Design And Implementation Of License Plate Recognition System Based On HAAR Cascade

With the increase in the motor vehicles and drivers in China,the traffic volume and vehicle load on highways are increasing,causing serious problems of urban traffic jam.Therefore,solving the problem of urban traffic jam becomes an urgent thing,which makes the rapid development of intelligent transportation systems become a trend.Intelligent transportation systems is as follow: advanced traffic management systems,electronic charging systems,emergency rescue systems,and so on.One of the most important technologies in the intelligent transportation system is the license plate recognition system.The license plate recognition system is an application of computer video image recognition technology in vehicle license plate recognition.Its main function is the License plate information identification in different complex backgrounds.The design flow of the traditional license plate recognition system is image processing,license plate positioning,character segmentation and character recognition.In order to enhance the accuracy of the measurement results,2 to 5 algorithms are used in almost every module.In order to simplify the program,the research of the license plate recognition system is divided into four modules: license plate coarse positioning,image processing,license plate fine positioning,character segmentation and recognition.The redundancy of program is reduced,making it more compact and improving the accuracy of license plate positioning.The system uses the AdaBoost tree classifier learning method,with the regression algorithm,and uses a convolutional neural network to identify characters.The car license plate characters are divided and then recognized one by one.This not only improves the problem of recognizing the license plate in various unfavorable environments in the conventional license plate recognition method,but also improves the accuracy of license plate recognition.In the process of obtaining the vehicle image,due to the change of the angle between the camera and the license plate,the image of the captured vehicle is often tilted,resulting in the coarse positioning of the license plate,and the rectangular area of the detected license plate is incomplete,which affects the subsequent license plate character segmentation and recognition accuracy,this paper uses a method of recognizing the area based on the diagonal of the rectangular area.The default maximum tilt angle is 45°.This algorithm can effectively recognize the rectangular area of the license plate even the degree of tilt is large.Also improves the accuracy of license plate recognition.Finally,the system uses 50 test pictures to demonstrate the test results and the final run results of each module and the output values of each parameter.Through the test,the final results are still in line with the expected results,and the accuracy is about 90%.