Roadside Camera Calibration Method Based on C-V2X Connected Vehicle
To safely and efficiently calibrate roadside cameras on the open road,the novel calibration method was proposed.The method did not require any prior knowledge or road closures to block traffic,overcoming the limitations of existing methods that required manual measurements.The method treated connected vehicles as markers.The WGS-84 coordinate of each position in the field of roadside camera's view was provided by the connected vehicle,and the coordinate of connected vehicle image for calibration was extracted by using the depth learning algorithm via cellular vehicle-to-everything(C-V2X)communication technology.After synchronizing the image coordinates and their WGS-84 coordinates with time stamps,based on the direct linear transformation,the mapping relation between connected vehicle image coordinate and its WGS-84 coordinates was obtained to calibrate the roadside camera.An experimental platform was set up at a real intersection in Changsha.There were 82 verification points collected to conduct feasibility analysis experiments.The result indicates that the errors of all 82 validation points are almost no more than 1.5 m,demonstrating that the proposed method can meet the lane-level positioning requirements for the cooperative intelligent transport system(C-ITS)application.Simultaneously,this method is compared with two representative roadside camera calibration methods,and the result shows that the proposed method has the minimum root mean square error of 1.80%.In addition,the influences of connected vehicle attitude,connected vehicle speed and GNSS positioning accuracy on the calibration accuracy were also discussed.The comparative experiments were designed and analyzed respectively.Finally,some practical suggestions were summarized,so that higher calibration accuracy can be easily obtained by using the proposed method.
intelligent transportroadside camera calibrationC-V2Xconnected vehicleWGS-84 coordinate
NETL
NSTL
万方数据
