Hand-Eye Calibration from Image Derivatives

In this paper it is shown how to perform hand-eye calibration using only the normal flow field and knowledge about the motion of the hand. The proposed method comprise a simple way to calculate the handeye calibration when a camera is mounted on a robot. Firstly, it is shown how the orientation of the optical axis can be estimated from at least two different translational motions of the robot. Secondly, it is shown how the other parameters can be obtained using at least two different motions containing also a rotational part. In both stages, only image gradients are used, i.e. no point matches are needed. As a by-product, both the motion field and the depth of the scene can be obtained. The proposed method is illustrated in experiments using both simulated and real data.