Automatic extrinsic calibration for an onboard camera

This paper presents an automatic extrinsic calibration method for an onboard camera in general drive condition, with some straight line and turning movements. Our method only needs the camera video and the synchronous vehicle poses, which can be estimated from Inertial Measurement Unit (IMU) data or four-wheel speeds of the vehicle. Basing on visual odometry, optical flow and motion stereo from the camera video, we introduce 3 specific constraints for automotive applications. Using these constraints, we propose two least-square optimization algorithms to solve the rotation matrix and the translation vector respectively. Our only assumption is the visible of the road surface, and we can achieve all extrinsic parameters (6 degrees of freedom) even if the vehicle only has planar movements. Experiments on simulative and real data demonstrate the effectiveness of our method.