Compared with traditional traffic detectors, new digital imaging sensors provide many additional features, such as simple installation and maintenance, access to spatial traffic variables, monitoring of large areas, and visual control. The use of specific methods to analyse road traffic images is a very powerful tool for road network traffic management, and it is increasingly appreciated by its still relatively few users. With this growing market in mind, and after ten years of research on image-processing algorithms, INRETS, the French national transport technology research centre, joined other private sector investors to launch Citilog, a new company marketing traffic management solutions based on image processing. INRETS has also developed fast, specific methods adapted to applications. For example, sampling techniques and scene modelling enabled drastic reduction of the amount of information to be handled and real-time processing capability on small and inexpensive machines. In INRETS-Citilog solutions, the software learns to recognise vehicles after a training period. The automatic incident detection (AID) video system offers a good incident response level with real operational improvement, and provides very valuable information in the video sequence immediately before an incident.
Samenvatting