本論文致力於改善可見光眼動儀在頭部移動時可能產生的凝視點計算錯誤。傳統眼動儀主要分為可見光眼動儀和紅外線眼動儀兩大類。雖然紅外線眼動儀技術已經成熟，但長時間使用可能導致眼睛疲勞。相反地，可見光眼動儀雖可長時間使用，但易受到環境光源或頭部移動等因素的影響，進而影響計算精確度。本論文提出了一種方法來改善可見光眼動儀在頭部移動時可能導致的凝視點計算錯誤。此方法包括計算頭部在相機空間中的三維位置，並透過幾何運算來補償頭部移動導致的眼睛旋轉向量偏移，將之校正到正確的凝視點位置。在眼球建模方面，由於眼球建模所需計算龐大，單純依賴軟體運算速度難以滿足消費型電子產品的需求，因此，本研究基於一個最佳化演算法，設計出對應的晶片架構，實現高度平行計算並結合管線式處理，有效提升計算效率。相較於傳統軟體方法，硬體架構在眼球建模時的運算速度提高了約 40 倍，從而增強眼動儀的效能，使其更加順暢和精確。總而言之，本論文提出了改進可見光眼動儀在頭部移動時的計算準確性，並提出了一個高效的晶片架構，使可見光眼動儀在實際應用中更為可行。

This thesis aims to improve the calculation accuracy of gaze points by visible spectrum eye trackers during head movement.Traditional eye trackers are mainly divided into two categories: visible-spectrum eye trackers and infrared eye trackers. While infrared eye tracker technology is mature, prolonged usage may lead to eye fatigue. In contrast, visible-spectrum eye trackers can be used for extended periods but are susceptible to factors such as environmental light sources or head movement, thereby affecting calculation accuracy.This paper proposes a method to address potential gaze point calculation errors caused by head movement in visible-spectrum eye trackers. The approach involves calculating the three-dimensional position of the head in camera space and compensating for eye rotation vector offsets caused by head movement through geometric operations, ultimately correcting it to the accurate gaze point position. In terms of eye modeling, due to the substantial calculations required for eye modeling, relying solely on software processing speed is challenging to meet the demands of consumer electronic products. Therefore, this study, based on an optimization algorithm, designs a corresponding chip architecture, achieving highly parallel computation combined with pipeline processing to effectively enhance computational efficiency. Compared to traditional software methods, the hardware architecture demonstrates a speed improvement of approximately 40 times during eye modeling,thereby enhancing the efficiency of the eye tracker for smoother and more precise performance. In conclusion, this paper presents improvements in the calculation accuracy of visible-spectrum eye trackers during head movement and proposes an efficient chip architecture, making visible-spectrum eye trackers more feasible in practical applications.

X. Li and W. G. Wee, “An efficient method for eye tracking and eye-gazed fov estimation,” in 2009 16th IEEE International Conference on Image Processing (ICIP),2009, pp. 2597–2600.
W.-C. Kao, J.-Y. Li, S.-C. Lin, and Y.-C. Chiu, “High precision canthus alignmentfor visible-spectrum gaze tracking system,” in 2019 IEEE International Conferenceon Consumer Electronics - Taiwan (ICCE-TW), 2019, pp. 1–2.
W.-C. Kao, K.-J. Huang, and Y.-C. Chiu, “Eyeball model construction with headmovement compensation for gaze tracking systems,” in 2020 IEEE InternationalConference on Consumer Electronics (ICCE), 2020, pp. 1–2.
J. Sigut and S.-A. Sidha, “Iris center corneal reflection method for gaze trackingusing visible light,” IEEE Transactions on Biomedical Engineering, p. 411–419,2011.
R. Valenti and T. Gevers, “Accurate eye center location through invariant isocentricpatterns,” IEEE Transactions on Pattern Analysis and Machine Intelligence, pp.1785–1798, 2012.
C. M. Y. H. K. S. J. Baek, K. A. Choi and S. J. Ko, “Eyeball model-based iris centerlocalization for visible image-based eye-gaze tracking systems,” IEEE Transactionson Consumer Electronics, pp. 415–421, 2013.
Z. Zhu and Q. Ji, “Novel eye gaze tracking techniques under natural head movement,” IEEE Transactions on Biomedical Engineering, vol. 54, no. 12, pp. 2246–2260, 2007.59
Y. Tan and Y. Zhu, Fireworks Algorithm for Optimization, Y. Tan, Y. Shi, and K. C.Tan, Eds. Berlin, Heidelberg: Springer Berlin Heidelberg, 2010.
D. Li, L. Huang, K. Wang, W. Pang, Y. Zhou, and R. Zhang, “A general frameworkfor accelerating swarm intelligence algorithms on fpgas, gpus and multi-core cpus,”IEEE Access, vol. 6, pp. 72 327–72 344, 2018.
Y.-H. Zou, J. Wen, H.-Y. Xing, and Y. Zhu, “Rapid eye movement tracking methodbased on fpga,” in 2016 International Conference on Machine Learning and Cybernetics (ICMLC), vol. 2, 2016, pp. 1021–1025.
A. L. X. Da Costa, C. A. D. Silva, M. F. Torquato, and M. A. C. Fernandes, “Parallelimplementation of particle swarm optimization on fpga,” IEEE Transactions onCircuits and Systems II: Express Briefs, vol. 66, no. 11, pp. 1875–1879, 2019.
J. Daugman, “How iris recognition works,” in Proceedings. International Conference on Image Processing, vol. 1, 2002, pp. I–I.
N. S. Nishino, K., Corneal imaging system: environment from eyes. Int J ComputVision 70, 2006.
N. M. T. Ohno and A. Yoshikawa, Freegaze:A gaze tracking system for everydaygaze interaction. Proc. Symp. ETRA 2002, 2002.
E. Guestrin and M. Eizenman, “General theory of remote gaze estimation using thepupil center and corneal reflections,” IEEE Transactions on Biomedical Engineering, vol. 53, no. 6, pp. 1124–1133, 2006.
G. J. . T. G. Garnier, S., “The biological principles of swarm intelligence,” p. 3–31,2007.
W.-C. Kao and Y.-C. Chiu, “Eyeball model construction and matching for visible60spectrum gaze tracking systems,” in 2018 IEEE 8th International Conference onConsumer Electronics - Berlin (ICCE-Berlin), 2018, pp. 1–2.
W.-C. Kao and S.-C. Lin, “Iris region matching for visible-spectrum gaze trackers,”in 2020 IEEE International Conference on Consumer Electronics (ICCE), 2020, pp.1–2.
B.-C. Chen, P.-C. Wu, and S.-Y. Chien, “Real-time eye localization, blink detection,and gaze estimation system without infrared illumination,” in 2015 IEEE International Conference on Image Processing (ICIP), 2015, pp. 715–719