Symbolic Image Understanding
Abstract
This report summarizes the work done at the University of Massachusetts in the area of Image Understanding of static scene domains. The goal of this research project was to demonstrate a practical knowledge-based approach to computer vision that utilized multiple levels of parallel processing. Some problems dealt with included extracting lines, grouping together regions and lines from the same surface, computing pose from geometry, and matching corresponding features in different views. Special attention was also given to the extraction of curved lines and curved surfaces. The static scene domains consisted of both ROAD and HOUSE scenes, each containing approximately fourteen (14) different objects per scene.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1991
- Accession Number
- ADA245661
Entities
People
- Al Hanson
- Ed Riseman
Organizations
- University of Massachusetts Amherst