本論文提出了一種根據平面物件其已知的內容及位置來求解此平面物件相對於相機的位姿。首先，我們推導出已知平面上及相片座標間的映射，透過最小平方法及牛頓法求解描述此映射的參數。引入多層感知機為牛頓法提供初值。因牛頓法本身並不穩定，導致此法並不是十分可靠，但我們提出了一些解決方案來修正此問題。最後，我們給出了修正後方法的結果，確認了牛頓法的穩定性問題是可以修復的。機器學習和牛頓法的結合是一種面對複雜模型的方法。

This paper propose a method to determine the pose of a planar target with known information with respect to the camera. First, we derive the mapping between the coordinates on the known plane and the photo and use the least square method and Newton's method to solve the parameters describing this mapping. Since Newton's method need an initial guess that is sufficiently close to the exact solution, we induce the method of machine learning to obtain a proper initial guess. Because Newton's method itself is not stable, this method is not very reliable, but we have proposed some solutions to correct this problem. Finally, we present the results of the corrected method, confirm the issue of the stability of Newton's method can be fixed.

Andrew Filisetti, Andreas Marouchos, Andrew Martini, Tara Martin, and Simon Collings. Developments and applications of underwater lidar systems in support of marine science. In OCEANS 2018 MTS/IEEE Charleston, pages 1–10. IEEE, 2018.

Lucas Afonso Casanova de Oliveira Nogueira. A unified approach to homography-based image registration.

Richard Hartley and Fredrik Kahl. Optimal algorithms in multiview geometry. In Asian conference on computer vision, pages 13–34. Springer, 2007.

Fredrik Kahl and Richard Hartley. Multiple-view geometry under the l∞ -norm. IEEE Transactions on Pattern Analysis and Machine Intelligence, 30(9):1603– 1617, 2008.

Zhijun Dai, Yihong Wu, Fengjun Zhang, and Hongan Wang. A novel fast method for l∞ problems in multiview geometry. In European Conference on Computer Vision, pages 116–129. Springer, 2012.

Ezio Malis. Vision-based estimation and robot control. PhD thesis, Université Nice Sophia Antipolis, 2008.

Yuchao Dai, Hongdong Li, Mingyi He, and Chunhua Shen. Smooth approximation of l_infinity-norm for multi-view geometry. In 2009 Digital Image Computing: Techniques and Applications, pages 339–346. IEEE, 2009.

Ty Nguyen, Steven W Chen, Shreyas S Shivakumar, Camillo Jose Taylor, and Vijay Kumar. Unsupervised deep homography: A fast and robust homography estimation model. IEEE Robotics and Automation Letters, 3(3):2346–2353, 2018.

K Zhang, W Luo, Y Zhong, L Ma, B Stenger, W Liu, and H Li. Proceedings of the ieee/cvf conference on computer vision and pattern recognition. 2020.

Daniel Pizarro, Rahat Khan, and Adrien Bartoli. Schwarps: Locally projective image warps based on 2d schwarzian derivatives. International Journal of Computer Vision, 119(2):93–109, 2016.

Timothy D Barfoot. State estimation for robotics, 2020.

David M Henderson. Euler angles, quaternions, and transformation matrices for space shuttle analysis. 1977.

Gaël Guennebaud, Benoit Jacob, et al. Eigen. URl: http://eigen. tuxfamily. org,
3, 2010.

Martín Abadi, Paul Barham, Jianmin Chen, Zhifeng Chen, Andy Davis, Jeffrey Dean, Matthieu Devin, Sanjay Ghemawat, Geoffrey Irving, Michael Isard, et al. Tensorflow: A system for large-scale machine learning. In 12th {USENIX} symposium on operating systems design and implementation ({OSDI} 16), pages 265–283, 2016.

F Chollet. others.(2015). keras. github, 2007.

Gary Bradski. The opencv library. Dr Dobb’s J. Software Tools, 25:120–125, 2000.