基于稳态视觉诱发电位的脑机接口性能与应用研究

发布时间：2018-03-16 08:35

本文选题：稳态视觉诱发电位　切入点：典型相关分析　出处：《南京邮电大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着计算机技术的崛起,脑电信号的研究步入了快速发展的阶段。目前,脑电信号的研究主要采用脑机接口(Brain Computer Interface,BCI)系统。该系统在大脑和外设之间建立连接通路并传送脑电信息,以直接控制外部设备的运转。其中,以稳态视觉诱发电位(Steady State Visual Evoked Potential,SSVEP)为典型响应信号的SSVEP系统应用最为广泛。本文在前人研究工作的基础上,基于BCI技术收集并提取SSVEP信号。通过分析SSVEP信号的特征频率,比较了在不同视觉诱发下SSVEP信号的特点,最终实现了由不同频率的黑白方块诱发SSVEP信号对手机音乐播放系统的实时控制。首先,本文采用Neuroscan系统采集了由黑白方块刺激所产生的SSVEP信号。SSVEP信号是人眼受到持续的视觉刺激而产生的节律性脑电信号。目前研究SSVEP的方法有很多,本文基于典型相关分析(Canonical Correlation Analysis,CCA)技术,提取了SSVEP信号特征频率等性能指标,分析了数据长度、信道数目、大脑区域、刺激图片性质与特征频率的提取准确性之间的关系,并得出了定量的结论。这为后续SSVEP控制音乐播放系统的研究奠定了理论基础。其次,基于改进的二维集合经验模态分解(Improved Two-Dimensional Ensemble Empirical Mode Decomposition,2D-EEMD)算法,将黑白方块刺激、棋盘翻转刺激、黑白横条纹刺激和黑白竖条纹刺激下分别收集到的SSVEP信号进行预处理,并综合运用傅里叶变换法,比较了在四种图形刺激下SSVEP信号响应性能的差异。结果表明,在黑白方块刺激下,SSVEP信号在各个脑区保持宏观一致;而在棋盘翻转刺激和黑白横竖条纹刺激下,SSVEP信号在额叶、颞叶、顶叶、枕叶的主响应频率不尽相同,区域差异明显。最后,本文设计并实现了基于SSVEP信号的脑电波无线控制手机音乐播放系统“SSVEPControl”:在黑白方块刺激下,将通过CCA技术所提取的SSVEP信号的特征频率作为指令,利用无线设备传送到手机客户端,实现在线音乐的“播放/暂停”、“保持”,“上一曲”,“下一曲”等四种操作功能。实验表明,该系统可做到5秒延迟的实时操作,准确率达到95%以上。
[Abstract]:With the rise of computer technology, the research of EEG has stepped into the stage of rapid development. Brain-computer interface brain Computer Interface (BCI) system is used in the study of EEG signals. The system establishes a connection between the brain and peripheral devices and transmits EEG information to directly control the operation of external devices. The SSVEP system with Steady State Visual Evoked potential signal as the typical response signal is the most widely used. Based on the previous research work, this paper collects and extracts the SSVEP signal based on BCI technology, and analyzes the characteristic frequency of SSVEP signal. This paper compares the characteristics of SSVEP signals under different visual evoked conditions, and finally realizes the real-time control of SSVEP signals induced by black and white squares with different frequencies to the music playing system of mobile phones. In this paper, Neuroscan system is used to collect SSVEP signal. SSVEP signal generated by black and white block stimulation is a rhythmic EEG signal produced by continuous visual stimulation in human eyes. There are many methods to study SSVEP at present. Based on canonical Correlation Analysis (CAA) technique, this paper extracts the characteristic frequency of SSVEP signal, and analyzes the relationship between the data length, the number of channels, the brain region, the nature of stimulus picture and the accuracy of feature frequency extraction. The quantitative conclusion is obtained, which lays a theoretical foundation for the subsequent research of SSVEP control music playback system. Secondly, based on the improved Two-Dimensional Ensemble Empirical Mode decomposition 2D-EEMDM algorithm, the black-and-white block stimulation and the chessboard flip stimulation are used. The SSVEP signals collected under black and white transverse stripe stimulation and black and white vertical stripe stimulation were preprocessed, and Fourier transform method was used to compare the response performance of SSVEP signals under four kinds of graphic stimuli. Under the stimulation of black and white squares, SSVEP signals remained consistent in all brain regions, while the main response frequencies of SSVEP signals in frontal lobe, temporal lobe, parietal lobe and occipital lobe were different under the stimulation of checkerboard flipping and black and white vertical stripes. Finally, there were significant regional differences in the main response frequency of SSVEP signal in the frontal lobe, temporal lobe, parietal lobe and occipital lobe. In this paper, we design and implement the brain wave wireless control handset music playing system "SSVEPControl" based on SSVEP signal. Under the stimulation of black and white square, the characteristic frequency of SSVEP signal extracted by CCA technology is taken as instruction, and the wireless device is used to transmit it to the mobile phone client. Four kinds of operation functions of online music, such as "play / pause", "hold", "last song" and "next song", are realized. The experiment shows that the system can achieve real-time operation of 5 seconds delay and the accuracy is more than 95%.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R318;TN911.7

【参考文献】