A two-step calibration method of a large FOV binocular stereovision sensor for onsite measurement

• A two-step calibration method of large FOV stereovision for onsite measurement is proposed.
• It can comprehensively and accurately calibrate camera-intrinsic parameters.
• A novel method of corner detection based on region of interest is proposed.
• The Epipolar constraint is introduced into the calibration of structure parameters.
• A case study indicates the calibration accuracy of our method is higher than the tradition method’s.

As there are some shortages with using both standard calibration method and self-calibration method to calibrate a large Field of View (FOV) binocular stereovision sensor, a two-step calibration method of a large FOV binocular stereovision sensor for onsite measurement is proposed. Furthermore, in order to improve the calibration efficiency, a novel method of corner detection based on region of interest (ROI) is also proposed to quickly detect the chessboard corner in the image of a small calibration target. Compared with the other two methods, the proposed method is more suitable for rapid onsite structure–parameter calibration of a large FOV binocular stereovision sensor. The experiments show that the onsite calibration process is efficient, and the average value of 10 measurements for a distance of 1071 mm is 1071.03 mm with a RMS less than 0.1 mm. Moreover, the application results indicate that the measurement accuracy of the stereovision prototype system calibrated by this method is improved by 13.56% compared with the other from standard calibration method.