Based on the invariance of cross-ratio and the property that all points of the one dimensional target are collinear

a global calibration method of Multi-vision system is proposed. The global coordinate frame is constructed based on the coordinate frame of one of the vision sensors

which is called the base vision sensor. The one dimensional target is positioned arbitrarily

for at least twice

in front of both the base sensor and the sensor to be calibrated. For each vision sensor

according to the invariance of cross-ratio

the image coordinates of the target feature points out of the view field are computed

utilizing the image coordinates of at least three feature points captured by the vision sensor. The transformation matrix from the coordinate frame of the sensors to be calibrated to that of the base sensor is solved

and optimized using non-linear optimization method. The global calibration of multi-vision sensors system is realized by pair-wise calibration between the base vision sensor and each of the vision sensors to be calibrated. The proposed method can be carried out without any high-accuracy 3d measuring equipments. It is simple

flexible and applicable in various situations. Experimental result shows that the error of the global calibration method is 0.041mm.