Multimodal sensor system for event detection at hot spots
The subject research project will break new ground by offering a completely new, not previously realized approach. The event detector will analyze optical signatures from video images, thermal images and noises altogether
Crime accumulation points – so called “hot spots” – have to be permanently monitored. At these hot spots crimes, such as theft, acts against the controlled substances law, personal injury with affray, property damage, moral hazards to persons under 16 as well as dangerous threats, rape and violations of the prohibition laws, are becoming more and more frequent. In order to keep this crime rate under control, it is absolutely necessary to use automatic detection systems. Thus, valuable staff resources can be better instated and furthermore suspicious activities are detected earlier and more likely.
The subject research project will break new ground by offering a completely new, not previously realized approach. The event detector will analyze optical signatures from video images, thermal images and noises altogether. By specific signal preprocessing, extraction of characteristic features from each sensor, sensor data fusion and classification of vectors, reliable decisions whether or not to set off the alarm are taken. The event detector will not replace but assist a human operator, and is expected to achieve a better "hit rate" than the pure video detection. The officer has a picture of the situation in front of him so that he can get immediately an overview over all events and ongoings.
Univ.-Doz. Dipl.-Ing. Dr. Heinrich Garn, M.Sc.
AIT Austrian Institute of Technology GmbH
Other Project or Cooperation partner:
Ulz Consulting GmbH
Bahnstraße 100, 7091 Breitenbrunn
Kapsch BusinessCom AG
Wienerbergstraße 53, 1120 Wien
vertreten durch den/die Bundesminister/in für Inneres
Abteilung IV/1 Technische Ausrüstung
Hohenbergstraße 1B, 1120 Wien