The Approach Of License Plate Localization Based On Connected Region Classification Algorithm

With the rapid development of the national economy, a sharp increase in the number of vehicles, that is imperative to improve the modernization level of urban traffic management. License Plate Recognition System is an important part of the Intelligent Traffic System, which is widely used in intelligent car park management, illegal vehicle identification, stolen vehicle detection, and other areas. It can enhance the level of intelligent management greatly, reduce human and material resources. It has an irreplaceable position in project management of the highway, urban roads and parking lots.License plate localization is the key of the License Plate Recognition System. In practical application, the license plate images are almost always conducted in a natural scene collection, a lot of factors affect the image content and quality. For example, weather factors (rain, snow), illumination change factors (direct sunlight, reflection, daytime, night, etc.), license plate image capture on-site defacement as well as a different landscape background, etc., the existing approach of license plate localization have to be further perfect.In this paper, the license plate localization problem is transformed into identification problem, using the approach of classification to achieve the license plate localization. The vehicle images of this article contain different lighting conditions of the day, night, and so on. I use anti-light feature - SIFT feature to reduce the lighting impact of the correct locating license plates, in order to improve the localization accuracy. This article uses the approach of extracting three features, and training three support vector machine classifiers separately, to distinct the two types of sample images that positive and negative, improve the robustness of classification algorithm. Through a variety of the experimental contrast of vehicle license plate localization, the algorithm proposed in this paper has great advantages of high localization accuracy and rapid localization speed. We can say that this algorithm has the robustness and adaptability to light, as well as real-timing.