Research And Design Of License Plate Recognition System For Gas Station With Large Inclination Angle

In this paper,the license plate recognition technology is used to identify the vehicle license plate.The experimental environment of this paper is based on the issue of license plate recognition system between Northeast Forestry University and a petrochemical company.In the gas station,the camera is suspended in the roof,the angle of inclination is relatively large,and the entrance of the gas station is wide,the disorder of the angle of the vehicle entering the vehicle is different,so that the difference between the camera and the license plate is different.There may be lanes around the building,high-rise buildings,all kinds of trees,advertising and other background,many license plate recognition method can not be applied;so how to identify in the upper and lower,such a large tilt angle of the vehicle license plate technology will be our research important topic.In this paper,through the review of literature and a number of field visits,the large tilt angle of the gas station license plate recognition system of technical difficulties to do a detailed analysis to determine the system composition,design,function and algorithm.Firstly,the image of the image and the information of the edge area are enhanced by image preprocessing,image grayscale,contrast enhancement,denoising,binarization and edge detection.,And the interference in the image of the source of multiple training,to exclude the impact of interference sources.Secondly,the license plate location method based on region amplification is used to adapt to the complicated environment.In the traditional edge detection method,the Sobel operator edge detection is carried out by adjusting the threshold value,and then the iterative fusion is carried out.Domain screening and multi-factor decision-making two rough positioning to determine the most likely to appear license plate area and remove the pseudo-license plate;for the license plate tilt angle,license plate useful information loss phenomenon,this article improved the regional amplification method for license plate positioning.Finally,the template matching algorithm and the neural network algorithm are used to identify the license plate.The two methods complement each other.In the template matching algorithm,after a lot of learning and training,and continue to add the database.,So the number of database templates(license plate variety of characters,characters and digital template)determines the template matching algorithm to identify the accuracy rate,so in the school and a gas station experimental environment after 2 years of simulation training and testing of thousands of license plates,Very good recognition effect.In order to ensure the stability of the system identification effect in a variety of weather and light(sunny,cloudy,rainy,haze,uniform,uneven)conditions,the system uses high-definition and resolution cameras to capture key pixels Point,through a number of edge detection,iterative fusion,accurate iterative filtering,stacking analysis algorithms and techniques to achieve more accurate license plate positioning and identification.Finally,for different weather and environment,especially under the angle of the license plate recognition system for multiple test analysis,and constantly improve the system algorithms and techniques to achieve the system program set the target.