一种新的基于显著图的视线跟踪方法

摘要

针对现有视线跟踪系统设备复杂、标定过程繁琐等方面的不足，提出了一种新的基于显著图的视线跟踪方法。通过红外光源设备在人眼角膜上产生的光斑中心与瞳孔中心建立瞳孔-角膜反射向量，然后将该向量作为视觉特征重构了基于显著图的视线跟踪算法。实验结果证明，提出的方法不仅缓解了视线跟踪系统标定过程繁琐的问题，而且对提高系统的精度和健壮性有一定的促进作用，这为面向人机交互的视线跟踪研究提供了可行的低成本解决方案。%For the deficiencies that existing eye gaze tracking devices are complex and calibration procedures are tedious, a novel eye gaze tracking method using saliency maps is proposed. With pupil center and reflection center on corneal generated by IR light device, a pupil-corneal reflection vector is constructed, which then acts as a kind of vision feature to reconstruct the eye gaze tracking algorithm based on saliency maps. The experiment result demonstrates that the proposed method not only can alleviate the tedious calibration of eye gaze tracking, but also has a little improvement in system accuracy and robustness, which provides a feasible low-cost eye gaze tracking research for human computer interaction.