Research And Implementation Of Real-time Fake License Plate Detection System Based On Big Data Analysis

With the development of economy and society,the number of motor vehicles is increasing,and the illegal acts involving vehicles are becoming more and more rampant at the same time.Among those acts,the vehicles with fake plate are common,which puts the authority in a tight spot.On the other hand,the monitor system of traffic captures billions of car monitoring records annually,which contain a large amount of information about the vehicle's space-time trajectory.Their value has not been effectively exploited.Aiming at solving the fake plate problem faced,this paper proposes a fake plate detection model.The model simulates vehicles' driving path with the information from traffic monitoring system,and calculates vehicles' interval velocity between two adjacent records.If the interval velocity is reasonable,it is regarded as a legitimate driving path.If the interval velocity exceeds the reasonable range,it's considered that an unlicensed vehicle disturbs the legitimate vehicle's driving path.License plate number with an unreasonable path is judged to be involved in fake license plate behavior.That's how the naive fake plate detection model works.The naive fake plate detection model fixes the desicion threshold,which determines whether a car is involved in misusing License plate,without taking the real-time traffic flow information into consideration.It causes bad performance of the model at some times.Based on the analysis of the traffic flow information extracted from the traffic monitoring history records,the interval velocity prediction model of the whole road network has been established.Based on the interval prediction model,two dynamic fake plate decision threshold models are proposed,which make the fake plate detection model be able to apperceive traffic status.According to the requirement analysis and the research of fake plate detection model,we have designed and implemented the whole fake plate detection system.Some necessary system tests have been conducted.At the same time,we have conducted a comparative experiment for a week on three fake plate detection models.To some extent,we have achived several remarkable results.Finally,the experimental results are shown and analyzed in this paper.The optimization direction of fake plate detection model is given after the analysis.