Research On Recognition Algorithm Of License Plate In Foggy Weather Based On Image Processing

Automobiles are used more and more in people's lives in recent years.Intelligent Transportation System(ITS)has become a common method to solve vehicle management problems.License Plate Recognition System(LPRS),an important component of ITS,has also developed rapidly.However,when the air quality drops and foggy weather forms,it is difficult to collect clear license plate images,thus reducing the success rate of license plate recognition.In view of this phenomenon,this paper studies the license plate image recognition technology in foggy weather.In this paper,the license plate images containing fog are defogged to effectively recover the fog images,and then the license plates are recognized to improve the success rate of license plate recognition in fog.The main research contents are as follows:In the aspect of image defogging,the common defogging algorithms are firstly studied and compared.Aiming at the shortcomings of traditional dark channel prior algorithms such as distortion when dealing with large areas of strong light such as sky and white vehicles,and the darker color of the image after defogging,the tolerance method is introduced into the dark channel prior algorithm and the estimation method of atmospheric light intensity is improved.At the same time,the brightness of the defogged image is increased,and the vehicle image is defogged by the improved dark channel prior algorithm.Finally,the simulation experiment is carried out and the defogging effect of the improved dark channel prior algorithm is analyzed.In the aspect of license plate location,the pretreatment process of license plate location is described.In view of the shortcomings of the fixed high and low threshold values in the traditional Canny operator,the high and low threshold values of the Canny operator are obtained in an adaptive manner,the improved Canny operator is used to detect the edge of the image.Aiming at the problem that the license plate image contains noise,several denoising algorithms are studied and compared,and the traditional median filtering algorithm is improved so that the size of the filter window can be adaptively changed.The adaptive median filtering method is used to denoise the license plate image.Finally,the location method based on mathematical morphology and pixel scanning method is used to accurately locate the license plate.In the phase of character segmentation,the method of license plate tilt correction is introduced,the selection of threshold in OTSU algorithm is optimized,and the improved OTSU algorithm is used to binarize the image.Aiming at the problem of low definition of license plate image,morphological method is used to process the license plate image,which improves the definition of the image and makes it easier to identify.After removing the upper and lower borders of the license plate,the characters of the license plate are segmented by vertical projection.In the aspect of character recognition,aiming at the characteristics of license plate characters,the improved BP neural network algorithm is adopted to recognize license plate characters,and the design of neural network and the selection of neural network related parameters such as genetic algorithm(GA)optimization parameters are elaborated.Finally,a simulation experiment is carried out on the flow of the license plate recognition algorithm in this paper.The experimental results show that the algorithm has achieved a high recognition rate.