Application of Fourier-Based Features for Classification of Synthetic Aperture Radar Imagery

Abstract

A method for segmenting synthetic aperture radar (SAR) images has been developed to operate primarily in the frequency domain. It is based on and was tested against a similar method which involves isolating information of the frequency-domain image that defines unique textural features within a class. The comparison involved classifying four simple vegetation SAR scenes with both segmentation methods. A statistical test was then performed against the null hypothesis that the new textural segmentation method is as accurate or more accurate than the original method based on random pixel classification results. All tests concluded that the texture extraction methods are not statistically different. Both methods were implemented on a mainframe computer and are computationally intensive, but the new method may be implemented optically more easily.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 14, 1992
Accession Number
ADA270450

Entities

People

  • David G. Ehrhard

Organizations

  • Air Force Institute of Technology

Tags

Communities of Interest

  • Air Platforms
  • Energy and Power Technologies
  • Space

DTIC Thesaurus Topics

  • Bandpass Filters
  • Computers
  • Detection
  • Digital Images
  • Frequency Domain
  • High Resolution
  • Image Classification
  • Image Processing
  • Information Processing
  • Information Science
  • Radar
  • Radar Reflections
  • Radar Signals
  • Remote Sensing
  • Statistical Tests
  • Synthetic Aperture Radar
  • Urban Areas

Readers

  • Computer Vision.
  • Control Systems Engineering.