The Research Of License Plate Recgnition System

LPR (License Plate Recognition Technique) in modern intelligent transportation’s role is very important. On the basis of research literature on the license plate recognition system itself from the departure of three parts, each optimized to improve, so the system can more accurately identify the license plate information.License plate location algorithm based on edge analysis method is a simple license plate location algorithm. License plate location is the first step in license plate recognition, license plate location method based on edge analysis of simple and easy computer processing, and the edges of license plate location algorithm analysis results are well positioned to meet the requirements of the follow-up work. First color image pre-processing, and use ROBERT detection operator image edge detection, edge information extraction, followed by expansion of corrosion, remove the loopholes, the image itself is characterized by a combination plate were analyzed to extract the license plate area. Experimental results show that the algorithm can be well positioning plate.According segmentation and image of the license plate license plate recognition requirements, combined with the method of partial differential equations, CV active contour model is proposed character segmentation algorithm. Algorithm improved CV model, strong growth curve of scope, so that you can quickly improve PDE and smooth segmented character image.Presents a template matching method based on texture feature extraction, the traditional method of using template matching pixel matching, texture feature template matching method, first using wavelet transform to extract the wavelet coefficients for each pixel, each pixel calculated the mean and variance of pixels used to describe the texture characteristics, a combination of pixel values of pixels of the template matching itself. Experimental results show that the improved algorithm in the character recognition process, a better recognition of the character segmentation, reducing false positives.