飞行员使用HUD的行为模式识别研究

为了研究飞行员在使用平视显示器执行不同飞行任务时的行为模式，提出了一种包括飞行员眼动、头部运动和手部运动多种特征的行为识别框架。首先，开展行为模式研究实验，通过眼动仪获取眼部运动和头部运动，通过基于视频的手动跟踪获取手部运动。之后采用实验得到的结果对模型进行训练和测试。最后，对比了条件随机场和隐动态条件随机场在不同特征下的识别效果。结果表明，采用眼动特征加手部特征时，隐动态条件随机场模型对不同飞行任务的识别效果较好。

The pilot use HUD's behavior pattern recognition research.

In order to study the behavior patterns of pilots using HUD in different stages of the flight, a behavior recognition framework including pilots'' eye movements, head movements and hand movements is proposed. Firstly, get eye movement data and head position data through the Smart-Eye, get hand movement data by hand tracking based on video. Secondly, design experiment and using the experimental results to train and test the model. Finally, we compare the recognition effect of the different characteristics of the pilot in different stages of the flight between CRF and LDCRF. Experimental results show that the recognition accuracy of pilot eye movement and head position, using LDCRF, is 91.28%.