Research On Algorithm Of Key Technologies Of Vehicle Plate Recognition In Intelligent Transportation System

With the progress of technology and popularization of private vehicle, thenetworked vehicle management is imperative, and so is the ITS (IntelligentTransportation System). As a link of ITS, VLPR (Vehicle License Plate Recognition)plays a crucial role in it. With deep analysis and research on the key technology, thispaper designed an overall procedure of VLPR, and simulated the algorithm of it.The VLPR composes of three parts, which are license plate positioning, charactersegmentation and character recognition respectively. From these problems, the thesisstudied the related technologies and made progress in it, which mainly includes thefollowing:Firstly, the algorithm of license plate positioning is improved. The plate is roughlypositioned by morphology algorithm; priori knowledge is then used to determine thoseimpossible regions, vertical and horizontal projection method is finally used to obtainthe accurate position of car plate.Secondly, segmentation algorithm is improved. On the basis of the traditionalprojection segmentation algorithm, connected domain method is combined to get bettereffect. After the slant correction process, connected domain method is used to reducethe noise of image, which lays the foundation for more precise projection. Projectionmethod is finally used to remove the license frame and segment characters one by one.Finally, in character recognition part, on the foundation of classical BP neuralnetwork algorithm, additional momentum factors and self-adaptive learning rate isadded to the algorithm. The experimental results show the improving accuracy andefficiency of recognition of this algorithm.