Author

Vakalis

Date of Award

4-1988

Degree Name

Master of Science

Department

Computer Science

First Advisor

Dr. Dionysios Kountanis

Second Advisor

Dr. Donna Kaminski

Third Advisor

Dr. Robert Trenary

Access Setting

Masters Thesis-Open Access

Abstract

The problem of reconstructing a three dimensional object from its two dimensional image is the most critical problem which any powerful visual system must solve. The human visual system is known to solve this problem with impressive range and resolution. Computer solutions have fallen far short of human performance mainly because the solution process is poorly understood.

The reconstruction problem is solved using stereo vision. The distance of points on the object from the viewing device and the orientation of specific patches of the visual surface with respect to the viewing device are computed. Discontinuities in distance and orientation are directly labelled and identified before the full representation is completed. A different approach to the interpolation problem is given, preserving the "wobbling" of the reconstructed visual surface as the viewpoint changes. The algorithm presented is able to reconstruct the whole surface of the object, even if it is not initially visible by the system.

Share

COinS