As for the one-dimensional multi-camera calibration algorithm of low precision, anti-manic and poor stability, a multi-camera calibration algorithm was proposed based on the mutual constraint among vanishing points. In order to avoid the impact of radial distortion caused by the imaging, the geometric characteristics of position constrains in Euclidean space were used to calculate the distortion parameters. By combining the constraints of target feature points with projective invariance of camera to eliminate interference of miscellaneous points, then the consistency of angles in the space vanishing points, line which consisted of target feature points and the parallelism of the reverse-ray of the vanishing points were used to calculate the camera parameters. When one-dimensional targets move randomly, a critical problem is existed which is inability to distinguish the target feature points corresponding to the image points, this problem can be avoided by taking use of the inverse theory of mathematical analysis. By constructing the multi-camera system to calibrate the experiments, we can see that this method has a higher precision calibration, the calibration results have a certain degree of noise immunity and stability with the increasing of noise. The relative error value shows that this method can be applied in multiple camera systems.