视线追踪系统中眼睛跟踪方法研究

The Research on Eye Tracking for Gaze Tracking System

为解决视线追踪系统中红外图像瞳孔跟踪鲁棒性差的问题,提出一种基于伪彩色图的粒子滤波瞳孔跟踪算法.利用亮暗瞳现象,提出三通道伪彩色图(Triple-channel pseudo-color map,TCPCM)的概念,并将其引入瞳孔跟踪过程.TCPCM充分利用了各通道信息,瞳孔区域的色彩明显与人脸其他部位不同,提高了跟踪的稳定性与精确性.采用了直方图相似性度量与几何相似性度量相结合的二次更新的瞳孔感知模型,提高了粒子权重的可信性.针对实时性要求,引入快速特征提取步骤,减少特征提取的时间,提高特征提取的可靠性.实验结果表明,该算法在瞳孔目标检测效果、跟踪稳定性和运行时间方面都有良好的性能.

To address the problem of poor robustness of pupil tracking in infrared images for gaze tracking systems,this paper presents a particle filter algorithm for pupil tracking based on pseudo-color map.This paper also presents a concept of triple-channel pseudo-color map （TCPCM） by use of the phenomena of bright and dark pupil and leads it into pupil tracking.TCPCM makes full use of the information of each channel.The color of pupil zone is distinct from that of other regions of face obviously,which will improve the stability and accuracy of tracking.A two-phase-updating method based on similarity measurements of histogram and geometry is adopted to improve the credibility of the particle weights.Fast feature extraction is introduced to improve system real-time performance,which not only increases the computational effciency,but also enhances the robustness of the system.Results from an extensive experiment show a significant improvement of the proposed algorithm over existing eye tracking techniques for gaze tracking systems.