Research and Design on License Plate Location in Complex Background
|School||South China University of Technology|
|Keywords||license plate location gray feature color feature fine location license plate recognition system|
As the core technology of modern intelligent traffic management, the license plate recognition system has been widely used in various traffic situations such as community vehicles regulation, traffic optimum and electronic police. Therefore, it is very meaningful to study plate location. This paper puts emphasis on efficient and accurate location in complex background to meet the requirements: universality, accuracy and real-time. In order to achieve this aim, the location in complex background still has many technical problems to be solved, so the research of the plate location in complex background is very significant.Accurate license plate location in system of license plate recognition at complex background is studied in this paper. The license plate location is divided into two parts: rough license plate location and fine license plate location. The former bases on bright and color feature, while for the fine plate location, the plate color is used for secondary location. In rough plate location, first, the pictures are intelligently separated in two types and the corresponding location method is selected, using the characters of picture. And then to get edges of the plate image based on bright feature, after that a threshold on the image is obtained, subsequently the binarization, clustering and corroding is performed, and the accurate plate location is achieved after plate selection algorithm. While for the location based on the color feature, the RGB color space of the picture is switched to LAB space, and the bule or yellow plate is located using B value of the LAB space. While for the fine plate location, a second location is employed, the main method is to highlight the edges of license plate in order to reduce the interference factors.At last, the paper introduces the whole system of license plate recognition which includes three modules: license plate location, character segement and character recognition. The research of this study is useful in various complex background and multi-license location, especially in complex conditions of multi-license, dusky situation in daytime or in the evening, uneven light, vague or contaminative plate.