Demisting License Plate Recognition Based On Retinex Algorithm Design And Implementation Of The System

In recent years,more and more attention has been paid to the development of smart city.As a new technology in the field of smart transportation,license plate recognition technology has been widely used in expressway management,parking lot management,traffic detection and other scenes.But up to now,most of the research on license plate recognition can only recognize the vehicle in a better weather environment.Once there is a fog environment,light is easy to be scattered and absorbed,resulting in image deterioration and low accuracy of the final recognition results,which leads to license plate recognition can not be completed.The recognition efficiency of license plate recognition in fog days is very important for the traffic law enforcement,which requires high accuracy and real-time performance.Therefore,the realization of the system of defog license plate recognition is the primary task of license plate recognition technology at present.At present,the image processing algorithm has gradually taken shape,but the research on the defog license plate recognition algorithm in the fog world is still not mature,for example,the color recovery of the image defog algorithm is poor and can not be well combined with the license plate recognition system,the license plate positioning effect is poor,and the image quality degradation leads to the low recognition rate of characters.Therefore,through the research of various image enhancement technologies in recent years,this paper deeply studies the application of multiple Retinex algorithms in image processing,focusing on the Retinex algorithm based on local features,and using a multi-scale Retinex algorithm based on HSV color space to design and implement a new type of license plate recognition system in fog days.The system implementation steps are as follows.Firstly,the moving vehicle area is separated by license plate detection and left-right regression algorithm,and the vehicle image is captured;secondly,the hsv-msr algorithm is used to defog and enhance the image;secondly,the connected domain analysis algorithm is used to locate the license plate,and then the top hat transformation,binarization and vertical projection method are used to segment the characters of the license plate.Finally,a three-layer BP neural network framework is established to complete the characters Based on Microsoft Visual Studio and opencv,a defog license plate recognition system is designed,This paper explores the advantages and disadvantages of Retinex algorithm and the important role of color space in image analysis.Finally,the HSV color space is used tomaintain the same hue and MSR algorithm is used to realize the defog license plate recognition system which can accurately recognize the license plate in the fog environment.After the function and performance test of the system and the effect test of the defog algorithm,it is confirmed that the defog license plate recognition system can meet the expected requirements,and has certain application value in the field of intelligent transportation.