Research And Design Of Korean License Plate Detection System

With rapid development of Chinese economic and social life, the higher efficiency and convenience of traffic system is required. Recently Intelligence Transportation System (ITS) has become a hot research topic for the road traffic administration. Vehicle related information is the main research part by traffic management system, such as license plate, collection maintain a road cost, violate traffic regulations, and so on. License plate which is the principal identification feature of the vehicle has become important method in the vehicle information acquisition and management. Hence, License Plate Recognition System (LPRS) plays a crucial role in the field of intelligent transportation.A LPR system consists of three parts:license plate detection (LPD), character segmentation and character recognition. LPD is the most important part, and it will directly affect the system overall performance. So for several kinds of Korean license plate type, the paper proposed a license plate detection algorithm.The proposed algorithm is shown as follow:firstly, image edge information is extracted from the real vehicle image on the road, Use the optical Canny edge detection to get the image edges approaching to the real edges as close as possible. Then the closing loop can be detected and the license plate features can be extracted by utilizing a specific algorithm. In the next step, the algorithm estimates the regions of license plate, and eventually locates the exact location of the license plate according to the features detected. The algorithm makes full use of various license plate features, such as geometry, edge, texture and so on, gray, after a series of steps: edge detection, noise removal, feature extraction, estimate candidates, binarization, plate calibration, and the position of license plate is located ultimately.Simultaneously, the thesis has presented the detailed analysis of different type's characteristic and structure in the South Korea license plates, and puts forward different detection algorithm for different type's Korea license plate, and propose some corresponding solving algorithms for unconnectedness of Korean characters.