NEURAL-VISION BASED APPROACH FOR REAL-TIME ROAD TRAFFIC APPLICATIONS

The authors describe a novel neural network and edge based image proce ssing approach for road traffic analysis. An edge detection technique to detect vehicles is used, while a back propagation neural network is used to track and count vehicles. The neural network is trained for v arious road traffic conditions and is able to analyse complex traffic conditions better than the heuristic approach. The results show that t his approach provides better results than the traditional image proces sing techniques.