Feature Selection for the Identification of Bauded Signals.

Abstract

This paper investigates some of the difficulties commonly encountered in the application of pattern recognition techniques to the identification of non-cooperative signals. Practical problems of feature selection and analysis are explored for a data base of actual bauded signal transition times. Analysis of a set of features based on the raster display of these transition times suggests that the use of fast Fourier transform techniques may yield useful identifying parameters not exploitable by the current process of visual comparison of signal raster patterns.

Document Details

Document Type
Technical Report
Publication Date
Mar 01, 1975
Accession Number
ADA009974

Entities

People

  • Thomas Arthur Reglein

Organizations

  • Naval Postgraduate School

Tags

DTIC Thesaurus Topics

  • Databases
  • Fast Fourier Transforms
  • Feature Selection
  • Identification
  • Pattern Recognition
  • Recognition
  • Transitions

Fields of Study

  • Engineering

Readers

  • Computer Vision.
  • Radio communications and signal processing.

Technology Areas

  • AI & ML