Research Of Related Algorithms And Implementation Of The License Plate Recognition System

License plate recognition system (LPRS) is an important application of computer visionand pattern recognition technology in intelligent transport system (ITS),it have been widelyused in the system of video monitoring on the highway、the parking intelligent managementsystem, and electronic police,etc. Therefore, the research of the license plate recognitionsystem not only has great theory significance, but also has important practical economic valueand greatly promoted the development of intelligent transport system.The related algorithm of the LPRS consists of three parts,i.e. license plate location，character segmentation and character recognition. the paper makes a further study andresearch on the algorithm of the LPRS, puts forward some effective improvement algorithm,and develops of the LPRS.It included contents as follows:1. Research on license plate location algorithm. This paper analysis of thecommonly used algorithm of license plate location, in order to detect the edgepixels in the image that license plate is in fuzzy situation or uneven illuminationenvironment, a new algorithms of license plate location based on an improvedIsotropic Sobel edge detection operator was proposed. The algorithm includesthe following steps, firstly, the edge pixels, that is in the horizontal, vertical anddiagonal directions，was detected by presenting the improvement Isotropic Sobeloperator. Secondly, the grayscale image was changed into binary image by usedOtsu method, and the candidate plate regions was acquired by using themathematical morphology operation. Finally, removing the false license plate.2. Research on license plate character segmentation algorithm. a new algorithm ofplate slant correction based on fitting a straight line was proposed, after Analysisof the common license plate character segmentation algorithm. Firstly, Thispaper used Otsu method to change grayscale image into binary image.Secondly,an algorithm of horizontal slant vehicle plates correction was proposed, which isadopted by the algorithm of fitting a straight line.Finally, characters segmentedwas got by removing the border and character segmentation. 3. Research on character recognition algorithm. After analyzing of plate characterrecognition algorithm based on template matching and classifier, the paperproposed an new method of license plate character recognition which is consistof three classifiers based on improved BP neural network.4. Realization of vehicle license plate system. This paper studies the relatedalgorithm of license plate recognition system, and develops the system of LPRSin VC++6.0platform and computer vision library OpenCV.