CN106580301B

CN106580301B - A kind of monitoring method of physiological parameter, device and handheld device

Info

Publication number: CN106580301B
Application number: CN201611189222.8A
Authority: CN
Inventors: 刘锦; 邹煜晖
Original assignee: Guangzhou Heart And Tide Mdt Infotech Ltd
Current assignee: Guangzhou Heart And Tide Mdt Infotech Ltd
Priority date: 2016-12-21
Filing date: 2016-12-21
Publication date: 2019-09-24
Anticipated expiration: 2036-12-21
Also published as: CN106580301A

Abstract

The invention discloses a kind of monitoring method of physiological parameter, device and handheld device, method is comprising steps of acquisition current face's image；Extract the cheek parts of images in facial image；The mean value for calculating cheek parts of images the channel Green, obtains equal value sequence；Sliding average value filtering is carried out to equal value sequence, section is filtered by adjusting it and obtains physiological change curve；Feature extraction is carried out to physiological change curve and obtains physiological parameter.By the different available different physiological parameter curves in filtering section, by carrying out respective handling, available multiple physiologic parameter values to curve.The invention also discloses the handheld devices of application this method.The present invention has the advantages that calculation amount is low, accuracy is high, low to hardware requirement.

Description

A kind of monitoring method of physiological parameter, device and handheld device

Technical field

The present invention relates to physiological compensation effects research field, in particular to a kind of monitoring method of physiological parameter, device and Handheld device.

Background technique

Human body physiological parameter monitoring includes in electrocardio, blood pressure, heart rate, blood oxygen saturation, respiratory rate, body temperature, breathing The parameters such as gas concentration lwevel monitoring, be now widely used for the fields such as clinical medicine, wearable device, mobile terminal.

For example, mature heart rate non-invasive measurement means have sphygmodynamometry, electrocardiosignal method, photoplethysmographic at present Graphical method (PhotoPlethysmoGraphy, PPG).Wherein:

One, sphygmodynamometry: this is most ancient method in fact, is exactly feeling the pulse for Chinese medicine.In wrist or neck two sides, The pressure regularly fluctuation of artery can be touched through skin.This signal can be become into heart rate by pressure sensor. This scheme is also that current commercialization is most jejune, and reason is first is that pressure sensor needs for a long time to press the artery of wearer half Compel, there is sense of discomfort.Second is that pressure sensor is difficult to be fixed on skin surface in an appropriate manner: fixed tightly will lead to very much blood flow It is unsmooth, it is fixed too loose and cannot achieve measurement.This problem shows especially obvious in sport wrist-watch design.So the party Method is general only within the hospital to use tranquillization patient postoperative in operation.Under normal conditions, pulse and heart rate are consistent, and are had Heart contraction will generate a pulse, but in the case where blood pressure is extremely low, and the hematopenia of heart contraction push is with energy It is enough that beating is perceived on blood vessel, then there is the case where pulse is lower than heart rate.

Two, electrocardiosignal method: antrum controls heart contraction diastole to pump blood to trunk with having the rhythm and pace of moving things.This control letter Number it is an electric signal (human nerve signal all shows as electric signal on nerve), body surface can be gradually diffused into, it can be in skin Skin passes through electrode measurement.The electrocardiogram equipment that Hospitals at Present uses is exactly to use this principle.This rhythm is exactly heart rate, except this it Outside, electrocardiosignal can also provide many reference informations for diagnosis.Current wearable heart rate measurement most accurate on the market Instrument, heart rate band, and use this method.But since the wavelength of electrocardiosignal is very long, in order to measure enough accuracy Signal, signal electrode and reference electrode just must be on torso spaces every enough to remote.Distant two o'clock usually on chest, or Person's left hand and the right hand or hand and foot etc..Watch is just more difficult to use this scheme, unless someone is ready while two tables of band.

Three, photoplethysmographic graphical method: this is a kind of simple lossless measurement heartbeat component and evaluation body surface circulation Method.According to the description of the interaction of propagation, distribution and light and tissue in organism optical to light in biological tissues and Research, when the light of some wavelength is by (injection) bio-tissue, bio-tissue (skin, fat, blood, muscle etc.) Scattering will be generated to light and will be absorbed, and the light intensity decays detected are made.Wherein its hetero-organization such as skin, muscle and bone is to light Absorb almost it is constant, but when transmission region there are arteries beat or vein blood vessel it is full when, with blood flow Increase and decrease, main extinction ingredient content of hemoglobin also accordingly increases and decreases in blood, and blood will change the uptake of light therewith.Blood When pipe is full, absorbing amount is maximum, and the luminous intensity detected is minimum, is believed using the intensity variation that photoelectric sensor will test It number is converted into electric signal detection, the variation that can trace out intravascular volume obtains pulse wave signal.

Currently, the implementation of PPG can be divided into contact and contactless.

Contact PPG realization needs photoelectric sensor to carry out data acquisition, for example that common is clipped in forefinger tip/ear Vertical Cardiotachometer, various wearable devices, sport wrist-watch etc..The some mobile phones of the S5/S6 of Samsung, ZukZ2Pro of association etc. Also infrared heart rate sensor is integrated at mobile phone back, it is only necessary to finger by can collect heart rate on a sensor.

Due to needing transmitting light source and receiving end, institute can also realize on mobile phone in this way.Also have at present many App utilizes this principle, it is only necessary to finger be pressed on camera, flash lamp is opened, pass through the image light and shade of acquisition camera Variation is to obtain the numerical value (a kind of form that can regard photoelectric signal transformation as) of heart rate.But current scheme is generally existing Following defects: 1, can wear camera lens, influence cam lens imaging effect；2, flash lamp is opened, finger may be scalded；3, day When gas is colder or in the case where body abnormality, blood possibly can not flow through finger, cause to can't detect heart rate；4, finger Permeability must be got well, and if there is spot or had covering, be will affect measurement result；5, there is no flash lamp or flash lamp distance The distant equipment of camera can not carry out heart rate acquisition (wherein 3 and 4 be also infrared sensor one of defect).

On wearable device (referring mainly to bracelet, wrist-watch here), typically at least there are a green light Led light source (general feelings Under condition, the Absorption by measuring green light can obtain more accurate data, drip with sweat in hot environment, such as gymnasium When, skin surface moisture increases, since more green lights have been predominantly absorbed, so needing to be switched to other light sources, such as white Light or infrared light supply), some bracelets use two kinds of light sources of green light and white light, and AppleWatch is then using green light and infrared Light.In addition to light source, it is also necessary to there is a photoelectric sensor to incude reflected light.Although this method can only obtain heart rate signal, no Cross comparatively to movement bring noise immunity it is stronger, so being well suited for current sport wrist-watch.Therefore each manufacturer It (to be removed integrated optic-electronic sensor on respective SOC (System-on-a-Chip integrating chip, usually bluetooth SOC) AppleWatch is oneself research and development, most of third-party, such as the manufacturers such as Philips, Kionix offer for being all ), by the algorithm of heart rate respectively researched and developed, the influence of compensation campaign noise is gone using accelerograph, to calculate heart rate.Due to being Contact PPG is guaranteeing steady, and under the premise of fitting skin is close, the interference being subject to is less, thus numerical value should all it is poor not It is more.

But even under ideal conditions, AppleWatch and other bracelet/wrist-watches cannot guarantee that all users are every It is secondary to obtain accurate heart rate data.The reason of causing unstable factor is mainly also: 1, such as under cold environment, hand Skin blood flow on wrist may be too low, causes heart rate sensor that can not read data；2, there is interference in the region of sensor covering, Such as tattoo, birthmark etc.；3, it is influenced caused by moving, for example arm bends and stretches the variation that can all cause blood flow.At this moment, then It needs to consider to get accurate heart rate by heart rate detectors such as third-party bluetooth chest straps.

Contactless PPG measurement method in practical applications, can bring great convenience to the life of user, Advantage is that the blood flow information of larger skin area can be monitored, so that calculated result is more accurate.In addition, relative to contact For PPG technology, contactless PPG technology, which has, is more widely applied field, for example, being used for some skin injury/injuries Crowd, it has not been convenient to the crowd etc. for physically thering is equipment to contact.Current implementation method is probably as follows: 1, utilizing photoelectric transfer Sensor (usually camera) to human body somewhere blood vessel concentrated area (any region all can, wrist or face covered without clothes Cover more convenient) carry out video data acquiring；Secondary light source or special light source emitter are usually required in collection process；2, ROI (interested, i.e., to the apparent region of light absorption reacting condition) region is isolated from each frame image of video data, The separation of RGB primary channel is carried out to the area image, and mean value is taken to each channel all pixels, as the frame image at this The characteristic value of Color Channel generates three new signal Xr, Xg, Xb, is standardized, obtains standardized to these three signals Signal Yr, Yg, Yb；3, blind source separating (ICAIndependent Component is carried out to the signal after standardization Correlation Algorithm, independent component analysis).Due to separation after independent signal source be it is unordered, can not directly select At this moment the number of winning the confidence just needs to carry out signal screening, by carrying out correlation analysis with green channel Xb, selection obtains final signal Zi；4, the signal processings such as band filtering are carried out to selected signal, cycle analysis (frequency spectrum point finally is done to filtered signal Analysis), that is, DFT transform is carried out, spectrogram is obtained, peak-peak frequency is chosen as palmic rate, heart rate can be obtained.

But there are the following problems for above scheme: 1, needing first to shoot one section of video, then could carry out to video Analysis, the video length generally shot needs 30 seconds or so, while shooting process has higher requirements to ambient light；2, needs pair The each frame of video carries out ROI region separation, while tested personnel should avoid being displaced as far as possible, in order to avoid influence calculated result；3, it needs Red/Green/Blue3 channel data mean value is calculated simultaneously；4, it needs to carry out blind source separating mixed signal using ICA, need Want more operation time；5, it needs to carry out spectrum analysis (DFT calculating) to final signal, it is same time-consuming more, and obtained after converting The spectrogram arrived is more demanding to data source, if there have more interference that can directly result in evaluation to be incorrect.

The maximum defect of above scheme is exactly hardware requirement height, and operand is big, and then there has been proposed many improved methods Such as: 1, ICA with fastICA jade algorithm is replaced, arithmetic speed is accelerated；2, DFT is replaced with FFT, reduces operation Amount accelerates arithmetic speed；3, the detection number for reducing ROI region, avoids excessive operation from consuming；4, increase high and low frequency filtering Device reduces noise, so that the calculated result is tended to be accurate.But effect is still not satisfactory.

Therefore, seek a kind of physiological parameter high, low to System Hardware Requirement, that result can be obtained in real time of accuracy of measurement The monitoring method and device of (especially heart rate, respiratory rate) are worth with important practical.

Summary of the invention

The shortcomings that it is a primary object of the present invention to overcome the prior art and deficiency, provide a kind of monitoring side of physiological parameter Method, this method have the advantages that calculation amount is low, accuracy is high, low to hardware requirement.

Another object of the present invention is to provide a kind of monitoring device of physiological parameter, which has calculation amount low, quasi- Exactness height, the advantage low to hardware requirement.

Another object of the present invention is to provide a kind of handheld devices of monitoring device including above-mentioned physiological parameter, here Handheld device can be one it is independent, be only used for carry out rhythm of the heart equipment, be also possible to one and not only carry out the heart Rate monitoring, also carries out the physiological parameter monitor of other parameters monitoring, it is of course also possible to be the prison for being integrated with above-mentioned physiological parameter Survey the mobile terminal, such as mobile phone, removable computer etc. of device.

The purpose of the present invention is realized by the following technical solution: a kind of monitoring method of physiological parameter, comprising steps of

Acquire current face's image；

Extract the cheek parts of images in facial image；

The mean value for calculating cheek parts of images the channel Green, obtains equal value sequence；

Sliding average value filtering is carried out to equal value sequence, section is filtered by adjusting it and obtains physiological change curve；

Feature extraction is carried out to physiological change curve and obtains physiological parameter.

Preferably, for the facial image of acquisition, diminution processing is first carried out, OpenCV then is passed through to the image after diminution The Face datection algorithm of offer detects face, after making face position, then carries out the amplification of equal proportion, obtains face area Domain.So as to the speed of improvement method.

Preferably, the location determining method of cheek parts of images is: human face region width is set as w, is highly h, start bit It is set to (x, y), the width range of two cheek regions is respectively 1/8w~3/8w and 5/8w~7/8w, altitude range 1/ 2h~3/4h.The region for probably omitting nose is limited by above-mentioned zone, only retains two cheek regions, and it is dry to reduce data It disturbs.

Preferably, the mean value in Green channel is calculated cheek parts of images, that is, the Green in each pixel that adds up is logical Road component, then obtained divided by the pixel quantity of image, the mode for taking interval to sample in calculating process, so as to improve Processing speed, and guarantee to get whole situation of change.

Preferably, the frame per second of current acquisition data is set as m, then is 3m/2 by the way that the filtering section of equal value sequence is arranged ~2m, available change of respiratory rate curve；The filtering section of equal value sequence is set as m/3~2m/3, available heart rate Change curve.Here it filters the setting range in section and similar corresponding change curve can be obtained in the minimum of computation period.

Further, it if physiological change curve is changes in heart rate curve, is extracted and is calculated using following fisrt feature Method carries out feature extraction:

Preferentially find the position of first wave crest；

Since the position of previous wave crest, the highest primary peak of value within the scope of R thereafter is searched for；

Then since the position of primary peak, the highest secondary peak of value within the scope of R thereafter is searched for；

Judge whether primary peak peak value is more than or equal to secondary peak peak value, if so, determining that primary peak is to work as prewave Then peak is continued searching since primary peak；If it is not, then determining that secondary peak is current wave crest, then from the second wave Peak starts to continue searching；

After complete changes in heart rate curve of search, a wave crest sequence is obtained, the wave crest sequence includes each wave crest Peak value and position, position i.e. eartbeat interval；

By the time interval of 2 wave crests of head and the tail divided by the number of wave crest, heart rate value is obtained.

Further, heart rate 30~220 is divided into n section, each R value is respectively adopted in n R value of corresponding setting Seek wave crest sequence；For each wave crest sequence, f value is calculated using following formula:

f_i=sd_i*sd_i/avg_i；

Wherein, i=1 ... ..., n；sd_iFor the standard deviation of eartbeat interval, avg_iFor the equispaced for calculating eartbeat interval, take Min f_iCorresponding i calculates heart rate value as the sequence currently selected, using the sequence data.

Further, it if physiological change curve is change of respiratory rate curve, is mentioned using following second feature Algorithm is taken to carry out feature extraction:

Step 1, according to the data sequence in change of respiratory rate curve, find first wave crest point O point；

Step 2, from O point, find the upward inflection point of next waveform, obtain first trough, label 1；

Step 3, from 1 point of label, it is continuous to find the downward inflection point of waveform, and be labeled as wave crest, then obtain wave Peak sequence O, A, B, C, D, E ... and trough sequence 1,2,3,4,5 ..., obtain the wave character of signal；

Step 4, according to wave character, it is peak-to-peak every the time difference and peak separation quantity to obtain head and the tail, pass through formula:

Respiratory rate br=head and the tail are peak-to-peak every the time difference/peak separation quantity；

Obtain current respiratory rate.Respiratory rate has hysteresis quality, when lag time is the half of preceding respiration duration Between.Under normal circumstances, primary complete waveform can judge the general frequency of breathing.

Further, since the oxygen whole content in blood changes a necessarily relatively slow process, so The wave crest point searched out is judged by choosing the r of a real-time change.Using as follows during finding wave crest Denoising method:

If peak to peak separation twice recently is respectively d and d1, calculate r=(d+d1)/2；

Judge whether the time where the wave crest currently found is greater than last time peak time+r, if it is, being judged as new Wave crest, if it is not, then being judged as noise spot.

Certainly, the computing interval also will appear some cases, for example the corresponding serial number of the minimum value got every time constantly becomes, when Preceding numerical value presence is abnormal, and the numerical value of heart rate is unstable etc., first to physiological change curve data in order to improve the accuracy of data Carry out following screening operation, the method is as follows:

Since experiment before has been proven that the length for calculating the time will not have an impact to the specific value of heart rate, here Take L seconds time spans as rate calculation section, and increase by one it is L1 seconds last in HR values refer to as a comparison, The numerical value of middle heart rate is denoted as hbr, and the HR values of last L1 seconds calculating are denoted as hbr L1, and hbr L1 also can be regarded as heart rate change The trend of change, if there is unexpected fluctuation or extraneous influence, hbr L1 and hbr have apparent numerical value difference.

Step 0, the last heart rate waveform serial number lastchoice selected of initialization, get the cumulative number of heart rate Count, as the cumulative number e of heart rate, (e is resisted extraneous dry when doing long-time rhythm of the heart to Current heart rate waveform serial number The parameter disturbed causes to choose other heart rate sequences for avoiding due to interference), recalculate number retrytime, assignment It is 0；Here count is that get the number of heart rate be all to use Minfi before not getting p1 heart rate for cumulative Corresponding sequence as heart rate sequence undetermined, behind just use averagely fluctuate it is the smallest as heart rate sequence undetermined；

Step 1, the n different filtering interval R according to setting, calculating heart rate hbr, hbr L1 and fluctuation situation f (if The external condition fluctuated in the excessive the currently monitored situation of explanation is undesirable, needs to abandon current sampling), and record storage, together When cumulative count；

If Step 2, count >=p1, calculateAs average fluctuation；

Step 3, g=Min g is taken_i(i=1~n), and corresponding i is recorded as the alternative of current most suitable heart rate

If lastchoice=0 (is not recorded) for the first time, lastchoice=i, e=1 are enabled；

If lastchoice=i, e=e+1, e is enabled to be not more than p2；

If lastchoice is not equal to i, e=e-1, e is enabled to be not less than 0；

Step 4, the difference d for calculating hbr and hbr L1, when heart rate stabilization, the numerical value of d does not exceed p4；

If Step 5, Current heart rate f > p3 (empirical value that many experiments obtain), illustrate currently by external interference It is bigger, or acquisition heart rate have abnormal conditions；

At this moment judge:

If f > g and f > p3, illustrate to influence increasing, needs to abandon some abnormal datas and re-start calculating, hold Row Step 6；If f≤g or f≤p3, Step 7 is executed；

Step 6, the data for abandoning current sequence of calculation the first half, enable lastchoice, count, e 0, retrytime =retrytime+1；If retrytime=2 abandons total data, lastchoice, count, e 0 is enabled；

If retrytime=3, prompt is provided, preset test environment is undesirable, and monitoring failure executes Step 8；

If difference d < p4 of Step 7, continuous L2 calculating, illustrate that rate calculation numerical value tends towards stability, at this moment judgement is exhaled The numerical value of suction whether calculated (can tend towards stability within the numerical value of heart rate most fast 7 seconds, the frequency of breathing then due to the period compared with It is long, typically at least need the time of a respiratory cycle that can just find out that waveform changes), and tend towards stability and (due to heart rate and exhale Suction is to calculate simultaneously, the consideration so the two is put together), if the numerical value of breathing is also stable, export heart rate and respiratory rate Numerical value executes Step 8；

Step 8, stop data acquisition, terminate monitoring process.

It is also possible to by this method, Lai Jinhang vivo identification (such as picture or image etc.), because of extraneous light Line reflection is substantially irregular, then can be with so if the fluctuation situation obtained is excessive (f > p3, under normal circumstances f≤p3/2) Judge that current light environment is undesirable, or tested cannot provide normal regular heart rate.

A kind of monitoring device of physiological parameter, comprising:

Image capture module, for acquiring current face's image；

Image interception module, for extracting the cheek parts of images in facial image；

Equal value sequence computing module obtains equal value sequence for calculating cheek parts of images the mean value in the channel Green；

Filter module filters section by adjusting it and obtains physiology for carrying out sliding average value filtering to equal value sequence Change curve；

Analysis and processing module obtains physiological parameter for carrying out feature extraction to physiological change curve.

Preferably, described image acquisition module includes Zoom module and face detection module, and Zoom module will be for that will acquire Facial image carry out diminution processing, then send face detection module for the image after diminution, and for examining in face After survey module orients face position, then the amplification of equal proportion is carried out, obtains human face region；Face detection module is for fixed Position goes out face position.

Preferably, image interception module specifically: human face region width is set as w, is highly h, initial position is (x, y), The width range of two cheek regions is respectively 1/8w~3/8w and 5/8w~7/8w, altitude range 1/2h~3/4h.

As a preference, the filter module adjustment filtering section is m/3~2m/3, m is current acquisition data Frame per second obtains changes in heart rate curve；The analysis and processing module includes fisrt feature extraction module, fisrt feature extraction module packet It includes:

First data read module, for finding the position of first wave crest；

Primary peak search module, for since the position of previous wave crest, search is worth thereafter highest the in R One wave crest；

Secondary peak search module, for searching for and being worth highest second within the scope of R thereafter since the position of primary peak Wave crest；

First judgment module, for judging whether primary peak is more than or equal to secondary peak, if so, determining primary peak For current wave crest, then continued searching since primary peak；If it is not, then determining that secondary peak is current wave crest, then It is continued searching since secondary peak；

Primary peak sequence preserving module obtains a wave crest sequence after searching for complete changes in heart rate curve, described Wave crest sequence include each wave crest peak value and position, position i.e. eartbeat interval；

Heart rate value computing module, for the time interval of 2 wave crests of head and the tail divided by the number of wave crest, to be obtained heart rate value.

Further, wave crest sequence preserving module includes several sub- preserving modules of wave crest sequence, by heart rate 30~220 It is divided into n section, n R value of corresponding setting is respectively adopted each R value and seeks wave crest sequence, be stored in a wave crest sequence respectively It arranges in sub- preserving module；

For each wave crest sequence, f value is calculated using following formula:

f_i=sd_i*sd_i/avg_i；

As another preferred embodiment, the filter module adjustment filtering section is 3m/2~2m, m is current acquisition data Frame per second obtains change of respiratory rate curve；The analysis and processing module includes second feature extraction module, and second feature extracts mould Block includes:

First crest seeking module, for finding first wave according to the data sequence in change of respiratory rate curve Peak dot O point；

First trough finds module, for finding the upward inflection point of next waveform, obtaining first from O point Trough, label 1；

Wave character determining module, for constantly finding the downward inflection point of waveform, and be labeled as from 1 point of label Wave crest, then obtain wave crest sequence O, A, B, C, D, E ... and trough sequence 1,2,3,4,5 ..., obtain the waveform of signal Feature；

Respiratory rate computing module, it is peak-to-peak from beginning to end every the time difference and peak separation quantity for obtaining according to wave character, lead to Cross formula: respiratory rate br=head and the tail are peak-to-peak every the time difference/peak separation quantity；Obtain current respiratory rate.

Further, the second feature extraction module further includes denoising module, and denoising module includes:

Step-size in search setting module calculates step-size in search r=(d+ if peak to peak separation twice recently is respectively d and d1 d1)/2；

Whether judgment module, time where the wave crest for judging currently to find are greater than last time peak time+r, if It is then to be judged as new wave crest, if it is not, then being judged as noise spot.

Further, the monitoring device includes data screening module, which includes:

Initialization module gets the tired of heart rate for initializing the heart rate waveform serial number lastchoice of last selection Metering number count, cumulative number e of the Current heart rate waveform serial number as heart rate, recalculates number retrytime, is assigned a value of 0；According to n of setting different filtering interval R, heart rate hbr, hbr L1 and fluctuation situation f, and record storage are calculated, is tired out simultaneously Add count；

Computing module is averagely fluctuated, when being used for count >=p1, is calculatedAs average fluctuation；

Alternate data module takes g=Min g_i, and corresponding i is recorded as the alternative of current most suitable heart rate；If Lastchoice=0 then enables lastchoice=i, e=1；If lastchoice=i, e=e+1, e is enabled to be not more than p2； If lastchoice is not equal to i, e=e-1, e is enabled to be not less than 0；

Difference calculating module, for calculating the difference d of hbr and hbr L1；

Second judgment module, if set up, executes third judgement for judging whether f≤g or f≤p3 is true Otherwise module executes data removing module；

Data removing module, for abandoning the data of current sequence of calculation the first half, initialization lastchoice, count, E is 0, enables retrytime=retrytime+1；If retrytime=2, total data is abandoned, is initialized Lastchoice, count, e 0；If retrytime==3, prompt is provided, preset test environment is undesirable, and monitoring is lost It loses, executes ending module；

Third judgment module, for judging whether the difference d of continuous L2 calculating is less than p4, if it is illustrating rate calculation Numerical value tends towards stability, then continues to judge whether the numerical value of breathing has calculated and tended towards stability, if it is, the output heart Rate and respiratory rate numerical value execute ending module；

Ending module terminates monitoring process for stopping data acquisition.

Compared with the prior art, the invention has the following advantages and beneficial effects:

1, computation complexity of the present invention is low, can obtain the physiology such as heart rate, the respiratory rate of current picker ginseng in real time Number, accuracy are high.By being compared with existing Medical Devices acquired results, discovery algorithm stability is very strong, and is not susceptible to To the interference of external various conditions, there is extremely strong applicability.

2, the present invention is by extracting cheek parts of images, denoising etc., it is possible to reduce the environmental disturbances factor in image, into one Step improves the accuracy of detection.

Detailed description of the invention

Fig. 1 is the flow chart of 1 method of the present embodiment.

Fig. 2 is the flow chart of 1 fisrt feature extraction algorithm of the present embodiment.

Fig. 3 is the flow chart of 1 second feature extraction algorithm of the present embodiment.

Fig. 4 is an original series curve of actual acquisition in embodiment 1.

Fig. 5 is the heart rate curve for obtain after smothing filtering to Fig. 3.

Fig. 6 is the schematic diagram that heart rate is calculated by time domain method.

Fig. 7 is the heart rate curve of perfect condition.

Fig. 8 is the respiratory rate curve for obtain after smothing filtering to Fig. 3.

Fig. 9 is the schematic diagram that respiratory rate is calculated by time domain method.

Specific embodiment

Present invention will now be described in further detail with reference to the embodiments and the accompanying drawings, but embodiments of the present invention are unlimited In this.

Embodiment 1

As shown in Figure 1, a kind of monitoring method of physiological parameter of the present embodiment, comprising steps of

One, current face's image is acquired

Since face is most often to be exposed to outer skin area and have high identification, it is easy to it positions, so Non-contact capture system usually chooses human face region as data collection zone domain.Acquisition face is also due to confession of the heart to human body Blood is preferentially to guarantee brain and important internal organs, followed by four limbs, so acquisition face can obtain than acquisition four limbs More reliable and accurate numerical value is obtained, because four limbs, in body tip, especially in winter sluggish metabolism, low temperature makes blood vessel It shrinks, blood backflow reduced capability, so that trick, especially finger tip, the positions such as tiptoe poor blood circulation, likewise, some People can present pale after strenuous exercise, and the ice-cold situation of four limbs, this, which is also due to heart blood supply scarce capacity, leads to blood Liquid circulation is unsmooth.And PPG is changed for detecting oxygen content of blood, is that can not detect accurately if poor blood circulation Numerical value.

The present embodiment only passes through natural light or the acquisition demand of PPG only can be realized by stable environment light, due to It is to calculate in real time, it is only necessary to save the channel the G mean value of ROI region as PPG signal, while not need to calculate the channel R and channel B Numerical value separate noise.

Two, the cheek parts of images in facial image is extracted

Due to needing what is obtained to be the part for the skin area that blood flows through, caused so needing to reduce to the greatest extent other regions Interference, such as eyes, nose, mouth, since brow portion may be covered by fringe, so the present embodiment proposition only takes face It compares available arrive of rounding face and more accurately changes numerical value in the region of cheek point.

It is not the emphasis that the present invention discusses about face recognition module, therefore can uses a variety of different existing Face recognition module is realized.In mobile platform, IOS and Android have respective face recognition module, naturally it is also possible to adopt With any third party library, such as OpenCV, Face++ etc..The examples herein is realized at the end PC, using open source The library OpenCV, version is 2.48.The library OpenCV is the method for detecting human face by comprehensive AdaBoost and Cascade algorithm (be otherwise known as Harr-Cascade face detection device) Lai Jinhang Face datection, calls trained cascade by OpenCV (classifier) carries out pattern match to realize recognition of face.OpenCV supports HAAR, LBP (Local Binary at present Patterns) and 3 kinds of training aids of HOG, compared with Haar feature, LBP feature is integer feature, therefore trained and detection process It will be several times faster than Haar feature.In actual test, it is time-consuming different to call the detection face of different feature databases, but detect knot Fruit does not have difference, difference such as following table.That choose herein is the haarcascade_frontalface_default of default.

The present embodiment experimental situation machine configuration: CPU:i3-4010u 1.7G, memory: 8GB, system: win7 Ultimate 64 Environment: Java8.0, OpenCV:2.48, developing instrument: Eclipse, resolution ratio of camera head: 640 × 480 is developed in position.

The present embodiment can first carry out image to be detected in carrying out face recognition process in order to improve detection efficiency Diminution processing carries out Face datection to the image after diminution, it is assumed here that the image size of initial acquisition is 640*480, is carried out Face datection time-consuming needs 0.12 second, and the image size after reducing 1/4 is 320*240, at this moment carries out Face datection and only needs It 0.03 second, is calculated by 30 frame of video is per second, it is in place also can to guarantee that each frame image all gets accurate face institute substantially It sets, (0.03 second × 30 frame=0.9 second, operand needed for the scaling of image are 320 × 240 × 30=2304000, when operation Between can be ignored substantially) only need to carry out equal proportion scaling, so that it may obtain the region in original image where face, Bu Guoshi In the operation of border, ROI region separation all being carried out to each frame image and is not necessarily to, 1~5 Face datection operation per second is Can meet the needs of real-time monitoring, 2 Face datections of progress per second in this example.

In the present embodiment, the location determining method of cheek parts of images is: setting the width of face region as w, highly For h, then the width range of cheek region is 1/8w~3/8w, 5/8w~7/8w, altitude range 1/2h~3/4h, the region The region of nose is probably omitted, two cheek regions are only retained, reduces data interference.

Three, the mean value for calculating cheek parts of images the channel Green obtains equal value sequence, the datagram ginseng of equal value sequence See Fig. 4.

Calculate the mean value of Green channel components in cheek region image, that is, the channel Green point in each pixel that adds up Amount, then can be obtained divided by the pixel quantity of image.In order to improve processing speed, the mode that interval can be taken to sample, i.e., Odd-numbered line takes even number column data, and even number line takes the data of odd column, while guaranteeing covering, gets whole situation of change.

When handling data, influenced to reduce baseline drift bring, to current data to be treated all into Row normalized, takes the processing method of zero-mean, i.e., all data all subtract data mean value as pending data.

Since different filtering sections can respectively obtain changes in heart rate curve and change of respiratory rate curve, below to two The detailed process of a change curve is described in detail respectively.

Four, sliding average value filtering is carried out to equal value sequence, filtering section by adjusting it is 4~6, obtains changes in heart rate Curve.As shown in Figure 5.

About digital filtering, there are many kinds of methods, such as arithmetic equal value filtering, weighted mean filter, median filtering etc.. Sliding average filter method is used herein, because comparing other Mean Filtering Algorithms, efficiently sampling value of every calculating must connect Continuous sampling n times, for sample rate compared with slow or require the higher real-time system of data computation rate, these methods are can not to make ?.And sliding average filtering method only needs to consider continuous N number of data, so that it may obtain new filter value.Sliding is flat Mean filter method has good inhibiting effect to PERIODIC INTERFERENCE, and smoothness is high, and sensitivity is low；But to the arteries and veins accidentally occurred The inhibiting effect of punching property interference is poor, is not easy the deviation of sampled value caused by eliminating impulse disturbances.Therefore it is not suitable for impulse disturbances More serious occasion.

Five, feature extraction is carried out to changes in heart rate curve and obtains heart rate value.

In common non-contact PPG method, after collecting PPG signal, need by amplifying circuit to PPG signal into Row amplification, is then successively filtered operation using low-pass filter circuit and high-pass filtering circuit, then utilizes the trap of 50hz Circuit for eliminating Hz noise, then just carrying out power spectral analysis operation to finally obtained signal, (method is shown in spectrum analysis side Method).Just comprising comparing clearly heart rate, breath signal in the PPG signal of script, why common methods are but most straight without use The temporal analysis (directly detecting maximum value, the method for calculating the peak separation time) connect, and frequency-domain calculations method is used, Mainly since time domain approach is less susceptible to identification noise and real heart rate signal, thus just need to pass through repeatedly filter again into Line frequency spectrum analysis.And judgement noise can be thus achieved by temporal analysis in the present embodiment the method, therefore no longer need into The multiple filtering of row and complicated frequency spectrum conversion, to promote the accuracy and operational efficiency of rate calculation.

Referring to fig. 2, and Fig. 6 is combined, the present embodiment carries out feature extraction using following fisrt feature extraction algorithm:

Since collected signal is exactly the data Y using time shaft as coordinate_i, therefore the waveform that first find out signal is special Sign, such as wave crest/trough/peak-to-peak away from etc..Steps are as follows:

Step 1, it finds the downward inflection point conduct of first waveform from first point according to sequence to be calculated and sets out Point, i.e. Y_i>Y_i+1(Y is sequential digit values, i=1,2,3 ...), if Y₁>Y₂(first point may not be highest point), then first find wave The upward inflection point of shape, i.e. Y_i<Y_i+1(i=1,2,3 ...), is further continued for the inflection point for looking for first waveform downward as starting point later, That is the O point in Fig. 6, label O point are first wave crest.

Step 2, from O point, find the upward inflection point of next waveform, i.e. Y_i<Y_i+1(i=1,2,3 ...), i.e. Fig. 6 In 1 point, 1 point of label be first trough.

Step 3, from 1 point, it is continuous to find the downward inflection point of waveform, and be labeled as wave crest, then obtain wave crest sequence O, A, B, C, D, E ... and trough sequence 1,2,3,4,5 ... (as shown in Figure 6) are arranged, has at this moment shown that the waveform of signal is special Sign can carry out feature calculation according to formula:

Heart rate hbr=head and the tail are peak-to-peak every the time difference/peak separation quantity.

It can be encountered during finding wave crest sequence between some noise points, such as two wave crests of C and D, J and K two Between wave crest, the noise of doubtful wave crest can all occur, need to filter out.At this moment can by find range R in highest point come Spurious peaks are filtered out, when finding a wave crest, it is assumed that the spurious peaks between C and D point are C1, are found within R backward from C1 Maximum value, if the not peak value than C1 high within the scope of R, C1 continue the search of next wave crest as true wave crest；Such as Fruit has found the peak value D than C1 high, then D continues the search of next wave crest as true wave crest.Here it sets between J and K Spurious peaks are J1, and when finding wave crest J, continuation has searched J1 within the scope of R backward, so J1 is considered as spurious peaks, are needed Continually look for true wave crest K.

Most suitable R value filters out noise data in order to obtain, obtains accurate heart rate.Here by Medical Devices into Row comparison, analyzes by mass data, different R values is arranged by cycle calculations, obtained and R when Medical Devices same heart rate Range, specific value such as table 1:

The relationship of table 1 heart rate range and R value interval

Heart rate range	30~60	60~80	80~110	110~150	150~220
						R value interval	R1	R1~R2	R2~R3	R3~R4	R4

By in table 1, obtaining 4 numerical value R1-R4, the section by the way that R is arranged can cover human normal from R1~R4 substantially The range intervals of heart rate.This method can also be in the denoising of other any waveforms.It much makes an uproar in this way, directly filtering Sound data reduce and calculate error.

As waveform such as Fig. 7, since without noise, the change of R will not influence calculated result substantially, but if the number of monitoring Value contains more noise (spurious peaks), will obtain 4 different heart rate datas, it is therefore desirable to determine by other methods Accurate heart rate.Since the heart rate of normal person is sinus property rule heart rate, the amplitude of variation of each eartbeat interval is not in Too big variation, and if being added or missing out spurious peaks, this amplitude of variation exception is inevitably resulted in, for this purpose, we draw Enter function:

F=sd*sd/avg；

The fluctuation situation of Current heart rate is calculated, wherein sd is the mark of computation interval eartbeat interval (corrugation pitch) each time Quasi- poor, avg is the equispaced (corrugation pitch mean value) of computation interval heartbeat, and emotionally condition is the smallest for average wave, is closest to true Real HR values.

Certainly, the computing interval also will appear some cases, for example the corresponding serial number of the minimum value got every time constantly changes, How to judge whether current numerical value is abnormal, how to judge that the numerical value of heart rate is stable etc., specific processing method is as follows:

Take 10 seconds time spans as rate calculation section, and increase the HR values in last 5 seconds as pair Than reference, wherein the numerical value of heart rate is denoted as hbr, and the HR values calculated are denoted as hbr5 within last 5 seconds, and hbr5 also can be regarded as the heart The trend of rate variation, if there is unexpected fluctuation or extraneous influence, hbr5 and hbr have apparent numerical value difference.

Step 0, initialization lastchoice (the heart rate waveform serial number of last time selection), count (gets the tired of heart rate Metering number), e (cumulative number of the Current heart rate waveform serial number as heart rate), retrytime (recalculating number) they are 0；

Step 1,4 difference filtering interval R (R1~R4) according to setting calculate heart rate hbr, hrv5 and fluctuation situation F, and record storage, while cumulative count (being all that the corresponding sequence data of Min fi is taken to calculate heart rate value in count < 5)；

If Step 2, count >=5, it calculatesAs average fluctuation；

Step 3, g=Min gi (i=1~n) is taken, and records corresponding i as the alternative of current most suitable heart rate

If lastchoice=i, e=e+1, e is enabled to be not more than 5；

If lastchoice is not equal to i, e=e-1, e is enabled to be not less than 0；

Step 4, the difference d for calculating hbr and hbr5, when heart rate stabilization, the numerical value of d does not exceed 3.

If f > 10 (empirical value that many experiments obtain) of Step 5, Current heart rate, illustrate currently dry by the external world Disturb it is bigger, or acquisition heart rate have abnormal conditions.

At this moment judge: if f≤g or f≤10, executing Step 7；Otherwise Step 6 is executed；；

Step 6, the data for abandoning current sequence of calculation the first half enable lastchoice, count, e 0, enable Retrytime=retrytime+1；

If retrytime=2 abandons total data, lastchoice, count, e 0 is enabled；

If Step 7, continuous 5 difference d < 3, illustrate that rate calculation numerical value tends towards stability, the number of breathing is at this moment judged Whether value has been calculated (can tend towards stability, the frequency of breathing is led to then since the period is longer for the numerical value of heart rate most fast 7 seconds Often at least need the time of a respiratory cycle that can just find out that waveform changes), and tend towards stability (since heart rate and breathing are same When calculate, the consideration so the two is put together), if breathing numerical value it is also stable, export heart rate and respiratory rate numerical value, hold Row Step 8.

Step 8, stop data acquisition, terminate monitoring process.

It is also possible to by this method, Lai Jinhang vivo identification (such as picture or image etc.), because of extraneous light Line reflection is substantially irregular, so can then sentence if the fluctuation situation obtained is excessive (f > 10, under normal circumstances f≤5) Disconnected current light environment is undesirable, or tested cannot provide normal regular heart rate.

In laboratory environments, the HR values that this method is calculated are basic identical with the equipment that Medical Devices obtain.

Respiratory rate calculating process is illustrated below.

Six, sliding average value filtering is carried out to equal value sequence, filtering section by adjusting it is 28~32, obtains breathing frequency Rate change curve.As shown in Figure 8.

Five, feature extraction is carried out to change of respiratory rate curve and obtains respiratory rate value.

Referring to Fig. 3, and Fig. 9 being combined, the method that the present embodiment calculates respiratory rate is identical with the method for heart rate is calculated, but Since respiratory rate is different with heart rate feature, so the setting of r is real-time change here.The step of second feature extraction algorithm As follows: Step 1, according to sequence to be calculated finds first downward inflection point of waveform as starting point from first point, That is Y_i>Y_i+1(Y is sequential digit values, i=1,2,3 ...), if Y₁>Y₂(first point may not be highest point), then first find waveform Upward inflection point, i.e. Y_i<Y_i+1(i=1,2,3 ...), the inflection point for being further continued for looking for first waveform downward later is as starting point, i.e., O point in Fig. 9, label O point are first wave crest.

Step 2, from O point, find the upward inflection point of next waveform, i.e. Y_i<Y_i+1(i=1,2,3 ...), i.e. Fig. 9 In 1 point, 1 point of label be first trough.

Step 3, from 1 point, it is continuous to find the downward inflection point of waveform, and be labeled as wave crest, then obtain wave crest sequence O, A, B, C, D, E ... and trough sequence 1,2,3,4,5 ... (as shown in Figure 9) are arranged, has at this moment shown that the waveform of signal is special Sign, can carry out feature calculation, such as peak to peak separation etc..

Step 4, some noise points, such as spurious peaks A1 between A and B can be encountered during finding wave crest, at this moment needed Will be according to the respiratory intervals obtained before, i.e. time interval d between O and A sets r, and the range of r is d/2.Due in blood An oxygen whole content variation necessarily relatively slow process, so the value of r should be according between preceding breathing twice Real-time adjustment is done every d and d1.

R=(d+d1)/2

Step 5, the particularity due to respiratory rate can directly pass through formula after filtering out noise data:

Respiratory rate br=head and the tail are peak-to-peak every the time difference/peak separation quantity

Current respiratory rate is calculated, respiratory rate has hysteresis quality, and lag time is the one of preceding respiration duration Half time.Under normal circumstances, primary complete waveform can judge the general frequency of breathing.In laboratory environments, the party The error of breathing numerical value and actual numerical value that method is calculated substantially conforms to subject actual conditions within positive and negative 2.

The present embodiment algorithm stability is strong, by different illumination intensity, face's pickup area, acquire distance test, Compared with existing medical equipment in hospital calculated result, it is not easy to by the interference of external various conditions, there is extremely strong applicability.

It is for statistical analysis by acquiring a large amount of data sample, obtain following result:

(1) accuracy of inventive algorithm --- (correlation analysis and regression analysis) is compared with Medical Devices

[data are Medical Devices test and the present invention under different time (15s, 60s) different light intensities (high, medium and low) All data of algorithm]

2 present invention of table and the correlation of Medical Devices compare

Pearson product moment correlation result: related significant in 0.01 bilateral level.Related coefficient is 0.906.

For table 3 using the present invention as dependent variable, constant, medical treatment are the model that predictive variable is established

4 Regression Analysis Result of table

Regression Analysis Result: significant, regression coefficient 0.906 is returned.The error estimated by two above result standard 2.92393 and regression coefficient 0.906 it is found that inventive algorithm result is close with authoritative Medical Devices test result, it is seen that it is quasi- True property.

(2) influence (two analysis of variance) that time and light intensity measure the present invention

The inspection of effect between 5 main body of table

Dependent variable: the present invention

The side a.R=.019 (the adjustment side R=- .033)

6 the result of multiple comparisons of table

Dependent variable: the present invention

Based on the mean value observed.

Error term is mean value side (mistake)=43.948.

By above-mentioned variance analysis, main effect analysis is shown with the result of multiple comparisons, as a result very not significant, illustrates difference Selecting time and different light intensities do not influence measurement result.

Embodiment 2

The monitoring device of physiological parameter described in the present embodiment corresponds to each method and step described in embodiment 1, each The specific works of module are referring to described in embodiment 1.Which is not described herein again.

A kind of monitoring device of physiological parameter, including

Image capture module, for acquiring current face's image；

The filter module adjustment filtering section is m/3~2m/3, and m is the frame per second of current acquisition data, obtains heart rate Change curve；The analysis and processing module includes fisrt feature extraction module, and fisrt feature extraction module includes:

First data read module, for finding the position of first wave crest；

Wave crest sequence preserving module includes several sub- preserving modules of wave crest sequence, and heart rate 30~220 is divided into n area Between, n R value of corresponding setting is respectively adopted each R value and seeks wave crest sequence, is stored in wave crest sequence respectively and saves mould In block；For each wave crest sequence, f value is calculated using following formula:

f_i=sd_i*sd_i/avg_i；

Wherein, i=1 ... ..., n；sd_iFor the standard deviation of eartbeat interval, avg_iFor the equispaced for calculating eartbeat interval, take The corresponding i of Min fi calculates heart rate value as the sequence currently selected, using the sequence data.

The monitoring device includes data screening module, which includes:

Initialization module gets the tired of heart rate for initializing the heart rate waveform serial number lastchoice of last selection Metering number count, cumulative number e of the Current heart rate waveform serial number as heart rate, recalculates number retrytime, is assigned a value of 0；According to n of setting difference filtering interval R (R1~Rn), heart rate hbr, hbr L1 and fluctuation situation f are calculated, and records and deposits Storage, while cumulative count (in count < 5, the corresponding sequence data of Min fi being taken to calculate heart rate value)；

Averagely fluctuate computing module, for count >=5, calculateAs average fluctuation；

Alternate data module takes g=Mingi (i=1~n), and records corresponding i as current most suitable heart rate Alternatively.If lastchoice=0 (is not recorded) for the first time, lastchoice=i, e=1 are enabled；If lastchoice =i then enables e=e+1, e be not more than 5；If lastchoice is not equal to i, e=e-1, e is enabled to be not less than 0；

Difference calculating module, for calculating the difference d of hbr and hbr L1, when heart rate stabilization, the numerical value of d does not exceed 3；

Second judgment module, if set up, executes data removing module for judging f > g and whether f > 10 are true, Otherwise, the second judgment module is executed；It also corresponds to, judges whether f≤g or f≤10 are true, if set up, execute the Otherwise three judgment modules execute data removing module；

Data removing module enables lastchoice, count, e be for abandoning the data of current sequence of calculation the first half 0, enable retrytime=retrytime+1；If retrytime=2 abandons total data, enable lastchoice, Count, e 0；If retrytime=3, prompt is provided, preset test environment is undesirable, and monitoring failure, executing terminates mould Block；

Whether third judgment module, the difference d for judging continuous L2 calculating exhale less than 3 if it is, continuing judgement Whether the numerical value of suction has calculated and has tended towards stability, if it is, output heart rate and respiratory rate numerical value, execution terminate Module；

Ending module terminates monitoring process for stopping data acquisition.

The filter module adjustment filtering section is 3m/2~2m, and m is the frame per second of current acquisition data, obtains breathing frequency Rate change curve；The analysis and processing module includes second feature extraction module, and second feature extraction module includes:

It can implement the technology that the present invention describes by various means.For example, these technologies may be implemented in hardware, consolidate In part, software or combinations thereof.For hardware embodiments, processing module may be implemented in one or more specific integrated circuits (ASIC), digital signal processor (DSP), programmable logic device (PLD), field-programmable logic gate array (FPGA), place Manage device, controller, microcontroller, electronic device, other electronic units for being designed to execute function described in the invention or In a combination thereof.

It, can be with the module of execution functions described herein (for example, process, step for firmware and/or Software implementations Suddenly, process etc.) implement the technology.Firmware and/or software code are storable in memory and are executed by processor.Storage Device may be implemented in processor or outside processor.

Those of ordinary skill in the art will appreciate that: realize that all or part of the steps of above method embodiment can pass through The relevant hardware of program instruction is completed, and program above-mentioned can store in a computer-readable storage medium, the program When being executed, step including the steps of the foregoing method embodiments is executed；And storage medium above-mentioned includes: ROM, RAM, magnetic disk or light The various media that can store program code such as disk.

The above embodiment is a preferred embodiment of the present invention, but embodiments of the present invention are not by above-described embodiment Limitation, other any changes, modifications, substitutions, combinations, simplifications made without departing from the spirit and principles of the present invention, It should be equivalent substitute mode, be included within the scope of the present invention.

Claims

1. a kind of monitoring method of physiological parameter, which is characterized in that comprising steps of

Acquire current face's image；

Extract the cheek parts of images in facial image；

Sliding average value filtering is carried out to equal value sequence, obtains physiological change curve by adjusting filtering section；

Feature extraction is carried out to physiological change curve and obtains physiological parameter；

Physiological change curve data is filtered, changes in heart rate curve and change of respiratory rate curve are respectively obtained, for upper It states physiological change curve data and first carries out following screening operation, the method is as follows:

Take L seconds time spans as rate calculation section, and increase by one it is L1 seconds last in HR values join as a comparison It examines, wherein the numerical value of heart rate is denoted as hbr, and the HR values of last L1 seconds calculating are denoted as hbr L1, is calculated for continuous L2 times The HR values that changes in heart rate is no more than a certain range and indicates tend towards stability, p1, p2, and p3, p4 are empirical parameter value；

Step0, the last heart rate waveform serial number lastchoice selected of initialization, get the cumulative number count of heart rate, when Cumulative number e of the preceding heart rate waveform serial number as heart rate, recalculates number retrytime, is assigned a value of 0；

Step1, the n different filtering interval R according to setting, calculate heart rate hbr, hbr L1 and fluctuation situation f, and record and deposit Storage, while cumulative count take the corresponding sequence data of Minfi to calculate heart rate value in count < p1；

If Step2, count >=p1, calculateAs average fluctuation；

Step3, g=Min g is taken_i, wherein i=1~n, and corresponding i is recorded as the alternative of current most suitable heart rate:

If lastchoice=0, lastchoice=i, e=1 are enabled；

If lastchoice=i, e=e+1, e is enabled to be not more than p2；

If lastchoice is not equal to i, e=e-1, e is enabled to be not less than 0；

Step4, the difference d for calculating hbr and hbr L1；

If Step5, f≤g or f≤p3, execute Step7；Otherwise Step6 is executed；

Step6, the data for abandoning current sequence of calculation the first half, enable lastchoice, count, e 0, enable retrytime= retrytime+1；

If retrytime=2 abandons total data, lastchoice, count, e 0 is enabled；

If retrytime=3, prompt is provided, preset test environment is undesirable, and monitoring failure executes Step8；

If difference d < p4 of Step7, continuous L2 calculating, at this moment judge whether the numerical value of breathing has calculated and become In stabilization, if the numerical value of breathing is also stable, heart rate and respiratory rate numerical value are exported, executes Step8；

Step8, stop data acquisition, terminate monitoring process.

2. the monitoring method of physiological parameter according to claim 1, which is characterized in that set the frame of current acquisition data Rate is m, sets the filtering section of equal value sequence as m/3~2m/3, obtains changes in heart rate curve, mentioned using following fisrt feature Algorithm is taken to carry out feature extraction:

Preferentially find the position of first wave crest；

Judge whether primary peak peak value is more than or equal to secondary peak peak value, if so, determining that primary peak is current wave crest, so It is continued searching since primary peak afterwards；If it is not, then determining that secondary peak is current wave crest, then since secondary peak It continues searching；

After complete changes in heart rate curve of search, a wave crest sequence is obtained, the wave crest sequence includes the peak value of each wave crest And position, the distance between wave crest and the position of wave crest are eartbeat interval；

3. the monitoring method of physiological parameter according to claim 2, which is characterized in that by entire heart rate interval 30~220 It is divided into n section, n R value of corresponding setting is respectively adopted each R value and seeks wave crest sequence；For each wave crest sequence, F value is calculated using following formula:

f_i=sd_i*sd_i/avg_i；

Wherein, i=1 ... ..., n；sd_iFor the standard deviation of eartbeat interval, avg_iFor the equispaced for calculating eartbeat interval, Min is taken f_iCorresponding i calculates heart rate value as the sequence currently selected, using the sequence data.

4. the monitoring method of physiological parameter according to claim 1, which is characterized in that set the frame of current acquisition data Rate is m, sets the filtering section of equal value sequence as 3m/2~2m, obtains change of respiratory rate curve, special using following second It levies extraction algorithm and carries out feature extraction:

Step1, according to the data sequence in change of respiratory rate curve, find first wave crest point O point；

Step2, from O point, find the upward inflection point of next waveform, obtain first trough, label 1；

Step3, from 1 point of label, it is continuous to find the downward inflection point of waveform, and be labeled as wave crest, then obtain wave crest sequence Arrange O, A, B, C, D, E ... and trough sequence 1,2,3,4,5 ..., obtain the wave character of signal；

Step4, according to wave character, it is peak-to-peak every the time difference and peak separation quantity to obtain head and the tail, pass through formula:

Obtain current respiratory rate；

Following denoising method is used during finding wave crest:

Judge whether the time where the wave crest currently found is greater than last time peak time+r, if it is, being judged as new wave Peak, if it is not, then being judged as noise spot.

5. a kind of monitoring device of physiological parameter characterized by comprising

Image capture module, for acquiring current face's image；

Filter module obtains physiological change song for carrying out sliding average value filtering to equal value sequence by adjusting filtering section Line；

Analysis and processing module obtains physiological parameter for carrying out feature extraction to physiological change curve；

The monitoring device includes data screening module, which includes:

Initialization module, for initializing the heart rate waveform serial number lastchoice of last selection, accumulative time for getting heart rate Count is counted, cumulative number e of the Current heart rate waveform serial number as heart rate recalculates number retrytime, be assigned a value of 0；Root According to n different filtering interval R of setting, heart rate hbr, hbr L1 and fluctuation situation f, and record storage are calculated, is added up simultaneously count；

Second judgment module, if set up, executes third judgment module for judging whether f≤g or f≤p3 is true, Otherwise, data removing module is executed；

Data removing module, for abandoning the data of current sequence of calculation the first half, initialization lastchoice, count, e are 0, enable retrytime=retrytime+1；If retrytime=2, total data is abandoned, initialization lastchoice, Count, e 0；If retrytime==3, prompt is provided, preset test environment is undesirable, and monitoring failure, execution terminates Module；

Third judgment module then shows for judging whether the difference d of continuous L2 calculating is less than p4 if it is less than p4 HR values tend towards stability, then continue to judge whether the numerical value of breathing has calculated and tended towards stability, if it is, Heart rate and respiratory rate numerical value are exported, ending module is executed；

Ending module terminates monitoring process for stopping data acquisition.

6. the monitoring device of physiological parameter according to claim 5, which is characterized in that the filter module adjustment filtering area Between be m/3~2m/3, m is the frame per second of current acquisition data, obtains changes in heart rate curve；The analysis and processing module includes the One characteristic extracting module, fisrt feature extraction module include:

First data read module, for finding the position of first wave crest；

Primary peak search module, for since the position of previous wave crest, searching for the highest first wave of value within the scope of R thereafter Peak；

Secondary peak search module, for since the position of primary peak, searching for highest second wave of value within the scope of R thereafter Peak；

First judgment module, for judging whether primary peak is more than or equal to secondary peak, if so, determining that primary peak is to work as Preceding wave crest, then continues searching since primary peak；If it is not, then determining that secondary peak is current wave crest, then from the Two wave crests start to continue searching；

Primary peak sequence preserving module obtains a wave crest sequence, the wave after searching for complete changes in heart rate curve Peak sequence includes peak value and the position of each wave crest, and the distance between wave crest and the position of wave crest are eartbeat interval；

7. the monitoring device of physiological parameter according to claim 5, which is characterized in that the filter module adjustment filtering area Between be 3m/2~2m, m is the frame per second of current acquisition data, obtains change of respiratory rate curve；The analysis and processing module packet Second feature extraction module is included, second feature extraction module includes:

First crest seeking module, for finding first wave crest point O according to the data sequence in change of respiratory rate curve Point；

First trough finds module, for finding the upward inflection point of next waveform from O point, obtains first trough, Label 1；

Wave character determining module, for constantly finding the downward inflection point of waveform, and be labeled as wave crest from 1 point of label, Then obtain wave crest sequence O, A, B, C, D, E ... and trough sequence 1,2,3,4,5 ..., obtain the wave character of signal；

Respiratory rate computing module, it is peak-to-peak every the time difference and peak separation quantity from beginning to end for obtaining according to wave character, pass through public affairs Formula: respiratory rate br=head and the tail are peak-to-peak every the time difference/peak separation quantity；Obtain current respiratory rate；

The second feature extraction module further includes denoising module, and denoising module includes:

Step-size in search setting module calculates step-size in search r=(d+d1)/2 if peak to peak separation twice recently is respectively d and d1；

Whether judgment module, time where the wave crest for judging currently to find are greater than last time peak time+r, if it is, It is judged as new wave crest, if it is not, then being judged as noise spot.

8. handheld device, which is characterized in that the monitoring device including the described in any item physiological parameters of claim 5-7.