CN111012345A

CN111012345A - Eye fatigue degree detection system and method

Info

Publication number: CN111012345A
Application number: CN201911405471.XA
Authority: CN
Inventors: 缪成同; 赵荣建; 刘标锋
Original assignee: Sinotech Shandong Intelligent Technology Co Ltd
Current assignee: Sinotech Shandong Intelligent Technology Co Ltd
Priority date: 2019-12-31
Filing date: 2019-12-31
Publication date: 2020-04-17

Abstract

The invention discloses a system and a method for detecting eye fatigue, which comprises a signal acquisition electrode, a signal analysis module, a communication module, a mobile device terminal and a data platform, wherein the signal acquisition electrode is a single-channel electrode and comprises three electrodes: the left electrode, the right electrode and the right leg driving electrode are respectively used for fixing the left position, the right position and the middle position of equipment, the analysis module comprises a primary signal amplification module, a secondary amplification module, a filter circuit module, a high-precision sampling module, a microprocessor, an external control module, a communication module and an eye fatigue detection method, and the eye fatigue detection method specifically comprises the following implementation steps: (1) acquiring an eye electrical signal; (2) preprocessing an original eye electric signal; (3) extracting a characteristic signal; (4) and scoring and classifying by using the characteristic value of the obtained eye movement signal. Seven characteristic values are extracted through the electro-oculogram signals, and the eye fatigue is scored and classified by using a decision tree and a support vector, so that the eye fatigue is concise and convenient to detect, targeted and suitable for large-scale application.

Description

Eye fatigue degree detection system and method

Technical Field

The invention relates to the technical field of fatigue detection, in particular to a system and a method for detecting eye fatigue.

Background

The eye fatigue is a common ophthalmic disease, and dry eyes, sour eyes, blurred vision and even visual deterioration caused by the eye fatigue directly influence the work and life of people. The eyestrain is mainly caused by that when people concentrate on watching a computer screen at ordinary times, the eye blinking frequency is reduced, the tear secretion is correspondingly reduced, and meanwhile, the blinking screen strongly stimulates eyes. It can cause pain in the neck, shoulders and other parts of human body, and also can cause and aggravate various eye diseases, for example, teenagers can cause myopia or deepen original myopia degree, and patients with eye diseases such as glaucoma can cause or aggravate original eye diseases due to eye over-fatigue. Therefore, in daily production and life, it is very important to clearly understand the degree of eyestrain of the user to implement measures for relieving the degree of eyestrain.

At present, the mainstream method for testing eye fatigue is mainly a detection method based on images and videos. The image and video detection mainly utilizes an infrared or color camera to collect images and video signals, adopts some computer vision technologies to position the positions of eyes, and extracts the characteristics of eyelid movement through the video signals so as to judge the fatigue state of people. The method needs image or video shooting equipment, is high in cost and cannot be widely applied to practice. The traditional method for testing fatigue by using electric signals aims at human fatigue, most of the fatigue is in the aspect of mental fatigue, and the fatigue is not in the aspect of eye fatigue. Based on the above problems in the prior art, a new system and method for detecting eye fatigue is urgently needed to be developed.

The eye potential (EOG) records the resting potential between the cornea of a human, which is generally positively charged, and the retina, which is negatively charged, has a weak potential change in the epidermis around the eye when the eyeball rotates, which can be picked up by electrodes placed around the human eye. The amplitude range of the general human eye electrical signal is 0.4-10 m V, the frequency is concentrated in 0.1-38 Hz, and the main component is below 10 Hz. Various diseases caused by retinopathy, such as nyctalopia caused by various causes, optic atrophy of the eye, congenital amaurosis, blood circulation disorder of the retina, and the like, can be evaluated using the ocular electrical signal.

Based on the eye fatigue detection system and method, the collected eye electrical signals are amplified and filtered, eye electrical signal characteristics related to fatigue degree are extracted from the eye electrical signals, and accurate prediction of eye fatigue is achieved through a fatigue detection algorithm by using the characteristic values. To solve the above-mentioned problems.

Disclosure of Invention

The present invention is directed to a system and a method for detecting eye fatigue, which solve the above problems of the prior art.

In order to achieve the purpose, the invention provides the following technical scheme: the eye fatigue detection system comprises a signal acquisition electrode, an analysis module, a communication module, a mobile equipment end and a data platform,

the signal acquisition electrode is a single-channel electrode and comprises three electrodes: the left electrode, the right electrode and the right leg driving electrode are respectively fixed at the left position, the right position and the middle position of the device, and can continuously record the eye electrical signals of the person to be measured in real time,

the analysis module comprises a primary signal amplification module, a secondary amplification module, a filter circuit module, a high-precision sampling module, a microprocessor, an external control module and a communication module, wherein the primary signal amplification module, the secondary amplification module, the filter circuit module and the high-precision sampling module are sequentially connected, the microprocessor is respectively connected with the high-precision sampling module, the external control module and the communication module, the primary signal amplification module, the secondary amplification module, the filter circuit module, the high-precision sampling module and the microprocessor are integrated in one chip,

the communication module is configured to communicate the calculated data and the collected data with the mobile equipment end through Bluetooth or WIFI,

the mobile equipment terminal transmits the data to a data platform through WIFI or 4G for storage and analysis to obtain long-term fatigue data,

the data platform is used for displaying data in real time, analyzing big data, summarizing and analyzing the data acquired for a long time, and tracking the eye fatigue health condition of the user for a long time.

Preferably, the signal collecting electrode is made of a material with good conductivity and is a dry electrode or a wet electrode, and the dry electrode is one of a metal, a metal alloy electrode or conductive cloth.

Preferably, the primary amplification module has a sufficiently high input impedance, at least 10M Ω or more, which is much greater than the impedance between the electrode and the skin.

Preferably, the high-precision sampling module adopts an oversampling technology, and has a very high sampling frequency and sampling precision, wherein the sampling frequency is above 250HZ, and the number of sampling bits is above 16 bits.

Preferably, the microprocessor is embedded with a high-performance main control chip, so that the eye fatigue algorithm can be smoothly operated, the characteristic signals of the eye electrical signals are extracted, and the eye fatigue degree is scored and classified according to the characteristic values in subsequent processing.

Preferably, the mobile device end is one of a mobile phone, a tablet or a computer, is configured with an APP, displays a calculation result in real time, and transmits data as a transfer device end.

Preferably, the detection method of the eye fatigue detection system is further included: extracting seven eye movement characteristic signals from the eye electrical signals, and extracting characteristic values after preprocessing; and taking different characteristic values of the electro-ocular signals as data to be processed next, and performing machine learning related algorithm processing to finally obtain the fatigue degree grading and classification of the eye fatigue.

Preferably, the method comprises the following specific steps:

(1) acquiring an ocular electrical signal: electrodes are arranged in the left eye, the right eye and the forehead area, and the most original eye electrical signals of the person to be tested are continuously recorded in real time;

(2) preprocessing of the original ocular electrical signal: low-pass filtering, down-sampling and wavelet analysis;

(3) extracting a characteristic signal; the method comprises various filtering methods, wavelet transformation, threshold interpolation extraction algorithm and Fourier transformation;

(4) and (3) scoring and classifying by using the characteristic value of the obtained eye movement signal: seven characteristic values are extracted by using the electro-oculogram signals, and the eye fatigue is scored and classified by using a decision tree and a support vector.

Preferably, in the step (2), the preprocessing method specifically includes performing digital filtering low-pass filtering to remove power frequency interference from the original signal, and for baseline wander, performing wavelet processing to analyze a baseline wander interference signal lower than 0.05HZ by using wavelet, performing multilayer wavelet decomposition on the filtered eye electrical signal to obtain a baseline component with a frequency lower than 0.1HZ, and subtracting the baseline part from the original signal to finally obtain a relatively pure EOG eye electrical signal with power frequency interference and baseline wander removed.

Preferably, in the step (3), the extraction method specifically includes the following steps:

3.1), extraction of slow eye movement: firstly, low-pass filtering of 25HZ is carried out on the preprocessed eye electric signals, the signals are classified by using a wavelet analysis method, classification similar to a frequency domain of slow eye movement signals (the frequency is between 0.2HZ and 0.6HZ) is found out, the classification is used as a base, and then wavelet coefficients are used for calculating energy of each step, wherein the calculation formula is as follows:

wherein E is_iRepresenting the total energy contained in the ith order wavelet, d_ijRepresents the ith order jth number. Therefore, the proportion of the slow eye movement to the whole eye movement process and the slow eye movement energy spectrum can be calculated;

3.2), rapid eye movement extraction:

first, the signal is band-pass filtered using eeg lab,

then, the difference of the signals is calculated, each point represents the change rate of the signals at the same time, and the formula is used:

D[i]＝(V[i+1]-V[i])×R

obtaining a differential signal, wherein D is the differential signal, V represents the eye movement signal after filtering, R is the sampling rate,

finally, counting the number of points exceeding a threshold value;

3.3), extracting eye movement energy: extracting energy characteristics by utilizing Fourier transform, wherein a slow eye movement frequency band is 0.1Hz-1Hz, and a fast eye movement frequency band is 1.5Hz-2 Hz; four frequency ranges of 2Hz-5Hz, 5Hz-8Hz and 8Hz-10Hz,

according to the pascal relation, the sum of squares of the time domain of the signal is equal to the sum of squares of the frequency domain, so that the energy of the frequency band is the sum of squares of the band-pass filtered signal values, and the energy values of different frequency bands within each ns can be expressed as:

3.4), calculating the eye movement peak value: taking upper and lower thresholds for the amplitude of the whole ocular electrical signal, making a sum after counting the number of all peaks and valleys exceeding the threshold,

N_[total]＝N_[peaks]+N_[valleys]

3.5), eye movement mean number: dividing the total number of the eye movement peak values obtained in the last step by the number of statistical periods in the period to obtain the average number of the eye movement peak values of each period in the measuring period,

3.6), standard deviation of eye movement: based on the number of eye movement peaks in each period and the number of eye movement averages in each period, the standard deviation of the eye movement signal is obtained by using the following formula, where N is_TIs the number of cycles in the measurement time; n is a radical of_IIs the number of eye movements per cycle; then N is_ATo average the number of eye movements per cycle over the measurement period,

preferably, the step (4) is implemented by the following steps:

4.1) calculating entropy by using seven characteristic values extracted by the electro-oculogram signals, thereby judging which characteristic is split first, wherein the calculation formula of the information entropy is as follows:

wherein p (x) represents the probability of occurrence of event x;

4.2) selecting the characteristic with the maximum information moisture change amplitude as a data set division basis;

4.3) recursively processing the divided sub data sets, continuously repeating the steps from the unselected features, selecting the optimal data division feature to divide the sub data sets to form a decision tree, and realizing the purpose of eye fatigue classification.

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

(1) the invention provides a device and a method for detecting eye fatigue degree by using an eye electrical signal, which is a solution scheme with convenient use, low cost and large-scale popularization.

(2) The method comprises the steps of amplifying and filtering collected electro-oculogram signals, extracting electro-oculogram signal characteristics related to fatigue degree from the electro-oculogram signals, and accurately predicting eye fatigue through a fatigue degree detection algorithm by using the characteristic values, wherein the algorithm complexity meets the requirement of low power consumption and can be implemented in portable equipment.

(3) Seven characteristic values are extracted through the electro-oculogram signals, and the eye fatigue is scored and classified by using a decision tree and a support vector, so that the eye fatigue is concise and convenient to detect, targeted and suitable for large-scale application.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without creative efforts.

FIG. 1 is a block diagram of a system for detecting eye fatigue according to the present invention;

FIG. 2 is a schematic diagram of the wearing position of the electrode during fatigue detection according to the present invention;

FIG. 3 is a flowchart of an algorithm for measuring eye fatigue according to the present invention;

FIG. 4 is a flow chart of a decision tree according to the present invention.

In the drawings, the components represented by the respective reference numerals are listed below:

100. a signal detection electrode; 110. an analysis module; 111. a primary signal amplification module; 112. a secondary amplification module; 113. a filter circuit module; 114. a high-precision sampling module; 115. a microprocessor; 116. a peripheral control module; 117. a communication module; 120. a mobile device end; 130. and a data platform.

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.

Fig. 1 is a block diagram of an eye fatigue detection system according to an embodiment of the present invention, which includes a signal collecting electrode 100, an analyzing module 110, a communication module 117, a mobile device 120, and a data platform 130. The signal detection electrode 100 is used for detecting an eye electrical signal by contacting a specific eye part of a human body; the analysis module 110 comprises a primary signal amplification module 111, a secondary amplification module 112, a filter circuit module 113, a high-precision sampling module 114, a microprocessor 115, an external control module 116 and a communication module 117; the primary signal amplification module 111 is a first-stage amplification circuit of a weak signal, has the characteristics of high input impedance, low noise, high common-mode rejection ratio, low drift, nonlinearity and the like, and can provide a proper dynamic range; the secondary signal amplifying module 112 is used for further amplifying the signal of 111 to increase the gain; the filter circuit module is used for eliminating power frequency interference, baseline drift and possible external noise; the high-precision sampling module 114 is used for acquiring 113 amplified signals, performing three-stage amplification and performing high-precision AD signal acquisition; the microprocessor 115 may be a single chip microcomputer system such as STM32, FPGA, DSP, etc. for extracting characteristic signals of the electro-oculogram signals and scoring and classifying the eye fatigue degree according to the characteristic values in the subsequent processing; the peripheral indication control module 116 is used for indicating the connection state of the communication module and other parts; the communication module 117 is used for communication between the microprocessor and the mobile device; the mobile device end 120 realizes data communication with the microprocessor through the communication module; the data platform 130 receives the data of the mobile device, stores and analyzes the data, tracks the eye fatigue information of the user and provides guidance suggestions.

In some embodiments, the signal collecting electrode 100 may be a dry electrode, such as a metal, metal alloy electrode, conductive cloth, or the like, which may be made of a conductive material, or may be a wet electrode.

In some embodiments, the wearing position of the electrodes in detecting fatigue is shown in fig. 2, and comprises three electrodes: a left electrode, a right electrode, and a right leg drive electrode. The electro-ocular electrodes are respectively fixed at the left, right and middle positions of the equipment and record the electro-ocular signals of the person to be measured in real time.

In some embodiments, the primary signal amplification module 111 primarily amplifies the most primitive ocular current signal and converts the ocular current signal into a voltage signal.

In some embodiments, the input impedance of the primary signal amplification module 111 is high enough, and generally should be above 10M Ω, much larger than the impedance between the electrode and the skin, so as to obtain a higher amplitude signal.

In some embodiments, the high-precision sampling module 114 employs an oversampling technique, and has a high sampling frequency and a high sampling precision, and the sampling frequency is generally at least above 250HZ, and at least above 16 sampling bits, so as to ensure that a valid EOG signal can be extracted.

In some embodiments, the primary signal amplification module 111, the secondary amplification module 112, the filter circuit module 113, the high-precision sampling module 114, and even the microprocessor 115 are integrated inside one chip to reduce the size and volume of the device, which is suitable for wearable devices.

In some embodiments, when the communication module 116 and the mobile client 120 are in a connected state, a connection indicator light prompts; when the quality of the signal obtained by the analysis module is unstable or no signal exists, the prompt is given through sound or an indicator light.

In some embodiments, the microprocessor 115 is embedded with a high-performance main control chip, and can operate an eye fatigue algorithm, extract characteristic signals of the electro-oculogram signals, and perform eye fatigue degree scoring and classification according to the characteristic values in subsequent processing;

in some embodiments, the main control chip communicates with the communication module through a serial port, and the serial port communication module communicates with the mobile device through bluetooth or WIFI.

In some embodiments, the mobile device end 120 may be a mobile device such as a mobile phone, a tablet, a computer, etc.

In some embodiments, the mobile device end 120 transmits the data to the data platform through WIFI or 4G, and stores and analyzes the data to obtain data of long-term fatigue.

As another aspect of the present invention, there is provided a method for detecting eye fatigue, which is implemented by applying the aforementioned apparatus. Fig. 3 is a flowchart of an algorithm for measuring and measuring eye fatigue, and the specific implementation steps and method are as follows:

step 1: acquiring an ocular signal

EOG is a technique for measuring resting potential of the retina, known as electrooculography. The cornea of the human eye appears positively with respect to the posterior end of the eyeball and is considered to be a static potential because its potential is not affected by the presence of light. This potential is not fixed, it changes extremely slowly, and provides a basis for the generation of electrooculogram. The eye electrical measurement is the cornea retina potential before and after the human eye, is commonly used for eye diagnosis and eye movement recording, and the common size is 0.4 mV-1.0 mV. To measure eye movement, typically one electrode is placed above and below and to the left and right of the eye. If the eyeball moves towards a certain direction, the electrodes can sense the positive and negative electrode movement of the retina, and the two electrodes show opposite electric potentials, so that a potential difference exists between the two electrodes. Using this principle, electrodes are placed in the left and right eye and forehead regions, resulting in the most primitive ocular electrical signals.

Step 2: pre-processing of raw ocular signals

When the electro-oculogram signal is obtained, the original signal is usually accompanied by signal interference such as power frequency interference and baseline drift, and the original signal also comprises signals such as myoelectricity and electroencephalogram. Preprocessing the raw signal to obtain a relatively pure electro-oculogram signal is the first step of the whole eye fatigue analysis. And removing baseline drift and power frequency interference by utilizing low-pass filtering, down sampling and wavelet analysis. The eye movement signals can be divided into slow eye movement signals and fast eye movement signals according to frequency, the frequency of the slow eye movement signals is usually 0.2HZ-0.6HZ, the frequency of the fast eye movement signals is 1HZ-8HZ, and the frequency of power frequency interference is usually above or below 50 HZ. Therefore, the digital filtering low-pass filtering is adopted to realize the removal of power frequency interference on the original signal. Compared with electroencephalogram and electromyogram, the amplitude of the electro-oculogram signal is relatively high, but the amplitude of the signal from the electroencephalogram and the electromyogram, which can be obtained by the electrode position for testing the electro-oculogram signal provided by the system, is very low, the interference on the electro-oculogram signal is very little, and therefore the noise can be reduced without processing. For baseline drift, wavelet processing is used. In ocular electrical signals, signals typically below 0.05HZ are considered baseline wander disturbances. Wavelet analysis is utilized, multi-layer wavelet decomposition is carried out on the filtered electro-ocular signals to obtain baseline components with the frequency lower than 0.1HZ, the baseline part is subtracted from the original signals, and finally the purer EOG electro-ocular signals without power frequency interference and baseline drift are obtained.

And step 3: feature signal extraction

The eye movement signal comprises a proportion of slow eye movement, a slow eye movement average amplitude, a slow eye movement amplitude variance, a fast eye movement proportion, a fast eye movement peak speed, a fast eye movement average amplitude, a fast eye movement amplitude variance, a blinking time period, an eye closing time period, an eye opening time delay, an eye opening time period to blinking time period ratio, a blinking interval, a blinking average amplitude, an eye closing peak speed, an eye opening peak speed, an eye closing average speed, an eye opening average speed, a horizontal eye electrical low-frequency high-frequency energy ratio and a vertical eye electrical low-frequency high-frequency energy ratio.

This patent adopts three electrode equipment of single channel, only adopts the horizontal eye movement signal of left and right eyes difference, has effectively simplified the equipment and has used the degree of difficulty, has reduced the algorithm dimension disaster because of using too much characteristic signal to cause. Selecting several types most related to fatigue from the characteristic values for extraction and processing, wherein the following steps are respectively adopted: slow eye movement signals (number and ratio column), fast eye movement signals (number and ratio column), eye movement energy, eye movement peak value number (corresponding to total eye movement times), eye movement average value number, eye movement standard deviation value and eye movement signal arrangement entropy. Since no electrode is placed on the upper and lower parts of the eye, the upper and lower eye movement signals and the blink signal are not extracted.

Extraction of slow eye movement: firstly, low-pass filtering of 25HZ is carried out on the preprocessed eye electric signals, the signals are classified by using a wavelet analysis method, and classification similar to a slow eye movement signal (the frequency is between 0.2HZ and 0.6HZ) in a frequency domain is found out and is used as a base. And then calculating the energy of each order by using the wavelet coefficient. The calculation formula is as follows:

wherein E is_iRepresenting the total energy contained in the ith order wavelet, d_ijRepresents the ith order jth number. From this, the occupancy of the slow eye movement over the entire eye movement and the slow eye movement power spectrum can be calculated.

Fast eye movement extraction: for fast eye movement signals, a threshold-based interpolation extraction algorithm is used. A threshold is set for the speed of horizontal eye movement, above which a rapid eye movement is deemed to have occurred. And performing 1-8HZ band-pass filtering on the horizontal eye movement signal, calculating a differential signal of the horizontal eye movement signal, taking an absolute value of the differential signal, and finally counting the number of points exceeding a threshold value. In a first step, the signal is band-pass filtered using eeglab. In the second step, the difference of the signals is calculated, and each point represents the change rate of the signals at the same time. Using the formula:

D[i]＝(V[i+1]-V[i])×R

and obtaining a differential signal, wherein D is the differential signal, V represents the eye movement signal after filtering, and R is the sampling rate.

Extracting eye movement energy: the energy of different frequency bands in the ocular electrical signal may express the intensity of different movements. This patent utilizes fourier transforms to extract energy features. The slow eye movement frequency band is 0.1HZ-1 HZ; in order to calculate the fast eye movement energy more accurately, the fast eye movement frequency band is 1.5Hz-2 Hz; 2Hz-5Hz, 5Hz-8Hz and 8Hz-10 Hz. According to the pascal relation, the sum of squares of the time domain of the signal is equal to the sum of squares of the frequency domain, so that the energy of the frequency band is the sum of squares of the band-pass filtered signal values, and the energy values of different frequency bands within each ns can be expressed as:

calculating the number of eye movement peaks: and taking upper and lower thresholds for the amplitude of the whole ocular electrical signal, and summing after counting the number of all peaks and valleys exceeding the thresholds.

N_[total]＝N_[peaks]+N_[valleys]

Eye movement mean number: and dividing the total number of the eye movement peak values obtained in the last step by the number of the statistical periods in the period to obtain the average number of the eye movement peaks per period in the measuring period.

Standard deviation of eye movement: based on the number of eye movement peaks in each period and the number of eye movement averages in each period, the standard deviation of the eye movement signal can be obtained by using the following formula. Wherein, N is_TIs the number of cycles in the measurement time; n is a radical of_IIs the number of eye movements per cycle; then N is_AThe average number of eye movements per cycle during the measurement period.

Eye movement signal arrangement entropy: the eye movement signal arrangement entropy is mainly used in a decision tree algorithm for subsequent eye fatigue measurement processing, so the work of extracting the eye movement signal arrangement entropy is explained in detail in the fourth step.

And 4, step 4: scoring and classifying by using the characteristic value of the obtained eye movement signal

The China Association of the electronic video industry issued ' rating method for visual fatigue testing and rating method of display terminal ' part 2 ' in 2019, 7, 18.A visual fatigue rating scale with accuracy and precision data is combined with domestic and foreign research experiences, and aims to standardize a method for evaluating eye fatigue of the public due to the use of a display terminal in the industry. The method adopts a questionnaire survey mode, obtains the fatigue score of the tested person by scoring a plurality of options, and is a unique authority evaluation method in the industry. The evaluation method is combined, before the tested person is tested by the EOG device, the eye fatigue of the tested person at the moment is scored and classified by the evaluation method, and the eye fatigue is used as a reference value to perform machine learning and classification on data.

The method utilizes a decision tree and a support vector machine to classify the fatigue degree of the electro-oculogram signals.

The decision tree (decision tree) is a tree structure, and the decision process of the decision tree is very intuitive and easy to understand. Each non-leaf node represents a test on a feature attribute, each branch represents the output of the feature attribute over a range of values, and each leaf node stores a category. The process of using the decision tree to make a decision is to start from the root node, test the corresponding characteristic attributes in the items to be classified, select an output branch according to the value of the characteristic attributes until the leaf node is reached, and take the category stored by the leaf node as a decision result. Information entropy is first calculated on the characteristic values of several ocular signals obtained in between, thereby judging which characteristic is split first. The calculation formula of the information entropy is as follows:

where p (x) represents the probability of occurrence of event x. After all characteristic values are calculated respectively, the change values of information moisture before and after the data set are divided by using the formula, and then the characteristic with the largest change amplitude of the information moisture is selected as a data set dividing basis. And finally recursively processing the divided sub data sets, continuously repeating the steps from the unselected features, and selecting the optimal data division feature to divide the sub data sets to form a decision tree. Finally, the purpose of classifying the eye fatigue is achieved.

For a clearer explanation of the eye fatigue decision tree algorithm, fig. 4 is a flow chart of the decision tree used in the present patent.

The linear Support Vector Machine (SVM) has similar action with the decision tree, and the algorithm is used for comparing the prediction result with the decision tree. It has been one of the most popular machine learning classifiers due to its high degree of flexibility, excellent computational efficiency and fast high dimensional data processing capabilities. The method is a supervised learning method and is widely applied to statistical classification and regression analysis. The specific idea is to map the vector into a higher-dimensional space, convert the linear indivisible sample of the low-dimensional input space into a high-dimensional feature space and make it linearly separable, and establish a maximum interval hyperplane in this space. The classification decision is to obtain a feature vector according to the mapping of the current sample on the feature space, and to use a classification model to classify the feature vector into a certain class. The linear support vector machine classifier model utilized by this patent is trained into a three classification model that can classify mild fatigue/moderate fatigue/severe fatigue.

In the description herein, references to the description of "one embodiment," "an example," "a specific example" or the like are intended to mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

The preferred embodiments of the invention disclosed above are intended to be illustrative only. The preferred embodiments are not intended to be exhaustive or to limit the invention to the precise embodiments disclosed. Obviously, many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the invention and the practical application, to thereby enable others skilled in the art to best utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. Eye fatigue degree detecting system, its characterized in that: comprises a signal acquisition electrode, an analysis module, a communication module, a mobile equipment end and a data platform,

the signal acquisition electrode is a single-channel electrode and comprises three electrodes: the left electrode, the right electrode and the right leg driving electrode are respectively fixed at the left position, the right position and the middle position of the device and continuously record the eye electrical signals of the person to be measured in real time,

the mobile equipment terminal transmits the data to the data platform through WIFI or 4G for storage and analysis,

2. The eye fatigue detection system according to claim 1, wherein: the signal acquisition electrode is a dry electrode or a wet electrode, and the dry electrode is one of metal, metal alloy electrode or conductive cloth.

3. The eye fatigue detection system according to claim 1, wherein: the primary amplification module is at least more than 10M omega and is far larger than the impedance between the electrode and the skin.

4. The eye fatigue detection system according to claim 1, wherein: the high-precision sampling module adopts an oversampling technology, the sampling frequency is more than 250HZ, and the sampling bit number is more than 16 bits.

5. The eye fatigue detection system according to claim 1, wherein: the mobile equipment terminal is one of a mobile phone, a tablet or a computer, is configured with an APP, displays a calculation result in real time, and transmits data as a transfer equipment terminal.

6. The system for detecting eye fatigue of any one of claims 1-5, wherein: the detection method of the eye fatigue detection system is also included: extracting seven eye movement characteristic signals from the eye electrical signals, and extracting characteristic values after preprocessing; and taking different characteristic values of the electro-ocular signals as data to be processed next, and performing machine learning related algorithm processing to finally obtain the fatigue degree grading and classification of the eye fatigue.

7. The detection method of the eye fatigue degree detection system according to claim 6, characterized in that: the method comprises the following specific steps:

8. The detection method of the eye fatigue degree detection system according to claim 7, characterized in that: in the step (2), the preprocessing method specifically includes digital filtering low-pass filtering to remove power frequency interference from the original signal, for baseline wander, wavelet processing is adopted, wavelet analysis is used for baseline wander interference signals lower than 0.05HZ, multi-layer wavelet decomposition is carried out on the filtered eye electric signals to obtain baseline components with frequency lower than 0.1HZ, and the baseline part is subtracted from the original signal to finally obtain relatively pure EOG eye electric signals with power frequency interference and baseline wander removed.

9. The detection method of the eye fatigue degree detection system according to claim 7, characterized in that: in the step (3), the extraction method specifically comprises the following steps:

3.2), rapid eye movement extraction:

first, the signal is band-pass filtered with eeglab,

D[i]＝(V[i+1]-V[i])×R

finally, counting the number of points exceeding a threshold value;

N_[total]＝N_[peaks]+N_[valleys]

10. the detection method of the eye fatigue detection system according to claim 9, characterized in that: the step (4) is realized by the following steps:

wherein p (x) represents the probability of occurrence of event x;