Fig. 1. System structure

Fig. 2. Three steps for determining the sub-pixel edge. (a) Calculating image gray gradient; (b) sampling along the gray gradient direction; (c) gray gradient amplitude curve fitting

Fig. 3. Edge comparison. (a) Ellipse image; (b) pixel edge; (c) sub-pixel edge

Fig. 4. Point-array coded target. (a) Point-array coded target template; (b)(c) images of point-array coded target from different perspectives

Fig. 5. Algorithm pseudo code

Fig. 6. Point-array coded target matching. (a)(b) Images of point-array coded targets in different perspectives

Fig. 7. Incremental SfM reconstruction

Fig. 8. Experimental device

Fig. 9. Images captured by two-camera vision system

Fig. 10. Error of extrinsic parameters when rotation and translation components optimize simultaneously. (a) Angle error; (b) position error; (c) reprojection root-mean-squared error

Fig. 11. Error of extrinsic parameters when only translation component optimizes. (a) Position error; (b) reprojection root-mean-squared error

Fig. 12. Error of fifteen calibration experiments. (a) Angle error; (b) position error; (c) reprojection root-mean-squared error

Fig. 13. Vision-based robotic arm system

Fig. 14. Schematic of experiment procedure. (a) Expected pose; (b) initial poses; (c) actual poses

Fig. 15. Positioning error of vision system. (a) Angle error; (b) position error