Parking Charge System Of License Plate Recognition Method Research

With the development of economy, car number increase sharply, intelligent transportation systems arises at the historic moment. This article main research license plate recognition method based on parking charge system of research, mainly adopts the relevant technology such as computer vision, digital image processing, aiming at the problems now the license plate recognition system and experimental analysis of image preprocessing, license plate locating, character segmentation algorithms of the three modules is optimized to improve.This thesis for the collected vehicle image preprocessing, image information including: vehicle image gray level, the filtering de-noising, histogram equalization, binarization process, etc. In the process of image binarization of binarization algorithm to optimize the a new binarization algorithm is proposed. In the global threshold method was studied on the basis of looking for threshold selection method of histogram trough, trough the corresponding grey value as a image of the overall threshold, to realize the image binarization and experiment analysis in MATLAB platform performance of the proposed algorithm, good results have been achieved.This thesis for the collected vehicle image preprocessing, image information including:vehicle image gray level, the filtering de-noising, histogram equalization, binarization process, etc. In the process of image binarization of binarization algorithm to optimize the a new binarization algorithm is proposed. In the global threshold method was studied on the basis of looking for threshold selection method of histogram trough, trough the corresponding grey value as a image of the overall threshold, to realize the image binarization and experiment analysis in MATLAB platform performance of the proposed algorithm, good results have been achieved.Second, the screen image after pretreatment of vehicle license plate location, this paper proposes a projection and edge detection and mathematical morphology based on pixel combination of license plate location algorithm. Before using the projection method for license plate location is roughly positioning, and then USES the canny edge detection operator to extract the edge, using mathematical morphology to form multiple connected regions, finally according to the geometrical characteristics of the license plate inherent for license plate positioning accurately. The algorithm in the experiment the location accuracy can reach 96%.The most commonly used in character segmentation algorithm is based on the projection of characters segmentation, clustering connected component method and template matching method, etc. But the study found that these algorithms under the condition of the external condition is not good will appear deviation. So this topic combines the three algorithm and improved, and achieved better character segmentation result.In license plate character recognition this thesis chose to use the method of template matching for license plate character recognition. How to identify the confusing characters are discussed, and studied the dual template matching method. Dual template matching method is verified by experimental analysis, and the characters of the easily confused with good effect. In This thesis, we study the content for vehicle license plate recognition in the future application and development has a certain significance.At the end of the article, on the charge of the whole system design are studied, and the existing problems are discussed.