A Parallel Pipeline Based Multiprocessor System For Real-Time Measurement of Road Traffic Parameters

Abstract

R eal-time measurement and analysis of road tra c¯ow parameters such as volume, speed and queue are increasingly required for tra c control and management. Image processing is considered as an attractive and¯exible technique for automatic analysis of road tra c scenes for the measurement and data collection of road tra c parameters. In this paper, the authors describe a novel image processing based approach for analysis of road tra c scenes. Combined background di erencing and edge detection techniques are used to detect vehicles and measure various tra c parameters such as vehicle count and the queue length. A RISC based multiprocessor system was designed to enable real-time execution of the authors algorithm. The multiprocessor system has nine processing modules connected in a parallel pipeline fashion. Results shows that the authors multiprocessor system is able to provide measurement of tra c parameters in real-time. Results are presented for real tests of our system by analysing tra c scenes on the highways of Singapore.