Arithmetic Research On License Plate Location And Character Segmentation By Image Processing

LPR(License Plate Recognition) system is a very important component of ITS (Intelligent Transportations System). It can extract the license plate area, segment the characters and recognize them only by the pictures taken in the real traffic environment. It refers to much subject knowledge, such as digital image processing, mathematical morphology, pattern recognition and digital signal processing.It has a very wide application in different occasion. Of the whole process, license plate location and character segmentation are the most important and critical step.According to the research object characteristic and the requirement of real- time, this paper mainly discusses several portions of the LPR system.Before locating plate, it is necessary to pre-process the images in order to be convenient to the next steps.In this paper, it designs three main steps:grey transformartion, noise filtration and contrast enhancement.Through these processes,it improves the quality effectively and lays a good foudation for the final goal.As the plate location is concerned, there are two methods about it in this paper. The one is based on the principle that there are a lot of corners in plate area compared to the other areas. First, detect the corners by Harris arithmetic operators.Then, sort these corners by the square model and make them divided into different areas. Choose some candidate ones which are likely to be the plate area according to the plate prior knowleges. Further more, eliminate the false and retain the true and get the real plate area. The other method is making use of Roberts operators to detect the object edge. The plate area is full of alternate variety between the black and white pixels. Because of this, there are several candidate ones can be considered as plate area. Then the connected region detection approach can accomplish the plate accurate location.With regard to the character segmentation, it is the step after the second location method. At first, correct the connected region position. Then analyze the character connected region distributing condition in the plate area. Segment the new character connected region from the previously old ones' edge until all the characters are segmented successfully.It proved that the arithmetic presented in this paper can be adapted to complex background, and they are excellent in accuracy and robustness.