ATR Performance Modeling and Estimation
Abstract
The purpose of this document is to assess the state of the art of performance modeling and estimation for synthetic aperture radar (SAR) automatic target recognition (ATR) algorithms. The study underlying this report is part of the Lincoln Laboratory effort under the OSD ATR Program. The intent is not to produce an exhaustive, detailed, voluminous report describing all ongoing efforts, but rather to capture in a succinct yet essentially complete way the approaches currently in use for modeling the performance of SAR ATR algorithms. To provide context for this document's assessment and recommendations, a brief discussion of the breadth of ATR problems and some of the resulting technical challenges will be conducted. This discussion will generally be couched in terms of the SAR ground surveillance problem of detecting and recognizing various military vehicles, though many of the statements are easily extended to other sensors and other applications. Basic performance metrics will be defined, and a discussion will ensue of the kinds of questions it would be useful to have addressed by a performance model. The principal approaches to modeling the performance of SAR ATR algorithms and estimating performance characteristics under a variety of conditions will then be outlined and assessed. The document concludes by recommending certain actions to encourage progress in the development of SAR ATR performance modeling and evaluation tools and methodologies.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 07, 1998
- Accession Number
- ADA357723
Entities
People
- Dan E. Dudgeon
Organizations
- Massachusetts Institute of Technology