基于数学形态学的眼电信号识别及其应用  

Mathematical morphology based electro-oculography recognition algorithm for human-computer interaction

作　　者：陈卫东[1,2] 李昕[1,2] 刘俊[1,3] 郝耀耀[1,3] 廖玉玺[1,3] 苏煜[1,2] 张韶岷[1,3] 郑筱祥[1,3] 

机构地区：[1]浙江大学求是高等研究院,浙江杭州310027 [2]浙江大学计算机科学与技术学院,浙江杭州310027 [3]浙江大学生物医学工程与仪器学院,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2011年第4期644-649,共6页Journal of Zhejiang University：Engineering Science

基　　金：国家自然科学基金资助项目(60873125)

摘　　要：通过分析眼电(EOG)信号可以识别人眼球的运动状态及眨眼情况,进而设计一种新型的人机交互(HCI)系统.眼电信号通常包含一些干扰信息,如漂移、肌电干扰、运动伪迹.为了去除这些干扰信息,提出一种利用数学形态学对眼电信号进行处理的方法;通过阈值检测可以准确识别使用者眼球的运动状态和有意识眨眼.设计一个基于眼电的人机交互系统并通过健康与残疾被试的测试.实验结果显示,眼电信号识别的平均正确率达到96.2%,表明该方法可以应用于临床人机交互领域.Electro-oeulography （EOG） signals can be used for recognizing the directions of eye movements and voluntary eye blinks, which can be used to develop a new human-computer interaction （HCI） system. A mathematical morphology based algorithm was presented to process the EOG signals, which always contain some interference components, such as baseline drift, EMG interference and movement artifacts. The new approach can effectively reduce the artifacts and recognize the directions of eye movements and voluntary eye blinks by using a set of thresholds. A HCI system for disabled using the method was designed and tested by both healthy and disabled people. Experimental results showed that the average correct rate was 96.2%. The system can he employed in clinical HCI fields.

关 键 词：眼电(眼电图)技术(EOG) 数学形态学 信号处理 人机交互(HCI) 

分 类 号：R318.04[医药卫生—生物医学工程]