Abstract: Based on the property that all feature points of 1D target are collinear, a global calibration method of multi-camera system is proposed. The global coordinate frame is constructed based on one of cameras, called the base camera. 1D dimensional target is positioned arbitrarily in front of both the base camera and the camera to be calibrated. The vanishing point of 1D target in the image is determined. According to the property that all feature points of 1D target are collinear, the coordinates of all feature points on 1D target in each camera coordinate frame are computed. The transformation matrix from the camera to be calibrated to the base camera is solved, and optimized by using the non-linear optimization method. The global calibration of multi-camera system is realized by pair-wise calibration between the base camera and each of the cameras to be calibrated. The proposed method can be carried out without any high-accuracy 3D measuring equipment. Experiment results show that the method is simple and practical.

Key words: 1D target, Global calibration, Machine vision, Multi-camera