Geometric calibration of a camera and a turntable system using two views.

A calibration algorithm for a system comprising a camera and turntable is provided, in which the camera captures three-dimensional point clouds while a target object rotates on the turntable to three-dimensionally scan the entire object. A mathematical approach is used to obtain the geometry between the coordinate systems of the camera and the turntable. As a minimal setup, the camera captures two views of a chessboard calibration pattern placed on the turntable. The coordinate transformation from the camera to the turntable is then calculated through geometric reasoning. First, two camera poses are computed with respect to the chessboard images taken before and after rotation. The direction of the rotation axis of the turntable coordinate system can be calculated without any ambiguity. However, there is one degree of freedom in the choice of the other two directions and one degree of freedom in choosing the location of the origin of the turntable coordinate system. We provide a practical method for these choices. Experimental results with a stereo-vision system and a turntable validate the proposed method.

Globally optimal camera-and-rotation-sensor calibration with a branch-and-bound algorithm.

This paper introduces a globally optimal algorithm for obtaining the rotational displacement between the coordinate frames of a rotation sensor and a camera that are rigidly attached. Our method minimizes the geometrically meaningful error using a branch-and-bound algorithm to find the global solution. For this, we derive a bounding inequality and corresponding feasibility problem for a top-down efficient search over the rotation space to minimize the L1-, L2-, or L∞-norm error function. Experiments are performed with synthetic and real data sets to show the efficacy of the algorithm.