Implementation of a C/C++ Library for Sonar Data Feature Extraction

Status: abgeschlossen
Betreuer: Horst Hellbrück , Torsten Teubler
Student: Matthieu Sion


In the context of the BOSS (Bionic Observation and Survey System) project we are developing an autonomous, intelligent, and cooperative behavior for AUV (Autonomous Underwater Vehicles). Therefore, we need an automatic examination and interpretation of sensor data. One sensor is the so called multibeam sonar (SOund NAvigation and Ranging) or multibeam echosounder (see multibeam echosounder on Wikipedia). Multibeam sonar is used to locate underwater objects and structures. Output of sonar systems are images built-up succesively while the sonar (e.g. on a ship) is moving towards. Sonar images are usually examined immediately by humans. As our aim is to automate the examination and interpretation we need a sonar data feature extraction.


There is already a plethora of methods and work done in the field of image feature extraction also for sonar images. In this work suitable methods for feature extraction have to be chosen. Then, a C/C++ library has to be implemented using the previously chosen methods for feature extraction from "Imagenex Model 837B" sonar data. Details of the implementation will be discussed together with the supervisor during the thesis work.


  • Getting familiar with the output data formats and then pick at least one suitable for feature extraction (using the Imagenex technical documentation).
  • Chosing suitable methods for sonar image feature extraction by a literature study.
  • Implementing a flexible C/C++ library for sonar data feature extraction.
  • For evaluation a command line program has to be implemented allowing configuration and use of the library.


  • Good knowledge of C/C++.
  • Very good skills in problem-abstraction.
  • Good self-organized working skills.
  • Knowledge of the English language.