UACE2017 Proceedings: Feature Selection and Classification for False Alarm Reduction on Active Diver Detection Sonar Data

• Session:
  Towards Automatic Target Recognition. Detection, Classification and Modelling
• Paper:
  Feature Selection and Classification for False Alarm Reduction on Active Diver Detection Sonar Data
• Author(s):
  Matthias Buß, Stephan Benen, Dietmar Stiller, Dieter Kraus, Anton Kummert
• Abstract:
  A requirement for many modern active sonar applications is to detect a threat fully automatically without a human operator. Regarding this, the major challenge is to achieve a high probability of detection and a low false alarm rate at the same time. In active sonar signal processing, often only the signal-to-noise ratio (SNR) and sometimes also the Doppler of echoes are used for detection. However, echoes have further characteristics that can be used to assess their relevance and hence improve the detection performance. \nIn this paper, a method for false alarm reduction by classification of contacts is presented. Initially an overview about the categories of used features is introduced and their individual suitability to distinguish between target contacts and false alarms is shown. Furthermore, one feature selection method is applied to determine the best subset of features for classification. Finally, two different classification algorithms are investigated regarding their performance and robustness for reducing the number of false alarms. \nThe algorithms are applied to recorded data of diver detection trials that were carried out in cooperation between the WTD71 and ATLAS ELEKTRONIK. The trials were conducted with the "Cerberus" diver detection sonar developed by ATLAS ELEKTRONIK UK.
•   Download the full paper

Contact details

• Contact person:
  Mr Matthias Buß
• e-mail:
  
• Affiliation:
  Bergische Universität Wuppertal (University of Wuppertal)
• Country:
  Germany