基于双眼同步运动特征约束的Kalman瞳孔跟踪算法

针对注视点研究中,红外光照明下双眼瞳孔运动的定位跟踪存在误差的问题,以双眼实时图像为研究对象,提出一种基于双眼同步运动特征约束的瞳孔跟踪算法.根据人类双眼在注视过程中的同步运动特征,把双眼瞳孔间距矢量作为隐式参数进行估计,简化包含左右眼位置、速度和双眼瞳孔间距的模型为统一的双眼同步跟踪模型,运用Kalman滤波器实现了运动特征估计和状态跟踪.实验采用自制的头戴式注视点传感装置进行眼部图像的采集.实验表明,该算法跟踪精度高,抗干扰能力强.相较于传统的以左右瞳孔位置与速度以及左右眼相对位置为状态量的算法,本文算法在位置跟踪和速度跟踪的鲁棒性上有明显改善,算法计算量也明显减少.另外,经过本文算法处理后的注视点位置估计精度大大提高,为注视点在人机交互领域中的进一步应用奠定了基础.

A Kalman pupil tracking algorithm based on binocular synchronous motion constraint

Aiming at the errors on localization and tracking of pupil motion under the infrared light illumination in the study of the gaze point,this paper presents a Kalman pupil tracking algorithm based on binocular pupil synchro-nous motion constraint. According to the synchronous motion characteristics of the human eyes,the binocular pupil spacing vector is used as an implicit parameter to be estimated,and the model including the left and right eye posi-tion,velocity and the binocular pupil distance is simplified as a unified binocular synchronization tracking model. Kalman filter is used to realize the motion feature estimation and state tracking. Experiments were performed using a self-made head-mounted sensor for eye image acquisition. Experiments show that the algorithm has high tracking precision and strong anti-interference ability. Compared with the traditional algorithm with the position and velocity of both eyes and their relative position,the proposed algorithm exhibits high robustness and low computational com-plexity. In addition,the accuracy of the gaze point is also improved greatly,which lays the foundation for the further study of the gaze point in the field of human-computer interaction.