An efficient algorithm on vehicle license plate location

Abstract

This paper presents an efficient and smart approach to the vehicle license plate location based on characteristics of fractal dimension. The proposed algorithm consists of three major parts: pretreatment of license plate, search of license plate and extraction of plate region. In this study, we firstly research the normalization of license plate image and the criteria how to select the image enhancement method of license plate. Then the fractal dimension of license region and the interval of fractal dimension are all calculated. Finally the ultimate license plate region can be determined. The outstanding advantage of this algorithm is simple, smart and fast. Moreover it not only can be used to all kind of license plate in China but also has a good robustness under complex background, nonuniform illumination conditions and inclined license plate conditions. The performance of the proposed algorithm has been tested on a large number of experimental data from random and real images. Based on the experimental results, our algorithm shows both the missing rate and false detection rate are all zero. The possibility that the candidate region is more than one is 20%, meanwhile the probability of correction for inspect is 100%. The fact that this approach has superior performance in car license plate location is worth to note.