Research On License Plate Recognition Algorithm Of New Energy Vehicle In Specific Environment

In the field of intelligent transportation,license plate recognition technology,as an important technical support,is a hot topic studied by experts and scholars at home and abroad.At the same time,in order to solve the problem of environmental pollution and the increasingly prominent contradiction between fuel supply and demand,China will take new energy vehicles as an important measure to develop low.carbon transportation.The research of license plate recognition of new energy vehicles has practical significance.The process of license plate recognition is divided into several stages: image preprocessing,license plate location,character segmentation and character recognition.This paper takes the new energy vehicle license plate as the research object,with the normal,smoggy and rainy environment as the main research background.In order to improve the accuracy of positioning,the main goal is to solve the problem that the license plate positioning is inaccurate or even impossible to locate.Firstly,the image is preprocessed to improve image contrast and recognition.Images acquired in a specific environment are generally characterized by ambiguity,low brightness,and color distortion,and analysis of homomorphic filtering and different Retinex algorithms and their effects on image processing.The homogeneity filtering,SSR,MSR,homomorphic filtering and SSR,homomorphic filtering and MSR are compared and analyzed in the haze environment to verify the superiority of the proposed algorithm.Secondly,the license plate is positioned for the image in the defogging.The license plate is used to realize the rough positioning of the license plate,remove the large area of the non.license plate area,and reduce the influence of insufficient details of the license plate area on the license plate location.The edge detection algorithm based on Kirsch operator is improved to improve the accuracy of license plate location in the smog environment.Thirdly,character segmentation is performed on the positioned image.The traditional LBF(Local Binary Fitting)model algorithm and GrabCut algorithm are used to carry out theoretical research and experimental analysis,and the deficiencies of the algorithm are improved.The time required for algorithm segmentation the complexity of the segmentation process,the integrity of the segmentation is used to verify the feasibility of the algorithm.Finally,using convolutional neural network technology,the convolutional neural network is established,and the relu function is used as the loss function.After training the established data set,the test set is tested to verify the model.Through the repeated comparison,elimination and improvement of the image algorithm,the quality of the degraded car image has been significantly improved,and it is suitable for the new energy vehicle license plate recognition system,and the improved algorithm improves the calculation efficiency of the license plate location.