Accession Number : AD0717754

Title :   Temporal Waveform Classification Using Selected Pattern Recognition Techniques.

Descriptive Note : Final rept.,

Corporate Author : INFOTON INC BURLINGTON MASS

Personal Author(s) : Henrichon,Ernest G. , Jr

Report Date : 03 NOV 1970

Pagination or Media Count : 93

Abstract : The report describes efforts aimed toward solving a particular acoustical waveform classification problem through the application of advanced pattern recognition procedures. The problem under investigation is similar to many surveillance and reconaissance tasks which may require special processing of such media modulatable energy signals as acoustic, electro-magnetic, seismic and electrostatic. The primary concern of this investigation was the application of classification procedures, as opposed to the development of pertinent feature extraction methods. The list of classification techniques tested included the Wiener Canonical Expansion Procedure, the Exponentially Mapped Past Method and the Empirical Processing Algorithm. The results of this investigation clearly indicate that, no matter what classification approach used, the end results are limited by the information contained in the selected feature set. (Author)

Descriptors :   (*SONAR SIGNALS, *PATTERN RECOGNITION), (*INFORMATION THEORY, PATTERN RECOGNITION), ELECTROMAGNETIC RADIATION, SEISMIC WAVES, CLASSIFICATION, MULTIVARIATE ANALYSIS, FOURIER ANALYSIS, DATA PROCESSING, DOPPLER EFFECT, CURVE FITTING, SPECTRUM ANALYZERS, DECISION MAKING, COMPUTER PROGRAMMING, STOCHASTIC PROCESSES

Subject Categories : Cybernetics
      Acoustic Detection and Detectors

Distribution Statement : APPROVED FOR PUBLIC RELEASE