The Study And Realization Of Key Algorithm In License Plate Recognition System

Along with the national economy rapid development, our country's automobile quantity increases rapidly. The Government also speeds up the step in infrastructure's construction. For example, the highway, the multi-level crossing bridge, the parking lot are getting more and more. But the corresponding vehicle management system actually has not followed the step, still pauses in primarily stage with the labor management, appears the relative backwardness. How to strengthen the traffic management, improve the use of highways efficiency are becoming more and more important. A variety of intelligent transportation systems have been produced. As a means of collecting information in intelligent traffic system, License Plate Recognition System has been developed very rapidly in recently years.In this paper, we investigate a variety of basic license plate recognition technology of domestic and foreign representatively. Based on the characteristics of the Chinese license plate, We study the existing algorithm of the license plate location algorithm, pre-processing algorithm as well as the license plate license plate character segmentation and recognition algorithm. In the light of the situation, an effective algorithm is presented with extensive study of some key techniques. Experiments show that compared with the existing algorithms, the new algorithm in terms of speed, algorithm complexity, or precision has an obvious superiority.The main innovation on the license plate recognition research in this paper has three aspects:1. Reviewed the existing classical car license localization method, and proposed a new algorithm based on multi-color models and integrated characteristics, which can satisfy the system timely request, and has the high robustness.2. Thoroughly analyzed the existing binarization method, and proposed one kind improved p-tile algorithm by combined with the inherent characteristics of plates. The experimental effect indicated that using this method to carry on the binarization of the car license image has the unique superiority. 3. Proposed a series of improving measure to the car license character segmentation and the recognition methods. Carried on the analysis and the simulation to the classics character division and the recognition algorithm, and analyzed the respective advantages and disadvantages.