基于人眼追踪的手机解锁系统 报告人:李映辉 指导老师:王继良 Unlock with A Glance 基于人眼追踪的手机解锁系统 报告人:李映辉 指导老师:王继良
Background Mobile devices hold sensitive information Authentication schemes are prone to shoulder-surfing attacks Unlock with biological information sometimes fails
Related Works GazeTouchPass EyeDent On mobile devices Use gaze direction and touch to unlock phone Only detect gaze directions EyeDent Based on computer Use dedicated eye tracker
Related Works Gaze Tracking Method Accuracy(°) Speed(fps) Limitation CHM Mapping (1.9, 2.2) NA Based on computers EyeTab Model Based 6.88±1.8 12 Low Accuracy Pupil-Canthi-Ratio Gaussian Process Regression 3.9 Not accurate in certain area GT on Tablet ANN 4.42 0.7 Too slow
Design Pupil Detection Gaze Estimation Mapping Authentication
Challenges & Potential Solutions Gaze estimation is not accurate enough Use Cluster Partition method to filter wrong data Speed is important Fine-grained pupil detection is the bottleneck Use mapping-based gaze tracking technique
Timeline 周次 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 工作安排 文献阅读 方案设计 眼球追踪算法实现 解锁功能实现 中期答辩 后期性能优化 撰写论文 期末答辩
THANKS FOR LISTENNING Q & A