Application of Wavelets to Automatic Target Recognition.

Abstract

'Application of Wavelets to Automatic Target Recognition,' is the second phase of multiphase project to insert compactly supported wavelets into an existing or near-term Department of Defense system such as the Longbow fire control radar for the Apache Attack Helicopter. In this contract, we have concentrated mainly on the classifier function. During the first phase of the program ('Application of Wavelets to Radar Data Processing'), the feasibility of using wavelets to process high range resolution profile (HRRP) amplitude returns from a wide bandwidth radar system was demonstrated. This phase obtained fully polarized wide bandwidth radar HRRP amplitude returns and processed, them with wavelet and wavelet packet or (best basis) transforms. Then, by mathematically defined nonlinear feature selection, we showed that significant improvements in the probability of correct classification are possible, up to 14 percentage points maximum (4 percentage points average) improvement when compared to the current classifier performance. In addition, We addressed the feasibility of using wavelet packets' best basis to address target registration, man made object rejection, clutter discriminations, and sythetic aperture radar scene speckle removal and object registration.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Mar 01, 1995
Accession Number
ADA294854

Entities

People

  • Charles Stirman

Organizations

  • Martin Marietta

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Attack Helicopters
  • Classification
  • Data Processing
  • Defense Systems
  • Department Of Defense
  • Feature Selection
  • Fire Control Radar
  • Machine Learning
  • Radar
  • Recognition
  • Target Recognition

Readers

  • Image Processing and Computer Vision.
  • Radar Systems Engineering.