Research On Key Technology Of License Plate Recognition System In Complex Environment

With the rapid development of China’s economy, people’s income has increased steadily. Accompanied with the process of urbanization speeding up, the automobiles, like other common household appliances, rapid popularization into personal family life, that results vehicle population rising quickly every year. But roads and related resources of China cannot increase sharply in supply, these leaded to some social problems needed to solve, such as traffic congestion, environmental pollution, parking difficult. To a certain extent, these problems restrict the social and economic development in China. In a short period, our country cannot provide a lot of road resources and auxiliary facilities, and then the intelligent transportation system came into being in the background.Intelligent Transportation is the developing direction of the present and the future of transportation systems, and it can effectively solve the traffic problems that we are facing. As one of the important research topics of the intelligent transportation system, license plate recognition technology is an important support in the field of intelligent transportation. This paper focuses on the research of several key technology of license plate recognition system in the complex environment. On the basis of summarizing the results of previous studies and work experience, the paper puts forward its own solutions and improved algorithm to the key technology of license plate recognition system.Firstly, the algorithm of pyramid transform was applied to vehicle image to enhance image detail, and reduce the impact of environment and lighting conditions change on the vehicle license plate location. Secondly, the candidate regions of license plate were identified by using image binarization and mathematical morphology technique. On the basis of above, the paper designs a new license plate extraction method which first eliminates surrounding interference areas and then center interference areas. By the method, the license plate location is accurately located and the license plate image is extracted. Tests are done on images collected from different places and different natural conditions. The test results showed that the success rate of license plate location is 99.2% and the average positioning time is 0.309s. And then, through an improved projection segmentation method, license plate characters are segmented by designed template box. Eventually, SURF algorithm is applied to extract feature vector of the license plate characters and template characters’ feature points, and feature matching. Purification of the match is being by nearest neighbor and next nearest neighbor method and the template character will be the output as a result that has most matching points.