Share Email Print
cover

Proceedings Paper

Target recognition, identification, and tracking using real and synthetic IR imagery
Author(s): H. Tim A. Pentecost; Andrew M. Rogoyski; Douglas R. McCann; Andrew J. Maurice
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

In the development of automatic target detection (ATD) and automatic target recognition/identification (ATR/I) systems, the issue of image data is commonly given inadequate consideration. All too often a poorly managed collection of real data, or unrealistic synthetic data is used during development, resulting in a loss of performance when used in the field with imagery having different characteristics. Some of the most promising approaches to ATD and ATR/I, such as neural networks, are particularly susceptible to this problem due to their direct dependence on the training data. This paper highlights the issues involved, with reference to a generic detection and classification approach and to the use of real and synthetic infra-red imagery at the Defence Research Agency at Fort Halstead in the United Kingdom.

Paper Details

Date Published: 27 June 1996
PDF: 7 pages
Proc. SPIE 2744, Infrared Technology and Applications XXII, (27 June 1996); doi: 10.1117/12.243492
Show Author Affiliations
H. Tim A. Pentecost, Defence Research Agency Kent (United Kingdom)
Andrew M. Rogoyski, Logica UK Ltd. (United Kingdom)
Douglas R. McCann, Logica UK Ltd. (United Kingdom)
Andrew J. Maurice, Logica UK Ltd. (United Kingdom)


Published in SPIE Proceedings Vol. 2744:
Infrared Technology and Applications XXII
Bjorn F. Andresen; Marija S. Scholl, Editor(s)

© SPIE. Terms of Use
Back to Top