The Design And Implementation Of Vehicle Identity Recognition Technology

With the development of highway and vehicle ownership increasesing, the congestion tollis increasingly becoming a serious traffic problems, based on long-range contactless smart ICcard electronic toll collection system (Electronic Toll Collection, ETC), can effectivelyimprove the traffic capacity of the toll station and this is an effective solution to this problem.The electronic toll collection system, the license plates and vehicle identification is to identifyvehicles that identity is an important and indispensable adjunct, which in recent years in theIntelligent Transportation Systems (ITS) in become a hot research related to recognitionalgorithms are emerging, however, very few algorithms to meet the actual needs.This article first electronic toll collection system model identification plate overall systemdesign has made a comprehensive overview, and then on one of image preprocessing, featureextraction and recognition algorithm for a detailed study, the analysis of the current status ofresearch on the basis of on some of the key technical issues put forward some new solutions.The main work is as follows:First, In the model identification, we propose a method using image difference, namely,by extracting the difference between two adjacent Xu images strike a contour of the vehicle,and as a basis for recognition software judgment models.Second, proposed the use of the license area boundary characteristic of regional search forlicenses using the template matching method for positioning area.Third, Characters in the license division, first contour projection method using charactersegmentation, and then the characters are normalized, and then using linear interpolationmethod to solve the normalized generated when the mosaic phenomenon. On the differentnature of the characters (such as Chinese characters, letters and data) to take a differentcharacter sizes.Forth, in the license plate character recognition, we propose two kinds of feature extraction method: crossover point feature extraction method and the mesh feature extractionmethod, and using the minimum distance method for character classification. Matchingrecognition, in accordance with the license plate characters in the order of characters ofdifferent characters (such as letters, numbers and characters) with a different priority.Fifth, In order to improve the reliability of the vehicle identity discrimination, willintroduce the idea of information fusion vehicle identity discrimination, that depending on themodel identification results and vehicle license plate recognition results are fused identitydiscrimination in order to get a more accurate result of discrimination.