一种新的基于瞳孔-角膜反射技术的视线追踪方法

A Novel Eye Gaze Tracking Technique Based on Pupil Center Cornea Reflection Technique

针对现有单相机单光源视线追踪系统存在的几个问题:精度不高、头动受限以及标定复杂,提出了一种新的基于瞳孔-角膜反射(PCCR)技术的视线追踪方法.通过提出的瞳孔边缘滤波算法(RDPEF)和三通道伪彩色图(TCPCM)解决了近红外条件下瞳孔定位误差较大、瞳孔跟踪鲁棒性较差的问题,进而提高了视线特征提取的精度.通过提出的头部位置补偿方法以及个体差异转化模型,使二维映射模型允许使用者头部运动并且只需要单点标定.该方法提高了单相机视线追踪的精度和应用范围,为面向人机交互的视线追踪系统提供了有效的低成本解决方案.

For several common drawbacks of existing gaze tracking systems using single camera and single light source. Low accuracy, restricted head movement and tedious calibration, this paper presents a novel gaze tracking method based on pupil center cornea reflection （PCCR） technique. Pupil location for gaze tracking has been well investigated and all methods worked on nearinfrared images. This makes the eye details small and accuracy is affected. This paper addresses these problems by proposed radial distance pupil edge filter algorithm （RDPEF） and triple-channel pseudo-color map （TCPCM）. Our pupil tracker improves the accuracy of gaze feature extraction. The 2-D mapping-based gaze estimation methods have two common drawbacks. This paper addresses these problems by proposed head position compensation method and individual difference transformation model. The advantage of the proposed method based on 2-D mapping-based gaze estimation is that it can estimate the gaze of user accurately under natural head movement. Furthermore, it minimizes the calibration procedure to only one calibration point for a new individual. The proposed methods will improve the accuracy and usability of the gaze tracking using single camera and single light source, and provides an effective low-cost gaze tracking for human computer interaction.