CN106128032A - A kind of fatigue state monitoring and method for early warning and system thereof - Google Patents
A kind of fatigue state monitoring and method for early warning and system thereof Download PDFInfo
- Publication number
- CN106128032A CN106128032A CN201610523738.5A CN201610523738A CN106128032A CN 106128032 A CN106128032 A CN 106128032A CN 201610523738 A CN201610523738 A CN 201610523738A CN 106128032 A CN106128032 A CN 106128032A
- Authority
- CN
- China
- Prior art keywords
- fatigue
- monitoring
- early warning
- state
- fatigue state
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/06—Alarms for ensuring the safety of persons indicating a condition of sleep, e.g. anti-dozing alarms
Landscapes
- Business, Economics & Management (AREA)
- Emergency Management (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention discloses a kind of fatigue state monitoring and method for early warning, comprise the following steps: (1) preset fatigue status monitoring and early warning system；(2) by the EEG signals of physiology information detecting module timing acquiring tested personnel, physiology information detecting module gathers the bioelectrical signals that brain produces, and send in intelligent chip, noise in bioelectrical signals and disturbance are filtered by intelligent chip, and the useful organisms signal of telecommunication after filtering is amplified；(3) the useful organisms signal of telecommunication is reached monitoring modular by wireless communication module；(4) the useful organisms signal of telecommunication is analyzed processing by monitoring modular, and differentiates whether tested personnel is in fatigue state；(5) when main control unit differentiates that tested personnel is in fatigue state, prewarning unit sends warning and reminds.Also disclose a kind of fatigue state monitoring and early warning system.The monitoring of this fatigue state and method for early warning and system thereof, capacity of resisting disturbance is strong, and reliability is high, improves the speed and accuracy differentiated.
Description
Technical field
The present invention relates to fatigue detecting field, particularly relate to a kind of fatigue state monitoring and method for early warning and system thereof.
Background technology
Along with the rapid development of China's economic, increasing people has purchased private car, but vehicle accident rises the most therewith,
Tracing it to its cause, the most serious fatigue driving of driver fatigue is the major incentive causing vehicle accident to take place frequently.Accident data
Statistics display, the traffic fatalities of 35%-45% are attributable to driving fatigue.Driving fatigue influences whether the note of driver
Meaning, sensation, consciousness, thinking, judgement, will, determine and the aspects such as motion.Driver is likely to be in fatigue driving, and
Being so extremely hazardous, and be currently used for preventing fatigue to be roughly divided into two classes, one is actively, and one is passive.Main
Flowing mode be exactly driver when perceive oneself drive to be in comparison tired time will ensure enough time of having a rest.But due to people
Rhythm of life very fast, operating pressure is big, causes driver the most less to take a good rest, and then causes the generation of accident.Separately
Outer one is passive mode, it is simply that judged by an external instrument.
At present both at home and abroad the product of preventing fatigue driving also exists corresponding defect the most simultaneously:
1, hanging on ear, function is bowed exactly and will be reported to the police；But tired dozing off is not necessarily intended to bow, and wait until
Bow the most a little evening；
2, using article of refreshing oneself, later physical and mental health can be caused bad consequence by the work of high intensity；
3, watch style, estimates the fatigue of people by beating of pulse；Single for tired judgement data source, thus may be used
Erroneous judgement can be there is；
4, human eye monitoring, captures eye image by photographic head, it is achieved fatigue monitoring；Same shortcoming is physiological data list
One, credibility needs to be investigated；
5. steering wheel is touch, and utilization is installed some sensors on the steering wheel and come whether perception driver holds direction
Dish.Principle is when driver fatigue, reacts slack-off, and the action of direction of operating dish also can slow down, the possible run-off-road of vehicle, or
Suddenly track is revised；Shortcoming to install a sensor at steering wheel, and so the meeting original structure of destruction direction dish, also has this
Unsuitable for various road conditions, as overtaken other vehicles, doubling, can there is wrong report in the situation such as turning.
Therefore, develop a kind of differentiate in time, fatigue state monitoring fast and accurately and method for early warning, it appears the most urgent
Cut.
Summary of the invention
For above-mentioned deficiency, it is an object of the invention to provide a kind of fatigue state monitoring and method for early warning and system thereof,
To solve the problem that existing fatigue state monitoring method distinguishing speed is slow, accuracy is low.The fatigue state monitoring that the present invention provides
And method for early warning and system thereof, capacity of resisting disturbance is strong, and reliability is high, substantially increases speed that fatigue state differentiates and accurately
Degree.
The present invention be the technical scheme is that by reaching above-mentioned purpose
A kind of fatigue state monitoring and method for early warning, it is characterised in that comprise the following steps:
(1) preset one fatigue state monitoring and early warning system, it include physiology information detecting module, wireless communication module with
Monitoring modular, wherein, described physiology information detecting module built-in dry state electrode sensor and intelligent chip, described monitoring modular bag
Including interconnective main control unit and prewarning unit, this main control unit includes that pretreatment subelement, model are set up subelement and melted
Zygote unit；
(2) by the EEG signals of described physiology information detecting module timing acquiring tested personnel, described physiologic information is adopted
Dry state electrode sensor in collection module gathers the bioelectrical signals that brain produces, and the bioelectrical signals collected is sent into intelligence
In energy chip, the disturbance that the noise mixed in bioelectrical signals and motion produce is filtered by this intelligent chip, and will filter
The useful organisms signal of telecommunication after removing is amplified processing；
(3) the useful organisms signal of telecommunication after processing and amplifying is reached monitoring modular by described wireless communication module；
(4) the useful organisms signal of telecommunication is analyzed processing by the main control unit in described monitoring modular, wherein, and described pre-place
Reason subelement carries out pretreatment to the useful organisms signal of telecommunication, and pretreatment includes useful organisms electrical signal data is resolved, gone
Make an uproar and Filtering Processing；Described model sets up subelement to pretreated useful organisms signal of telecommunication founding mathematical models, i.e. distinguishes
Set up frequency band power value, frequency band power percentage ratio, frequency band power ratio and frequency band power to add and ratio, and calculate brain electric fatigue
Value, and differentiate whether tested personnel is in fatigue state；
(5) when described main control unit differentiates that tested personnel is in fatigue state, described prewarning unit sends warning and reminds；
Otherwise, step (2)-(5) are repeated.
As a further improvement on the present invention, the main control unit in described step (1) also include feature extraction subelement with
Fusant unit；Described step (2) also includes: the eye image letter of described physiology information detecting module timing acquiring tested personnel
Number；Described step (3) also includes: eye image signal is reached monitoring modular by described wireless communication module；Described step (4) is also
Including: described pretreatment subelement carries out pretreatment to eye image signal, and pretreatment includes entering eye image signal data
Row denoising, gray processing and equalization processing；Described feature extraction subelement is to pretreated eye image signal extraction human eye
Feature, and it is calculated human eye fatigue data；Brain electric fatigue value and human eye fatigue data are merged by described fusant unit, and point
Analysis calculates level of fatigue, it determines whether tested personnel is in fatigue state.
As a further improvement on the present invention, described step (4) also includes: described feature extraction subelement is with Adaboost
Algorithm is that core realizes human eye location, intercepts out human eye parts of images, increases Otsu adaptive two-tone images algorithm, and employing is averaging
The algorithm of value extracts human eye feature, designs dynamic sliding window human eye state evaluation algorithm, thus judges eyes closed threshold value
State, finally by improve Perclos human eye fatigue detecting algorithm, obtain human eye fatigue data.
As a further improvement on the present invention, described step carries out gray processing process to eye image signal data in (4),
The formula used is: Y=0.30R+0.59G+0.11B, is weighted averagely to RGB three-component in coloured image, obtains rationally
Gray level image.
As a further improvement on the present invention, the Otsu adaptive two-tone images algorithm in described step (4) specifically includes: right
In image I, (x, y), the segmentation threshold of foreground and background is T, belongs to foreground pixel and counts that to account for the ratio of whole eye image be ω
0, its average gray is μ 0；Background pixel is counted and accounted for the ratio of whole eye image is ω 1, and its average gray is μ 1, people's eye pattern
The grand mean gray scale of picture is μ, and inter-class variance is g, and eye image size is M × N, and in eye image, the gray value of pixel is less than
The number of pixels of threshold value T is denoted as N0, and pixel grey scale is denoted as N1 more than the number of pixels of threshold value T, then draw below equation:
N0+N1=M × N, ω 0+ ω 1=1,
μ=ω 0 × μ 0+ ω 1 × μ 1, g=ω 0 × (μ 0-μ)2+ω1×(μ1-μ)2,
Draw: g=ω 0 × ω 1 × (μ 0-μ 1)2；The method using traversal obtains the maximum threshold making inter-class variance g corresponding
Value T, finally according to threshold value T, carries out binaryzation to gray level image, i.e. when image intensity value is more than threshold value T, is entered as 255, no
Then it is entered as 0.
As a further improvement on the present invention, in described step (4), eye image is believed by described feature extraction subelement
The step number extracting human eye feature is: calculate the average of black pixel point number shared by black line segment in eye image vertical direction
It is worth, the number=(bar of the number of total black pixel point/vertical black line segment of black pixel point shared by averagely every black line segment
Number), it is more than at eye closing shape according to tested personnel's number of black pixel point shared by averagely every black line segment under eyes-open state
The number of black pixel point shared by averagely every black line segment under state, and combine black pixel point shared by averagely every black line segment
Number, it is judged that tested personnel is eyes-open state or closed-eye state；Then, within the unit interval, carry out multiple repairing weld, to gather
To frame number be unit, calculate eyes closed state frame number and account for the percentage ratio of totalframes, formula is:Wherein, in P represents a period of time, the frame number of eyes closed state accounts for gathered total
The ratio of frame number.
As a further improvement on the present invention, the pre-treatment step of described step (4) including: uses constant compensation algorithm pair
Previous value is repaired, and the value of next point is determined by a upper point, the value of next point takes=currently this point value-
(value of a upper point) 1/60+ adds the value of a point.
As a further improvement on the present invention, the main control unit in described step (1) also includes syndrome unit；Described step
Suddenly (4) also include: described syndrome unit carries out T inspection according to focus, the meditation degree of tested personnel with the mathematical model of foundation
Testing process, this T inspection formula is:
Wherein, d ' is sample average, and Sd is sample variance, and n is number of samples, and α is confidence interval.
A kind of fatigue state monitoring implementing said method and early warning system, it is characterised in that include physiology information detecting
Module, wireless communication module and monitoring modular, wherein, described physiology information detecting module built-in dry state electrode sensor and intelligence
Chip, for the physiologic information of timing acquiring tested personnel；Described wireless communication module is for sending physiologic information to monitoring
Module；Described monitoring modular includes interconnective main control unit and prewarning unit, and described main control unit is for physiologic information
Carrying out processing and carrying out degree of fatigue differentiation, described prewarning unit is used for sending warning and reminds；Described main control unit includes pre-place
Reason subelement, model set up subelement and fusant unit.
As a further improvement on the present invention, described main control unit also include feature extraction subelement, fusant unit with
Syndrome unit.
The invention have the benefit that the tired feelings that can reflect human body according to brain wave data and human eye feature value in real time
The characteristic of condition, is acquired with eye image tested personnel's EEG signals, it is achieved the fatigue state prison of multi-modal biological characteristic
Surveying, multicomponent data source determines the monitoring of this fatigue state and early warning reliability promotes, and the ability of anti-objective factor interference strengthens, no
Easily directly result in the accuracy that final fatigue data differentiates because part data are disturbed, substantially increase the speed that fatigue state differentiates
Degree and accuracy, safety is high；Only characteristics of human body's value is acquired, not by the normal driving action of driver in driving procedure
Interference, the impacted degree of data is close to zero.
Above-mentioned is the general introduction of inventive technique scheme, below in conjunction with accompanying drawing and detailed description of the invention, is the present invention further
Explanation.
Accompanying drawing explanation
Fig. 1 is the untreated brain wave broken line graph of the embodiment of the present invention；
Fig. 2 is the brain wave broken line graph after removing and making an uproar of the embodiment of the present invention；
Fig. 3 is the brain wave broken line graph through sliding window algorithm of the embodiment of the present invention；
Fig. 4 is the brain wave broken line graph through learning algorithm at a slow speed of the embodiment of the present invention；
Fig. 5 is the comparison diagram of the embodiment of the present invention and tradition brain wave algorithm；
Fig. 6 is the eye opening of the embodiment of the present invention, eye closing binaryzation picture comparison diagram；
Fig. 7 is that the human eye feature of the embodiment of the present invention extracts flow chart；
Fig. 8 is the feed-forward type neutral net schematic diagram of the BP learning algorithm of the embodiment of the present invention；
Fig. 9 is the mobile terminal application flow schematic diagram of the embodiment of the present invention；
Figure 10 is the algorithm flow chart of the embodiment of the present invention.
Detailed description of the invention
By further illustrating the technological means and effect that the present invention taked by reaching predetermined purpose, below in conjunction with accompanying drawing
And preferred embodiment, the detailed description of the invention of the present invention is described in detail.
From the point of view of medical science, some physiological feature, the change such as physiological signals such as nictation, heart beating, blood pressure, brain waves can
To reflect the degree of driver fatigue.And the present invention is by adopting with eye image signal the EEG signals of tested personnel
Collection, in the case of using multi-mode signal to gather, it determines whether tested personnel is in fatigue state.
Embodiment one:
The present embodiment provides a kind of fatigue state monitoring and method for early warning, comprises the following steps:
(1) preset one fatigue state monitoring and early warning system, it include physiology information detecting module, wireless communication module with
Monitoring modular, wherein, described physiology information detecting module built-in dry state electrode sensor and intelligent chip, described monitoring modular bag
Including interconnective main control unit and prewarning unit, this main control unit includes that pretreatment subelement, model are set up subelement and melted
Zygote unit；
(2) by the EEG signals of described physiology information detecting module timing acquiring tested personnel, described physiologic information is adopted
Dry state electrode sensor in collection module gathers the bioelectrical signals that brain produces, and the bioelectrical signals collected is sent into intelligence
In energy chip, the disturbance that the noise mixed in bioelectrical signals and motion produce is filtered by this intelligent chip, and will filter
The useful organisms signal of telecommunication after removing is amplified processing；
(3) the useful organisms signal of telecommunication after processing and amplifying is reached monitoring modular by described wireless communication module；
(4) the useful organisms signal of telecommunication is analyzed processing by the main control unit in described monitoring modular, wherein, and described pre-place
Reason subelement carries out pretreatment to the useful organisms signal of telecommunication, and pretreatment includes useful organisms electrical signal data is resolved, gone
Make an uproar and Filtering Processing；Described model sets up subelement to pretreated useful organisms signal of telecommunication founding mathematical models, i.e. distinguishes
Set up frequency band power value, frequency band power percentage ratio, frequency band power ratio and frequency band power to add and ratio, and calculate brain electric fatigue
Value, and differentiate whether tested personnel is in fatigue state；
(5) when described main control unit differentiates that tested personnel is in fatigue state, described prewarning unit sends warning and reminds；
Otherwise, step (2)-(5) are repeated.
In this embodiment, can use earphone that the EEG signals of tested personnel is acquired, it is preferred that can use
MindSet earphone, intelligent chip can use ThinkGear chip.Described intelligent chip will mix in bioelectrical signals
The disturbance that noise and motion produce filters, and meanwhile, interprets description tested personnel by NeuroSky eSense algorithm and works as
The eSense parameter (focus, meditation degree) of the front mental status.
Described monitoring modular is integrated on mobile terminals, mobile terminal or make mobile communication terminal refer to move
The computer equipment of middle use, even includes vehicle-mounted computer including mobile phone, notebook, panel computer, POS.Along with network and
Technology will move towards the real mobile message epoch towards the development in the most broadband direction, Mobile Communication Industry.The opposing party
Face, along with developing rapidly of integrated circuit technique, the disposal ability of mobile terminal has had powerful disposal ability.Mobile
Terminal becomes an integrated information processing platform from simple call instrument.This also adds broader to mobile terminal
Development space.
See Fig. 1 to Fig. 5, for using instrument Mind Sensor II for Windows Version 4.0 to measure brain electricity
Ripple signal, obtains tested personnel's brain wave data under different fatigue state.Brain wave data is carried out by the software of application own
Denoising Processing, then carry out fast fourier transform it is converted into frequency domain data with time domain for leader's wave data.Extract successively
Brain wave data segment under sample different fatigue state.Analyze the variation characteristic of different-waveband under different fatigue state, mainly from
Four aspects are analyzed below: frequency band power value, frequency band power percentage ratio, frequency band power ratio, frequency band power add and ratio.Phase
To prior art, prior art is analyzed by two numerical value, i.e. focus and meditation degree the most intuitively.The present embodiment
Analysis is based on actual experiment, by mathematical modeling mode, draws several proportionate relationship, the most satisfactory for result.By many
Group experiment, also indicates that with estimated value relatively, meets under driving environment, the numeric state of brain fag.
Embodiment two:
Seeing Fig. 6 and Fig. 7, the present embodiment differs primarily in that with embodiment one: the main control unit in described step (1)
Also include feature extraction subelement and fusant unit；Described step (2) also includes: the timing of described physiology information detecting module is adopted
The eye image signal of collection tested personnel；Described step (3) also includes: eye image signal is reached by described wireless communication module
Monitoring modular；Described step (4) also includes: described pretreatment subelement carries out pretreatment, pretreatment bag to eye image signal
Include and eye image signal data is carried out denoising, gray processing and equalization processing；After described feature extraction subelement is to pretreatment
Eye image signal extraction human eye feature, and be calculated human eye fatigue data；Described fusant unit by brain electric fatigue value and
Human eye fatigue data merges, and analysis meter calculates level of fatigue, it determines whether tested personnel is in fatigue state.
Described step (4) also includes: described feature extraction subelement realizes human eye location with Adaboost algorithm for core,
Intercept out human eye parts of images, increase Otsu adaptive two-tone images algorithm, use the algorithm averaged to extract human eye feature, if
Count out dynamic sliding window human eye state evaluation algorithm, thus judge the state of eyes closed threshold value, finally by improvement
Perclos human eye fatigue detecting algorithm, obtains human eye fatigue data.
Described step carries out gray processing process to eye image signal data in (4), and the formula of employing is: Y=0.30R+
0.59G+0.11B, is weighted averagely, obtaining rational gray level image to RGB three-component in coloured image.Owing to photographic head is adopted
Collect to image be coloured image, and human eye location algorithm needs to process gray level image, so needing collecting
Image carries out gray processing.Each pixel color in coloured image is determined by tri-components of R, G, B, and each component span is
0 to 255, gray level image is a kind of special coloured image that tri-components of R, G, B are identical, and gray value span is similarly 0
To 255.
And image equilibration is by the image of a known gray scale probability density distribution, convert through certain, become a width tool
Having the new images of inhomogeneous intensity probability distribution, its result is the dynamic range extending pixel value, thus has reached enhancing figure
Effect as overall contrast.
Otsu adaptive two-tone images algorithm in described step (4) specifically includes: for image I (x, y), foreground and background
Segmentation threshold be T, belong to foreground pixel and count that to account for the ratio of whole eye image be ω 0, its average gray is μ 0；Background picture
It is ω 1 that vegetarian refreshments number accounts for the ratio of whole eye image, and its average gray is μ 1, and the grand mean gray scale of eye image is μ, side between class
Difference is g, and eye image size is M × N, and in eye image, the gray value of pixel is denoted as N0 less than the number of pixels of threshold value T, as
Element gray scale is denoted as N1 more than the number of pixels of threshold value T, then draw below equation:
N0+N1=M × N, ω 0+ ω 1=1,
μ=ω 0 × μ 0+ ω 1 × μ 1, g=ω 0 × (μ 0-μ)2+ω1×(μ1-μ)2,
Draw: g=ω 0 × ω 1 × (μ 0-μ 1)2；The method using traversal obtains the maximum threshold making inter-class variance g corresponding
Value T, finally according to threshold value T, carries out binaryzation to gray level image, i.e. when image intensity value is more than threshold value T, is entered as 255, no
Then it is entered as 0.Use Otsu Threshold Segmentation Algorithm can be effectively improved the impact that light brings to a certain extent, strengthen system pair
The adaptability of complex environment.
In described step (4), the step of eye image signal extraction human eye feature is by described feature extraction subelement:
Calculate the meansigma methods of black pixel point number shared by black line segment in eye image vertical direction, shared by averagely every black line segment
The number of black pixel point=(the bar number of the number of total black pixel point/vertical black line segment), is opening eyes according to tested personnel
Under state, the number of black pixel point shared by averagely every black line segment is more than averagely every black line segment institute under closed-eye state
Account for the number of black pixel point, and combine the number of black pixel point shared by averagely every black line segment, it is judged that tested personnel is
Eyes-open state or closed-eye state；Then, within the unit interval, carry out multiple repairing weld, in units of the frame number collected, calculate eye
Eyeball closure state frame number accounts for the percentage ratio of totalframes, and formula is:Wherein, P represents one
In the section time, the frame number of eyes closed state accounts for the ratio of gathered totalframes.
It is contemplated that performance factor, the judgement of hundreds of time within the unit interval one second, can not be carried out.So the present embodiment
PERCLOS eye detection method is improved.It is not in units of the time, but in units of the frame number collected.I.e. one
The frame number that interior collection of fixing time is certain, eyes closed frame number accounts for the percentage ratio of totalframes.
The state of each frame data is can determine whether out according to above method, but can not be only according to when former frame driver's eyes
State judge whether fatigue, need the continuous data of driver's eyes state in a period of time is added up, it is judged that
The fatigue state of driver.The some frame image datas of native system continuous acquisition, if P is more than or equal to 90%, then judge that driver is
Major fatigue state, if P is between 60%～90%, then judge driver as slight fatigue state, if P is less than 60%, then
Judge that driver is as normal condition.Specifically it is referred to following three standards: (P70: refer to that eyelid covers the area of pupil and exceedes
70% is just calculated as eyes closed, adds up time scale shared during eyes closed within a certain period of time.P80: refer to that eyelid covers pupil
The area in hole is just calculated as eyes closed more than 80%, adds up time scale shared during eyes closed within a certain period of time.EM: refer to
Eyelid covers the area of pupil and is just calculated as eyes closed more than 50%, adds up the time shared during eyes closed within a certain period of time
Ratio.
The pre-treatment step of described step (4) including: uses constant compensation algorithm to repair previous value, next
The value of point is determined by a upper point, the value of next point takes=currently value-(value of a upper point) 1/60+ of this point add again
The value of a upper point.
Main control unit in described step (1) also includes syndrome unit；Described step (4) also includes: described syndrome
Unit carries out T inspection process according to focus, the meditation degree of tested personnel with the mathematical model of foundation, and this T inspection formula is:
Wherein, d ' is sample average, and Sd is sample variance, and n is number of samples, and α is confidence interval.Through multiple test number
According to combination, be finally concluded that attention with meditation is corresponding when confidence interval a=0.025n=500
T value be smaller than 2.4, illustrate that observed value | t | does not falls in region of rejection, if therefore accept H, it is believed that use two kinds of pretreatment
The result obtained is without significant difference.
The present embodiment uses mobile terminal camera to be acquired the eye image of tested personnel, mobile terminal camera
Refer to mobile terminal carry out shooting static images or short-movie shooting by built-in or external digital camera.
The present embodiment difference with the prior art is to have employed Adaboost algorithm as core, Adaboost be a kind of repeatedly
For algorithm, its core concept is the grader (Weak Classifier) different for the training of same training set, then these weak point
Class device gathers, and constitutes a higher final grader (strong classifier).Its algorithm itself is by changing data distribution
Realizing, it is the most correct according to the classification of sample each among each training set, and the general classification of last time is accurate
Rate, determines the weights of each sample.Give following grader by the new data set revising weights to be trained, finally will be every
The grader that secondary training obtains finally merges, as last Decision Classfication device.Adaboost grader is used to arrange
Except the training data feature that some are unnecessary, and it is placed on above the training data of key.This algorithm is by Freund and Schapire
A kind of popular processing method proposed according to online allocation algorithm, this algorithm has characteristics that and need not pre-prophet
The error of road Weak Classifier；Can deeply excavate the ability of Weak Classifier algorithm.Preferably, the present embodiment uses OpenCV image
Function library develops human eye fatigue monitoring system, is judged the fatigue state of driver by behavior nictation.
As seen in figs. 8-10, the feed-forward type neutral net of BP learning algorithm, BP neural network model topological structure bag are used
Include input layer (input), hidden layer (hide layer) and output layer (output layer).Input layer, from definition, quite
In extraneous stimulation, it is the source stimulated, and stimulation is passed to neuron.
For the fatigue driving algorithm of this information fusion designs, above-mentioned brain wave fatigue information and people eyestrain letter
Breath is all the input layer of this algorithm, namely neuron.For the autgmentability of this algorithm, therefore this algorithm not only supports both god
Through unit, it is also possible to support the input of multiple neuron.As except brain wave fatigue information, people's eyestrain's information, also steering wheel
Other information such as action status information, continuous driving time information, pulse beat information, body temperature information, or surrounding are all
It is can be as the input layer (neuron) of this algorithm.
Hidden layer, represents that neuron transmits stimulation and signal each other.And the quantity of stimulus accumulation of neuron is by adjacent
Neuron pass over.Represent the accumulation of semaphore with Xj, Yi represents the quantity of stimulus that neuron passes over, and Wi represents god
Through the weight of unit's quantity of stimulus, obtain formula:
Yj=∑ YiWi=(Y1*W1)+(Y2*W2)+...+(Yi*Wi)+...+(Yn*Wn)
After once Xj completes accumulation, this neuron can think that the neuron being the most also to accumulate sends signal, by it
It is expressed as Yj can obtain: Yj=F (Xj)
Input sample, uses back-propagation algorithm that weights and the deviation of network are carried out adjusting training repeatedly.First, will
Training input feeding network is to obtain exciter response, secondly, with the target output of training input correspondence, exciter response is asked poor, from
And obtain the response error of hidden layer and output layer, then input stimulus is multiplied with response error, thus obtains the gradient of weight,
Finally, it is added in weight after this gradient being multiplied by a ratio and negating.Make the vector of output with Mean Vector as much as possible
Close, train when the error sum of squares of network output layer is less than the error specified, preserved weights and the deviation of network.
Through above analysis, for obtaining the time interval of data, it is possible to there be different following the example of, theoretically
Value at every moment.But, accounting for the most very much system cpu, that is that how long to take a secondary data relatively good on earth,
Try 1ms, different the fetching data such as 10ms, 100ms, 500ms, 1s, 5s, then analyzes.In TG_BAUD_9600 baud rate feelings
Under condition, general every each value of 500ms will change.
The data that a secondary data obtains are taken, the most corresponding attention meditation raw delta at 100ms
These collection all effective data of data of theta alpha1alpha2...... are the most all once to refresh for 5 times, and wherein
Raw row refresh the most every time, this raw wave band, experiment this time will not relate to, simply explanation be every time have actual from firmly
Part gets value, if to improve after project, it is possible to carries out the most original of data from this Raw wave band and adopts
Collection, then recalculates the non-Raw data got above again according to this Raw.
For taking how many problems, after contrast experiment, this time experiment is that 1s takes once the when of test.If (should
Algorithm, for reality, thinks that the time interval of this value can be changed according to actual needs.) here occur as soon as one
Individual problem, the number every time got from hardware is the most different, it is assumed that 1s takes once, can or can not lose the effective of tired information
Data？Also having is exactly to be to launch brain wave the most again by reason human brain.If said accurately, 1s has many times
If sampling, have countless versions result.The most whether can take once with 1s？The most how long take a sub-value.(this time interval
Do the sampling interval).Whether have an impact, through consulting reference materials, this is relevant with frequency domain with time domain.Intercept one section therein, it can be seen that
The fluctuation of these lines is very big, is substantially not visible effective relation.
The embodiment of the present invention additionally provides fatigue state monitoring and the early warning system implementing said method, and it includes that physiology is believed
Breath acquisition module, wireless communication module and monitoring modular, wherein, described physiology information detecting module built-in dry state electrode sensor
With intelligent chip, for the physiologic information of timing acquiring tested personnel；Described wireless communication module is for sending physiologic information
To monitoring modular；Described monitoring modular includes interconnective main control unit and prewarning unit, and described main control unit is for life
Reason information carries out processing and carrying out degree of fatigue differentiation, and described prewarning unit is used for sending warning and reminds；Described main control unit bag
Include pretreatment subelement, model sets up subelement and fusant unit.Described main control unit also includes feature extraction subelement, melts
Zygote unit and syndrome unit.
The monitoring of the present embodiment fatigue state and early warning system are by being carried out the brain wave number to people by MindSet earphone
According to collection, and by video head, the feature of human eye is acquired.Data after gathering carry out pretreatment, respectively at brain wave
Branch mainly by noise analysis, slip Mean Filtering Algorithm, at a slow speed self-learning algorithm, two kinds of algorithms carry out T inspection process right
The brain wave data gathered carries out data analysis calculating.In people's eye portion by the pictorial information of camera collection by image in advance
Reason, the extraction of Otsu adaptive two-tone images algorithm, human eye feature value, PERCLOS human eye fatigue detecting also quantify to enter collection data
Row processes, and the data after being calculated by brain wave the most again and human eye data carry out fusion and obtain final fatigue data.
Meanwhile, " car-mounted terminal layer--high in the clouds processes layer--is mobile for the monitoring of the present embodiment fatigue state and early warning system employing
End supervisory layers " framework is design and the realization of the fatigue driving algorithm of information fusion.The vehicle-mounted layer of information gathering, is by multiple
Sensor forms, and is responsible for each characteristic parameter of detected personnel, and this has brain wave acquisition equipment, video capture device.High in the clouds
Processing layer is to receive the pretreated information of multiple equipment, is analyzed and integrates, and system, automatically according to different parameters, is made
Comprehensive evaluation.Tired information characteristics broken line graph is drawn in mobile terminal supervisory layers display.
The present invention gathers multicomponent data source, and based on BP neutral net, proposes a kind of multi-modal biological characteristic (brain wave
And human eye detection) the fatigue driving detection algorithm framework that merges, break away from the defect that conventional data source is single, compare more reliability
And accuracy.Utilize the feedback mechanism of BP neutral net, when error exceedes the value specified, then will return in last layer by backstepping
The semaphore (weight) of neuron in hidden layer is modified, thus reaches the process of study, and then reduce error, improve
The accuracy of whole fatigue data.
The emphasis of the present invention essentially consists in: can reflect the fatigue of human body in real time according to brain wave data and human eye feature value
The characteristic of situation, is acquired with eye image tested personnel's EEG signals, it is achieved the fatigue state of multi-modal biological characteristic
Monitoring, multicomponent data source determines the monitoring of this fatigue state and early warning reliability promotes, and the ability of anti-objective factor interference strengthens,
It is difficult to directly result in the accuracy that final fatigue data differentiates because part data are disturbed, substantially increases what fatigue state differentiated
Speed and accuracy, safety is high；Only characteristics of human body's value is acquired, is not moved by the normal driving of driver in driving procedure
Disturbing, the impacted degree of data is close to zero.
The above, be only presently preferred embodiments of the present invention, not impose any restrictions the technical scope of the present invention,
Therefore use or the technical characteristic of approximation identical with the above embodiment of the present invention, all within protection scope of the present invention.
Claims (10)
1. a fatigue state monitoring and method for early warning, it is characterised in that comprise the following steps:
(1) presetting fatigue state monitoring and an early warning system, it includes physiology information detecting module, wireless communication module and monitoring
Module, wherein, described physiology information detecting module built-in dry state electrode sensor and intelligent chip, described monitoring modular includes phase
The main control unit connected and prewarning unit, this main control unit includes that pretreatment subelement, model set up subelement and fusant
Unit；
(2) by the EEG signals of described physiology information detecting module timing acquiring tested personnel, described physiology information detecting mould
Dry state electrode sensor in block gathers the bioelectrical signals that brain produces, and the bioelectrical signals collected is sent into intelligence core
In sheet, the disturbance that the noise mixed in bioelectrical signals and motion produce is filtered by this intelligent chip, and after filtering
The useful organisms signal of telecommunication be amplified process；
(3) the useful organisms signal of telecommunication after processing and amplifying is reached monitoring modular by described wireless communication module；
(4) the useful organisms signal of telecommunication is analyzed processing by the main control unit in described monitoring modular, wherein, and described pretreatment
Unit carries out pretreatment to the useful organisms signal of telecommunication, pretreatment includes resolving useful organisms electrical signal data, denoising with
Filtering Processing；Described model sets up subelement to pretreated useful organisms signal of telecommunication founding mathematical models, sets up the most respectively
Frequency band power value, frequency band power percentage ratio, frequency band power ratio and frequency band power add and ratio, and calculate brain electric fatigue value,
And differentiate whether tested personnel is in fatigue state；
(5) when described main control unit differentiates that tested personnel is in fatigue state, described prewarning unit sends warning and reminds；No
Then, step (2)-(5) are repeated.
Fatigue state the most according to claim 1 monitoring and method for early warning, it is characterised in that the master in described step (1)
Control unit also includes feature extraction subelement and fusant unit；Described step (2) also includes: described physiology information detecting module
The eye image signal of timing acquiring tested personnel；Described step (3) also includes: eye image is believed by described wireless communication module
Number reach monitoring modular；Described step (4) also includes: described pretreatment subelement carries out pretreatment to eye image signal, in advance
Process includes eye image signal data is carried out denoising, gray processing and equalization processing；Described feature extraction subelement is in advance
Eye image signal extraction human eye feature after process, and it is calculated human eye fatigue data；Described fusant unit is by tired for brain electricity
Labor value and human eye fatigue data merge, and analysis meter calculates level of fatigue, it determines whether tested personnel is in fatigue state.
Fatigue state the most according to claim 2 monitoring and method for early warning, it is characterised in that described step (4) also includes:
Described feature extraction subelement realizes human eye location with Adaboost algorithm for core, intercepts out human eye parts of images, increases
Otsu adaptive two-tone images algorithm, uses the algorithm averaged to extract human eye feature, designs dynamic sliding window people's eye
State evaluation algorithm, thus judge the state of eyes closed threshold value, finally by improving Perclos human eye fatigue detecting algorithm,
To human eye fatigue data.
Fatigue state the most according to claim 2 monitoring and method for early warning, it is characterised in that to people in described step (4)
Eye image signal data carries out gray processing process, and the formula of employing is: Y=0.30R+0.59G+0.11B, in coloured image
RGB three-component is weighted averagely, obtaining rational gray level image.
Fatigue state the most according to claim 3 monitoring and method for early warning, it is characterised in that in described step (4)
Otsu adaptive two-tone images algorithm specifically includes: for image I, (x, y), the segmentation threshold of foreground and background is T, belongs to prospect
It is ω 0 that pixel number accounts for the ratio of whole eye image, and its average gray is μ 0；Background pixel is counted and is accounted for whole eye image
Ratio is ω 1, and its average gray is μ 1, and the grand mean gray scale of eye image is μ, and inter-class variance is g, and eye image size is
M × N, in eye image, the gray value of pixel is denoted as N0 less than the number of pixels of threshold value T, and pixel grey scale is more than the pixel of threshold value T
Number is denoted as N1, then draw below equation:
N0+N1=M × N, ω 0+ ω 1=1,
μ=ω 0 × μ 0+ ω 1 × μ 1, g=ω 0 × (μ 0-μ)2+ω1×(μ1-μ)2,
Draw: g=ω 0 × ω 1 × (μ 0-μ 1)2；The method using traversal obtains making max-thresholds T corresponding for inter-class variance g,
Afterwards according to threshold value T, gray level image is carried out binaryzation, i.e. when image intensity value is more than threshold value T, is entered as 255, otherwise assignment
It is 0.
Fatigue state the most according to claim 2 monitoring and method for early warning, it is characterised in that in described step (4), institute
Stating feature extraction subelement to the step of eye image signal extraction human eye feature is: calculate black in eye image vertical direction
The meansigma methods of black pixel point number shared by line segment, number=(total black of black pixel point shared by averagely every black line segment
The bar number of the number of pixel/vertical black line segment), according to tested personnel under eyes-open state shared by averagely every black line segment
The number of black pixel point is more than the number of black pixel point shared by averagely every black line segment under closed-eye state, and combines flat
The number of black pixel point shared by all every black line segment, it is judged that tested personnel is eyes-open state or closed-eye state；Then, at list
Carry out multiple repairing weld in bit time, in units of the frame number collected, calculate eyes closed state frame number and account for the percentage of totalframes
Ratio, formula is:Wherein, in P represents a period of time, the frame number of eyes closed state accounts for
The ratio of the totalframes gathered.
Fatigue state the most according to claim 2 monitoring and method for early warning, it is characterised in that the pre-place of described step (4)
Reason step includes: use constant compensation algorithm to repair previous value, and the value of next point is determined by a upper point, under
The value of one point takes=currently value-(value of a upper point) 1/60+ of this point add the value of a point.
Fatigue state the most according to claim 2 monitoring and method for early warning, it is characterised in that the master in described step (1)
Control unit also includes syndrome unit；Described step (4) also includes: described syndrome unit according to the focus of tested personnel,
Meditation degree carries out T inspection process with the mathematical model of foundation, and this T inspection formula is:
Wherein, d ' is sample average, SdBeing sample variance, n is number of samples, and α is confidence interval.
9. implementing fatigue state monitoring and the early warning system of method described in the claims 1-8 at least, its feature exists
In, including physiology information detecting module, wireless communication module and monitoring modular, wherein, described physiology information detecting module is built-in
Dry state electrode sensor and intelligent chip, for the physiologic information of timing acquiring tested personnel；Described wireless communication module is used for
Physiologic information is sent to monitoring modular；Described monitoring modular includes interconnective main control unit and prewarning unit, described master
Control unit is for processing to physiologic information and carry out degree of fatigue differentiation, and described prewarning unit is used for sending warning and reminds；
Described main control unit includes that pretreatment subelement, model set up subelement and fusant unit.
Fatigue state the most according to claim 9 monitoring and early warning system, it is characterised in that described main control unit also wraps
Include feature extraction subelement, fusant unit and syndrome unit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610523738.5A CN106128032A (en) | 2016-07-05 | 2016-07-05 | A kind of fatigue state monitoring and method for early warning and system thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610523738.5A CN106128032A (en) | 2016-07-05 | 2016-07-05 | A kind of fatigue state monitoring and method for early warning and system thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106128032A true CN106128032A (en) | 2016-11-16 |
Family
ID=57469423
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610523738.5A Pending CN106128032A (en) | 2016-07-05 | 2016-07-05 | A kind of fatigue state monitoring and method for early warning and system thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106128032A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106691443A (en) * | 2017-01-11 | 2017-05-24 | 中国科学技术大学 | Electroencephalogram-based wearable anti-fatigue intelligent monitoring and pre-warning system for driver |
| CN107358785A (en) * | 2017-08-25 | 2017-11-17 | 成都大学 | A kind of bus driver fatigue detection device |
| CN107550495A (en) * | 2017-07-20 | 2018-01-09 | 防城港市港口区天平电子科技有限公司 | The process adjustments system of control person's fatigue in a kind of detection |
| CN107623842A (en) * | 2017-09-25 | 2018-01-23 | 西安科技大学 | A kind of video monitoring image processing system and method |
| CN108305681A (en) * | 2018-02-01 | 2018-07-20 | 四川东鼎里智信息技术有限责任公司 | It is a kind of based on wearable female mammary gland monitoring and warning system |
| CN109464152A (en) * | 2019-01-09 | 2019-03-15 | 浙江强脑科技有限公司 | Detection method, equipment and the computer readable storage medium of brain fatigue state |
| CN109934189A (en) * | 2019-03-19 | 2019-06-25 | 何世全 | A kind of driving warning method, system and terminal |
| CN110232327A (en) * | 2019-05-21 | 2019-09-13 | 浙江师范大学 | A kind of driving fatigue detection method based on trapezoidal concatenated convolutional neural network |
| CN110544041A (en) * | 2019-09-06 | 2019-12-06 | 马鞍山市清和财务咨询有限公司 | employee data management system for financial consulting company |
| CN111209815A (en) * | 2019-12-28 | 2020-05-29 | 杭州电子科技大学 | Non-contact fatigue driving detection method of BP neural network based on momentum optimization |
| CN112286775A (en) * | 2020-10-30 | 2021-01-29 | 深圳前海微众银行股份有限公司 | Method, equipment and storage medium for detecting fatigue state |
| CN114327046A (en) * | 2021-11-30 | 2022-04-12 | 北京津发科技股份有限公司 | Multi-mode man-machine interaction and state intelligent early warning method, device and system |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4502122A (en) * | 1981-06-03 | 1985-02-26 | Nissan Motor Company, Limited | Method and apparatus for measuring vehicle driver's fatigue to give an alarm |
| CN102271584A (en) * | 2009-10-29 | 2011-12-07 | 松下电器产业株式会社 | Organism fatigue evaluation device and organism fatigue evaluation method |
| CN103505203A (en) * | 2013-09-30 | 2014-01-15 | 西安交通大学 | Method for detecting human metal states based on brain electrical source locating method |
| CN103606245A (en) * | 2013-11-08 | 2014-02-26 | 北京工业大学 | Fatigue driving detection early warning system based on Bluetooth brain wave earphone and Android mobile phone |
| CN104463102A (en) * | 2014-11-07 | 2015-03-25 | 中国石油大学(华东) | Human eye positioning method based on point-by-point scanning |
| CN104504856A (en) * | 2014-12-30 | 2015-04-08 | 天津大学 | Fatigue driving detection method based on Kinect and face recognition |
| CN104778473A (en) * | 2014-01-09 | 2015-07-15 | 深圳市中瀛鑫科技股份有限公司 | Image binarization method, device and video analysis system |
| CN104881955A (en) * | 2015-06-16 | 2015-09-02 | 华中科技大学 | Method and system for detecting fatigue driving of driver |
| CN105011932A (en) * | 2015-08-11 | 2015-11-04 | 西安科技大学 | Fatigue driving electroencephalogram monitoring method based on degree of meditation and degree of concentration |
-
2016
- 2016-07-05 CN CN201610523738.5A patent/CN106128032A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4502122A (en) * | 1981-06-03 | 1985-02-26 | Nissan Motor Company, Limited | Method and apparatus for measuring vehicle driver's fatigue to give an alarm |
| CN102271584A (en) * | 2009-10-29 | 2011-12-07 | 松下电器产业株式会社 | Organism fatigue evaluation device and organism fatigue evaluation method |
| CN103505203A (en) * | 2013-09-30 | 2014-01-15 | 西安交通大学 | Method for detecting human metal states based on brain electrical source locating method |
| CN103606245A (en) * | 2013-11-08 | 2014-02-26 | 北京工业大学 | Fatigue driving detection early warning system based on Bluetooth brain wave earphone and Android mobile phone |
| CN104778473A (en) * | 2014-01-09 | 2015-07-15 | 深圳市中瀛鑫科技股份有限公司 | Image binarization method, device and video analysis system |
| CN104463102A (en) * | 2014-11-07 | 2015-03-25 | 中国石油大学(华东) | Human eye positioning method based on point-by-point scanning |
| CN104504856A (en) * | 2014-12-30 | 2015-04-08 | 天津大学 | Fatigue driving detection method based on Kinect and face recognition |
| CN104881955A (en) * | 2015-06-16 | 2015-09-02 | 华中科技大学 | Method and system for detecting fatigue driving of driver |
| CN105011932A (en) * | 2015-08-11 | 2015-11-04 | 西安科技大学 | Fatigue driving electroencephalogram monitoring method based on degree of meditation and degree of concentration |
Non-Patent Citations (1)
| Title |
|---|
| 杨才坤: "疲劳驾驶检测方法的研究及其嵌入式实现", 《中国优秀硕士学位论文全文数据库》 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106691443A (en) * | 2017-01-11 | 2017-05-24 | 中国科学技术大学 | Electroencephalogram-based wearable anti-fatigue intelligent monitoring and pre-warning system for driver |
| CN107550495A (en) * | 2017-07-20 | 2018-01-09 | 防城港市港口区天平电子科技有限公司 | The process adjustments system of control person's fatigue in a kind of detection |
| CN107358785A (en) * | 2017-08-25 | 2017-11-17 | 成都大学 | A kind of bus driver fatigue detection device |
| CN107623842A (en) * | 2017-09-25 | 2018-01-23 | 西安科技大学 | A kind of video monitoring image processing system and method |
| CN108305681A (en) * | 2018-02-01 | 2018-07-20 | 四川东鼎里智信息技术有限责任公司 | It is a kind of based on wearable female mammary gland monitoring and warning system |
| CN109464152B (en) * | 2019-01-09 | 2021-11-26 | 浙江强脑科技有限公司 | Method and device for detecting brain fatigue state and computer readable storage medium |
| CN109464152A (en) * | 2019-01-09 | 2019-03-15 | 浙江强脑科技有限公司 | Detection method, equipment and the computer readable storage medium of brain fatigue state |
| CN109934189A (en) * | 2019-03-19 | 2019-06-25 | 何世全 | A kind of driving warning method, system and terminal |
| CN109934189B (en) * | 2019-03-19 | 2021-08-13 | 山东同云网络科技有限公司 | Driving early warning method, system and terminal |
| CN110232327A (en) * | 2019-05-21 | 2019-09-13 | 浙江师范大学 | A kind of driving fatigue detection method based on trapezoidal concatenated convolutional neural network |
| CN110544041A (en) * | 2019-09-06 | 2019-12-06 | 马鞍山市清和财务咨询有限公司 | employee data management system for financial consulting company |
| CN110544041B (en) * | 2019-09-06 | 2022-05-17 | 马鞍山市清和财务咨询有限公司 | Employee data management system for financial consulting company |
| CN111209815A (en) * | 2019-12-28 | 2020-05-29 | 杭州电子科技大学 | Non-contact fatigue driving detection method of BP neural network based on momentum optimization |
| CN111209815B (en) * | 2019-12-28 | 2023-08-22 | 杭州电子科技大学 | Non-contact fatigue driving detection method based on BP neural network with momentum optimization |
| CN112286775A (en) * | 2020-10-30 | 2021-01-29 | 深圳前海微众银行股份有限公司 | Method, equipment and storage medium for detecting fatigue state |
| CN114327046A (en) * | 2021-11-30 | 2022-04-12 | 北京津发科技股份有限公司 | Multi-mode man-machine interaction and state intelligent early warning method, device and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106128032A (en) | A kind of fatigue state monitoring and method for early warning and system thereof | |
| Sikander et al. | Driver fatigue detection systems: A review | |
| CN105769120B (en) | Method for detecting fatigue driving and device | |
| CN110276273B (en) | Driver fatigue detection method integrating facial features and image pulse heart rate estimation | |
| CN109389806B (en) | Fatigue driving detection early warning method, system and medium based on multi-information fusion | |
| CN101583313B (en) | Awake state judging model making device, awake state judging device, and warning device | |
| Ahmad et al. | Drowsy driver identification using eye blink detection | |
| CN101593425B (en) | Machine vision based fatigue driving monitoring method and system | |
| CN104720796B (en) | A kind of automatic checkout system and method for epileptic attack time section | |
| CN103340637A (en) | System and method for driver alertness intelligent monitoring based on fusion of eye movement and brain waves | |
| CN109726771A (en) | Abnormal driving detection model method for building up, device and storage medium | |
| CN113743471B (en) | Driving evaluation method and system | |
| CN107007257A (en) | The automatic measure grading method and apparatus of the unnatural degree of face | |
| CN111753674A (en) | Fatigue driving detection and identification method based on deep learning | |
| CN104068868A (en) | Method and device for monitoring driver fatigue on basis of machine vision | |
| CN110334592A (en) | A kind of monitoring of driver's abnormal behaviour and safety control system and safety control method | |
| CN109875583A (en) | A kind of fatigue driving detecting system and method based on AR technology | |
| Tang et al. | Real-time image-based driver fatigue detection and monitoring system for monitoring driver vigilance | |
| CN109953763A (en) | A kind of vehicle carried driving behavioral value early warning system and method based on deep learning | |
| CN110020632A (en) | A method of the recognition of face based on deep learning is for detecting fatigue driving | |
| CN106691440A (en) | Controller fatigue detection method and system based on BP neural network | |
| Huang et al. | P-FDCN based eye state analysis for fatigue detection | |
| CN107563346A (en) | One kind realizes that driver fatigue sentences method for distinguishing based on eye image processing | |
| CN106667506A (en) | Method and device for detecting lies on basis of electrodermal response and pupil change | |
| CN110309813A (en) | A kind of model training method, detection method, device, mobile end equipment and the server of the human eye state detection based on deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161116 |