Monitoring traffic by optical sensors

Optical systems are well suited for traffic observation and management. The real-time requirements can be met by implementation of appropriate image processing algorithms in hardware. The use of programmable logic structures offers advantages of high flexibility and low costs. Instead of transmitting a lot of images only the data of extracted objects has to be transferred from the cameras to local and to regional computer networks. Primary and secondary traffic features can be calculated from the object information and further sources as weather forecast, geographical information or knowledge about automobile characteristics. A variety of problems can be solved in this way, e.g., dynamic local and wide area traffic signal management, or optimisation of traffic flow in busy periods. The design of such a system is very complex; it is still under development. Three examples of image processing procedures are given: sensor fusion, stereo image analysis by epi-polar geometry, and estimation of traffic related air pollution.