基于Electrooculogram的眼动信息识别

眼动电图（Electrooculogram, EOG）是无创记录眼球运动的一种有效方法,它能记录到眼球的任何微小动作,甚至闭眼状态下眼球的动作.为能自动识别其所包含的眼动信息,本文从时域分析出发,提出了幅度阈值法和微分法2种眼动信息识别方法,并重点实现了微分算法.微分法识别眼动信息的基本思想是,对基线调整、平滑处理后的EOG信号进行求一阶微分的运算,然后对微分结果求极值,最后根据不同眼动模式下EOG波形的微分极值分布特征进行眼动信息的识别.通过该法对19个EOG样本中所包含的眼动信息的识别,最终得到了总误判率小于0.02的结果.最后,本文还给出了从眨眼EOG波形分析得来的眨眼时的眼球动作的现象.

Identification of eye-movement information based on Electrooculogram

EOG is a very effective eye-movement recording method, which is not only noninvasive, but also can record any small eye movements even if the eyes are closed. In order to detect the eye-movement information contained in EOG automatically, this paper gives two identification algorithms, the amplitude-threshold algorithm and the differentiation algorithm, which are all based on time-domain analysis. In differentiation algorithm, the EOG signal is base line adjusted, smoothed and 1-order differentiated. Then the extrema of the differentiation result are obtained. By comparing the differentiation extremum distribution of EOG waveform under different eye-movement modes, the characteristics of eye-movement can be detected. Furthermore, 19 EOG samples were identified in a specific eyemovement test experiment using the proposed differentiation algorithm. Experimental results indicate that differentiation algorithm can identify eye-movement effectively, and the overall error rate is less than 0.02. Finally, the paper also presents the phenomenon of eyeball action at the time of blinking.