Research Of Recognition Technology Of Vehicle VIN Based On Android Platform

With the increasing number of automobiles in recent years,business volume of automobile sales,lease,security check,maintenance,insurance claims,second-hand transactions and other auto services industry are also increasing.Vehicle identification number(VIN)as an automobile sole identification,is the basic information which shall be registered during the automobile services.At present,registering of the vehicle's VIN adopts the traditional manual mode,which is not only inefficient,but also is very prone to errors because of handwriting irregularity.To this end,research and development of vehicle VIN identification system based on the Android platform is proposed in this thesis.This thesis introduces the background and significance of the research project at first,and studies and analyzes the current situation of related technologies at home and abroad by reading a large number of related technical literatures,then puts forward the research contents of this project and analyzes the technical difficulties of the project.Secondly,the background knowledge of vehicle VIN is introduced,and the preprocessing process and principle of VIN original image are described emphatically.The aim is to reduce the noise in the image,keep the image characteristic information and improve the processing speed of the image.And the image preprocessing methods are simulated based on MATLAB,the feasibility and effectiveness of the scheme have been verified.In addition,the Java programming technology,Tesseract-OCR(optical character recognition)technology and the building process of Android development environment used in development of the vehicle VIN recognition system are introduced.Then,the work flow of the vehicle VIN recognition system is designed through analyzing the VIN identification operation process.The specific process is to call the Android camera to complete the location and acquisition image of the vehicle VIN.Next,the acquired original image of vehicle VIN is been processed by graying,Top-hat transformation,image enhancement and binarization.character features are extracted from the preprocessed image of the vehicle VIN based on the open source Tesseract-OCR identification library provided by Google,and match the vehicle VIN character library generated by the training.This dissertation also explains the training method of vehicle VIN character library in detail.In order to acquire correct result,the recognition outputs need to be verified automatically according to the VIN coding rules or corrected manually.Finally,system operation and functional testing for the vehicle VIN identification system designed and developed based on the Android platform are executed,during the testing,the system is in smooth operation,no system halt and breakdown occure;The desiged and developed identification system can realize the basic functions of vehicle VIN original image acquisition,preprocessing,recognition,automatic calibration and manual correction.The experimental results show that the image preprocessing method adopted and the vehicle VIN character library generated by training contribute to improve the recognition rate of vehicle VIN.In order to improve the recognition rate further,need to strengthen research about the image preprocessing algorithm and vehicle VIN character library training.