Dissertation
Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device

The Research and Implementation on Character Recognition of License Plate Recognition

Author TangHongZhen
Tutor PanJianShou
School Northwestern University
Course Circuits and Systems
Keywords Pattern Recognition Feature Extraction BP neural network Support Vector Machine License plate character recognition
CLC TP391.41
Type Master's thesis
Year 2008
Downloads 695
Quotes 8
Download Dissertation

The license plate recognition system is a core part of the Intelligent Transportation Systems , violation detection , highway automatic toll collection and intelligent parking management plays an important role in the transport sector . The license plate recognition system , including the license plate location, character segmentation , license plate character recognition three parts , the license plate character recognition is one of the key technologies of the license plate recognition , is currently one of the hot topics of research in this field . In this paper, the license plate character recognition and related technologies , and research topics include : . Transition characteristics based on the shape of the plate area and grayscale positioning license plate location ; proposed an improved based on the the projector distance between the characters tilt correction method for correcting the tilt license plate , after the removal of the border on the license plate , license plate character segmentation using the projection method . License plate characters smoothing, normalization and other preprocessing . Extract improved elastic mesh feature numeric and alphabetic characters , Kanji characters , the strokes of a superimposed grid weighted feature extraction method to extract the directional line element feature . Focuses on character recognition method based on BP neural network , a three -layer BP neural network structure design , parameter settings , as well as the design of the training set method ; BP neural network has slow convergence , easy to fall into the local minimum of defects introduced momentum term and adaptive learning rate and steepness factor and integrated with the improved method . The method enables network convergence to more advantage of the convergence rate has also been accelerated to improve the recognition rate . . Focuses on the character recognition method based on support vector machines , the use of \SVM method in the case of feature extraction can get high recognition rate .

Related Dissertations
More Dissertations