Framework for Automatic Target Recognition Optimization

Abstract

We have designed a framework for the optimization of Automatic Target Recognition (ATR) algorithms. Successful ATR algorithms are complex, with non-linear components and feedback between components, and thus do not lend themselves to traditional analytical optimization methods. A prototype of the designed framework has been implemented with a visual programming interface that simultaneously aids design decisions and provides opportunities for improvements and optimizations. This framework is applicable to individual algonthms, groups of algorithms, and whole ATR suites. The framework can accommodate larger systems where the ATR algonthm is but one part; it is also possible to embed the framework into a larger system. We established concept feasibility in Phase I, which specified a design and implemented a prototype for the ATR optimization framework entirely in Java. The Phase I effort included a built-in ATR taxonomy to aid algorithm design and successfully demonstrated algorithm optimization.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Oct 31, 1997
Accession Number
ADA332739

Entities

People

  • David Shue
  • Harald Ruda
  • Magnus Snorrason

Tags

Communities of Interest

  • Energy and Power Technologies
  • Materials and Manufacturing Processes
  • Sensors
  • Weapons Technologies

DTIC Thesaurus Topics

  • Artificial Intelligence
  • Artificial Intelligence Software
  • Bayesian Networks
  • Computational Science
  • Computer Languages
  • Computer Programming
  • Computer Vision
  • Computers
  • Detection
  • Detectors
  • Neural Networks
  • Ontologies
  • Pattern Recognition
  • Supervised Machine Learning
  • Target Recognition
  • Three Dimensional
  • Two Dimensional

Readers

  • Computer Vision.
  • Operations Research
  • Software Engineering