Research On Key Technologies Of License Plate Character Recognition System

The license plate recognition system(LPR)identifies license plate images through computer technology and image processing technology,it can obtain license plate number and can store number in the database,which is convenient for management and convenient.This paper studies the algorithms used by each module of the system and compares their advantages and disadvantages.Finally,the algorithm used by each module is determined,and programmed in the MATLAB environment,the specific functions of license plate position,character segmentation and character recognition of this system are realized,and the license plate recognition system was completed.1.In license plate position part,the morphological features and region of the image are experimented.However,because of the influence of redundant factors,the license plate position method based on features of color is combined to make the effect of these two method added to get license plate image.2.The obtained picture of plate is slightly deformed due to problems such as the shooting angle,and the inclined license plate is performed tilt correction by Radon transform.The resulting image is then subjected to character segmentation.The license plate is projected according to characteristics of each character and interval between each character and frame,and then license plate is divided into separate character images according to the projection result,and the characters are normalized.3.In the part of license plate character recognition,the template matching method and BP neural network method are adopted.When the BP neural network method is used to recognize the license plate characters,a hidden layer is added based on the original traditional network,which effectively improves the recognition rate.The results of the two algorithms are compared and analyzed.The BP neural network method is better than template matching method.