Research On License Plate Segmentation Under Complex Background

With the development of the economy and the increase in the number of vehicles, the highway communication becomes an important means of transportation. Also crowded urban traffic needs more advanced and more effective traffic administration and control system. Intelligent Transportation System(ITS) which makes use of electronic information technology to raise management efficiency, traffic efficiency and traffic security becomes main direction of traffic administration. License Plate Recognition System (LPR) is the core of Intelligent Transportation System (ITS). The system can automatically register, verify, monitor vehicle or report to the police with automatic recognition for vehicle license plates. So it can be used in many kinds of occasions, such as the charges system of expressway, monitoring system at road and roll-gate, charge and monitoring system at the district, parking area, guide system, the system counting the quantity of vehicle passing in a certain period time, and so on.In recent years, using the computer to deal with and analyze image has already made remarkable progress. The study on the critical techniques for LPR has already became an important research field in scientific community.License plate location, license plate character segmentation, license plate recognition are the three essential components of the License Plate Recognition System. In this paper we discussed the tow parts of License Plate Recognition System: license plate location and character segmentation, and made a deep research about some important and key technology.A synthetic license plate segmentation scheme is presented in this paper. This algorithm makes full use of the texture and geometric characteristics of the license plate .The mainly stages of this approach are designed to deal with images taken under various real world conditions. Generally, the images have complex background. First we wipe off the complex scenes and then we carry on the license localization. In this step the image projection method is used to determine the horizontal boundary roughly first, then the line-scanning method is implemented to count the jump variables to locate horizontal boundary accurately. Next, the sliding window is adopted to designate vertical boundary according to the ratio of the length and width of the plate.In the character segmentation stage, after binarizing the image of license plate and correcting the tilt, the Mathematical Morphology operation is applied to the image processing to eliminate the noise and to fill the hollows. After extracting the characters of license plate with the H-S method, we segment each character through projection.In view of the character such as'川',this article presents a new method to determine the vertical border of the character according to the projection.The experimental results indicate that the proposed method can effectively locate license plate under complex background and segment characters correctly. The performance of the system is promising.