When using cameras to accomplish sensor-based control tasks in robotics, e.g. object recognition, tracking, or grasping, the pose of the camera w.r.t. scene has to be determined. Moreover, several characteristics of the lens system and the image sensor have to be considered to determine the real-world dimensions of objects in the images.
The setup and camera characteristics are captured by so-called extrinsic and intrinsic camera parameters.
The extrinsic parameters specify the position and orientation of the camera w.r.t. scene whereas the intrinsic parameters comprise the focal length, principal point, and the image sensor format.
Over the years, different approaches to camera calibration, i.e. to estimating these parameters, have been proposed. Among the most popular ones are Tsai's[1] and Zhang's[2] method.
The talk shall first give an introduction to camera models, extrinsic and intrinsic camera parameters and then outline and compare(cf. e.g.[3,4]) Tsai's and Zhang's methods.
Hinweis: Dieses Thema kann von Bachelor-Studierenden auf ausdrücklichen Wunsch auch in deutscher Sprache bearbeitet werden.
Sources:
