The line structured light plane calibration method using a plane target cannot produce satisfactory calibration results due to inaccurate positioning of the calibrated points. Field of view noise and sensor noise affect the target light stripe extraction and camera parameter calculation during the calibration process. These factors will cause the calculation of the coordinates of the calibrated point to deviate, and thus affect the light plane calibration. To solve this problem, we propose a new method to calculate the calibrated point based on spatial geometry. Firstly, for the projection line corresponding to the feature point on the light stripe and the corresponding line on the target, a common perpendicular of these two lines above is established, and since the sum of the squares of the distances from the midpoint to the two straight lines is the smallest, the midpoint of the common perpendicular is taken as the calibrated point. Secondly, the target is moved to different positions, and the non-collinear calibrated points are calculated. Finally, the parameters of the light plane are obtained by fitting these calibrated points. This method requires only a checkerboard target, and has a simple calibration process. The experimental results show that the average error of the calibration method proposed in this paper is 0.011 mm, which is less than the 0.031 mm of the calibration method based on the plane target with cross-ratio invariant.
Keywords: least square principle; plane target; spatial geometry; structured light plane calibration.