Aiming at the difficulty in locating vehicle plate, an algorithm of vehicle plate's location based on fixed color collocation is put forward in this paper. Edges in chromatic image are detected by this algorithm on the basis of fixed color collocation. Then, the plate's region is located by the texture of vehicle plate. This Algorithm is fast and its accuracy is high. The accuracy is 96.8% in complicated conditions. In order to obtain gradient angle to emend the image, borders of vehicle plate are detected by random algorithm of lines. In comparison with Hough method, this algorithm has less calculation and takes less time. After analyzing the advantages and disadvantages of global and local algorithms, an algorithm which is derived from the combination of them is put forward in this paper. It not only can retain image's details, but also can eliminate disturbances of artifacts. After that, vehicle plate will be cut into several images with single character. At last, single character is recognized by BP neural network, average rate of recognition is 95.7%. |