Abstract: As a new way of information collecting and human-computer interaction, the research of eye tracking technology has become a hot direction in computer vision. The core technology of gaze tracking is the gaze estimation. To solve problems of existing gaze estimation method including complex calibration and restricted head movement, an improved algorithm based on two-dimensional pupil and corneal reflection technique is proposed. Under the condition of single camera and single light source, gaze estimation by single point calibration is implemented through the establishment of the pupils-corneal reflection model, individual difference error compensation, and head motion error compensation. The experimental results show that the algorithm has greater accuracy when head moves in a certain range.

Key words: gaze estimation, corneal reflection, neural networks, particle swarm

摘要： 作为信息获取与人机交互的一种新型方式，视线跟踪技术已经成为计算机视觉领域的热门研究方向。视线跟踪的核心技术是视线估计。针对现有视线估计方法标定复杂、限制头部运动等问题，提出了一种改进的基于二维瞳孔角膜反射技术的视线估计方法。在单相机单光源条件下，通过建立瞳孔角膜反射模型、补偿个体差异误差、补偿头部运动误差等步骤实现单点标定视线估计。实验结果表明，用该算法估计视线，在一定范围内，头部移动不会带来精度的明显下降。

关键词: 视线估计, 角膜反射, 神经网络, 粒子群