A novel method for license plate localization

A novel method for license plate localization based on texture and edge information is proposed in this paper. The whole process is divided into two parts: candidate extraction and candidate verification. In the first part, to ensure the true license plate being extracted from complex environment, several candidate areas instead of one with the max texture information are extracted. In the second part, auto-correlation based binary image and projection algorithm are used to verify the plate candidates. For a Chinese license plate, the auto-correlation curve of license plate shows approximately 13 peaks. And the real plate applied with our projection algorithm behaves about 7 areas which have the similar size to plate characters. The above characteristics combined with different binary methods to verify the Chinese license plate from all candidates. The experiments show the superiority of our proposed method.