The Study Of Intelligentized Method Application In The License Plate Location And Character Segmentation

With the fast development of transportation, more and more attention has been paid to its automation. Automatic Vehicle Identification system based on videos and still images, with its advantages in system installation, image search and image retrieval, seems to be an extremely useful method to solve these problems, and thus appeals to many people.In this thesis, we mainly study three parts of the automatic vehicle identification system on the basis of the past experiences. These critical parts are respectively the license plate location, the threshold segmentation of license plate and the character segmentation. This paper presents the theoretical investigation and practice methods.Two algorithms is presented in this paper for the license plate location: the first based on Hough transform and the second based on symmetry Hough transform. The experience results show that the second method is better, which decreases the dimensions with the high veracity and robustness.We select the threshold by minimizing the measure of fuzziness in the license plate segmentation. Moreover, we preprocess the image with the gray stretch and smoothing in order to conquer the light asymmetry. Comparing with the other methods the results prove the validity of our method.For the character segmentation, the results in the paper improves that the method of maximum width retrospect for the single character is effective.