Development of Algorithms and Hardware for Structured-Light Vision.
Abstract
Computer vision is finding increased application in robotics and for shipbuilding, construction, machining, sorting, and inspection. An essential part of a computer vision system is the ability to measure range (distance) to points on the surfaces of three-dimensional objects. This information allows the description of such intrinsic characteristics of objects as orientation, surface condition, and dimensions. A goal of the work described here was the development, testing, and evaluation of a real-time system for measuring range, suitable for the kinds of applications described above. A second goal was the development of algorithms for description of objects, based on their range-image representations. We have shown the feasibility, and measured the performance, of a prototype ranging system that uses a technique called structured light; GWU has also derived and implemented algorithms for rapid description of a limited class of objects using only the data that would be supplied by such a system. Note: Included are a FORTRAN computer program and the following papers: (1) PIPE: A Specialized Computer Architecture for Robot Vision; (2) A New Method for Characterization of Shape; (3) The Quadcode and Its Arithmetic; (4) Adjacency Detection Using Quadcodes.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1986
- Accession Number
- ADA168197
Entities
People
- Gao Shangkai
- James Hargrove
- Murray H. Loew
- Reza Momenan
Organizations
- George Washington University