基于多传感器信息融合和感知的人体行为识别（HAR）技术已成为模式识别和物联网等领域的研究热点。现阶段，HAR的主要实现方法还停留在一些通用传感器层面上，而越来越多的研究表明，在不同的视觉行为状态下所引发的眼电信号（EOG）将会呈现出相应的特征模式。因此，EOG可以作为一种新的手段进行HAR研究。本项目以“基于EOG的人体行为识别算法及其应用研究”为课题，开展相关的理论、算法和应用研究，具体内容可概括为三个主要方面，即：1) 针对HAR的EOG信号处理新方法研究，主要包括：EOG采集范式的设计、行为相关EOG成分的获取和典型EOG单元的自动识别和分割等；2) EOG与视觉行为的关系模型建立及其鲁棒性研究；3) 基于EOG的HAR仿真实验平台的构建。通过本项目的研究，我们希望能够进一步丰富和完善EOG信号处理方法，有效拓展其应用范围。

Human Activity Recognition (HAR) technologies, which based on multi-sensor information fusion and perception, have become a hot topic in pattern recognition, Internet of Things, etc. However, the main methods of HAR implementation still rely on universal sensors. Recently, numerous studies have revealed that different EOG signals, caused by different specific visual activities, will present corresponding pattern characteristics. Thus, EOG signals could be used as a new method for HAR research. . In this project, “a research on EOG-based human activity recognition algorithm and its application” will be carried out. The main work can be summarized as three aspects, namely: 1) research on new EOG signal processing methods aimed to HAR, including EOG acquire experiments design, activity related EOG components identification, automatic recognition and segmentation of typical EOG unit; 2) establishment of activity relationship model between EOG and visual behaviors and its robustness studies; 3) implementation of an EOG-based HAR simulation platform. During the research, we hope to further enrich and improve EOG signals processing methods and effectively expand its applications field..