Global Terrain Registration of LiDAR and Camera Fusion Using Multiple Calibrators

The fusion of LiDAR and camera sensors is extensively applied in the field of 3D reconstruction. LiDAR provides high-accuracy measurements and strong anti-interference capabilities, while cameras offer rich color and texture information. This paper presents a high-precision portable integrated solution spanning the entire process from extrinsic calibration of LiDAR and camera to comprehensive 3D terrain scene reconstruction. First, a 3D calibration object system is designed, allowing for precise extraction of camera feature points and utilizing the RANSAC algorithm and ICP algorithm to achieve high-precision extrinsic matrix calibration. Secondly, the frame-to-frame registration method is employed to achieve globally consistent 3D terrain scene reconstruction, with loop closure optimization further enhancing point cloud reconstruction accuracy. Finally, by incorporating color information, a high-precision, colorized point cloud terrain reconstruction is accomplished.