Radar Descriptors for the Classification of Terrain Features.

Abstract

An approach toward the automated extraction of terrain features from synthetic aperture radar (SAR) imagery is the development of sets of descriptors that uniquely and unambiguously characterize each feature. This investigation involves a detailed examination of 701 SAR image examples covering 29 types of man-made and natural terrain features. The descriptors represent attributes of the radar signatures from terrain features. The descriptors are developed by means of which image analysts identify the terrain features. The development of descriptors is guided by the objectives that the descriptors can be easily recognized and identified by untrained personnel and that they provide a baseline for interactive and automated feature extraction. The number of selected descriptors at this point is 52 and may change as the research progresses. A feature class is characterized by specific sets of descriptors. Because radar signatures of the same type of terrain feature will vary it is necessary to establish combination rules for descriptor sets.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 16, 1987
Accession Number
ADA188145

Entities

People

  • Frederick W. Rohde

Organizations

  • Geospatial Research Laboratory

Tags

Communities of Interest

  • Human Systems
  • Space

DTIC Thesaurus Topics

  • Acquisition
  • Army Intelligence
  • Computer Vision
  • Data Acquisition
  • Detection
  • Expert Systems
  • Feature Extraction
  • Graphics
  • Identification
  • Image Processing
  • Materials
  • Mental Processes
  • Radar
  • Radar Images
  • Radar Signatures
  • Security
  • Synthetic Aperture Radar

Fields of Study

  • Computer science

Readers

  • Neural Network Machine Learning.
  • Sensor Fusion and Tracking Systems.
  • Systems Analysis and Design

Technology Areas

  • AI & ML