Research On License Plate Recognition System Based On Character Area Segmentation

With the development of urbanization, automotive vehicle is getting to a part of our life day by day. The universalness of automotive vehicle bring us more convenience to our life, meanwhile it brings more difficulties to the trip and the management of the traffic. To make the automotive vehicle server better, people strengthen the management and invented the Intelligent Transport System. License Plate Recognition System (LPRS) is one of the most important kinds of Intelligent Transport System, LPRS always contains four steps: Digital Image Acquisition, Location of License Plate, Segmentation of Characters and Recognition of Characters. Segmentation of Characters area is the key step of this process; it is the bridge between the image data and the characters data. The result of LPRS generally depends on the quality of the segmentation algorithm. So the research on segmentation of character area is quite significant to the development of Intelligent Transport System.On the basis of researching the'License plate of motor vehicle of the People's Republic of China GA36-2007', this article has designed the overall architecture of segmentation of character area and made the requirements to the step of Location of License Plate .For the pretreatment of image, this article chose to deal with the grayscale images. This article didn't chose those common methods as image smoothing, image sharpening and edge detection, but adopt more simple methods to keep the integrality of the image data and to decrease the complexity of the system. The pretreatment part of this article contains four steps as: grayscale of image, binaryzation of image, conditional negative phase transformation and plate tilt correction. The article chose the method of iteration threshold value for binaryzation and Hough transform method for tilt correction.The segmentation of character area part of this article was divided into three parts: the first is removing the borders and rivets of the license plate, followed by the primary segmentation and the detailed segmentation. The primary segmentation is added to solve the problem of bistratal plates, which is different from the traditional methods. To give the solution of the Chinese characters, the article adopts dividing the image form right to left and setting a counter to accumulation, remaining the last Chinese character.This article is aimed at designing a segmentation algorithm with strong universality, real-time and veracity. This algorithm gave a segmentation solution to the plate of compact cars, large cars, police cars, embassies cars and motor cars. Regretfully, the method can't meet the requirement of the plate of armed police cars and the military cars. To achieve the aim of universality this algorithm had abdicated some cost of time, but it also can meet the requirement of the instantaneity of the system. Basically, this article has met the anticipative purpose.