Distortion of Depth Perception in a Virtual Environment Application

by

Jonathan D. Pfautz

Submitted to the Department of Electrical Engineering and Computer Science in partial fulfillment of the requirements for the degrees of

Bachelor of Science in Computer Science and Engineering and
Master of Engineering in Electrical Engineering and Computer Science

at the

MASSACHUSETTS INSTITUTE OF TECHNOLOGY

May 28th, 1996

(c) Jonathan D. Pfautz, MCMXCVI. All rights reserved.

The author hereby grants to MIT permission to reproduce and distribute publicly paper and electronic copies of this thesis document in whole or in part, and to grant others the right to do so.

Certified by
Nathaniel I. Durlach
Senior Research Scientist, MIT Research Laboratory of Electronics
Thesis Supervisor


Accepted by
F. R. Morgenthaler
Chairman, Departmental Committee on Graduate Theses


Abstract:

Virtual Environment (VE) technology suffers from a variety of hardware and software problems that interfere with the ability to present the experience of being immersed in a synthetic world. Most notably, the poor quality of the viewing surfaces in head-mounted displays has led to many complications in the development of VE systems. These visual displays have a low spatial or pixel resolution that creates several side effects. Because the ability to present depth cues is hampered by the low resolution, the presentation of a sufficiently realistic three-dimensional world is very difficult. Depth perception is distorted by the lack of resolution in two ways. One, the threshold distance at which an object is just visible is much closer than in the real world. Two, the error and variance in estimating the depth of an object are significantly greater than in the real world. This thesis presents a series of mathematical analyses of these distortions, including a model of perspective geometry and the relevant human psychophysical biases. Algorithms are presented that improve both distorted threshold depth perception and distorted depth estimation. Experiments have been conducted to evaluate the usefulness of the algorithms in terms of both perceptual performance and training performance. The ability of the geometric models to predict human performance in the experiments serves as a measure of the models' effectiveness in devising solutions. The findings suggest that the visible range can be extended using the proposed algorithms without significantly affecting the ability to estimate depth.


Thesis Supervisor: Nathaniel I. Durlach
Title: Senior Research Scientist, M.I.T. Research Laboratory of Electronics

THIS IS STILL UNDER CONSTRUCTION: figures and formulae have not yet been added

Contents

1. Introduction

2. Background

3. Experiment

3.1 Experiment Background 3.2 Method 3.3 Results
3.4 Discussion

4. Future Work

5. Conclusion

References

Appendix A: Hark Grammar

Appendix B: Subject Instructions
  Main Page