Classification experiments with the PONDEX and TREX13 datasets


  1. Fawcett, J.A.
  2. LeBlanc, T.
Corporate Authors
Defence Research and Development Canada, Atlantic Research Centre, Halifax NS (CAN)
The classification of an object from its scattered signal (echo) has been a research topic of much interest in the underwater sonar community. High-frequency sidescan or synthetic aperture sonars (SAS) can provide an image of the objects on the seabed but it can still be difficult to distinguish minelike objects from similar sized clutter objects (such as rocks, crab pots, barrels, etc). At high frequencies, the amplitude of the echo return from a target is dominated by its specular and diffraction returns. The relative amplitudes of these returns are in turn related to the geometrical properties of the scattering object. Hence objects with the same dimensions and texture will yield similar sonar images. Scattered signals at lower frequencies, than are typically used by imaging sonars, may contain useful classification information due to internal structural/elastic scattering effects. At lower frequencies the incident energy will penetrate more into the interior than in the high-frequency case. Recently DRDC – Atlantic Research Centre has acquired two, unclassified, wideband scattering data sets from the U.S., the PONDEX and TREX13 datasets. The data is from a number of clutter and Unexploded Ordnance (UXO) objects. The frequency of the scattered energy is approximately in the interval [4,28] kHz for a full 360° of aspects. In addition, modelled signals for a number of the objects were also provided. A number of binary-classification (clutter vs. UXO) and multi-class metho

Il y a un résumé en français ici.

Report Number
DRDC-RDDC-2018-R153 — Scientific Report
Date of publication
01 May 2018
Number of Pages
Electronic Document(PDF)

Permanent link

Document 1 of 1

Date modified: