The Key Technology Research On Location And Tilt Correction Of The License Plate

License plate recognition system is an import part of the Intelligent Transportation Systems andit has a great applications. The license palte recognition system includes three modules: location,character segmentation and recognition. The paper is made a research on the three modules, and thespecific contents are as follows:1. In the license palte location, it is proposed a method based on HSI color model and the coloredge to locate the car plate. Firstly, the color image is converted from the RGB color space to theHSI color space, and according to the color characteristics of the license plate to set the thresholdvalue of the HSI components. Removing the region that dose not comply with the conditions, andthen it can get the candidate region of the license plate. Then detecting the edges of the image bythe features of the character color and the background color in the license plate image, and thepseudo-plate areas are removed through the geometric characteristics of the license plate. Finally,the license plate image is projected to obtain the starting position of the license plate, and thus itcould get the final positon of the license plate. The method can locate the license plate accurately inthe image with the complex background, and it has strong robustness.2. In the part of the character segmentation, the paper first corrects the tilt license plate image.According to the feature of the horizontal and vertical projection on the license plate image, it isproposed an improved method that based on rotation projection. The method can reduce thecomputation time and detect the tilt angle accurately. In the character segmentation, the paper usedthe feature of the character size on the license plate and the size of the vertical projection to split thecharacters. The method is also made effective process on the situation as the congutination character,fracture character and so on, and it also improves the accuracy of the character segmentation.3. In the part of the character recognition, it is used the method of the support vector machine toidentify the license plate characters. First, it is used the method that is based on pixel by pixel andcounting the blocked features to extract the character features. Then according the characteristics ofthe character features to constructe the classifiers, and by the bilinear method to determine the bestclassifier. Finally, the characters are recognized through the features of the arrangement on thecharacters, and the recognized results are put together to obtain the license plate number. Theexperimental results show that the method has a high recognition rate and recognition speed.