稳态视觉诱发电位(SSVEP)技术是脑机接口的重要部分,当脑电图信号在强干扰与低信噪比环境时识别率较低。针对该问题,文中提出一种基于副瓣相消(SLC)与典型相关分析(CCA)的SSVEP识别方法。首先,利用希尔伯特变换将全部通道的脑电信号转换为解析信号;然后,将枕区的多个通道设为主瓣通道,其他含SSVEP较弱的多个通道设为副瓣通道,利用副瓣相消技术抑制主瓣通道脑电信号的干扰;最后,根据不同刺激图像的闪烁频率构造复数域单频率信号模板,利用CCA求解不同模板的最大典型相关系数,并根据其最大值确定SSVEP的分类结果。实测实验中,分别在环境安静状态与较大外界干扰状态下进行了多次试验,结果表明SLC处理对SSVEP的频谱增益有较大提升,相较于传统CCA方法,分类识别率更高,证明了本文方法的有效性和良好性能。

The steady-state visual evoked potential (SSVEP) technique is an important part of brain-computer interface, while its accuracy always suffers from external interferences and low signal-to-noise ratio of electroencephalogram signals. Focus on this issue, the recognition method of the SSVEP using the side-lobe cancellation (SLC) and canonical correlation analysis (CCA) is proposed. First, the analytic signals of the electroencephalogram signals of all channels are generalized by the Hilbert transform. Then, the main-lobe signals are formed by the signals of the channels in the occipital region, and the side-lobe signals are selected from other regions, thus the SLC technique can be achieved to suppress interference of the main-lobe signals. Finally, the complex-domain templates are constructed according to the flicker frequencies of different stimulus images, and the maximum canonical correlation coefficient is solved by the CCA method and used to classify the SSVEP. The experiments are executed several times in the interference-free and strong external inference environment respectively. The results show that the SLC step improves the frequency spectrum of the SSVEP signals, and the proposed method has better classification accuracy compared with the traditional CCA method, which verify the effectiveness and superior performance.

TN911.7

国家自然科学基金资助项目