The vehicle number plate recognition system is expected to have numerous applications in many fields. In this paper the full system of the vehicle number plate recognition is discussed in detail. The full system can be divided into four parts: Picture capture, Number plate locating, Character segmentation, Character recognition. In this paper we discussed the last three parts. Number plate locating: we use the method based on the idea of two dimenion color histogram. First, to reduce the redundancy color information we cluster the color of the whole vehicle image into several colors. Second, we get the two dimenion color histogram from the vehicle image. Third, we search the vehicle licence based on the geometric and color property of licence automatically from the vehicle image. Character segmentation: first, the color of the licence image must be simplified into binarization. Second we adjust the inclination of the licence to keep it in the horizontal level. Third, we get the two dimenion color histogram of the licence image and use a threshold to cut the licence character based on the histogram. Character recognition: we use the K-L transform to reduce the complexity of the calculation. Then we use a method of HMM to recognize the character.
|