摘要
近视会导致用户眼球发生形状畸变,近视用户眼球与视线估计中眼球参考模型的差异导致基于固定模型法得到的视线位置与用户实际视线位置不一致,这影响了视线跟踪的数据精度。针对以上问题,文中提出了一种基于眼球误差模型的眼动数据校正方法。结合眼动仪使用中例行的视线标定,通过计算视线标定过程中参考点与对应的视线估计点之间关系求得近视用户眼球与视线跟踪中眼球参考模型之间的眼球误差,然后根据眼球误差计算跟踪数据误差,最后将误差量补偿于眼动数据,并进行了计算机仿真验证。实验结果表明:近视用户眼球发生形状畸变时其眼动数据偏差不仅与近视程度有关而且与用户观察点位置密切相关。数据校正前,视线估计点与真实值平均偏离误差为48. 66pixels;数据校正后,视线估计点均匀分布在真实值周围,平均跟踪误差为28. 02pixels,并且可以验证,随着观察对象位置不同及近视程度变化,矫正后的数据偏差始终小于48pixels。说明以上提出的方法可以很好校正近视造成的眼动数据偏差。
Myopia can cause the distortion of eyeball. Differences of eyeballs between myopia and the reference eyeball model integrated in the eye-tracker can cause the deviations between the eye-tracking data derived by eye tracker and the actual gaze data. This paper aims to propose an eyeball model-based gaze data correction method for myopia. It obtains the systematic error of eyeball based on the actual gaze data and the eye-tracking data firstly; then calculates the data error based on systematic error model of eyeball; and corrects the eye-tracking data finally. This method is validated by computer simulations. The results show that the data deviation is affected by the distortions of eyeball model caused by myopia and the positions of gaze points. The mean deviation is 28. 02 pixels and the maximum deviation is less than 48 pixels. The results show that this method can correct the inaccurate eye-tracking data effectively.
引文
[1]李晨辉,庞志兵,张艳君,等.眼动测量在人机界面作业优化评估系统中的应用[J].技术与创新管理,2017,38(01):41-45.
[2]郑秀娟,栗战恒,张昀.基于视觉特性的驾驶安全眼动研究进展[J].技术与创新管理,2018,39(01):50-59.
[3]Morimoto C H,Koons D,Amir A,et al.Pupil detection and tracking using multiple light sources[J].Image&Vision Computing,2000,18(04):331-335.
[4]Shao G,Che M,Zhang B,et al.A novel simple 2D model of eye gaze estimation[C]//International Conference on Intelligent Human-Machine Systems and Cybernetics,2010:300-304.
[5]李作勇.关于近视[J].中国眼镜科技杂志,2000(06):30.
[6]Zhang Y,Hornof A J.Mode-of-disparities error correction of eye-tracking data[J].Behavior Research Methods,2011,43(03):834-842.
[7]Hornof A J,Halverson T.Cleaning up systematic error in eye-tracking data by using required fixation locations[J].Behavior Research Methods Instruments&Computers,2002,34(04):592.
[8]Schenk S,Dreiser M,Rigoll G,et al.Gaze Everywhere:Enabling Gaze-only User Interaction on an Unmodified Desktop PC in Everyday Scenarios[C].CHI Conference,2017:3034-3044.
[9]Gibaldi A,Vanegas M,Bex P J,et al.Evaluation of the tobii eye X eye tracking controller and Matlab toolkit for research[J].Behavior Research Methods,2017,49(03):923-946.
[10]Zhang Y,Hornof A J.Easy post-hoc spatial recalibration of eye tracking data[C].Symposium on Eye Tracking Research and Applications,2014:95-98.
[11]Biele C,Kobylinski P.Surface recalibration as a new method improving gaze-based human-computer interaction[C].International Conference on Intelligent Human Systems Integration,2018:197-202.
[12]Binaee K,Diaz G,Pelz J,et al.Binocular eye tracking calibration during a virtual ball catching task using head mounted display[C].ACM Symposium on Applied Perception,2016:15-18.
[13]Vadillo M A,Street C N,Beesley T,et al.A simple algorithm for the offline recalibration of eye-tracking data through best-fitting linear transformation.[J].Behavior Research Methods,2015,47(04):1365-1376.
[14]Zhang Y,Hornof A J.Easy post-hoc spatial recalibration of eye tracking data[C].Symposium on Eye Tracking Research and Applications,2014:95-98.
[15]Kyung-Nam K,Ramakrishna R S.Vision-based eye-gaze tracking for human computer interface[C].IEEE International Conference on Systems.IEEE,1999(02):324-329.
[16]余沁,刘龙繁,熊艳,等.文字强调方式对多媒体学习的影响---基于眼动追踪技术[J].生物医学工程研究,2018,37(01):96-100.
[17]Atchison D A,Jones C E,Schmid K L,et al.Eye shape in emmetropia and myopia[J].Investigative Ophthalmology&Visual Science,2004,45(10):3380.
[18]Mutti D O,Sholtz R I,Friedman N E,et al.Peripheral refraction and ocular shape in children[J].Investigative Ophthalmology&Visual Science,2000,41(05):1022-1030.
[19]周跃华,李志辉.近视眼角膜屈光力及眼轴的测定分析[J].中华眼科杂志,1995(05):356-358.