DSP based acoustic vehicle classification for multi-sensor real-time traffic surveillance

Vehicles may be recognized from the sound they emit when driving along a road. Characteristic acoustic finger prints and audio features can be used to increase the ro-bustness of existing video based vehicle tracking and clas-sification algorithms. Using this information in a multi-sensor surveillance system helps to improve various param-eters such as recognition rates, detection times and robust-ness. We propose a two-fold approach, where vehicle detec-tion and classification are handled separately. We demon-strate the feasibility of the proposed method using outdoor audio sequences of traffic situations. 1