A Novel Method of Spatial Calibration for Camera and 2D Radar Based on Registration

The fusion of sensor data is important to improve the accuracy of vehicle detection in traffic surveillance system. For the sensor fusion, spatial calibration is required between each coordinate system of sensor. In this paper, we propose a new method of spatial calibration between mono camera and 2D radar. Using an AR-marker, which is designed to be detectable by radar, we simultaneously measure the position of marker with respect to the coordinate system of camera and radar. Based on multiple measurements of marker position, the position data is obtained from it. By applying the paired point registration, the transformation is acquired between the coordinate system of radar and camera. The experimental results confirm that the proposed method shows around 0.3 m registration error, which is acceptable for detecting vehicles. We believe that our method can be helpful to enhance the accuracy of vehicle detection in traffic surveillance system and other application such as autonomous driving vehicle.