The Study Of Vehicle License Plate Recognition Based On Digital Image Processing

Intelligent Transportation System(ITS), such as Automatic Traffic Signal Control System, Global Position System(GPS), Navigation System, Intelligent Residential Area Management System, plays a increasingly important role in the management of transportation. Because license plate is efficient in identifying vehicles, Vehicles License Plate Recognition System (VLPRS) is the main part of the contemporary ITS.On the status of present transportation system, some details of VLPRS are discussed and a simulator system can be applied in real situations in this paper. The system includes five parts, including image collection input, image preprocessing, license plate locating, characters getting and characters recognition. Image collection input module is mainly consisted of hardware such as cameras, switches, video card and cables. It monitors the video signal and cuts frame pictures at appropriate time. Image preprocessing module, license plate locating module, character getting module and characters recognition module, which are the key parts of the prototype of VLPRS, are realized by software. To realize them in a simpler way, it is coded in the environment of Matlab.In image preprocessing module, whole dynamic threshold value method is used in the binarization processes. And this paper proposes an improved image noise reduction using mathematical morphology. Canny Operator and Hough transformation are employed to adjust the gradient of image.In license plate locating module, the locating is realized in level direction in the way of scanning bean according to the frequency characters of the plate and the vertical direction is defined by using both the vertical projective image and frequency characters of the plate.Finally the license plate recognition is researched. In this paper, the image which formed after positioning, locating and character division is recognized and the characters is outputted. This process includes the establishment of the Character Feature Library, characters getting and characters recognition. The main recognition method based on the K-L transform is proposed to the characters of vehicle license plate. The experiments show that the recognition method proposed by the dissertation can attain a quite finer vehicles license plate recognition effect, whose precision rates are 87.9%.