Object Recognition Using Range Images.
Abstract
This study involved forming synthetic range images and investigating correlation as a target recognition technique. The synthetic range images were formed with a computer program, which models the laser scanning of an air-to-ground missile. With this range imaging program, a model of a Russian T-72 tank was formed, which was subsequently used in investigating image correlation. In the investigation into image correlation, the effects of background clutter and target rotation on a range image's correlation coefficient were examined, as well as possible methods of correcting for these effects. Other factors affecting the correlation coefficient that were considered were pixel dropouts and the beam spot size of the laser. Pixel dropouts were shown to be detrimental to a range image's correlation coefficient, but could be corrected by using a 'median replacement' technique. Also shown was that for a certain of 'on target' beam diameters, as the diameter of the scanning laser beam increased, so did the correlation's discerning ability.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1985
- Accession Number
- ADA167148
Entities
People
- Jeffrey W. Grantham
Organizations
- Air Force Institute of Technology