Segmentation And Recognition In The Vehicle License Plate Automatic System

The vehicle license plate automatic recognition system is one of the important research subjects in the traffic department. Because of the variety of the types and conditions in different vehicle license plates, existing system can't solve all the problems. As the result, this subject is still intractable.The color images we acquired from the fixed CCD camera in highway station. We take the following step after the locating the plate .The whole system includes the unifying the background color, binarization, segmentation and recognition. This paper presents 4 parameters which contain the information of distribution of characters and background, realizing the unifying the background color. Then the author brings forward a mended binarization method based on mean of gray value. In the segmentation the paper presents a arithmetic taking the projection, character width and the shape of the character in to consider. Finally we use BP neural network to recognize the characters.Testing more than 200 representative vehicle license plates, we can achieve the accuracy of 99% in unifying the background color, 96% in segmentation and 90% in recognition.