skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Feature discovery via neural networks for object recognition in SAR imagery

Abstract

A two-stage self-organizing neural network architecture has been applied to object recognition in Synthetic Aperture Radar imagery. The first stage performs feature extraction and implements a two- layer Neocognitron. The resulting feature vectors are presented to the second stage, an ART 2-A classifier network, which clusters the features into multiple target categories. Training is performed off-line in two steps. First, the Neocognitron self-organizes in response to repeated presentations of an object to recognize. During this training process, discovered features and the mechanisms for their extraction are captured in the excitatory weight patterns. In the second step, Necognitron learning is inhibited and the ART 2-A classifier forms categories in response to the feature vectors generated by additional presentations of the object to recognize. Finally, all training is inhibited and the system tested against a variety of objects and background clutter. In this paper we report the results of our initial experiments. The architecture recognizes a simulated tank vehicle at arbitrary azimuthal orientations at a single depression angle while rejecting clutter and other object returns. The neural architecture has achieved excellent classification performance using 20 clusters. 14 refs.

Authors:
; ; ;  [1];  [2]
  1. Sandia National Labs., Albuquerque, NM (United States)
  2. New Mexico Univ., Albuquerque, NM (United States). Dept. of Electrical and Computer Engineering
Publication Date:
Research Org.:
Sandia National Labs., Albuquerque, NM (United States)
Sponsoring Org.:
USDOE; USDOE, Washington, DC (United States)
OSTI Identifier:
5727446
Report Number(s):
SAND-92-0134C; CONF-9206102-1
ON: DE92007046
DOE Contract Number:  
AC04-76DP00789
Resource Type:
Conference
Resource Relation:
Conference: International joint conference on neural networks, Baltimore, MD (United States), 7-11 Jun 1992
Country of Publication:
United States
Language:
English
Subject:
42 ENGINEERING; 99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE; SYNTHETIC-APERTURE RADAR; PATTERN RECOGNITION; AERIAL SURVEYING; COMPUTERIZED SIMULATION; IMAGE PROCESSING; NEURAL NETWORKS; MEASURING INSTRUMENTS; PROCESSING; RADAR; RANGE FINDERS; SIMULATION; 420200* - Engineering- Facilities, Equipment, & Techniques; 990200 - Mathematics & Computers

Citation Formats

Fogler, R J, Koch, M W, Moya, M M, Hostetler, L D, and Hush, D R. Feature discovery via neural networks for object recognition in SAR imagery. United States: N. p., 1992. Web.
Fogler, R J, Koch, M W, Moya, M M, Hostetler, L D, & Hush, D R. Feature discovery via neural networks for object recognition in SAR imagery. United States.
Fogler, R J, Koch, M W, Moya, M M, Hostetler, L D, and Hush, D R. Wed . "Feature discovery via neural networks for object recognition in SAR imagery". United States.
@article{osti_5727446,
title = {Feature discovery via neural networks for object recognition in SAR imagery},
author = {Fogler, R J and Koch, M W and Moya, M M and Hostetler, L D and Hush, D R},
abstractNote = {A two-stage self-organizing neural network architecture has been applied to object recognition in Synthetic Aperture Radar imagery. The first stage performs feature extraction and implements a two- layer Neocognitron. The resulting feature vectors are presented to the second stage, an ART 2-A classifier network, which clusters the features into multiple target categories. Training is performed off-line in two steps. First, the Neocognitron self-organizes in response to repeated presentations of an object to recognize. During this training process, discovered features and the mechanisms for their extraction are captured in the excitatory weight patterns. In the second step, Necognitron learning is inhibited and the ART 2-A classifier forms categories in response to the feature vectors generated by additional presentations of the object to recognize. Finally, all training is inhibited and the system tested against a variety of objects and background clutter. In this paper we report the results of our initial experiments. The architecture recognizes a simulated tank vehicle at arbitrary azimuthal orientations at a single depression angle while rejecting clutter and other object returns. The neural architecture has achieved excellent classification performance using 20 clusters. 14 refs.},
doi = {},
url = {https://www.osti.gov/biblio/5727446}, journal = {},
number = ,
volume = ,
place = {United States},
year = {1992},
month = {1}
}

Conference:
Other availability
Please see Document Availability for additional information on obtaining the full-text document. Library patrons may search WorldCat to identify libraries that hold this conference proceeding.

Save / Share: