Extrinsic Calibration between Camera and LiDAR Sensors by Virtual Planar Junctions Matching

We propose an accurate calibration method for extrinsic parameters of LiDAR and camera by virtual planar junctions matching. It is difficult to find reliable feature correspondences between the LiDAR and the camera due to the sparsity and centimeter-level measurement errors of LiDAR point clouds. In our approach, a large size board with sparse corner markers is used as a calibration board to estimate the extrinsic parameters between a stereo pair and a multi-scanline LiDAR. In data acquisition stage, the board is placed in different positions around the LiDAR-cameras and virtual junctions are computed by each set of three planes for the calibration process. Based on matching virtual planar junctions between camera and LiDAR coordinate system, an ICP algorithm is used to solve the extrinsic parameters iteratively. Our experiments demonstrate that our method can get the rotation error is less than 0.15 degrees, and the translation error is less than 1.5mm.