Research And Realization Of The Vehicle License Plate Automatic Recognition System

In recent years,with the development of the computer and the intelligence traffic technology,vehicle license plate recognition is regarded as a branch of the picture recognition technology,and it is applied extensively day by day.With the development of the expressway and vehicle management system,it provides with a chance to the intelligence traffic system base on picture handle can be used on the practical application.Vehicle license plate automatic recognition is just developed under this kind of application background.Nowadays,some vehicle license plate recognition products have been put into market in China.However they can only achieve a high recognition rate base on the hardware.The problem is solved from three subsystem which exist the vehicle license plate recognition system.They are license plate location,character segmentation and character recognition,which can be adaptable to the environment.Meanwhile,the design on the application program is completed,and has a practical application value.For the license plate location problem,the thesis puts forward a new method to measure the edge base on the two value picture,which breaks through two defects of the traditional edge detection method,and the more abundant characteristic information on the level and vertical direction is withdrawn.This method stresses the vehicle license plate on the edge picture and locates the vehicle license plate. Meanwhile,the color and size characteristic of vehicle license plate is used to inspect the vehicle license plate which has located,increasing the rate of the license plate location.For the character segmentation problem,the thesis provides a new correcting slant method about the problem of the more and more metal aluminum frames installed on vehicle license plate,and the parallel can not be withdrawn form the edge picture.This method is called the horizontal line projection,it has a good fault-tolerant and the vehicle license plate can be corrected whether the metal aluminum frame exists or not in the actual conditions.On the basis of traditional character segmentation,this thesis puts forward the concept of projective characteristic value and a new character segmentation algorithm base on the concept of projection eigenvalue,fully utilizing the priori knowledge of vehicle license plate, this algorithm has an ideal effect to the character segmentation.For the character recognition problem,the vehicle license plate can be imperfection or contamination because of some factors on function of CCD camera, mud on the vehicle license plate and light.In order to improve the rate of correctly recognizing character,the thesis expounds a new vehicle license plate recognition method base on kinds of license-plate characters combining together,which includes three characters,they are the statistics characteristic,structure characteristic, projective characteristic,and then the number sorter,the alphabet sorter and Chinese character sorter are used to attain to the characters characteristic,and lastly,the thesis use a method of fuzzy match to recognize character,this method increases the recognition rate.