Design And Implementation Of License Plate Recognition System

Automatic License Plate Recognition System(ALPRS),which is a system with extensive uses and application prospects,is an important part in the intelligent traffic system.At present,there are many researchers who have been engaged in this field and achieved some positive results through efforts.This paper aims to make scientific researches on the four key technologies of ALPRS, namely,the license plate preprocessing,plate orientation,plate character segmentation and plate character recognition,as well as their respective realization methods.At the processing stage,the system firstly adopts the image denoising method with the combination of the piece-wise linear gray-scale stretch and Kalman filter during rainy days. The experimental results show that this is an effective method to help filter out the white noise in images.At the stage of the plate orientation and character segmentation after preprocessing, this paper adopts an image location method based on mathematical morphology and interconnected areas labeling,which conducts the image erosion by using 9×1 longitudinal probe to filter out the noises.Based on this,the system will perform the close operation on the images by using 19×17 probe according to the width-to-length ratio of the plate characters,in order to disconnect the interconnected areas the plate is in from the relatively independent ones that are possibly merged with it,and to segment the characters in the plate image through the algorithm for judging interconnected areas..The character recognition method,which is based on the improved BP Artificial Neutral Network,is introduced in this paper to design the plate recognition system.There are 128 nodes in the input layer,6 in the output layer,and 18 in the hidden layer on the network structure.The accuracy of capital letters and that of Arabic numbers are 91.85% and 96%respectively. An ALPRS was built on the Visual C++ 6.0 platform through the above-mentioned methods.In the end,a personal view on the further development of ALPRS was put forward in this paper.