With analyzing the defect of the methods based on vanishing point with radial distortion, a calibration algorithm is proposed. The fundamental matrix of binocular set is estimated, which allows to perform a projective calibration of each camera. The calibration is updated for the Euclidean space. The calibration is possible without imposing any restriction on the movement of the pattern and without any prior information about the cameras or calculating vanishing point. It can calibrate all the cameras at the same time. This method is suitable for the fast camera calibration in a non-laboratory environment. Finally, the experiments on synthetic images validate that the proposed method is superior to the calibration method based on the vanishing point when the radial distortion cannot be ignored. The experiments on real images show that its accuracy makes it suitable for practical applications.