Research On License Plate Recognition Technology In Haze Weather

The daily life of human society and license plate recognition technology are inextricably linked.Human society has transportation in all directions.From the early days of the founding of the People’s Republic of China to the present,our country’s transportation system has undergone earth-shaking changes.With the growing and perfecting of the highway system,the total number of vehicles in our country is also increasing year by year.The coordination of the relationship between people and vehicles,especially the various problems of vehicles,such as speeding,running red lights,hit-and-run,and Vehicle management systems of various parking lots involve license plate recognition,especially in open-air scenarios,license plate recognition will be greatly disturbed by weather,such as haze weather.Every year,about one-quarter to one-third of the road blockage incidents reported by local governments in our country,are caused by foggy weather.Photographing equipment takes pictures in smoggy weather or foggy weather,due to the absorption and scattering of light by haze,fog,and smoke,the irradiance received by the camera from the scene will be attenuated along the line of sight,so the reception of light by the imaging device will change accordingly.Image information such as details and distortion levels will be weakened accordingly.This will directly increase the difficulty of license plate recognition,so first dehaze the picture and then identify the license plate,the effect will be ideal.This paper conducts in-depth research on image dehazing and image license plate recognition after dehazing.The main contributions are as follows:1.Rapid defogging for license plate recognition in foggy weather,a lightweight network SNAOD-Net for image dehazing is proposed,which is a new dehaze network based on the idea of AOD-Net dehazing algorithm and combined with the network structure of squeeze-net lightweight network.Image dehazing has a significant effect,and the speed is greatly improved compared to the traditional dehazing algorithm.2.The path signature theory is introduced into the field of license plate recognition,the path signature is used as the core of the theory to extract the character features of the license plate,and combine the K-nearest neighbor,random forest and XGBoost classification algorithms in machine learning to perform the character recognition of the single character image of the license plate,including the recognition of the Chinese image characters in the license plate,have a good recognition effect.In haze weather,the recognition rate can reach 95.3%.