The high-precision calibration technique of stereo photography systems plays a crucial role in determining the three-dimensional position of LAMOST fibers. This paper proposes an improved calibration method for stereo photography systems. Firstly, building upon four sets of purely translational vertical movements, precise camera intrinsic parameters can be obtained by compensating for verticality errors. Secondly, a principal point estimation model with varying focal lengths effectively reduces the strong coupling of camera intrinsic parameters. Finally, a precise global optimization method based on the coplanarity of chessboard grid corner points is proposed. Experimental results demonstrate that the single-target calibration reprojection error of this method can be reduced to as low as 0.0697 pixels, with a stereo distance measurement accuracy of up to 99.9806%.
|