CN103584856A - Algorithm for identifying blinking force by processing brain waves - Google Patents

Abstract

In the present invention, an algorithm for recognizing the intensity of blinking by processing brain waves, first measures the amplitude x of brain waves in each working cycle, and uses the iterative formula Y _n = (1-k) Y _n-1 + kx _n and Z _n = (1-k) Z _n-1 +kx _n ² Calculate the approximate mean Y and mean square Z, and calculate the variance Var according to the formula Var(x)=E(x ² )-[E(x)] ² (x), and then according to the formula

or the formula

Computes the normalized magnitude S of x. And compare the amplitude S with the set positive threshold value and negative threshold value to judge event A or B. If the A event occurs, it is considered that a blink action has occurred, and the blinking force at this time is the amplitude when the A event occurs. The algorithm for identifying blink force by processing brain waves of the present invention has the advantages of being able to identify rapid and continuous blinks and blink force, relatively small amount of calculation, good algorithm adaptability, and the like.

Description

An Algorithm to Identify Blink Strength by Processing Brainwaves

technical field

The invention relates to an algorithm for identifying blink strength by processing brain waves.

Background technique

By observing the original amplitude images of brain waves, we found that for any person, when his body is still, blinking will trigger fluctuations in frontal brain waves, and the magnitude of the fluctuations is proportional to the intensity of the blink. However, because each person's brain wave image has subtle differences, if the brain wave is processed by setting a threshold, this will lead to a lack of universality in the designed product. However, if the mean and variance are directly calculated, it will lead to a huge amount of calculation and calculation space.

In the prior art, the TGAM module of NeuroSky Company can measure the original brain wave data and blink strength, but the TGAM module has a very slow response speed in recognizing the blink strength, and cannot recognize continuous rapid blinks.

Contents of the invention

The present invention aims to avoid the shortcomings in the above-mentioned prior art, and provides an algorithm that has a small amount of calculation and can identify rapid and continuous blinking by processing brain waves to identify blinking force, so as to be able to identify rapid and continuous blinking and blinking force.

The present invention adopts the following technical solutions to solve the technical problems.

An algorithm for identifying blink strength by processing brain waves, which includes the following steps:

Step 1: Use the earlobe as a reference ground, measure the voltage of the forehead, and observe the waveform of the brain wave;

Step 2: Set a positive threshold value and a negative threshold value; the brain wave amplitude higher than the positive threshold is called event A, and the brain wave amplitude lower than the negative threshold is called event B;

Step 3: In each working cycle of the waveform diagram of the brain wave, measure the brain wave amplitude x;

Step 4: Iterate through the formula (1) to find the approximate value Yn of the mean value of the brain wave amplitude x

The iteration formula (1) is: Y _n = (1-k) Y _n-1 + kx _n (1)

In the formula (1), x _n is the amplitude of the brain wave measured at the nth moment; Y _n is the approximate value of the mean value of the amplitude x at the nth iteration; k is a constant, and n is a natural number; (if k is smaller , the closer Y _n is to the approximate value of the real x mean, but the slower the adaptive speed. Usually k is 0.005.)

Step 5: Iterate through the formula (2) to find the approximate value of the mean square value Z _n of the brain wave amplitude x;

The iteration formula (2) is: Z _n = (1-k) Z _n-1 + kx _n ² (2)

In the formula (2), x _n is the amplitude of the brain wave measured at the nth moment; Z _n is the approximate value of the mean square value of the amplitude x at the nth iteration; k is a constant, and n is a natural number; (if k The smaller the value, the closer Z _n is to the approximate value of the real x-mean-square value, but the slower the adaptive speed. Usually k is 0.005.)

Step 6: Calculate the variance Var(x) of the approximate value of x by formula (3);

Var(x)=E(x ² )-[E(x)] ² (3);

Formula (3), x is the amplitude of the brain wave, E(x) is the mean value of the brain wave amplitude, E(x ² ) is the mean square value of the brain wave amplitude, and Var(x) is the variance of the brain wave amplitude ;

Step 7: Standardize x by formula (4);

$\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}{\sqrt{\mathrm{<span\; class="notranslate">\; Var</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span>$

Formula (4), x is the amplitude of the brain wave, E(x) is the mean value of the brain wave amplitude, Var(x) is the variance of the brain wave amplitude; S is the standardized value of x.

A kind of algorithm of the present invention recognizes the blink strength by processing brain wave is also characterized in that:

In the step 7, the formula (5) is used to standardize x;

${\mathrm{<span\; class="notranslate">\; the\; <figure-callout\; id="2"\; label="s"\; filenames="HDA0000426355510000021.png"\; state="\{\{state\}\}">s</figure-callout></span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; sgn</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>\frac{{<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; Var</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>$

Among them, x is the amplitude of the brain wave, E(x) is the mean value of the brain wave amplitude, Var(x) is the variance of the brain wave amplitude; S is the standardized value of x; the function sgn(x) expresses: when When x is greater than or equal to 0, its value is 1; when x is less than or equal to 0, its value is -1.

Compared with the prior art, the beneficial effects of the present invention are reflected in:

In the present invention, an algorithm for recognizing the intensity of blinking by processing brain waves, first measures the brain wave amplitude x in each working cycle, and calculates the approximate mean value Y and mean square value through iterative formulas (1) and (2) Z, calculate the variance according to formula (3), and then calculate the standardized amplitude S of x according to formula (4) or formula (5). And compare the amplitude S with the set positive threshold value and negative threshold value to judge event A or B. If the A event occurs, and the time difference between the occurrence moments of the two events does not exceed the preset value, it is considered that a blink has occurred, and the maximum value of S in the entire blink process is taken as the blink force.

The algorithm for identifying blink force by processing brain waves of the present invention has the advantages of being able to identify rapid and continuous blinks and blink force, relatively small amount of calculation, good algorithm adaptability, and the like.

Description of drawings

FIG. 1 is a flow chart of an algorithm for identifying blink strength by processing brain waves according to the present invention.

FIG. 2 is an identification diagram of the value range of the parameter i in FIG. 1 .

Fig. 3 is a Matlab simulation result diagram of an algorithm for recognizing blink strength by processing brain waves according to the present invention.

The present invention will be further described below through specific embodiments and in conjunction with the accompanying drawings.

Detailed ways

An algorithm for identifying blinking strength by processing brain waves is characterized in that it includes the following steps:

Step 1: Use the earlobe as a reference ground, measure the voltage of the forehead, and observe the waveform of the brain wave;

Step 2: Set a positive threshold value and a negative threshold value; the brain wave amplitude higher than the positive threshold is called event A, and the brain wave amplitude lower than the negative threshold is called event B;

Step 3: In each working cycle of the waveform diagram of the brain wave, measure the brain wave amplitude x;

Step 4: Iterate through the formula (1) to find the approximate value Yn of the mean value of the brain wave amplitude x

The iteration formula (1) is: Y _n = (1-k) Y _n-1 + kx _n (1)

In the formula (1), x _n is the amplitude of the brain wave measured at the nth moment; Y _n is the approximate value of the mean value of the amplitude x at the nth iteration; k is a constant, and n is a natural number; (if k is smaller , the closer Y _n is to the real x-mean approximation, but the slower the adaptive speed. Usually k is 0.005.)

Step 5: Iterate through the formula (2) to find the approximate value of the mean square value Z _n of the brain wave amplitude x;

The iteration formula (2) is: Z _n = (1-k) Z _n-1 + kx _n ² (2)

In the formula (2), x _n is the amplitude of the brain wave measured at the nth moment; Z _n is the approximate value of the mean square value of the amplitude x at the nth iteration; k is a constant, and n is a natural number; (if k The smaller the value, the closer Z _n is to the approximate value of the real x-mean-square value, but the slower the adaptive speed. Usually k is 0.005.)

Step 6: Calculate the variance Var(x) of the approximate value of x by formula (3);

Var(x)=E(x ² )-[E(x)] ² (3);

Formula (3), x is the amplitude of the brain wave, E(x) is the mean value of the brain wave amplitude, E(x ² ) is the mean square value of the brain wave amplitude, and Var(x) is the variance of the brain wave amplitude ;

Step 7: Standardize x by formula (4);

$\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}{\sqrt{\mathrm{<span\; class="notranslate">\; Var</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span>$

Formula (4), x is the amplitude of the brain wave, E(x) is the mean value of the brain wave amplitude, Var(x) is the variance of the brain wave amplitude; S is the standardized value of x.

Figure 1 is a flowchart of a specific embodiment of the algorithm of the present invention. During initialization, the value of each parameter is: Y=0; Z=10000; k=0.005; maxI=100; i=maxI; maxS=0; where: i is the number of periods between events A and B; as mentioned above , the maximum waiting period constant maxI is 100. Refer to Figure 1 for the specific algorithm calculation process.

As shown in Figure 2, it is a value range identification diagram of the parameter i in Figure 1. Taking the brain wave waveform of a blink in Figure 2 as an example, the time i cannot be too long, because if i is too long, a continuous blinking process cannot be formed, so an upper limit maxI needs to be set. This method has been tested and can identify eye blinks and their strength well when the body is still. However, in the process of human movement, muscle electricity will interfere with brain wave signals, resulting in calculation distortion, so filtering algorithms should also be used to recognize blinking and its strength during movement.

After obtaining the variance Var(x) of the approximate value of x, x is standardized by the variance Var(x) of the approximate value of x. In the prior art, the mean and variance of n times of EEG data are directly calculated, and the calculation of the mean at n times requires a total of n+1 operations (addition and multiplication), and the variance requires 2n+3 operations. After adopting the algorithm of the present invention, it only needs 3 calculations to calculate the mean value and variance, which greatly reduces the calculation amount of the whole calculation process, and facilitates the recognition of blink strength through brain waves.

Experiments have found that the amplitude of brain waves and the intensity of blinking are approximately proportional, so the amplitude of brain waves can be used to represent the intensity of blinking. However, since the wave amplitude and range of each person's brain waves are different, the algorithm needs to be adaptive. Moreover, the calculation amount of this algorithm must be small in order to realize real-time processing. The algorithm of the invention has a small amount of computation and can identify quick and continuous blinks.

Through experiments, it was found that the blinking action can be divided into two stages: closing eyes and opening eyes. Use the earlobe as a reference ground, measure the voltage of the forehead, and observe the waveform of the brain wave. When it is regarded as an eye-closing action, a pulse in the same positive direction as the vertical axis of the waveform image will be generated, and the maximum amplitude of the pulse is approximately proportional to the eye-closing force. When it is regarded as an eye-opening action, a pulse in the opposite direction to the positive direction of the vertical axis of the waveform image will be generated. The maximum amplitude of the pulse is approximately proportional to the eye-opening force.

Set a positive threshold value and a negative threshold value, call the brain wave amplitude higher than the positive threshold value event A, and call the brain wave amplitude lower than the negative threshold value event B. Only when event B occurs after event A, and the difference between the occurrence times of the two events is less than a preset value, is it considered that an eye blink has occurred. If the A event occurs after the A event or the B event occurs directly, it is not considered that an eye blink has occurred.

Since the method of standardizing the amplitude of brain waves is adopted, both the positive and negative thresholds are constant. After experimental verification, the positive threshold is 3, and the negative threshold is -2.

It needs to be explained that when people blink on purpose, they almost always close their eyes with force, and do not use force when they open their eyes, so the force when closing eyes should be used as the force of blinking.

In the step 7, the formula (5) is used to standardize x;

${\mathrm{<span\; class="notranslate">\; the\; <figure-callout\; id="2"\; label="s"\; filenames="HDA0000426355510000021.png"\; state="\{\{state\}\}">s</figure-callout></span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; sgn</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>\frac{{<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; Var</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>$

Among them, x is the amplitude of the brain wave, E(x) is the mean value of the brain wave amplitude, Var(x) is the variance of the brain wave amplitude; S is the standardized value of x; the function sgn(x) expresses: when When x is greater than or equal to 0, its value is 1; when x is less than or equal to 0, its value is -1.

In formula (4), the square root sign is required. However, since the calculation of root sign is very heavy, if the algorithm is used on a device with poor computing power, it will easily affect the calculation speed. However, if formula (5) is used, there is no process of calculating the square root, and the calculation speed is relatively fast. This can greatly reduce the amount of computation. When comparing the magnitude, it is only necessary to compare S ² with the square of the magnitude x. Since the comparison is the square value of the amplitude, when comparing the values, it should also be noted that the amplitude needs to retain the original sign.

The algorithm is simulated by matlab. The simulation results are shown in Figure 3 below (the brain wave signals are all collected by the TGAM module), where the value of k is 0.00015. The black band in the figure represents the EEG, and the white dot in the middle of the EEG represents the recognition of eye blinking at this moment. The straight line corresponding to the white dot and the number next to the straight line represent the blink force measured at that moment.

The calculation process of the algorithm in each working cycle is as follows: in each working cycle, first measure the brain wave amplitude x, and calculate the approximate mean value Y and mean square value Z through iterative formulas (1) and (2), according to Formula (3) calculates the variance, and then calculates the standardized amplitude S of x according to formula (4) or formula (5). And compare S with the set positive threshold value and negative threshold value to judge event A or B.

Use S to judge whether event A or B has occurred. If B time has occurred, look at what event occurred when the event occurred last time. If it is A event that occurred, and the difference between the occurrence times of the two events is less than the preset value, it is considered to have occurred. Blinking action, take the maximum value of S in the whole blinking process as the blinking force.

The algorithm of the present invention is an algorithm for identifying blink strength by measuring forehead brain waves. The algorithm has been tested and is self-adaptive, and the algorithm is simple and easy to implement. The blink strength obtained through algorithmic processing has sufficient precision.

The algorithm for identifying the intensity of blinking by processing brainwaves of the present invention includes a method for standardizing digitized brainwave amplitude data and how to process the standardized data, and provides a real-time algorithm that can identify continuous blinking and eye blinking with a small amount of calculation and a fast calculation speed. The algorithm of blinking strength can distinguish between intentional blinking and unconscious blinking.

The algorithm of the invention has been tested and can well identify blinking and its strength when the body is still.

Claims (2)

1. An algorithm for identifying blink strength by processing brain waves, is characterized in that it comprises the steps: Step 1: Use the earlobe as a reference ground, measure the voltage of the forehead, and observe the waveform of the brain wave; Step 2: Set a positive threshold value and a negative threshold value; the brain wave amplitude higher than the positive threshold is called event A, and the brain wave amplitude lower than the negative threshold is called event B; Step 3: In each working cycle of the waveform diagram of the brain wave, measure the brain wave amplitude x; Step 4: Iterate through the formula (1) to find the approximate value Yn of the mean value of the brain wave amplitude x The iteration formula (1) is: Y _n = (1-k) Y _n-1 + kx _n (1) In the formula (1), x _n is the amplitude of the brain wave measured at the nth moment; Y _n is the approximate value of the mean value of the amplitude x at the nth iteration; k is a constant, and n is a natural number; (if k is smaller , the closer Yn is to the real x-mean approximation, but the slower the adaptive speed. Usually k is 0.005.) Step 5: Iterate through the formula (2) to find the approximate value of the mean square value Z _n of the brain wave amplitude x; The iteration formula (2) is: Z _n = (1-k) Z _n-1 + kx _n ² (2) In the formula (2), x _n is the amplitude of the brain wave measured at the nth moment; Z _n is the approximate value of the mean square value of the amplitude x at the nth iteration; k is a constant, and n is a natural number. Step 6: Calculate the variance Var(x) of the approximate value of x by formula (3); Var(x)=E(x ² )-[E(x)] ² (3); Formula (3), x is the amplitude of the brain wave, E(x) is the mean value of the brain wave amplitude, E(x ² ) is the mean square value of the brain wave amplitude, and Var(x) is the variance of the brain wave amplitude ; Step 7: Standardize x by formula (4); $\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}{\sqrt{\mathrm{<span\; class="notranslate">\; Var</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span>$ Formula (4), x is the amplitude of the brain wave, E(x) is the mean value of the brain wave amplitude, Var(x) is the variance of the brain wave amplitude; S is the standardized value of x. 2. An algorithm for identifying blink strength by processing brain waves according to claim 1, characterized in that, in step 7, formula (5) is used to standardize x; ${\mathrm{<span\; class="notranslate">\; the\; s</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; sgn</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>\frac{{<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; Var</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; .</span>$ Among them, x is the amplitude of the brain wave, E(x) is the mean value of the brain wave amplitude, Var(x) is the variance of the brain wave amplitude; S is the standardized value of x; the function sgn(x) expresses: when When x is greater than or equal to 0, its value is 1; when x is less than or equal to 0, its value is -1.

Images