Fig. 1. Schematic diagram of ideal pinhole camera model

Fig. 2. Integer pixel detection process of target marking points. （a） Original calibration picture； （b） mixed edges； （c） edges of target marking points

Fig. 3. Sub-pixel edge estimation method for circular features

Fig. 4. Solution process of edge line

Fig. 5. Coordinate position correction of marking points

Fig. 6. Summary of proposed calibration method

Fig. 7. Size drawings of two different specifications of calibration board. (a) Calibration board 1; (b) calibration board 2

Fig. 8. Calibration pictures in normal environment

Fig. 9. Calibration results of four methods in normal environment. (a) Global threshold centroid detection method; (b) centroid detection method based on watershed image segmentation; (c) traditional ellipse fitting detection method; (d) proposed method

Fig. 10. Comparison of calibration results of four different detection methods in conventional environment

Fig. 11. Calibration pictures in complex environment

Fig. 12. Calibration results of three methods in complex environment. (a) Centroid detection method based on watershed image segmentation; (b) traditional ellipse fitting detection method; (c) proposed method

Fig. 13. Comparison of calibration results of three different detection methods in complex environment