基于空间滤波器组典型相关分析的SSVEP信号处理  

SSVEP Signal Processing Based on Spatial Filter Banks Canonical Correlation Analysis

作　　者：唐世泽 张学军[1] 杨忆 TANG Shize;ZHANG Xuejun;YANG Yi(Nanjing University of Posts and Telecommunications,College of Electronic and Optical Engineering,College of Flexible Electronics(Future Technology),Nanjing Jiangsu 210023)

机构地区：[1]南京邮电大学电子与光学工程学院、柔性电子(未来技术)学院,江苏南京210023

出　　处：《软件》2023年第4期26-34,共9页Software

基　　金：国家自然科学基金(61977039),“教学情境下的EEG情绪脑网络机理与分类研究”。

摘　　要：稳态视觉诱发电位(Steady-State Visual Evoked Potential,SSVEP)作为一种特征较为明显的脑电信号,有着训练迅速、数据集要求少、分类准确率高等优点,近年来广受关注。SSVEP信号采集的过程中会受到环境、设备、人工操作等因素的影响,因此模型对不同信噪比信号的稳定性显得尤为重要。本文提出一种空间滤波器组典型相关分析模型,在原始信号、人工合成信号、平均特征信号三者之间提取四种空间滤波器用于后续分类,并将该模型与另外四种模型在不同信噪比情况下进行对比分析。六组低信噪比数据集为采集数据、六组高信噪比数据集为清华数据集,实验证明该模型在使用高低信噪比数据时均有优秀的分类性能,最高可达99.24%,且对不同数据长度有较高鲁棒性,同时信息传递速率(Information Translate Rate,ITR)最高可达105.1bits/min。Steady-State Visual Evoked Potential,as an EEG signal with obvious characteristics,has the advantages of rapid training,less data set requirements,and high classification accuracy.It has been widely concerned in recent years.The SSVEP signal acquisition process will be affected by the environment,equipment,manual operation and other factors,so the stability of the model for signals with different SNR is particularly important.In this paper,a model of spatial filter banks canonical correlation analysis is proposed.Four spatial filters are extracted from the original signal,artificial reference signal and average characteristic signal for subsequent classification,and the model is compared with the other four models under different SNR.Six sets of low SNR datasets are collected data,and six sets of high SNR datasets are Tsinghua datasets.Experiments show that the model has excellent classification performance when using high and low SNR data,up to 99.24%,and has high robustness to different data lengths.At the same time,the ITR can reach up to 105.1 bits/min.

关 键 词：脑机接口技术 稳态视觉诱发电位 空间滤波器组典型相关分析 

分 类 号：TN911.7[电子电信—通信与信息系统]