一种基于几何形状特征的实时瞳孔定位追踪技术

A Real-Time Pupil Positioning and Tracking Technology Based on Geometric Shape Features

在基于视觉的定位分析过程中,瞳孔作为重要的生理参数,易受鬓角、眉毛和光照强度等因素干扰,导致定位结果准确率不高,因此本文提出一种基于几何形状特征的实时瞳孔定位追踪技术.该技术方法结合两种不同定义的连通域外接矩形,通过面积阈值、角度差进行初步筛选,利用外接矩形长宽比进行二次筛选,最终实现瞳孔的精确定位.将该技术方法在5种不同数据集上进行准确性检测,定位结果的最高准确率提高到99.5%,平均误差减少到7.54%.由85名被试者在不同眼球注视情况下进行瞳孔实时定位追踪检测,结果表明该方法具有良好的鲁棒性,能够实现对瞳孔的实时追踪定位.

In the process of vision-based positioning analysis,the pupil,as an important physiological parameter,is easily disturbed by such factors as sideburns,eyebrows and light intensity,resulting in low accuracy of positioning results.There-fore,a real-time pupil positioning and tracking technology based on geometric shape features is proposed in this study.This method combines two differently defined circumscribed rectangles of connected domains,conducts preliminary screening through area threshold and angle difference,and uses the circumscribed rectangle aspect ratio for secondary screening,which finally achieves the precise pupil positioning.The accuracy of the technical method was tested on five different data setswith the highest accuracy of the positioning results increased to 99.5%,and the average error reduced to 7.54%.Real-time detec-tion was performed by 85 subjects under different eye fixation conditions,and the results show that this method has good robustness and can realize real-time pupil tracking and positioning.