The Application And Research Of Neural Network In License Plate Recognition

In recent years, image processing, communication technology andcomputer technology have been widely used in the transportation field, andlicense plate identification technology is one of the hot spot in the research.With the construction of high-grade residential area and the improvement ofpeople's living standard, more and more people move into high-gradevillage, and there are more and more cars come into our daily lives. Thehigh quality property management level to people's daily life, it is veryimportant to the security of the respondent in district residents andconvenience plays an important role. District vehicle intelligentmanagement is an important part of property management, and license platerecognition technology is the key technology of intelligent management ofvehicle license plate recognition accuracy, directly determine these systemswork performance and application effect. Therefore, studying a kind ofefficient and high accuracy of license plate recognition technology hasimportant practical value.Based on a lot of documents and data analysis and comparison on thebasis of the image processing, license plate localization, license platecharacter segment and the field of license plate character recognition,finally several methods to optimize the induction and feasible scheme. Inthe vehicle plate recognition system in each specific steps required toachieve, through specific processes the effect of the algorithm and methodfor a lot of the reasonable selection and improvement, and use Matlabsimulation is conducted to important steps, and by comparison got us afterimprovement and optimization of important data and parameters.This paper focus on license plate identification technology research,license plate pre-processing verified the gray change, filtering, binary andedge detection, and the main algorithms based on Roberts operator of thelicense plate pre-processing algorithm experimental results were analyzed;This paper introduces some common license plate localization andsegmentation technology, the simulation tests the vertical segmentationmethod validity; Based on artificial neural network is introduced, thelicense plate identification method for China respectively from the license plate feature BP neural network design, and character characteristicextraction, sample training discussed.