The Study Of Segmentation And Recognition In The Vehicle License Plate Automatic System

With the increment of vehicle all over the world, the situation of city traffic is attracted the recognition of people. How to manage the traffic effectively is become the focus of people more and more. License Plate Recognition System is the computer vision and one of the important subjects for research that recognition technology employ in the field of intellectual traffic system, realize the intelligent important links of traffic administration.This topic is a part of License Plate Recognition System, has established one completely the license plate character recognition system which realizes by the software. This system will input the car license color image through recognition processing, the string of character which the output might edit. The entire treating processes divide into the image pretreatment, the license plate correction for grade, the character division and the character recognition four parts.The image pretreatment stage mainly studies the image gradation processing, denoising processing and the image binaryzation operation. In view of because reasons and so on illumination creates the image black and white is not distinct, this paper has used the improvement binaryzation algorithm, in the OSTU algorithm foundation, improved the image processing effect. The inclined correction stage mainly examines that tilt angle of car license image because of shooting angle and so on questions, first , then uses the method which two times transform to the image to carry on correction processing. In the character division stage, this paper mainly uses the way guarantee division success ratio which divides first and second confirms, then will divide the character image will make normalized processing, unified size. The final character recognition part first uses in mathematics morphology refinement processing, extraction the character skeleton, then use template match method recognition character.This paper proved that, the improvement binaryzation algorithm can retain the good retention image detail, eliminate the noise, also has strong anti-chirp ability; Based on the Hough transformation seeking angle of tilt, after uses the method which two times transform to cause the distortion of correction the image to be small; After refinement processing, uses the template match method to be possible distinguish the car license character fast and accurately and so on.