The University of Southampton
University of Southampton Institutional Repository

Three-dimensional target recognition via sonar: a neural network model

Three-dimensional target recognition via sonar: a neural network model
Three-dimensional target recognition via sonar: a neural network model
A neural network was trained to recognize two three-dimensional shapes independent of orientation, based on echoes of ultrasonic pulses similar to those used by an echolocating bat, Eptesicus fuscus. Following supervised learning, the network was required to generalize and recognize echoes from the shapes at novel orientations. The representation of the echo was manipulated to explore how information about target shape may be encoded in sonar echoes. Three types of input representations were used: time domain (waveform and cross-correlation), frequency domain (power spectrum), and time frequency (spectrogram). The probability of correctly recognizing the novel echoes by chance was only 25%. The network using the spectrogram representation recognized 90% of the echoes from novel orientations. The bat Eptesicus fuscus uses a multiple echolocation cry, and we explored the relative contribution of low and high frequencies for carrying information about target shape. We presented the network with low-pass and high-pass filtered spectrogram representations, preserving sound energy in frequency bands roughly corresponding to either the first or the second harmonic of the bat's echolocation sounds. The network was able to recognize 90% and 95% of the novel echoes using only the first or only the second harmonic, respectively. We continued by examining whether the network could perform the task using only time domain or only frequency domain information. When presented with a time waveform representation, the network was unable to perform the task. Similar results were obtained with other time domain representations, the cross-correlations between the emitted sound and its returning echo. However, when using only frequency domain information, the network was able to recognize 70% of the echoes from novel orientations. Again, we explored the relative contribution of the frequency bands corresponding to the first and second harmonics used by the bat Eptesicus fuscus. We found that the network was able to recognize 70% of the novel echoes using the first harmonic, and only 55 % of the novel echoes using the second harmonic. The results are discussed in light of studies on echolocation of bats and models of sonar processing.
modeling, representation, shape recognition, object recognition, sonar, echolocation, bats
149-160
Dror, Itiel E.
4d907da2-0a2e-41ed-b927-770a70a35c71
Zagaeski, Mark
c01e2f66-bbab-4a7d-95d6-a277f20c046f
Moss, Cynthia F.
5a92abf6-dfac-43f8-8efd-33d9d01b5318
Dror, Itiel E.
4d907da2-0a2e-41ed-b927-770a70a35c71
Zagaeski, Mark
c01e2f66-bbab-4a7d-95d6-a277f20c046f
Moss, Cynthia F.
5a92abf6-dfac-43f8-8efd-33d9d01b5318

Dror, Itiel E., Zagaeski, Mark and Moss, Cynthia F. (1995) Three-dimensional target recognition via sonar: a neural network model. Neural Networks, 8 (1), 149-160. (doi:10.1016/0893-6080(94)00057-S).

Record type: Article

Abstract

A neural network was trained to recognize two three-dimensional shapes independent of orientation, based on echoes of ultrasonic pulses similar to those used by an echolocating bat, Eptesicus fuscus. Following supervised learning, the network was required to generalize and recognize echoes from the shapes at novel orientations. The representation of the echo was manipulated to explore how information about target shape may be encoded in sonar echoes. Three types of input representations were used: time domain (waveform and cross-correlation), frequency domain (power spectrum), and time frequency (spectrogram). The probability of correctly recognizing the novel echoes by chance was only 25%. The network using the spectrogram representation recognized 90% of the echoes from novel orientations. The bat Eptesicus fuscus uses a multiple echolocation cry, and we explored the relative contribution of low and high frequencies for carrying information about target shape. We presented the network with low-pass and high-pass filtered spectrogram representations, preserving sound energy in frequency bands roughly corresponding to either the first or the second harmonic of the bat's echolocation sounds. The network was able to recognize 90% and 95% of the novel echoes using only the first or only the second harmonic, respectively. We continued by examining whether the network could perform the task using only time domain or only frequency domain information. When presented with a time waveform representation, the network was unable to perform the task. Similar results were obtained with other time domain representations, the cross-correlations between the emitted sound and its returning echo. However, when using only frequency domain information, the network was able to recognize 70% of the echoes from novel orientations. Again, we explored the relative contribution of the frequency bands corresponding to the first and second harmonics used by the bat Eptesicus fuscus. We found that the network was able to recognize 70% of the novel echoes using the first harmonic, and only 55 % of the novel echoes using the second harmonic. The results are discussed in light of studies on echolocation of bats and models of sonar processing.

This record has no associated files available for download.

More information

Published date: 1995
Keywords: modeling, representation, shape recognition, object recognition, sonar, echolocation, bats

Identifiers

Local EPrints ID: 40152
URI: http://eprints.soton.ac.uk/id/eprint/40152
PURE UUID: ac01f584-12dd-4da6-9d14-219a51d87a97

Catalogue record

Date deposited: 27 Jul 2006
Last modified: 15 Mar 2024 08:17

Export record

Altmetrics

Contributors

Author: Itiel E. Dror
Author: Mark Zagaeski
Author: Cynthia F. Moss

Download statistics

Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.

View more statistics

Atom RSS 1.0 RSS 2.0

Contact ePrints Soton: eprints@soton.ac.uk

ePrints Soton supports OAI 2.0 with a base URL of http://eprints.soton.ac.uk/cgi/oai2

This repository has been built using EPrints software, developed at the University of Southampton, but available to everyone to use.

We use cookies to ensure that we give you the best experience on our website. If you continue without changing your settings, we will assume that you are happy to receive cookies on the University of Southampton website.

×