Further Development and New Concepts for Bionic Sonar. Volume 1. Software Processors.

Abstract

This report documents theoretical studies of models of echolocation signal processing of dolphins and the testing of the effectiveness of these models for detecting and classifying small objects in introduced noise and reverberation. This volume describes the theoretical and software development of three models. The first, bionic components, is based on recognizing target features predicted by the geometric theory diffraction. The second method was a principal component analysis of a spectrogram, and the third was the principal component analysis of a Fourier-Mellin transform. These algorithms have been tested by computing misclassification probabilities in noise and bottom reverberation. These tests indicate that the spectrogram model gives the best performance, with a probability of misclassification of < or = of 0.1 occurring for SNR > or = dB and SCR (signal to clutter ratio) > or = 6 dB. (Author)

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Oct 01, 1978
Accession Number
ADA072940

Entities

People

  • R. W. Floyd
  • Richard A. Altes
  • W. J. Faust

Tags

Communities of Interest

  • Energy and Power Technologies
  • Space

DTIC Thesaurus Topics

  • Algorithms
  • Biosonar
  • Counter Countermeasures
  • Databases
  • Detection
  • Detectors
  • Diffraction
  • Eigenvalues
  • Eigenvectors
  • Frequency
  • Ions
  • Pattern Recognition
  • Recognition
  • Reverberation
  • Scattering
  • Signal Processing
  • Target Classification

Readers

  • Atmospheric Science / Meteorology, specifically Wind Wave Turbulence.
  • Sensor Fusion and Tracking Systems.
  • Speech Processing/Speech Recognition.

Technology Areas

  • Biotechnology