Recognizing 3 D Objects from 2D Images Using Structural Knowledge Base of Genetic Views
Abstract
Model-based object recognition is an essential task for mobile robotics and assembly. Given an image of a scene containing one or more objects from unknown viewpoints, the goal is to efficiently recognize those objects for which there is sufficient evidence. At the University of Massachusetts, we are developing a model-based object recognition system which is capable of recognizing objects from a large data base of models and from arbitrary viewpoints. Contents: Overview; Extraction of Straight Lines; The View Sphere for Curved Surfaces; Reconstruction of Surfaces From Multiple Views; Predictions and the Prediction Hierarchy Compiler.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 31, 1988
- Accession Number
- ADA207875
Entities
People
- Allen R. Hanson
Organizations
- University of Massachusetts Amherst