringe pattern technique. The training set of the neural network is the phase recovered of several calibration planes whose locations in space are known. An advantage of our method is that knowledge of the optical parameters of the experiment is not explicitly required.
Fast relative depth computation for an active stereo vision system
1999, Real-Time Imaging
Real-time computation of exact depth is not feasible in an active vision setup. Instead, reliable relative depth information which can be rapidly computed is preferred. In this paper, a stereo cue for computing relative depth obtained from an active stereo vision system is proposed. The proposed stereo cue can be computed purely from the coordinates of points in the stereo pair. The computational cost required is very low. No camera calibration or prior knowledge of the parameters of the stereo vision system is required. We show that the relationship between the relative depth cue and the actual depth in the three-dimensional (3D) space is monotonic. Such a relation is maintained even when the focal length and the vergence angle are changed, so long as the focal lengths of the two cameras are similar. Therefore, real-time implementation in an active vision setup can be realized. Stability analysis shows that the proposed method will be stable in practical situations, unless the stereo camera diverges. Experimental results are presented to highlight the properties and advantages of the proposed method.
A computational model of depth perception based on headcentric disparity
1998, Vision Research
It is now well established that depth is coded by local horizontal disparity and global vertical disparity. We present a computational model which explains how depth is extracted from these two types of disparities. The model uses the two (one for each eye) headcentric directions of binocular targets, derived from retinal signals and oculomotor signals. Headcentric disparity is defined as the difference between headcentric directions of corresponding features in the left and right eye’s images. Using Helmholtz’s coordinate systems we decompose headcentric disparity into azimuthal and elevational disparity. Elevational disparities of real objects are zero if the signals which contribute to headcentric disparity do not contain any errors. Azimuthal headcentric disparity is a 1D quantity from which an exact equation relating distance and disparity can be derived. The equation is valid for all headcentric directions and for all binocular fixation positions. Such an equation does not exist if disparity is expressed in retinal coordinates. Possible types of errors in oculomotor signals (six) produce global elevational disparity fields which are characterised by different gradients in the azimuthal and elevational directions. Computations show that the elevational disparity fields uniquely characterise both the type and size of the errors in oculomotor signals. Our model uses a measure of the global elevational disparity field together with local azimuthal disparity to accurately derive headcentric distance throughout the visual field. The model explains existing data on whole-field disparity transformations as well as hitherto unexplained aspects of stereoscopic depth perception.
Optical Flow Estimation Using Discontinuity Conforming Filters
1997, Computer Vision and Image Understanding
Discontinuities and large image displacements pose some of the hardest problems in flow estimation. This paper proposes a set of filters that change shape to avoid blending of the constraints across discontinuity boundaries according to an incompatibility measure of the constraints of neighboring pixels. The algorithm is embedded in a coarse to fine multigrid scheme to address the problem of large displacements. We report results on real and synthetic images which show that the algorithm works very well.

View all citing articles on Scopus

View full text

Qualitative depth from stereo, with applications

Abstract

Vision Res.

Vision Res.

Vision Res.

Finding the orientation of a cone or cylinder

The interpretation of a moving retinal image

Effects of voluntary eye movement and convergence on the binocular appreciation of depth

Percept. Psychophys.

Robot Vision

A computer algorithm for reconstructing a scene from two projections

Nature