Study Of Vehicle License Plate Recognition

License plate recognition system is an important part of the current intelligent transportation system, which is generally used at highway bayonet charges, car park, petrol filling stations, automated monitoring, management to capture traffic intersections and other irregularities. License Plate Recognition system includes three parts:plate pretreatment and positioning,character segmentation and identification of a single character. License plate location and character are the difficulties dealt in the identification license plate recognition technology. This article on the license plate recognition technology in all processes discusses deeply, and explores pretreatment license plate, license plate location and license plate recognition process in some of the processing technology.Firstly, a variety of current license plate image preprocessing techniques used in the vehicle license plate location are meticulously analyzed and compared; Then an improved mix of space law is proposed; Finally, a variety of image preprocessing techniques are used to preprocess the image on the plate. Through comparison of a variety of experimental results of pretreatment, the use of improved hybrid space after the license plate image pretreatment not only remove Gaussian noise, but also more clearly show the edges of the plate, which has a very good preparation for the follow-up plate location.As it is better to eliminate the Gaussian noise by using of preconditioning methods for improving the mix of space, image edge information can be kept completely and the license plate can be clearly and accurately located by using Prewitt operator, image segmentation method and the method of combining morphological processing in the license plate location.In character recognition, the paper details the principles of neural network identification and the comparison of various networks, and introduces the structural model of BP algorithm, learning steps. For low character recognition rate and high error rate problem, Chinese character is input into module of identification of neural network to match, the letters and numbers is input into the template matching module to identify to match, which meet the requirements of accuracy and real-time at the same time.Finally, the paper uses matlab simulation to achieve the various algorithms. Simulation experiments include the license plate image preprocessing and orientation, the calibration plate, license plate character segmentation and character identification, which tested a lot of license plates pictures which are took really and got relatively satisfying results.