Abstract
Abstract: (shorter) We designed and implemented a traffic monitoring system, which can be used to compute the micro- and macro-parameters that characterize a car traffic flow. We used a surveillance video camera for the image acquisition and we implemented our own application that detects and tracks the vehicles. For a specific video sequence we compute offline the average velocity of the vehicles. For the vehicle detection we used a robust algorithm based on feature extraction and learning. Once detected, a vehicle is tracked in the following video frames by computing the cross-correlation between the vehicle image and the image regions in a pre-defined search area. The maximum cross-correlation value indicates the new position of the vehicle. We present our results, conclusions and future plans.
Keywords: vehicle motion tracking, traffic monitoring system, cross-correlation.