Research On The License Plate Recognition

The license plate recognition plays an important role on improving transport efficiency and traffic safety. It is an important technology for intelligent transportation systems. Researches on the license plate location, license plate character segmentation and recognition are given in this thesis.First, a new method of license plate location is designed in the thesis and it is divided into two steps, which are rough location step based on gray image processing and precise location one based on color image processing. Vertical and horizontal edge images are extracted from gray car image using Haar wavelet and Gabor filter. The rough location is achieved based on the specific projection features of these edge images. The precise location is achieved through making projection analysis of the binary image obtained from the binarization of the roughly located color plate image.In the license plate character segmentation, the license plate characters region is extracted using an improved projection analysis. License plate character image is obtained based on the vertical projection features of license plate characters region image.In the license plate character recognition, 73-dimensional feature is extracted from the license plate character image. Then, both the license plate Chinese character recognition method based on support vector machine and the recognition method for number and letter character based on feature matching are given.The simulation results prove that the method of license plate recognition in the thesis is effective.