earticle

영어

With high definition cameras are widely used in urban traffic monitoring, the resolution of vehicle and traffic monitoring image rises. As the main component of intelligent traffic system, the vehicle license plate recognition technique faces new challenges and some new demands are put forward. Aiming at the problems of cutting down the time delay and eliminating characters outside of vehicle license plate, a set of license plate recognition algorithm is developed in this paper. The process of recognition in this paper is divided into four parts: image preprocessing, license plate location, license plate segmentation and character recognition. In license plate location, an accelerating matching method based on particle image correlation method is employed. In character recognition, an improved BP neural network algorithm is employed. The principle, procedure and simulation of the algorithm is described in details. High definition images with 2 mega pixels are used as the input of simulation.