面向人机交互的眼动跟踪方法研究  

Eye-gaze tracking method for HCI system

作　　者：杨庆华[1] 张达磊[1] 荀一[1] 鲍官军[1] 

机构地区：[1]浙江工业大学机械工程学院,浙江杭州310014

出　　处：《机电工程》2016年第7期904-908,共5页Journal of Mechanical & Electrical Engineering

基　　金：国家科技支撑计划资助项目(2015BAF01B02)

摘　　要：针对患有严重身体残疾和行动不便的特殊人群难以与计算机进行交互的问题,对基于眼动跟踪的人机交互系统进行了研究,采用可穿戴式的系统硬件方案:将连接个人电脑的微型CCD摄像头通过3D打印成型的载具安装在普通眼镜上实现了眼图采集功能,运用基于梯度值和圆特性的改进的椭圆拟合方法检测定位瞳孔中心,并进行了基于二阶模型的注视点的标定,最终实现了对眼睛视线的跟踪检测。实验测试结果表明,该系统的水平方向视线检测误差和垂直方向视线检测误差分别在3.74%和5.34%,验证了所提出的眼动跟踪人机交互系统的可行性和有效性。Aiming at the difficulty of interaction with computer for people with severe physical disabilities and special people with mobility impairments,study on an interactive system based on eye-tracking was carried out. Wearable systems hardware is introduced as the solution：the CDD camera that connected with the PC is mounted on the spectacles via the carrier made through 3D printing. Through eye diagram capturing,utilizing the improved elipse fitting method to detect the center of the pupil,and calibrating the gaze point based on second-order model,eventually,the eye gaze tracking and detection in natural light is achieved. The user tests show that the error of eye-gaze detection on horizontal and vertical is 3. 74% and 5. 34% respectively,verifying the feasibility and effectiveness of the proposed interactive eye-tracking system.

关 键 词：眼动跟踪 人机交互 视线检测 

分 类 号：TP314.7[自动化与计算机技术—计算机软件与理论]