阅读说明：本技术 一种基于脑电波信号特征的眼部噪声去除方法 (Eye noise removing method based on brain wave signal characteristics ) 是由 王晓岸 马鹏程 张乾坤 于 2020-06-23 设计创作，主要内容包括：本发明公开了一种基于脑电波信号特征的眼部噪声去除方法。本方法包括了特制的脑电数据处理模块、特制的眼部噪声分析模块和特制的噪声去除模块三部分。本方法的工作原理为：(1)特制的脑电数据处理模块对输入的原始脑电数据进行预处理,得到标准化的信号数据；(2)特制的眼部噪声分析模块对输入的信号数据进行眼部噪声识别和噪声波形特征计算分析；(3)特制的噪声去除模块依据计算出的眼部噪声基础波形特征值进行信号噪声去除。本发明实现了脑电信号眼部噪声的自动化去除,降低了眼部噪声去除的成本,提高了信号处理效率。(The invention discloses an eye noise removing method based on brain wave signal characteristics. The method comprises a specially-made electroencephalogram data processing module, a specially-made eye noise analysis module and a specially-made noise removal module. The working principle of the method is as follows: (1) the method comprises the steps that a specially-made electroencephalogram data processing module preprocesses input original electroencephalogram data to obtain standardized signal data; (2) a specially-made eye noise analysis module carries out eye noise identification and noise waveform feature calculation analysis on input signal data; (3) and the special noise removing module removes signal noise according to the calculated eye noise basic waveform characteristic value. The invention realizes the automatic removal of the eye noise of the electroencephalogram signal, reduces the cost of removing the eye noise and improves the signal processing efficiency.)

1. The eye noise removing method based on the brain wave signal features is characterized by comprising the following steps: the method comprises the steps of normalization processing of brain wave signal data, eye noise identification analysis and noise removal, wherein the steps of normalization processing of brain wave signal data, eye noise identification analysis and noise removal are as follows:

the first step is as follows: carrying out standardization processing on original electroencephalogram signal data through a special electroencephalogram data processing module, and transmitting the processed data to a special eye noise analysis module;

the second step is that: a specially-made eye noise analysis module carries out eye noise identification on the standardized signal data, determines the data position with noise, carries out noise basic waveform data calculation and determines the noise data characteristic and numerical value;

the third step: in a special noise removal module, data containing noise is removed according to the characteristics and the numerical values of a noise basic waveform, and brain wave signal data with the noise removed is generated.

2. The method for removing eye noise based on brain wave signal features of claim 1, wherein the pre-processing of the original brain wave signal data by the brain wave data processing module specially made in the first step comprises:

the first step is as follows: carrying out Z-score unified standardization on the original electroencephalogram signal data, and improving the comparability of the signal data;

the second step is that: and outputting the normalized data to a specially-made eye noise analysis module.

3. The method for removing eye noise based on brain wave signal features of claim 1, wherein the flow of the eye noise recognition analysis of the brain wave signal data by the eye noise analysis module of the first and second steps is as follows:

the first step is as follows: calculating a linear correlation coefficient and a peak value of the signal data to identify blink noise and eye movement noise, and confirming a data position containing the noise;

the second step is that: respectively carrying out zero phase shift filtering on signal data containing the blink noise and the eye movement noise to obtain basic waveforms and data characteristic values of the blink noise and the eye movement noise;

the third step: and outputting the position information containing the eye noise data and the data characteristic values of different noises to a specially-made noise removing module.

4. The method for removing eye noise based on brain wave signal features according to claim 1, wherein the procedure of removing eye noise from the brain wave signal data by the noise removing module specially prepared in the third step is as follows:

the first step is as follows: using signal data which are identified by a special eye noise analysis module and contain eye noise as a subtree, using a calculated eye noise characteristic value as a subtree, and calculating a difference;

the second step is that: and replacing the corresponding signal data value containing the eye noise by the obtained difference value serving as data after noise removal to generate brain wave data after the eye noise removal.

5. The method for removing eye noise based on brain wave signal features of claim 3, wherein the first and second steps of calculating the linear correlation coefficient and peak value of the signal data for blink noise identification to obtain the basic waveform and the data feature value comprise the steps of:

the first step is as follows: calculating correlation and peak value by taking 500ms as time window length and 100ms as window step length;

the second step is that: the calculated linear correlation coefficient exceeds 0.85, and the peak-to-peak value exceeds 300 mu v, and the signal data contains the blink noise;

the third step: carrying out a zero phase shift filter of 1-15Hz on data containing the blink noise, wherein the data range is that the left and right of the peak value are respectively extended by 150ms, and the basic waveform of the blink noise can be obtained;

the fourth step: the data of the basic waveform of the blink noise is taken as the characteristic value of the noise data in the data range.

6. The method for removing eye noise based on brain wave signal features of claim 3, wherein the steps of calculating the linear correlation coefficient and peak value of the signal data for eye movement noise identification, and obtaining the basic waveform and data feature value in the first and second steps are:

the first step is as follows: calculating correlation and peak value by taking 1000ms as the length of a time window and 200ms as the window step length;

the second step is that: if the calculated linear correlation coefficient is lower than-0.5, the signal data containing the eye movement noise is obtained;

the third step: carrying out a 1-5Hz zero phase shift filter on data containing eye movement noise, wherein the range is that the peak value is delayed for 100ms and is delayed for 800ms, and the basic waveform of the eye movement noise can be obtained;

the fourth step: the data of the basic waveform of the eye movement noise is taken as the characteristic value of the noise data in the data range.

Technical Field

The invention belongs to the field of brain wave signal processing, and particularly relates to an eye noise removing method based on brain wave signal characteristics.

Background

Brain wave equipment is applied to daily life of people more and more, the application and processing quantity of brain wave data is larger and more, and interference noise mainly comprises eye electric waves, facial muscle electric waves, heart electric waves and the like in the application of the brain wave equipment. The amplitude energy of these noises is much higher than the brain waves themselves, which are easily completely submerged by these disturbances.

The existing mature method for preprocessing brain waves basically aims at multi-channel design analysis, needs a large amount of manual operation and cannot realize automatic noise removal.

Therefore, the eye noise removing method based on the brain wave signal features is designed, is used for brain wave equipment with few channels, and can achieve full-automatic eye noise removal of the brain wave signals.

Disclosure of Invention

The invention aims to provide an eye noise removing method based on brain wave signal characteristics, and aims to solve the problems that the existing brain wave preprocessing method on the market cannot realize full-automatic eye noise removal, and a large amount of manpower is consumed for eye noise analysis and removal. The linear correlation coefficient and the peak value are calculated through a special eye noise analysis module, so that the position of eye noise in the electroencephalogram signal can be effectively confirmed, and the eye noise signal can be identified; the zero phase shift filtering process can calculate the basic waveform and characteristic value of the eye noise, and automatic eye noise analysis and removal are realized.

In order to achieve the above purpose, the invention provides the following technical scheme: an eye noise removing method based on brain wave signal characteristics. The method comprises a set of brain wave signal data standardization processing, eye noise identification analysis and noise removal steps, wherein a specially-made brain wave data processing module, a specially-made eye noise analysis module and a specially-made noise removal module are applied in the steps. The method comprises the following steps of brain wave signal data standardization processing, eye noise identification and analysis and noise removal:

the first step is as follows: carrying out standardization processing on original electroencephalogram signal data through a special electroencephalogram data processing module, and transmitting the processed data to a special eye noise analysis module;

the second step is that: a specially-made eye noise analysis module carries out eye noise identification on the standardized signal data, determines the data position with noise, carries out noise basic waveform data calculation and determines the noise data characteristic and numerical value;

the third step: in a special noise removal module, data containing noise is removed according to the characteristics and the numerical values of a noise basic waveform, and brain wave signal data with the noise removed is generated.

Preferably, the first step of preprocessing the raw brain wave signal data by the specially-made brain wave data processing module comprises the following steps:

the first step is as follows: carrying out Z-score unified standardization on the original electroencephalogram signal data, and improving the comparability of the signal data;

the second step is that: and outputting the normalized data to a specially-made eye noise analysis module.

Preferably, the eye noise identification and analysis of the brain wave signal data by the specially-made eye noise analysis module in the first step and the specially-made eye noise analysis module in the second step comprises the following procedures:

the first step is as follows: calculating a linear correlation coefficient and a peak value of the signal data to identify blink noise and eye movement noise, and confirming a data position containing the noise;

the second step is that: respectively carrying out zero phase shift filtering on signal data containing the blink noise and the eye movement noise to obtain basic waveforms and data characteristic values of the blink noise and the eye movement noise;

the third step: and outputting the position information containing the eye noise data and the data characteristic values of different noises to a specially-made noise removing module.

Preferably, the procedure of removing the eye noise of the brain wave signal data by the noise removing module specially manufactured in the third step is as follows:

the first step is as follows: using signal data which are identified by a special eye noise analysis module and contain eye noise as a subtree, using a calculated eye noise characteristic value as a subtree, and calculating a difference;

the second step is that: and replacing the corresponding signal data value containing the eye noise by the obtained difference value serving as data after noise removal to generate brain wave data after the eye noise removal.

Preferably, the step of calculating the linear correlation coefficient and the peak value of the signal data to perform blink noise identification to obtain the basic waveform and the data characteristic value comprises:

the first step is as follows: calculating correlation and peak value by taking 500ms as the length of a time window and 100ms as the window step length;

the second step is that: the calculated linear correlation coefficient exceeds 0.85, and the peak-to-peak value exceeds 300 mu v, and the signal data contains the blink noise;

the third step: carrying out a zero phase shift filter of 1-15Hz on data containing the blink noise, wherein the data range is that the left and right of the peak value are respectively extended by 150ms, and the basic waveform of the blink noise can be obtained;

the fourth step: the data of the basic waveform of the blink noise is taken as the characteristic value of the noise data in the data range.

Preferably, the step of calculating the linear correlation coefficient and the peak value of the signal data to perform eye movement noise identification to obtain the basic waveform and the data characteristic value comprises:

the first step is as follows: calculating correlation and peak value by taking 1000ms as the length of a time window and 200ms as the window step length;

the second step is that: if the calculated linear correlation coefficient is lower than-0.5, the signal data containing the eye movement noise is obtained;

the third step: carrying out a 1-5Hz zero phase shift filter on data containing eye movement noise, wherein the range is that the peak value is delayed for 100ms and is delayed for 800ms, and the basic waveform of the eye movement noise can be obtained;

the fourth step: the data of the basic waveform of the eye movement noise is taken as the characteristic value of the noise data in the data range.

Compared with the prior art, the invention has the beneficial effects that:

1. the eye noise removing method based on the brain wave signal features is constructed, so that the eye noise of the brain wave signals can be completely and automatically removed without manual operation;

2. eye noise identification is carried out on the brain wave signals by applying a linear correlation and peak value calculation method, so that blink noise and eye movement noise can be accurately identified;

3. the zero phase shift filtering is adopted to calculate the eye noise basic wave row and the characteristic value, so that the analysis efficiency of the eye noise is improved.

Drawings

FIG. 1 is a flow chart of the present invention;

FIG. 2 is a schematic diagram of the principle steps of the present invention;

FIG. 3 is a comparison graph of blink noise removal data for brain wave signals in accordance with the present invention;

fig. 4 is a comparison graph of eye movement noise removal data of brain wave signals according to the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The invention provides a technical scheme that: an eye noise removing method based on brain wave signal characteristics. The method comprises a set of brain wave signal data standardization processing, eye noise identification and analysis and noise removal steps, wherein a specially-made supervisor mechanism module, a specially-made deep learning algorithm model module and a filter module are applied in the steps as shown in figures 1-2. The embodiment specifically comprises the following steps:

the first step is as follows: carrying out standardization processing on original electroencephalogram signal data through a special electroencephalogram data processing module, and transmitting the processed data to a special eye noise analysis module;

the second step is that: a specially-made eye noise analysis module carries out eye noise identification on the standardized signal data, determines the data position with noise, carries out noise basic waveform data calculation and determines the noise data characteristic and numerical value;

the third step: in a special noise removal module, data containing noise is removed according to the characteristics and the numerical values of a noise basic waveform, and brain wave signal data with the noise removed is generated.

In this embodiment, preferably, the first specially-made electroencephalogram data processing module performs preprocessing on the original electroencephalogram signal data by the following procedure:

the first step is as follows: carrying out Z-score unified standardization on the original electroencephalogram signal data, and improving the comparability of the signal data;

the second step is that: and outputting the normalized data to a specially-made eye noise analysis module.

In this embodiment, preferably, the flow of the eye noise identification analysis of the brain wave signal data by the eye noise analysis module specially manufactured in the first step and the second step is as follows:

the first step is as follows: calculating a linear correlation coefficient and a peak value of the signal data to identify blink noise and eye movement noise, and confirming a data position containing the noise;

the second step is that: respectively carrying out zero phase shift filtering on signal data containing the blink noise and the eye movement noise to obtain basic waveforms and data characteristic values of the blink noise and the eye movement noise;

the third step: and outputting the position information containing the eye noise data and the data characteristic values of different noises to a specially-made noise removing module.

In this embodiment, preferably, the procedure of removing the eye noise of the brain wave signal data by the noise removing module specially manufactured in the third step is as follows:

the first step is as follows: using signal data which are identified by a special eye noise analysis module and contain eye noise as a subtree, using a calculated eye noise characteristic value as a subtree, and calculating a difference;

the second step is that: and replacing the corresponding signal data value containing the eye noise by the obtained difference value serving as data after noise removal to generate brain wave data after the eye noise removal.

In this embodiment, as shown in fig. 3, preferably, the step of calculating a linear correlation coefficient and a peak value of the signal data to perform blink noise identification to obtain a basic waveform and a data characteristic value includes:

the first step is as follows: calculating correlation and peak value by taking 500ms as the length of a time window and 100ms as the window step length;

the second step is that: the calculated linear correlation coefficient exceeds 0.85, and the peak-to-peak value exceeds 300 mu v, and the signal data contains the blink noise;

the third step: carrying out a zero phase shift filter of 1-15Hz on data containing the blink noise, wherein the data range is that the left and right of the peak value are respectively extended by 150ms, and the basic waveform of the blink noise can be obtained;

the fourth step: the data of the basic waveform of the blink noise is taken as the characteristic value of the noise data in the data range.

In this embodiment, preferably, as shown in fig. 4, the step of calculating the linear correlation coefficient and the peak value of the signal data to perform eye movement noise identification, and obtaining the basic waveform and the data characteristic value includes:

the first step is as follows: calculating correlation and peak value by taking 1000ms as the length of a time window and 200ms as the window step length;

the second step is that: if the calculated linear correlation coefficient is lower than-0.5, the signal data containing the eye movement noise is obtained;

the third step: carrying out a 1-5Hz zero phase shift filter on data containing eye movement noise, wherein the range is that the peak value is delayed for 100ms and is delayed for 800ms, and the basic waveform of the eye movement noise can be obtained;

the fourth step: the data of the basic waveform of the eye movement noise is taken as the characteristic value of the noise data in the data range.