Intelligent Traffic Monitoring System Based On Fusion Of Video And Radar Detection Information

China has quickly entered the era of automotiveization,and the number of vehicles has increased dramatically in recent years,which has caused the reliability of the traditional transportation systems to face huge challenges.The proposal of the intelligent transportation system conforms to the development requirements of the world's information and alleviates the problem of traffic congestion.In the intelligent transportation system,vehicle detection is an important development direction.From the perspective of multi-target detection,low cost,high precision and visualization,this thesis proposes a method to fuse video and radar detection information to realize vehicle detection.The work is as follows.Firstly,vehicle detection is performed on the video using the YOLO deep learning framework.The Tiny-YOLOv3 model is improved and trained.Experimental results show that the improved algorithm has higher vehicle detection accuracy.Secondly,this thesis analyzes the characteristics of radar waveforms,and proposes the concept of Multi-stage Linear Frequency Shift Keying(MS-LFSK)waveform based on Multiple Frequency Shift Keying(MFSK)waveform.Under the condition that the radar's horizontally polarized beam and vertically polarized beam work together,the estimation of vehicle distance,speed,azimuth and pitch angle can be realized.The detection performance of MS-LFSK waveform is verified by simulation results.Finally,the three-dimensional spatial coordinate information of the vehicle detected by the radar is projected into a two-dimensional image coordinate system according to camera calibration,and the Euclidean distance between the pixel coordinates of the video and the radar is calculated.At the same time,the video detection target is tracked by Kalman filter,and the European detection distance is used to match the video detection target with the radar detection target.The experimental results confirm that the method of fusion of video and radar can realize the functions of high-precision vehicle speed measurement,traffic flow statistics,and over-speed snapshot,which have certain guiding significance for intelligent travel of urban traffic.