License Plate Recognition Algorithm Based On BP Neural Network

With the rapid growth in the number of cars,how to solve the problem of traffic congestion has become the most difficult problem in traffic management.Therefore,intelligent transportation system should be born and become the most effective way to solve the traffic problem.License plate recognition system is an important application of intelligent transportation system,used in district intelligent management,electronic police and parking management and so on Therefore,in-depth study of license plate image recognition has a high social value.The paper includes five modules:license plate location,tilt correction,character segmentation,character recognition and GUI interface design.License plate positioning module based on color edge detection positioning method,the use of color jump point to determine the license plate area.The tilt correction module selects the Radon transform to correct the tilting license plate.The character segmentation module divides the character by looking for the maximum spacing of the license plate according to the spacing between the second character of the license plate and the third character,which is the maximum spacing among all license plate characters.In the process of segmentation,if the width of a character in the last 5 characters is too small,it can be determined that the character is 'l'.The character recognition module is a combination of template matching and BP neural network two character recognition methods.The template matching party is used to identify Chinese characters.Multiple BP neural networks cooperate with each other to identify letters and digital characters.GUI interface design module is the design of the license plate recognition system for interface design.The design of the license plate image recognition system with 60 car photos to test,successfully identify the 56 car license plate number.The test results show that the design of the license plate image recognition system can accurately identify most of the license plate number.