This paper evaluates the effect of mixing the distortion centre, principal point and arithmetic image centre on the distortion correction, focal length determination and resulting real-world stereo vision triangulation. A robotic arm is used to generate a ground truth set of known positions resulting in 2078 measurements per cameras. It is seen that compared to the naive use of the arithmetic image centre improvements of 10% to 27% in triangulation accuracy can be made by determining an optimal principal point. An optimal distortion centre has a smaller but still beneficial effect.
Reference:
De Villiers, J. Jermy, R. and Nicolls, F. 2015. A study on the effect of different image centres on stereo triangulation accuracy. In: Proceedings of the 2015 Pattern Recognition Association of South Africa and Robotics and Mechatronics International Conference (PRASA-RobMech), 26-27 November 2015, Port Elizabeth, South Africa
De Villiers, J., Jermy, R., & Nicolls, F. (2015). A study on the effect of different image centres on stereo triangulation accuracy. IEEE Xplore. http://hdl.handle.net/10204/8732
De Villiers, J, R Jermy, and F Nicolls. "A study on the effect of different image centres on stereo triangulation accuracy." (2015): http://hdl.handle.net/10204/8732
De Villiers J, Jermy R, Nicolls F, A study on the effect of different image centres on stereo triangulation accuracy; IEEE Xplore; 2015. http://hdl.handle.net/10204/8732 .
Proceedings of the 2015 Pattern Recognition Association of South Africa and Robotics and Mechatronics International Conference (PRASA-RobMech), 26-27 November 2015, Port Elizabeth, South Africa.
ResearchSpace
