Improvement Research And Implementation Of License Plate Recognition

With the rapid development of world economic, traffic in every country is becoming worse and worse. Through a long period of exploration, people found intelligent transportation system is the best acknowledged method to solve problems in traffic field at present. The license plate recognition is the key technique in intelligent transportation system.In this paper, plate location, slant correction, character segmentation and character recognition are studied.For plate location, this paper firstly convert the color image in RGB color format to HSL color format, then the color image is changed into the one which only has four color by K-means clustering. At last, vehicle license plate can be located by its own characteristics. For slant correction, Hough transform is used to estimate the skew angle in this paper. Then the license plate is rotated to achieve the tilt correction. For character segmentation, vertical projection based on plate characteristics is used in this paper.For character recognition, template matching and neural network are used in this paper. In order to improve recognition rate of Chinese character, this paper uses character thinning. In order to reduce the training time of neural network, this paper extracts the feature of numbers and letters.All the methods mentioned in this paper were achieved based on Matlab and Delphi.