CN113180706B - FHN stochastic resonance-based SSVEP characteristic frequency extraction method - Google Patents
FHN stochastic resonance-based SSVEP characteristic frequency extraction method Download PDFInfo
- Publication number
- CN113180706B CN113180706B CN202110419436.4A CN202110419436A CN113180706B CN 113180706 B CN113180706 B CN 113180706B CN 202110419436 A CN202110419436 A CN 202110419436A CN 113180706 B CN113180706 B CN 113180706B
- Authority
- CN
- China
- Prior art keywords
- fhn
- stochastic resonance
- frequency
- ssvep
- noise
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000605 extraction Methods 0.000 title claims abstract description 13
- 238000012545 processing Methods 0.000 claims abstract description 21
- 238000001514 detection method Methods 0.000 claims abstract description 11
- 238000001914 filtration Methods 0.000 claims abstract description 9
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 4
- 238000000034 method Methods 0.000 claims description 21
- 238000004364 calculation method Methods 0.000 claims description 6
- 239000012528 membrane Substances 0.000 claims description 6
- 230000005284 excitation Effects 0.000 claims description 5
- 210000002569 neuron Anatomy 0.000 claims description 5
- 238000005070 sampling Methods 0.000 claims description 5
- 230000000638 stimulation Effects 0.000 claims description 5
- 210000004556 brain Anatomy 0.000 claims description 4
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000010304 firing Methods 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 230000003321 amplification Effects 0.000 claims description 2
- 238000005311 autocorrelation function Methods 0.000 claims description 2
- 210000000624 ear auricle Anatomy 0.000 claims description 2
- 210000001061 forehead Anatomy 0.000 claims description 2
- 238000003199 nucleic acid amplification method Methods 0.000 claims description 2
- 230000000737 periodic effect Effects 0.000 claims description 2
- 150000002500 ions Chemical class 0.000 claims 1
- 238000007781 pre-processing Methods 0.000 abstract description 3
- 238000001228 spectrum Methods 0.000 description 6
- 230000000007 visual effect Effects 0.000 description 6
- 230000000694 effects Effects 0.000 description 5
- 230000004044 response Effects 0.000 description 4
- 230000001629 suppression Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000009774 resonance method Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000000763 evoking effect Effects 0.000 description 2
- 210000000869 occipital lobe Anatomy 0.000 description 2
- 206010033799 Paralysis Diseases 0.000 description 1
- 238000002266 amputation Methods 0.000 description 1
- 230000002238 attenuated effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 210000003169 central nervous system Anatomy 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 210000003128 head Anatomy 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000017074 necrotic cell death Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000002269 spontaneous effect Effects 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Landscapes
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
Description
技术领域technical field
本发明涉及脑-机接口技术领域，具体涉及一种基于FHN随机共振的SSVEP特征频率提取方法。The invention relates to the technical field of brain-computer interface, in particular to an SSVEP feature frequency extraction method based on FHN stochastic resonance.
背景技术Background technique
脑-机接口技术通过对脑电信号中包含的运动意图进行解码，并转化为不同的驱动命令来实现人脑对外部设备的直接控制。作为一种新型的人机交互手段，脑-机接口给部分神经坏死、脑中风、高位截肢、重度瘫痪等患者带来了自主生活的新希望。稳态视觉诱发电位是人眼在接受视觉刺激后在大脑枕叶区产生的一组特定的脑电信号，与P300、运动想象信号及自发脑电相比，具有周期稳定、特征明显且无需训练的特点，已成为脑-机接口最常用的控制信号之一。Brain-computer interface technology realizes direct control of external devices by the human brain by decoding the motor intention contained in the EEG signal and converting it into different driving commands. As a new type of human-computer interaction method, the brain-computer interface has brought new hope for independent life to some patients with nerve necrosis, stroke, high amputation, and severe paralysis. Steady-state visual evoked potentials are a set of specific EEG signals generated by the human eye in the occipital lobe area of the brain after receiving visual stimulation. Compared with P300, motor imagery signals and spontaneous EEG, it has stable cycles, obvious characteristics and does not require training. It has become one of the most commonly used control signals for brain-computer interfaces.
目前，绝大多数SSVEP提取方法均建立在线性框架下，将噪声视为有害信息，通过对抑制噪声来突出信噪比，提高微弱信号的检测能力。虽然，这些方法均能在低信噪比状况下不同程度的提取出原始EEG中包含的信息，体现出一定的SSVEP检测能力，却都不能避免以下问题：(1)为了消除EEG中包含的多尺度噪声，需要选择合适的带通滤波器，而带通滤波器的边缘效应会减少有效数据长度，并增加检测时间，此外，还需要考虑带通滤波器的通带范围与信号特征频率的自适应匹配。(2)使用线性方法来提取具有明显非线性和非平稳特征的SSVEP，有用信号在噪声被抑制的同时也会衰减或丢失。当诱发信号的稳定性不足时，对有用信号的抑制程度甚至会远远超过抑制噪声。因此，原始EEG中包含的信息无法被完全利用，影响检测灵敏度和识别精度。At present, most SSVEP extraction methods are based on the linear framework, and noise is regarded as harmful information, and the signal-to-noise ratio is highlighted by suppressing noise to improve the detection ability of weak signals. Although these methods can extract the information contained in the original EEG to varying degrees under the condition of low signal-to-noise ratio, and reflect a certain SSVEP detection ability, they cannot avoid the following problems: (1) In order to eliminate the multiple information contained in the EEG scale noise, it is necessary to select an appropriate band-pass filter, and the edge effect of the band-pass filter will reduce the effective data length and increase the detection time. Adapt to match. (2) Using a linear method to extract SSVEP with obvious nonlinear and non-stationary features, the useful signal will be attenuated or lost while the noise is suppressed. When the stability of the evoked signal is insufficient, the degree of suppression of the useful signal is even far greater than that of suppressing the noise. Therefore, the information contained in the original EEG cannot be fully utilized, affecting detection sensitivity and recognition accuracy.
发明内容Contents of the invention
为了克服上述现有技术的缺点，本发明的目的在于提供一种基于FHN随机共振的SSVEP特征频率提取方法，利用多通道EEG中包含的噪声增强SSVEP的频谱图，并结合FHN输出频率响应相当于一组非线性带通滤波器且通带范围可调的特性，将原始EEG信号送入FHN非周期随机共振模型进行噪声增强，来保留SSVEP的全部信息，从而实现特征频率的高精度识别。In order to overcome the above-mentioned shortcoming of prior art, the object of the present invention is to provide a kind of SSVEP characteristic frequency extraction method based on FHN stochastic resonance, utilize the noise enhancement SSVEP spectrogram contained in the multi-channel EEG, and combine FHN output frequency response to be equivalent to A set of non-linear band-pass filters with adjustable pass-band range, the original EEG signal is sent to the FHN aperiodic stochastic resonance model for noise enhancement to retain all the information of SSVEP, so as to achieve high-precision identification of characteristic frequencies.
为了达到上述目的，本发明采取的技术方案是：In order to achieve the above object, the technical scheme that the present invention takes is:
一种基于FHN随机共振的SSVEP特征频率提取方法，包括以下步骤：A kind of SSVEP characteristic frequency extraction method based on FHN stochastic resonance, comprises the following steps:
1)多通道数据采集：对被试者进行多通道EEG信号采集；多通道EEG信号经过放大、滤波与数模转化处理；1) Multi-channel data collection: Collect multi-channel EEG signals from the subjects; the multi-channel EEG signals are amplified, filtered and digital-to-analog conversion processed;
2)信号预处理：2) Signal preprocessing:
2.1)多通道信号降维：采用共平均参考算法来降低多信道信号的维度；2.1) Dimensionality reduction of multi-channel signals: the common average reference algorithm is used to reduce the dimensionality of multi-channel signals;
2.2)低通滤波处理：用巴特沃斯滤波器滤除低频噪声；2.2) Low-pass filter processing: filter out low-frequency noise with a Butterworth filter;
3)FHN随机共振参数初始化及模型处理：设置计算参数，包括模型参数ε以及需要识别的最大峰值阶次N；3) FHN stochastic resonance parameter initialization and model processing: set calculation parameters, including model parameter ε and the maximum peak order N to be identified;
将预处理后带有噪声干扰的SSVEP信号送入到相应的模型以进行FHN随机共振处理，再通过快速傅里叶变换计算噪声增强的SSVEP的频谱图以识别目标频率；Send the preprocessed SSVEP signal with noise interference to the corresponding model for FHN stochastic resonance processing, and then calculate the spectrogram of the noise-enhanced SSVEP by fast Fourier transform to identify the target frequency;
4)峰值频率识别：从步骤3)中获得的输出信号的频谱图中，分别提取第N阶主峰对应的特征频率；4) peak frequency identification: from the spectrogram of the output signal obtained in step 3), respectively extract the characteristic frequency corresponding to the Nth order main peak;
5)频率匹配检测：将识别频率与所有刺激频率进行匹配，如果匹配成功，则目标频率被有效识别；如果匹配失败，则检测当前识别的阶次是否大于设定的最大阶次；如果终止条件满足，则检测结束，表明目标频率标识失败；否则，计算返回到步骤4)。5) Frequency matching detection: match the recognition frequency with all stimulation frequencies, if the matching is successful, the target frequency is effectively recognized; if the matching fails, then detect whether the current recognized order is greater than the set maximum order; Satisfied, the detection ends, indicating that the target frequency identification fails; otherwise, the calculation returns to step 4).
所述的步骤1)多通道EEG信号采集中采集电极按照10/20电极分布标准设置，参考电极(Ref)位于大脑前额(FPz)，接地电极(GND)位于单侧左耳垂(A1)，用OZ、O1、O2、POZ、PO3、PO4、PO5、PO6八个通道来记录脑电信号，各导联的采样频率为250Hz。In the step 1) in the multi-channel EEG signal acquisition, the acquisition electrodes are set according to the 10/20 electrode distribution standard, the reference electrode (Ref) is located on the forehead of the brain (FPz), and the ground electrode (GND) is located on the unilateral left earlobe (A1). Eight channels of OZ, O1, O2, POZ, PO3, PO4, PO5, and PO6 are used to record EEG signals, and the sampling frequency of each lead is 250Hz.
所述的步骤2.1)中以OZ通道为基准通道，选取PO5、PO3、PO6、O2四个通道平均值作为共同平均的参考通道。In the step 2.1), the OZ channel is used as the reference channel, and the average values of the four channels PO5, PO3, PO6, and O2 are selected as the common average reference channel.
所述的步骤2.2)中通带纹波设置为1，阻带纹波设置为10。In the step 2.2), the passband ripple is set to 1, and the stopband ripple is set to 10.
所述的步骤3)中FHN随机共振模型的数学表达式为：Described step 3) in the mathematical expression of FHN stochastic resonance model is:
式中：v(t)——细胞膜电压，是一个快变量；w(t)——膜内离子浓度，是一个慢变量；A——为常数表示激励幅值，促使神经元定期点火；ε——时间参数常量，决定了神经元点火的速率，此处取值为0.04；b——参数常量，值为0.15；n(t)——高斯白噪声，均值为零且自相关函数满足<n(t)n(s)>＝2Dδ(t-s)；<.——求整体均值；s(t)——输入的非周期激励信号，该微分方程组求解时采用四阶Runge—Kuta方法；In the formula: v(t)—cell membrane voltage, which is a fast variable; w(t)——membrane ion concentration, which is a slow variable; A——a constant representing the excitation amplitude, which prompts neurons to fire regularly; ε ——Time parameter constant, which determines the firing rate of neurons, the value here is 0.04; b——parameter constant, the value is 0.15; n(t)——Gaussian white noise, the mean value is zero and the autocorrelation function satisfies < n(t)n(s)>=2Dδ(t-s); <.——seeking the overall mean value; s(t)——input aperiodic excitation signal, and the fourth-order Runge-Kuta method is used when solving the differential equations;
当a＝0.5时，令v(t)＝v(t)′+1/2，w(t)＝w(t)′-b+1/2，A＝A′-b+1/2，FHN随机共振模型被简化为以下形式：When a=0.5, let v(t)=v(t)'+1/2, w(t)=w(t)'-b+1/2, A=A'-b+1/2, The FHN stochastic resonance model is simplified to the following form:
式中：——阈值电压；B——信号幅值到阈值电压的距离；In the formula: ——threshold voltage; B——distance from signal amplitude to threshold voltage;
令AT-B＝0，则只需要设置和调整模型参数ε和需要识别的最大峰值阶次N。Let A T -B=0, then It is only necessary to set and adjust the model parameter ε and the maximum peak order N to be identified.
本发明的有益效果为：The beneficial effects of the present invention are:
(1)本发明FHN随机共振处理考虑了随机噪声对系统输出的影响，进一步加强了对噪声的抑制，得到的信号更加平滑。(1) The FHN stochastic resonance processing of the present invention considers the influence of random noise on the system output, further strengthens the suppression of noise, and obtains a smoother signal.
(2)本发明FHN随机共振处理的输出频率响应类似于一组非线性带通滤波器，只有一个可调模型参数ε且通带范围可调，适合用于SSVEP多尺度噪声的抑制。(2) The output frequency response of FHN stochastic resonance processing in the present invention is similar to a group of nonlinear bandpass filters, with only one adjustable model parameter ε and adjustable passband range, which is suitable for the suppression of SSVEP multi-scale noise.
(3)本发明利用噪声能量来增强SSVEP，避免有用信号受损及滤波器边缘效应，提高SSVEP特征频率的识别正确率。(3) The present invention utilizes noise energy to enhance SSVEP, avoids useful signal damage and filter edge effect, and improves the identification accuracy rate of SSVEP characteristic frequency.
附图说明Description of drawings
图1是本发明的流程图。Fig. 1 is a flow chart of the present invention.
图2是本发明模型参数ε对随机共振输出影响时得到的不同参数下FHN随机共振输出信号曲线。Fig. 2 is the FHN stochastic resonance output signal curves under different parameters obtained when the model parameter ε of the present invention influences the stochastic resonance output.
图3为采用常规CCA方法识别特征频率时，得到的滤除8Hz以下低频成分的EEG信号与模板信号的CCA系数谱。Figure 3 is the CCA coefficient spectrum of the EEG signal and the template signal obtained by filtering the low frequency components below 8 Hz when the conventional CCA method is used to identify the characteristic frequency.
图4为采用本发明FHN随机共振方法识别特征频率时，得到的滤除8Hz以下低频成分EEG信号频谱图。Fig. 4 is a spectrum diagram of the EEG signal obtained by filtering out low-frequency components below 8 Hz when the FHN stochastic resonance method of the present invention is used to identify the characteristic frequency.
图5为15名正常被试通过常规CCA方法和本发明FHN随机共振方法得到的特征频率识别正确率的结果图。Fig. 5 is a result graph of the correct rate of characteristic frequency recognition obtained by 15 normal subjects through the conventional CCA method and the FHN stochastic resonance method of the present invention.
具体实施方式Detailed ways
以下结合附图对本发明进一步的详细说明。The present invention will be further described in detail below in conjunction with the accompanying drawings.
参照图1，一种基于FHN随机共振的SSVEP特征频率提取方法，包括以下步骤：With reference to Fig. 1, a kind of SSVEP characteristic frequency extraction method based on FHN stochastic resonance comprises the following steps:
1)多通道数据采集：通过g.USBamp(g.tec Inc.,Austria)脑电采集系统对被试者进行多通道EEG信号采集，EEG采集过程中以单侧耳垂接地，在正面位置(FPz)作为参考电极。靠近枕叶的OZ、O1、O2、POZ、PO3、PO4、PO5、PO6八个通道来记录脑电信号，各导联的采样频率为250Hz；电极与脑电采集系统的输入连接，经过放大、滤波与数模转化处理，脑电采集系统输出脑电信号数据，与数据处理模块的输入连接；1) Multi-channel data acquisition: The subjects were collected multi-channel EEG signals through the g.USBamp (g.tec Inc., Austria) EEG acquisition system. ) as a reference electrode. The eight channels of OZ, O1, O2, POZ, PO3, PO4, PO5, and PO6 near the occipital lobe are used to record EEG signals. The sampling frequency of each lead is 250Hz; the electrodes are connected to the input of the EEG acquisition system, and after amplification, Filtering and digital-to-analog conversion processing, the EEG acquisition system outputs EEG signal data, and is connected to the input of the data processing module;
模型参数ε对随机共振输出的影响分析：参照图2，建立一组标准正弦仿真信号(正弦信号幅值A＝5，频率f＝0.5HZ，采样频率fs＝1000HZ，采样点数N＝10000)并加入一定D＝20的噪声作为模型的输入，进行FHN随机共振处理，调节系统参数可获取不同表现的系统输出，由此得到的不同模型参数输入下FHN随机共振系统的输出信号，可以从图中看出，当ε＝0.01时，输出信号存在较大的随机波动，此时噪声的随机干扰起到了主导作用，得到信号的毛刺很大；随着模型参数ε增大到0.04，输出信号中的波动成分被逐步抑制，系统的响应得到改善；但过大的模型参数ε会使得系统输出状态在转移过程中无法跟上输入信号的响应速度，输出信号波形产生畸变；同时，噪声和驱动信号的幅值也被大幅滤除，造成输出信号失真；因此，对于不同的输入信号，会存在一个最佳模型参数ε，使得FHN随机共振系统具有最好的滤波效果；Analysis of the influence of model parameter ε on stochastic resonance output: with reference to Figure 2, set up a group of standard sinusoidal simulation signals (sinusoidal signal amplitude A=5, frequency f=0.5HZ, sampling frequency fs=1000HZ, sampling point number N=10000) and Add a certain noise of D=20 as the input of the model, perform FHN stochastic resonance processing, and adjust the system parameters to obtain system outputs with different performances. The output signals of the FHN stochastic resonance system obtained by inputting different model parameters can be obtained from the figure It can be seen that when ε=0.01, the output signal has large random fluctuations, and the random interference of noise plays a leading role at this time, and the glitch of the obtained signal is very large; as the model parameter ε increases to 0.04, the output signal The fluctuation component is gradually suppressed, and the response of the system is improved; but too large a model parameter ε will make the output state of the system unable to keep up with the response speed of the input signal during the transfer process, and the output signal waveform will be distorted; at the same time, noise and drive signal The amplitude is also largely filtered out, resulting in distortion of the output signal; therefore, for different input signals, there will be an optimal model parameter ε, which makes the FHN stochastic resonance system have the best filtering effect;
2)信号预处理：2) Signal preprocessing:
2.1)多通道信号降维：为了完全利用每个信道中包含的信息，数据处理模块采用共平均参考算法来降低多信道信号的维度，以OZ通道为基准通道，选取PO5、PO3、PO6、O2四个通道平均值作为共同平均的参考通道；2.1) Dimensionality reduction of multi-channel signals: In order to fully utilize the information contained in each channel, the data processing module adopts a common average reference algorithm to reduce the dimension of multi-channel signals. Taking the OZ channel as the reference channel, select PO5, PO3, PO6, O2 The average of the four channels is used as the reference channel for the common average;
2.2)低通滤波处理：用巴特沃斯滤波器滤除低频噪声，通带纹波设置为1，阻带纹波设置为10，防止低频噪声对识别特征频率的干扰；2.2) Low-pass filtering processing: filter out low-frequency noise with a Butterworth filter, set the passband ripple to 1, and the stopband ripple to 10 to prevent low-frequency noise from interfering with the identification feature frequency;
3)FHN随机共振参数初始化及模型处理：根据采集到的信号特点和实际分析需要设置计算参数，包括模型参数ε以及需要识别的最大峰值阶次N；3) FHN stochastic resonance parameter initialization and model processing: set the calculation parameters according to the collected signal characteristics and actual analysis needs, including the model parameter ε and the maximum peak order N to be identified;
FHN随机共振模型的数学表达式为：The mathematical expression of the FHN stochastic resonance model is:
式中：v(t)——细胞膜电压，是一个快变量；w(t)——膜内离子浓度，是一个慢变量；A——为常数表示激励幅值，促使神经元定期点火；ε——时间参数常量，决定了神经元点火的速率，此处取值为0.04，下同；b——参数常量，值为0.15；n(t)——高斯白噪声，均值为零且自相关函数满足<n(t)n(s)>＝2Dδ(t-s)；<.>——求整体均值；s(t)——输入的非周期激励信号，该微分方程组求解时采用四阶Runge—Kuta方法；In the formula: v(t)—cell membrane voltage, which is a fast variable; w(t)——membrane ion concentration, which is a slow variable; A——a constant representing the excitation amplitude, which prompts neurons to fire regularly; ε ——Time parameter constant, which determines the firing rate of neurons, the value here is 0.04, the same below; b——parameter constant, the value is 0.15; n(t)——Gaussian white noise, the mean value is zero and autocorrelation The function satisfies <n(t)n(s)>=2Dδ(t-s); <.>—to find the overall mean value; s(t)——input non-periodic excitation signal, and the fourth-order Runge is used to solve the differential equations — Kuta method;
当a＝0.5时，令v(t)＝v(t)′+1/2，w(t)＝w(t)′-b+1/2，A＝A′-b+1/2，FHN随机共振模型被简化为以下形式：When a=0.5, let v(t)=v(t)'+1/2, w(t)=w(t)'-b+1/2, A=A'-b+1/2, The FHN stochastic resonance model is simplified to the following form:
式中：——阈值电压；B——信号幅值到阈值电压的距离；In the formula: ——threshold voltage; B——distance from signal amplitude to threshold voltage;
令AT-B＝0，则只需要设置和调整模型参数ε和需要识别的最大峰值阶次N；Let AT-B=0, then Only need to set and adjust the model parameters ε and the maximum peak order N to be identified;
将预处理后带有噪声干扰的SSVEP信号送入到FHN随机共振模型进行FHN随机共振处理，再通过快速傅里叶变换计算噪声增强的SSVEP的频谱图以识别目标频率；Send the preprocessed SSVEP signal with noise interference to the FHN stochastic resonance model for FHN stochastic resonance processing, and then calculate the spectrogram of the noise-enhanced SSVEP by fast Fourier transform to identify the target frequency;
4)峰值频率识别：从步骤3)中获得的输出信号的频谱图中，分别提取第N阶主峰对应的特征频率；4) peak frequency identification: from the spectrogram of the output signal obtained in step 3), respectively extract the characteristic frequency corresponding to the Nth order main peak;
5)频率匹配检测：将识别频率与所有刺激频率进行匹配，如果匹配成功，则目标频率被有效识别；如果匹配失败，则有必要检测当前识别的阶次是否大于设定的最大阶次；如果终止条件满足，则检测结束，表明目标频率标识失败；否则，计算返回到步骤4)。5) Frequency matching detection: match the recognition frequency with all stimulation frequencies, if the matching is successful, the target frequency is effectively recognized; if the matching fails, it is necessary to detect whether the current recognized order is greater than the set maximum order; if If the termination condition is satisfied, the detection ends, indicating that the target frequency identification fails; otherwise, the calculation returns to step 4).
下面再结合实施例对本发明进行说明。The present invention will be described below in conjunction with the embodiments.
采用本发明方法对15名正常被试者(10男，5女，均为20-26岁)进行实验。将光闪烁作为视觉刺激范式，视觉刺激范式包含40个目标，刺激频率分别为8-16Hz，间隔0.2Hz；在Dell-S2409W电脑的24英寸LCD显示器上以75Hz的刷新率同时渲染40个视觉闪烁(长宽为3cm*3cm)；两次刺激之间的水平间隔和垂直间隔分别为2cm和3cm；使用者头部距离计算机屏幕150cm。分别输出常规CCA方法、本发明FHN随机共振方法下SSVEP信号特征频谱图，并计算识别正确率，得到的提取效果分别如图3、图4所示(图3中8.4Hz、8.6Hz、12.4Hz、14.6Hz、14.8Hz及15Hz表示特征频率识别错误、图3中9.4Hz、11.2Hz、12.8Hz及15.4Hz表示特征频率周围存在大量干扰峰、图4中8.4Hz、8.6Hz、12.4Hz、14.8Hz及15Hz表示特征频率的提取结果得到纠正、图4中9.4Hz、11.2Hz、12.8Hz及15.4Hz表示干扰峰得到较好的抑制)。15名被试者的特征频率识别正确率如图5所示，与CCA方法相比，本发明FHN随机共振对EEG信号特征频率的识别效果得到大幅提升。对于目标频率周围存在较大干扰峰而信号,如9.4Hz、11.2Hz、12.8Hz及15.4Hz所对应的频谱图，经过FHN共振处理后，干扰频率得到了更为明显的抑制，目标频率在频谱图中的主导地位进一步凸显；尤其是对于CCA无法识别的信号,如8.4Hz、8.6Hz、12.4Hz、14.8Hz及15Hz所对应的频谱图，FHN随机共振则可以有效的识别出相应的目标频率。同时，利用CCA识别SSVEP特征频率的平均处理时间为2.79s，而利用本发明FHN随机共振识别SSVEP特征频率的平均处理时间为1.24s，大大压缩了提取方法的处理速度。Adopt the method of the present invention to carry out experiment to 15 normal subjects (10 males, 5 females, all 20-26 years old). Using light flicker as a visual stimulus paradigm, the visual stimulus paradigm contains 40 targets, the stimulation frequency is 8-16Hz, and the interval is 0.2Hz; 40 visual flickers are simultaneously rendered on a 24-inch LCD monitor of a Dell-S2409W computer with a refresh rate of 75Hz (Length and width are 3cm*3cm); the horizontal and vertical intervals between the two stimuli are 2cm and 3cm respectively; the distance between the user's head and the computer screen is 150cm. Output conventional CCA method, SSVEP signal feature spectrum diagram under the FHN stochastic resonance method of the present invention respectively, and calculate recognition correct rate, the extraction effect that obtains is shown in Fig. 3, Fig. 4 respectively (8.4Hz, 8.6Hz, 12.4Hz in Fig. 3 , 14.6Hz, 14.8Hz, and 15Hz indicate that the characteristic frequency is incorrectly identified. In Figure 3, 9.4Hz, 11.2Hz, 12.8Hz, and 15.4Hz indicate that there are a large number of interference peaks around the characteristic frequency. In Figure 4, 8.4Hz, 8.6Hz, 12.4Hz, 14.8 Hz and 15Hz indicate that the extraction results of the characteristic frequencies have been corrected, and 9.4Hz, 11.2Hz, 12.8Hz and 15.4Hz in Figure 4 indicate that the interference peaks have been better suppressed). The correct rate of eigenfrequency recognition of 15 subjects is shown in Fig. 5. Compared with the CCA method, the recognition effect of the FHN stochastic resonance of the present invention on the eigenfrequency of EEG signals is greatly improved. For signals with large interference peaks around the target frequency, such as 9.4Hz, 11.2Hz, 12.8Hz and 15.4Hz corresponding to the spectrum diagram, after FHN resonance processing, the interference frequency has been more obviously suppressed, and the target frequency is in the spectrum The dominant position in the figure is further highlighted; especially for signals that cannot be identified by CCA, such as the frequency spectrum corresponding to 8.4Hz, 8.6Hz, 12.4Hz, 14.8Hz and 15Hz, FHN stochastic resonance can effectively identify the corresponding target frequency . At the same time, the average processing time for identifying the SSVEP characteristic frequency by using CCA is 2.79s, while the average processing time for identifying the SSVEP characteristic frequency by using the FHN stochastic resonance of the present invention is 1.24s, which greatly reduces the processing speed of the extraction method.
本发明能够从视觉中枢神经出发，应用脑机接口技术，在SSVEP多尺度噪声抑制及特征频率提取方法实现了更高的识别精度和更快的处理速度，有效的增加了基于SSVEP的BCI系统的信息传输率，为SSVEP的特征频率提取提供了有效的手段。The present invention can start from the visual central nervous system, apply the brain-computer interface technology, realize higher recognition accuracy and faster processing speed in the SSVEP multi-scale noise suppression and feature frequency extraction method, and effectively increase the BCI system based on SSVEP The information transmission rate provides an effective means for the feature frequency extraction of SSVEP.
Claims (5)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110419436.4A CN113180706B (en) | 2021-04-19 | 2021-04-19 | FHN stochastic resonance-based SSVEP characteristic frequency extraction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110419436.4A CN113180706B (en) | 2021-04-19 | 2021-04-19 | FHN stochastic resonance-based SSVEP characteristic frequency extraction method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113180706A CN113180706A (en) | 2021-07-30 |
| CN113180706B true CN113180706B (en) | 2023-08-15 |
Family
ID=76978027
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110419436.4A Active CN113180706B (en) | 2021-04-19 | 2021-04-19 | FHN stochastic resonance-based SSVEP characteristic frequency extraction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113180706B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113673440B (en) * | 2021-08-23 | 2024-04-05 | 西安交通大学 | FHN-CCA fusion-based SSVEP multi-scale noise transfer and characteristic frequency detection method |
| CN115040143B (en) * | 2022-06-13 | 2024-07-26 | 西安交通大学 | FHN-PSD or FHN-STFTCSP motor imagery time-frequency energy enhancement and characteristic response detection method |
| CN115192044B (en) * | 2022-07-28 | 2024-08-20 | 西安交通大学 | Single-target SSVEP recognition system and method based on data continuation |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102499675A (en) * | 2011-10-27 | 2012-06-20 | 杭州电子科技大学 | Feedback system random resonance intensification method of electro-corticogram signal |
| CN103970273A (en) * | 2014-05-09 | 2014-08-06 | 西安交通大学 | Steady motion visual evoked potential brain computer interface method based on stochastic resonance enhancement |
| CN104951082A (en) * | 2015-07-09 | 2015-09-30 | 浙江大学 | Brain-computer interface method for intensifying EEG (electroencephalogram) signals through stochastic resonance |
| US10849526B1 (en) * | 2016-10-13 | 2020-12-01 | University Of South Florida | System and method for bio-inspired filter banks for a brain-computer interface |
-
2021
- 2021-04-19 CN CN202110419436.4A patent/CN113180706B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102499675A (en) * | 2011-10-27 | 2012-06-20 | 杭州电子科技大学 | Feedback system random resonance intensification method of electro-corticogram signal |
| CN103970273A (en) * | 2014-05-09 | 2014-08-06 | 西安交通大学 | Steady motion visual evoked potential brain computer interface method based on stochastic resonance enhancement |
| CN104951082A (en) * | 2015-07-09 | 2015-09-30 | 浙江大学 | Brain-computer interface method for intensifying EEG (electroencephalogram) signals through stochastic resonance |
| US10849526B1 (en) * | 2016-10-13 | 2020-12-01 | University Of South Florida | System and method for bio-inspired filter banks for a brain-computer interface |
Non-Patent Citations (1)
| Title |
|---|
| 用于稳态视觉诱发电位特征频率提取的同步压缩短时傅里叶变换方法;张黎明;张小栋;陆竹风;李睿;;西安交通大学学报(第02期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113180706A (en) | 2021-07-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113180706B (en) | FHN stochastic resonance-based SSVEP characteristic frequency extraction method | |
| CN104965584B (en) | Mixing brain-machine interface method based on SSVEP and OSP | |
| CN105496363B (en) | The method classified based on detection sleep cerebral electricity signal to sleep stage | |
| JP7372358B2 (en) | Association between intentional and unintentional changes in brain state and brain signals | |
| CN108888280B (en) | Evaluation method of students' attention in class based on EEG signal analysis | |
| CN101339455B (en) | Brain-computer interface system based on the N170 component of the specific wave for face recognition | |
| CN102063180B (en) | HHT-based high-frequency combined coding steady state visual evoked potential brain-computer interface method | |
| CN102436302B (en) | Method for brain-computer interface based on amplitude modulated visual evoked potential | |
| CN113288181A (en) | Individual template reconstruction method based on steady-state visual evoked potential electroencephalogram signal identification | |
| CN109770900B (en) | Brain-computer interface instruction delivery method, system and device based on convolutional neural network | |
| CN106502410A (en) | Improve the transcranial electrical stimulation device of Mental imagery ability and method in brain-computer interface | |
| CN114081506B (en) | Steady-state visual evoked potential classification method based on long-term and short-term memory network | |
| CN117064409B (en) | Method, device and terminal for evaluating transcranial direct current intervention stimulation effect in real time | |
| CN113673440B (en) | FHN-CCA fusion-based SSVEP multi-scale noise transfer and characteristic frequency detection method | |
| CN115192907A (en) | Real-time biofeedback percutaneous vagus nerve electronic acupuncture device | |
| CN113378737A (en) | Implanted brain-computer interface neuron spike potential classification method | |
| Wu et al. | Fast, accurate localization of epileptic seizure onset zones based on detection of high-frequency oscillations using improved wavelet transform and matching pursuit methods | |
| CN113017653A (en) | Steady-state visual evoked potential identification system and method based on chaos detection principle | |
| CN113662558B (en) | Intelligent classification method for distinguishing brain electrical blink artifacts and forehead electrode epilepsy-like discharge | |
| CN113180707B (en) | Visual electrophysiological rapid detector for graphic evoked potential | |
| Xiao et al. | A data expansion technique based on training and testing sample to boost the detection of SSVEPs for brain-computer interfaces | |
| CN114052752B (en) | Method for filtering interference of multi-channel surface electromyographic signal center electric QRS complex | |
| CN116870360A (en) | Evaluation method of transcranial electrical stimulation effect | |
| CN107411737A (en) | A kind of across the time recognition methods of mood based on resting electroencephalogramidentification similitude | |
| CN115040143B (en) | FHN-PSD or FHN-STFTCSP motor imagery time-frequency energy enhancement and characteristic response detection method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |