Content of the invention
The purpose of the present invention is the defect for prior art, provides a kind of image processing method, thus economical and low becomeThe detection data of this acquisition human body.
For achieving the above object, the invention provides a kind of image processing method, methods described includes：
Step 1, the image acquiring device using mobile terminal obtains body surface image；
Step 2, extracts image intensity value and chromatic value from described body surface image；
Step 3, generates the first image intensity value curve and the first image chroma according to image intensity value and chromatic value respectivelyValue curve；
Step 4, carries out curve data and divides to described first image gray value curve and the first pattern colour angle value curve respectivelyAnalysis, thus obtaining human detection data, described human detection data includes heart rate value, pressure value, hyperlipidemia angle value and blood oxygenIntensity value；
Step 5, when described human detection data exceedes first threshold, or during less than Second Threshold, generates the first reportAlarming information；
Step 6, calculates the first rate of change of described human detection data, if described first rate of change is more than the 3rd thresholdValue, then generate the second warning message；
Step 7, according to repeatedly described human detection data, calculates variation tendency, if described variation tendency meets firstDuring variation tendency, then generate the 3rd warning message.
Further, described step 1 specifically includes：Step 1, obtains the body of human body finger tip using the photographic head of mobile terminalTable image.
Further, described step 2 specifically includes, and described body surface image is converted to the numeral of YUV color spaceImage, extracts Y value and UV value from described digital picture.
Further, obtain heart rate value in described step 4 to specifically include：
Small echo is selected to carry out wavelet decomposition to described first image gray value curve；
Threshold value quantizing is carried out to the wavelet conversion coefficient after decomposing；
According to the coefficient reconstruct small echo after described threshold value quantizing, form the second image intensity value curve；
Real-time heart rate data is calculated according to the crest number of described second image intensity value curve.
Further, obtain pressure value in described step 4 to specifically include：
Select small echo that described first image gray-value variation curve is carried out with wavelet decomposition, and to the wavelet transformation after decomposingCoefficient carries out threshold value quantizing；
According to the coefficient reconstruct small echo after described threshold value quantizing, form the second image intensity value change curve；
Real-time heart rate data is calculated according to the crest number of described second image intensity value change curve；
Blood pressure data is calculated with the relation of blood pressure and described real-time heart rate data according to heart rate.
Further, obtain hyperlipidemia angle value in described step 4 to specifically include：
Determine the relative value of hematocrit according to described first image chromatic value curve, relative according to hematocritIt is worth to hyperlipidemia angle value.
Further, obtain oximetry value in described step 4 to specifically include：
Light intensity rate of change is obtained according to described first image gray value curve, blood oxygen is calculated according to described light intensity rate of change and satisfiesAnd angle value.
Further, the first rate of change in described step 6 is mean square deviation rate of change.
Further, in described step 7, calculate variation tendency specifically, being spaced according to detection time, the unit of account timeAmplitude of variation, if the first variation tendency continuing to increase or persistently reducing, will surpass when reaching next unit intervalCross the 4th threshold value or be less than the 5th threshold value, then generate the 3rd warning message.
Further, also include after described step 7：Step 8, described variation tendency formation visualized graphs are represented.
Image processing method of the present invention it is achieved that get the detection number of human body with low time cost and Financial costAccording to.
Specific embodiment
Below by drawings and Examples, technical scheme is described in further detail.
Fig. 1 is the flow chart of image processing method of the present invention, as illustrated, the present invention specifically includes following steps：
Step 101, the image acquiring device using mobile terminal obtains body surface image；
Specifically utilize the surface images of the photographic head acquisition human body finger tip of mobile terminal.
In order to save the Financial cost of image acquisition, so make use of the photographic head of mobile terminal, photographic head can be movedPhotographic head that dynamic terminal has been configured is it is also possible to by other photographic head external for mobile terminal (such as high-definition camera or special take the photographAs head) gathering body surface image.
Preferably, in order to not make, the brightness of surface images is too low or gray value is too high, so body surface imageCollection point preferably can be higher with light transmittance, and this allows for, and body surface is thin as far as possible, and preferably collection point is finger tip.
Step 102, extracts image intensity value and chromatic value in body surface image；
Specifically include, body surface image is converted to the digital picture of YUV color space, extract Y from digital pictureValue and UV value.
Because being analogue signal using the body surface image that photographic head collects, need to convert analog signals intoThe image of digital signal.And Digital Image Data can be a lot of image spaces, such as RGB image space or YUV image space.If for the gradation data easily gathering image, certainly preferably with YUV image space, such Y value is exactly imageGray value, UV value is chromatic value.
Step 103, generates the first image intensity value curve and the first pattern colour according to image intensity value and chromatic value respectivelyAngle value curve；
The foundation of the first image intensity value curve needs the gray value of continuous multiple images just can generate, therefore in stepIn rapid 101, the body surface image of collection should be continuous multiple images frame.Thus identification obtains the gray value of multiple images,Thus generate the first image intensity value curve.Equally, the first pattern colour angle value curve is generated according to pattern colour angle value.
Specifically, body surface image is mainly the gray scale gathering blood color.
Step 104, carries out curve data to described first image gray value curve and the first pattern colour angle value curve respectivelyAnalysis, thus obtain human detection data；
First image intensity value curve and the first pattern colour angle value curve carry out curve data analysis it is simply that to blood colorThe pulse wave that obtains of light and shade mutation analysises, by heart rate value, pressure value, blood viscosity are finally given to the analysis of pulse waveValue and oximetry value.
Specifically, obtain heart rate value to specifically include：Small echo is selected to carry out wavelet decomposition to the first image intensity value curve；RightWavelet conversion coefficient after decomposition carries out threshold value quantizing；According to the coefficient reconstruct small echo after threshold value quantizing, form the second image ashAngle value curve；Real-time heart rate data is calculated according to the crest number of the second image intensity value curve.
Video data, during collection, is inevitably disturbed by all kinds noise, and common noise is doneDisturbing source mainly has following three kinds：The first is myoelectricity noise, is that the frequency being caused by physical activity or muscular tone is usualBetween 5 hertz to 2000 hertz；Second is power frequency noise, is the spatial electromagnetic interference being produced by supply network and its equipmentIn the reaction of human body, it is the interference of fixed frequency, frequency is typically more than 50 hertz；The third is baseline drift, is by human bodyThe low-frequency disturbance that breathing, limb activity etc. cause, somewhat violent limb motion is serious by causing heart rate waveform signal to occurChange, frequency is typically between 0.05 hertz to 2 hertz.Myoelectricity noise and baseline drift are important interference sources, in this exampleMethod using wavelet threshold denoising.Wavelet function changes in the range of finite time, and meansigma methodss are 0.
Choose a wavelet function and determine the level N of a wavelet decomposition, then the first image intensity value curve is enteredRow N shell wavelet decomposition, obtains wavelet coefficient, and wherein N is positive integer.Specifically, the first image intensity value curve is averagely decomposedBecome the part gray value curve of several times；Small echo is alignd with the starting point of part gray value curve, calculates very first time portionDivide the approximation ratio of gray value curve and wavelet function, that is, calculate wavelet conversion coefficient, wavelet conversion coefficient means that more greatlyPart gray value curve is more close with the waveform of selected wavelet function；When then by mobile along time shafts for a wavelet function unitBetween, calculate the wavelet conversion coefficient of the part gray value curve of next time, bent until covering whole first image intensity valueLine.
For each layer of high frequency coefficient, select a threshold value to carry out quantification treatment, obtain new wavelet coefficient.
Low frequency coefficient according to wavelet decomposition n-th layer and the 1st layer of high frequency coefficient to n-th layer after quantification treatment,Carry out the wavelet inverse transformation of the first image intensity value curve, obtain the second image intensity value curve.
Because cardiac motion result in the waveform running of a blood arrival fingerstick capillary each time, work as blood capillaryDuring congestive state, in blood, oxygen content increases, and blood color is in cerise, and average gray value is relatively low, and consumes blood in bodyAfter oxygen in liquid, blood becomes kermesinus, and the period of change therefore calculating blood color just can calculate the week of cardiac motionPhase, i.e. heart rate.Crest number according to the second image intensity value change curve and the ratio of acquisition time, can calculate blood per secondThe number of times of liquid color change, then be multiplied by 60 number of times being blood color per minute change, thus to real-time heart rate data.
Obtain pressure value to specifically include：Small echo is selected to carry out wavelet decomposition to the first image intensity value change curve, and rightWavelet conversion coefficient after decomposition carries out threshold value quantizing；According to the coefficient reconstruct small echo after threshold value quantizing, form the second image ashAngle value change curve；Real-time heart rate data is calculated according to the crest number of the second image intensity value change curve；According to heart rateIt is calculated blood pressure data with the relation of blood pressure and real-time heart rate data.
Blood pressure data includes systolic pressure data and diastolic blood pressure data.Systolic pressure, diastolic pressure and heart rate have dependency, according toSystolic pressure=0.1736 × heart rate+105.34 and real-time heart rate data calculate systolic pressure data, and according to diastolic pressure=0.378 × heart rate-(20-M) and real-time heart rate data calculate diastolic blood pressure data, and wherein M is the integer in the range of 0 to 5.
Obtain hyperlipidemia angle value to specifically include：The relative of hematocrit is determined according to the first pattern colour angle value curveValue, relative according to hematocrit is worth to blood viscosity value.
Because user's blood viscosity has dependency with the relative value of hematocrit, hematocrit change causes fingerThe colourity change of sharp color, the colourity change of finger tip color is by the every two field picture in human body finger tip surface images video dataColourity embodied.Therefore, it can determine that the hemocyte in user's blood accounts for whole blood by the colourity change of finger tip colorIn percent volume to volume, that is, obtain hematocrit, so that it is determined that the data of blood viscosity.
Obtain oximetry value to specifically include：Light intensity rate of change is obtained according to the first image intensity value curve, according to lightStrong rate of change calculates oximetry value.
In near-infrared region, when two-beam detects tissue, only consider the impact of reduced hemoglobin and HbO2 Oxyhemoglobins,Therefore blood oxygen can be carried out using the secondary light source signal of the first light signal of the R component in rgb color space and G componentThe measurement of saturation.
Because the optical signal that light is returned after blood can be attenuated, and the rate of change of transmitted light intensity and reflective light intensityIt is directly proportional to absorptance, therefore can calculate blood oxygen saturation using light intensity rate of change.
Step 105, when human detection data exceedes first threshold, or during less than Second Threshold, generates the first warningInformation；
Specifically, if in human detection data (heart rate value, pressure value, hyperlipidemia angle value and oximetry value)Certain or a few data exceed first threshold set in advance, or less than Second Threshold predetermined in advance, then generate theOne warning message.
Because by each one high first threshold of different human detection data settings, if the human detection detectingData then needs to generate the first warning message higher than first threshold；In the same manner by one ratio of each different human detection data settingRelatively low Second Threshold, if the human detection data detecting is less than Second Threshold, needs to generate the first warning message.
Specifically, the generation of the first warning message is because that human detection data is excessive or too small, might have risk,So the first warning message generating can be to remind to check UP, remind and remove examination in hospital etc..
Step 106, calculates the first rate of change of human detection data, if the first rate of change, more than the 3rd threshold value, is given birth toBecome the second warning message；
Specifically, the first rate of change is mean square deviation rate of change.
Human detection data (heart rate value, pressure value, hyperlipidemia angle value and oximetry value) each time is all lonelyVertical data, such that it is able to carry out time-domain analyses by human detection data.
For example, obtain, using mean square deviation rate of change, the first rate of change that human detection data detects every time, if the first changeRate excessive (the such as first rate of change is more than the 3rd threshold value), then need to generate the second warning message.
Specifically, the generation of the second warning message is because that the rate of change of human detection data is too big, might have risk,So the second warning message generating can be to remind to check UP, remind and remove examination in hospital etc..
Step 107, according to multiple human detection data, calculates variation tendency, if variation tendency meets the first change and becomesDuring gesture, then generate the 3rd warning message.
What the data processing of step 106 obtained is the situation of change of human detection data, but does not learn specific changeChange trend.And the purpose of the data processing of step 107 is desirable to obtain the variation tendency of human detection data.
Calculate variation tendency specifically, according to detection time interval, the amplitude of variation of unit of account time, if persistently increasedPlus or the first variation tendency of persistently reducing, the 4th threshold value will be exceeded or be less than the 5th threshold when reaching next unit intervalValue, then generate the 3rd warning message.
It is possible to carry out time-domain analyses when having multiple human detection data, specific time-domain analyses need to pressAccording to the concrete time of multiple human detection data, because the time interval of the different generations of human detection data is different,So being accomplished by calculating the first variation tendency according to time interval.Such as variation tendency is that human detection data becomes larger, meterCalculate the next unit interval detection time when, human detection data is likely to be breached more than the 4th threshold value；Or human detection numberAccording to being gradually reduced, calculate the next unit interval detection time when, human detection data is likely less than the 5th threshold value, thusNeed to generate the 3rd warning message.
Specifically, the generation of the 3rd warning message is because the too Gao Houtai that the variation tendency of human detection data can becomeLow it is possible to can occurrence risk, so generate the 3rd warning message can be remind check UP, remind remove examination in hospital etc.Deng.
Step 108, variation tendency formation visualized graphs are represented.
The step for meaning be according to time domain distribution, representing of multiple human detection data visualization thus may be usedIntuitively to be observed it is seen that the situation of change/rate of change of specific qualitative/quantitative by people, and possible variation tendency.
Image processing method of the present invention it is achieved that get the detection number of human body with low time cost and Financial costAccording to.
Professional should further appreciate that, each example describing in conjunction with the embodiments described hereinUnit and algorithm steps, can be hard in order to clearly demonstrate with electronic hardware, computer software or the two be implemented in combination inPart and the interchangeability of software, generally describe composition and the step of each example in the above description according to function.These functions to be executed with hardware or software mode actually, the application-specific depending on technical scheme and design constraint.Professional and technical personnel can use different methods to each specific application realize described function, but this realizationIt is not considered that it is beyond the scope of this invention.
The step of the method in conjunction with the embodiments described herein description or algorithm can be with hardware, computing deviceSoftware module, or the combination of the two is implementing.Software module can be placed in random access memory (RAM), internal memory, read only memory(ROM), electrically programmable ROM, electrically erasable ROM, depositor, hard disk, moveable magnetic disc, CD-ROM or technical fieldIn interior known any other form of storage medium.
Above-described specific embodiment, has been carried out to the purpose of the present invention, technical scheme and beneficial effect furtherDescribe in detail, be should be understood that the specific embodiment that the foregoing is only the present invention, be not intended to limit the present inventionProtection domain, all any modification, equivalent substitution and improvement within the spirit and principles in the present invention, done etc., all should compriseWithin protection scope of the present invention.