Research On Key Techniques Of License Plate Image Restoration And Recognition In Haze Environment

With the rapid development of information technology and the annual increase of automobile sales,under the background of globalization,License plate recognition has been applied in many fields such as intelligent transportation and security monitoring.Due to the interference of haze,sand,cloudy,etc.,the imaging quality of the visual system is severely limited,which in turn causes different degrees of degradation.Especially in road traffic,haze weather has caused great interference to the complete acquisition of license plate information,which seriously reduces the validity and robustness of license plate recognition.The research of restoration and recognition of license plate images under hazy conditions has been full of profound significance for academic research and application development.This paper focuses on four key technologies of license plate image defogging,license plate location,license plate segmentation and license plate recognition.Firstly,The dark channel prior theory is introduced and an improved image dehazing method is proposed based on the theory,the AECES edge-preserving filter is used to estimate the dark channel image,and an image sharpening system based on DSP+FPGA is developed.This method not only has good edge retention for images in a suddenly changed environment,but also can save tedious soft matting optimization process,which can greatly improve the effect of fog removal.Second,the edge detection and prior knowledge methods are applied to location,segmentation and normalization of license plates,and then the normalized characters are transformed into coarse grid matrix,in order to calculate the feature vectors.Finally,BP neural network is used to recognize the character of the license plates.Experiments on 66 license plate images randomly selected show that the accuracy rate can reach 91%,which can achieve the expected effect in general.In summary,the proposed method for license plate image restoration and identification can effectively reduce the influences on outdoor roads such as real-time monitoring,investigation,and other weather conditions,which has certain practicality and important research value in intelligent transportation.