In order to solve the problem of low accuracy and poor robustness due to the interference of reflected light spot in the positioning of pupil center, this paper presented an algorithm for accurate positioning of pupil center for embedded system. Firstly, the Average of Synthetic Exact Filters (ASEF) was constructed by pre-processing the human eye image with contrast enhancement and marking the central position of the pupil. Next, the average synthetic precision filter was used to determine the location of the monocular pupil, and the binocular pupil was determined by the “three courts and five eyes” face ratio algorithm to realize the coarse positioning of the pupil. Then, taking the coarse positioning coordinates as the center, the region of interest was selected several times for projection operation to reduce the interference of reflected light spots, so as to corrected the pupil center coordinates, and finally achieved the accurate positioning of the binocular pupil center. The fitting ellipse method, Starburst model method and the proposed algorithm were used to detect simulated human eyes and real human eyes. The experimental results show that the proposed algorithm can effectively avoid the interference of reflected light spots on the positioning of pupil center, accurately obtain the central position of pupil, and the running time reaches 0.035 s. The detection error is controlled within 1 pixel. Compared with elliptic fitting method and Starburst model method, the proposed algorithm has better performance in detection speed and accuracy, and can meet the practical application requirements of real-time ophthalmic instrument and embedded system deployment.
CAIHuaiyu, SHIYu, LOUShiliang, et al. Pupil location algorithm applied to infrared ophthalmic disease detection[J]. Chinese Optics, 2021, 14(3): 605-614. (in Chinese)
[3]
XIAY, LOUJ, DONGJ, et al. Hybrid regression and isophote curvature for accurate eye center localization[J]. Multimedia Tools and Applications, 2020, 79(1): 805-824.
[4]
HSUW Y, CHUNGC J.A novel eye center localization method for multiview faces[J].Pattern Recognition, 2021, 119: 108078.
WANGJunxiang, MAJianwei, SONGDiqiu, et al. Research on facial action control mouse operationbased on opencv and media pipe[J]. Computing Technology and Automation, 2024, 43(2): 177-181. (in Chinese)
[7]
POULOPOULOSN, PSARAKISE Z. A real-time high precision eye center localizer[J]. Journal of Real-Time Image Processing, 2022, 19(2): 475-486.
ZHANGHongwei, WANGShiyang, LIXianlong, et al. Research and implementation of pupil recognition based on Hough transform[J]. Chinese Journal of Liquid Crystals and Displays, 2016, 31(6): 621-625. (in Chinese)
ZHANGJing, XUXuyun, LIUGuannan . et al. A pupil location method based on hybrid projection and iris location[J]. Electronics World, 2017(13): 175. (in Chinese)
WANGPeng, WENHongtao, WangShilong, et al. High-precision pupil center positioning method based on near-eye infrared image[J]. Journal of Harbin University of Science and technology, 2022, 27(5): 38-46.(in Chinese)
[14]
LUC, XIAS, SHAOM, et al.Arc-support line segments revisited: an efficient high-quality ellipse detection[J] in IEEE Transactions on Image Processing, 2019, 29: 768-781.
[15]
DEWID A S, WIBIRAMAS, ARDIYANTOI, et al. Robust pupil localization algorithm under off-axial pupil occlusion[C]//2nd International Conference on Bioinformatics, Biotechnology and Biomedical Engineering (BioMIC)-Bioinformatics and Biomedical Engineering.IEEE, 2019: 1-6.
[16]
YUP, DUANW, SUNY, et al. A Pupil-positioning method based on the starburst model[J]. Computers, Materials & Continua, 2020, 64(2): 1199-1217.
[17]
CHENJ C, YUP Q, YAOC Y, et al. Eye detection and coarse localization of pupil for video-based eye tracking systems[J]. Expert Systems with Applications, 2024, 236: 121316.1-121316.12.
[18]
GOUC, ZHONGR, YUY Z, et al. MAL-Net: multiscale attention link network for accurate eye center detection[J]. Computer Vision and Image Understanding, 2023, 234: 103750.
FANShunyi, NILei, LIUBinbin, et al. Robust visual tracking algorithm based on peak characteristics to determine model updating[J]. Journal of Air & Space Early Warning Research, 2024, 38(1): 50-56. (in Chinese)
TANGYunmin. Research on digital media image information recognition of smart terminal based on contour projection algorithm[J]. Automation & Instrumentation, 2024(2): 20-23.(in Chinese)