Research On Vehicle License Plate Recognition

With development of information technology and intelligence technology,Intellectualization and informatization become development trend of traffic management system.License Plate Recognition(LPR) system is the core component of Intelligent Traffic System(ITS).It is very important in modern traffic management systems.Studies on key technology of LPR have recently became a focus in intelligent traffic field.This paper gives deep and comprehensive discussion on image pre-processing,plate location,character segmentation and character recognition included in LPR,and a new attempt on plate location is also discussed.Plate location is the first step in LPR,a wrong location will cause a wrong recognition.The method based on jumping features of horizontal gray scales is adopted in this paper.It includes two steps of sketchy locating and precision locating.In sketchy locating,a search from bottom to top combined with prior knowledge is applied;In precision locating,a multi-lines detection by using Hough transformation is executed.The quality of character segmentation effects the difficulty of character recognition.Before segmentation,the lower pixels are firstly separated from the image which has been binaried,then the inclination angle is detected by Hough transformation and then corrected.Projection method is improved to eliminate influences caused by rims and rivets.Character recognition is the core and also the purpose of LPR.According to features of character,BP neural network is used in this paper.A test is done after Character images which have been binaried are respectively trained.The results show that this method is correct,so the research target is reached.