Parallel Vision Algorithms
Annual technical rept. no. 2, 1 Oct 1987-28 Dec 1988
COLUMBIA UNIV NEW YORK DEPT OF COMPUTER SCIENCE
Pagination or Media Count:
This Second Annual Technical Report covers the project activities during the period from October 1, 1987 through December 31, 1988. The objective of this project is to develop and implement, on highly parallel computers, vision algorithms that combine stereo, texture, and multi-resolution techniques for determining local surface orientation and depth. Such algorithms can serve as front-end components of autonomous land vehicle vision systems. During the second year of the project, efforts have concentrated on the following first, implementing and testing on the Connection Machine the Parallel programming environment that will be used to develop, implement and test our parallel vision algorithms second, implementing and testing primitives for the multi-resolution stereo and texture algorithms, in this environment. Also, we continued our efforts to refine techniques used in our texture algorithms, and to develop a system that integrates information from several shape-from-texture methods. This report describes the status and progress of these efforts. We describe first the programming environment implementation, and how to use it. We summarize the results for multi-resolution based depth interpolation algorithms on parallel architectures. Then, we present algorithms and test results for the texture algorithms. Finally the results of the efforts of integrating information from various shape-from-texture algorithms are presented.
- Computer Programming and Software