Camera self-calibration based on single rectangle

The main difference between visual servo control and computer vision is the requirement of real-time control, which is often higher for the former. Considering the image-based visual control system, camera is included in the closed loop, so the precision is not sensitive to the error of camera parameters. A camera self-calibration method based on a single rectangle is proposed in the paper according to the fact that visual servo system emphasizes real-time property over accuracy for camera models. The image information of the four rectangular vertexes is obtained by the method based on the corner detection in curvature scale space, according to the geometric relationship constructing constraint equations to calibrate camera intrinsic parameters. The advantages of this method are its linear calculation and simple calibrating procedure, avoiding the complex problems of non-linear optimization. The feasibility of this method is shown through simulation and practicality experiments. Regardless of the lower robustness of the proposed method comparing to traditional calibration methods, it is more adaptable to visual servo control system due to its excellent real-time property.