基于稳态视觉诱发电位频相特征提取的脑机接口算法设计

发布时间：2018-02-06 00:48

本文关键词： 脑机接口稳态视觉诱发电位谱校正点通滤波全相位FFT　出处：《天津大学》2014年硕士论文　论文类型：学位论文

【摘要】：随着计算机科学、信号处理技术的飞速发展,以及残疾患者的生活质量提高意识不断加强,近年来在临床治疗中涌现出大量使用脑机接口(BCI)的康复辅助手段。其中基于稳态视觉诱发电位(SSVEP)的BCI因具有非侵犯性、系统配置简单及其高信息转化率的优势,故被选作大脑命令的良好载体。而衡量SSVEP-BCI系统性能的标准之一就是该系统可产生的命令数(即目标激励块数目),命令数越多,对应的执行动作也越多,系统也越完善。目前最常用的SSVEP命令识别方法是通过提取脑电信号的频率信息来实现的,而对于采用液晶显示器(LCD)激励产生的SSVEP信号,由于其激励频率是通过对LCD的刷新频率整数分频得到的,因而其激励频率数目受到限制,而且还不得不排除一些因非同步采样而无法直接准确检测到的激励频率。此外,因电极采集的脑电信号非常微弱及存在伪迹成分的干扰,都会影响检测精度,需要引入新的信号处理措施给予改善。为提高所容许的窄频带内可检测的命令数目,本文提出结合频率与相位差混合编码的BCI设计,在编码环节,通过赋予激励块以频率与相位相结合的特征而增大了目标激励数,有效地提高了可用频带的利用率。为提高目标检测精度,在解码环节,本文引入了频谱校正理论、全相位点通滤波器设计理论、全相位FFT谱分析理论和模式分类技术进行频率、相位信息提取。其中,借助频谱校正,检测出非同步采样下SSVEP的频率和相位值;借助全相位点通滤波,滤除了激励频率的带外干扰,提高了信号质量;利用全相位FFT谱分析的“相位不变性”和良好的抑制谱泄漏性能,进一步抑制了临近频率干扰,提升了相位估计精度,降低了计算复杂度;最后,结合最近邻分类法,完成了基于相位特征的目标识别。本文不但借助理论分析和仿真实验论证了以上方法在提取SSVEP频率与相位特征的可行性,而且通过现场数据采集与后续实验处理,结果表明:综合以上方法能够解决SSVEP激励信号在非同步采样下引起的测不准相位的问题,从而大大放宽了对目标激励频率的限制,增加了激励目标数目,具有很高的临床利用价值。
[Abstract]:With the rapid development of computer science, signal processing technology, and the quality of life of patients with disabilities to improve awareness is constantly strengthened. In recent years, a large number of rehabilitation aids using brain-computer interface (BCI) have emerged in clinical treatment. Among them, BCI based on steady-state visual evoked potential (SSVEP) is non-invasive. The system configuration is simple and has the advantage of high information conversion rate. Therefore, it is chosen as a good carrier of brain commands. One of the criteria for measuring the performance of SSVEP-BCI system is the number of commands that can be generated by the system (that is, the number of target excitation blocks, the number of commands). The more corresponding actions, the more perfect the system. At present, the most commonly used SSVEP command recognition method is to extract the frequency information of EEG. For the SSVEP signal excited by liquid crystal display (LCD), the number of the excitation frequency is limited because the excitation frequency is obtained by the integer division of the refresh frequency of the LCD. Moreover, some excitation frequencies which can not be detected directly and accurately because of asynchronous sampling have to be eliminated. In addition, the detection accuracy will be affected by the very weak EEG signals collected by electrodes and the interference of artifacts. New signal processing measures need to be introduced to improve it. In order to increase the number of detectable commands in the narrow band, this paper proposes a hybrid BCI design combining frequency and phase difference coding in the coding link. In order to improve the target detection accuracy, the target excitation number is increased by giving the excitation block the characteristic of combining frequency and phase, and the efficiency of available frequency band is improved effectively. In this paper, the spectrum correction theory, the design theory of all phase point pass filter, the all phase FFT spectrum analysis theory and the pattern classification technology are introduced to extract the frequency and phase information. The frequency and phase of SSVEP are detected under asynchronous sampling. With the help of all-phase point-pass filtering, the out-of-band interference of excitation frequency is filtered, and the signal quality is improved. Using the "phase invariance" of all phase FFT spectrum analysis and good spectral leakage suppression performance, the near frequency interference is further suppressed, the accuracy of phase estimation is improved, and the computational complexity is reduced. Finally, combined with the nearest neighbor classification, the target recognition based on phase features is completed. This paper not only uses theoretical analysis and simulation experiments to demonstrate the feasibility of the above methods in extracting SSVEP frequency and phase features. And through the field data acquisition and subsequent experimental processing, the results show that the synthesis of the above method can solve the problem of uncertain phase caused by asynchronous sampling of SSVEP excitation signal. Therefore, the limit of motivation frequency is greatly relaxed, and the number of incentive goals is increased, which is of high clinical value.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R318;TN911.7

【参考文献】