CN102488508A - Heart rate measuring method based on image capture - Google Patents

Abstract

The invention relates to a heart rate measuring method based on image capture, and relates to a heart rate measuring method, which solves the problems of inconvenience for measurement in the conventional electrocardiogram contact measuring method and low measuring accuracy in a method for measuring indirectly by using a sensor. The method comprises the following steps of: acquiring a characteristic vector T of detection space and a threshold value epsilon converted towards a binary image; acquiring a heart rate detection video; extracting all frame images F1, F2,...,Fx from the video; acquiring a grayscale image Gx of the Fx in the detection space; acquiring a grayscale image Ry of a differential value of Gx and G1; extracting a variation region My of each blood vessel image; extracting a value Ky of the variation region in each image; extracting a variable characteristic value from the variable Ky to acquire a period Tr; and acquiring a heart rate value HR. By the heart rate measuring method, the defect of high noise caused by the contact of the conventional measuring instrument and human bodies is overcome; and the heart rate measuring method based on the image capture has a bright application prospect, for an example, heart rates can be measured directly, conveniently and quickly by utilizing mobile camera and the like.

Description

A kind of method for measuring heart rate based on image capturing

Technical field

The present invention relates to a kind of method for measuring heart rate.

Background technology

Heart rate (Heart Rate): be used for describing the technical term of cardiac cycle, be meant the dancing number of times of heart per minute.Heart rate when normal adult is quiet has significant individual variation, on average about 75 times/minute (between 60-100 time/minute).Heart rate can be different because of age, sex and other physiological conditions.The heart rate of neonatal is very fast, can reach more than 130 times/minute.Same individual, decreased heart rate when quiet or sleep, heart rate is accelerated during motion or when excited, under the influence of some drugs or neuro humor factor, can make heart rate take place to accelerate or slow down.Adult's per minute heart rate surpasses 100 times or infant surpasses 105 times/minute persons, is called tachycardia.Adult's heart beating number of times just is called bradycardia less than 60 times/minute.Tachycardia and bradycardia all can influence people's health, and especially concerning the problematic people of heart, heart rate measurement is a very important index.Therefore, heart rate measurement is one of medical inspection project of using always, and heart rate measurement all has a wide range of applications at aspects such as patient's monitoring, clinical treatment, body-building and athletic competitions accurately in real time.

At present the method in heart rate measurement field mainly contains two big types, and one type is to measure through electrocardiogram, and another kind of is to be measured signal aroused in interest indirectly and counted and calculate heart rate by pick off.The former measures accurately and can show other Human Physiology index such as heart rate intuitively, but cost is very high, and carries inconvenience.Latter's cost is low, but because pick off must closely contact with human body, causes noise bigger, and measurement result is unreliable.Therefore, a kind of cost is low, and contactless method for measuring heart rate is demanded urgently proposing.

Summary of the invention

The present invention causes measuring inconvenience in order to solve the graphic measuring method that is measured as contact of existing employing electrocardio, adopts the low problem of accuracy of measurement of pick off indirect measurement method, thereby a kind of method for measuring heart rate based on image capturing is provided.

A kind of method for measuring heart rate based on image capturing, it is realized by following steps:

Step 1, initialization: set and detect spatial characteristic vector T, said characteristic vector T is 3 * 1 vector; Setting is to the threshold epsilon of bianry image conversion;

Step 2, employing photographic head are taken the video of one section heart rate detection; And from video, extract each two field picture, and be labeled as F successively ₁, F ₂..., F _x, x ∈ [1, n], n is the frame number of image in this section video, F _xBe the expression of x two field picture under rgb space, that is: F _xIt is the matrix of M * N * 3;

If F _x ¹Be that x opens the gray value of image on the R space, F _x ²Be that x opens the gray value of image on the G space, F _x ³Be that x opens the gray value of image on the B space;

Use F _x ¹(i, j), F _x ²(i, j), F _x ³(i j) representes that respectively x opens the element value of image at the capable j row of the i on the R space, on the G space, on the B space, i ∈ [1, M], j ∈ [1, N];

Interval between adjacent two two field pictures does

T _vDuration for video;

Step 3, with image F _xProject to and detect in the space, obtain the gray-scale map G under the described detection of step 1 space _x, G _xMatrix for M * N; G _x(i j) is G _xIn the element value of the capable j of i row, get:

$G_x(i,j)={F_x}^1(i,j)\×\frac{T_{11}}{\sqrt{{T_{11}}^2+{T_{21}}^2+{T_{31}}^2}}+{F_x}^2(i,j)\×\frac{T_{21}}{\sqrt{{T_{11}}^2+{T_{21}}^2+{T_{31}}^2}}$

$+{F_x}^3(i,j)\×\frac{T_{31}}{\sqrt{{T_{11}}^2+{T_{21}}^2+{T_{31}}^2}}$

Wherein, T ₁₁, T ₂₁, T ₃₁Be respectively vector T first to the third line element;

Step 4, with every gray-scale map G of gained in the step 3 _xRespectively with first gray-scale map G ₁It is poor to do, and obtains n-1 after taking absolute value and open new difference gray-scale map R _y, y ∈ [1, n-1], R _y(i j) is R _yIn the element value of the capable j of i row, that is: R _y(i, j)=| G _y(i+1, j+1)-G ₁(i, j) |;

Step 5, extract the region of variation M of each two field picture medium vessels image _y

Step 6, extract the region of variation M of each two field picture medium vessels image in the step 5 _yImage K _y, detailed process does; Region of variation M with each the two field picture medium vessels image that obtains in the step 5 _yOne by one with step 4 in the gray-scale map R that obtains _yCarry out NAND operation, obtain K _y

K _y(i j) is K _yIn the element value of the capable j of i row, i.e. K _yIn the value of each element be: K _y(i, j)=M _y(i, j) * R _y(i, j);

Step 7, at image K _yThe middle eigenvalue that changes that extracts is designated as

Image K in step 8, the obtaining step seven _yThe middle eigenvalue that changes that extracts

Cycle T _r

Step 9, according to formula:

$\mathrm{HR}=\frac{60}{T_r}$

Obtain heart rate value HR.

Extract the region of variation M of each two field picture medium vessels image described in the step 5 _yMethod be:

Steps A 1, with the difference gray-scale map R that obtains in the step 4 ₁, R ₂..., R _N-1Being converted into ε is the binary image encoder B that Threshold Segmentation forms ₁, B ₂.., B _N-1, in the formula: B _y(i, j) expression B _yThe element value of the capable j row of i among the figure, that is:

$B_y(i,j)=\left\{\begin{array}{c}1,R_y(i,j)\&GreaterEqual;\mathrm{\&epsiv;}\\ 0,R_y(i,j)<\mathrm{\&epsiv;}\end{array}\right.;$

Steps A 2, to B in the steps A 1 _yExecution is the opening operation of structural element with Q, obtains C ₁, C ₂..., C _N-1, wherein:

$Q=\left(\begin{array}{ccccccc}0& 0& 0& 1& 0& 0& 0\\ 0& 0& 1& 1& 1& 0& 0\\ 0& 1& 1& 1& 1& 1& 0\\ 1& 1& 1& 1& 1& 1& 1\\ 0& 1& 1& 1& 1& 1& 0\\ 0& 0& 1& 1& 1& 0& 0\\ 0& 0& 0& 1& 0& 0& 0\end{array}\right)$

C _y＝B _yοQ；

Steps A 3: to the Elements C in the steps A 2 _yExecution is the closed operation of structural element with Q, obtains the region of variation M of each two field picture medium vessels image _y, that is: M _y=B _yQ.

Described in the step 7 from new image K _yThe middle method of extracting the eigenvalue that changes is:

Gained image K in step B1, the calculation procedure six _yElement in minima and maximum, be designated as K respectively _y(i, j) _MinAnd K _y(i, j) _MaxAnd with [K _y(i, j) _Min, K _y(i, j) _Max] interval on average is divided into 7 minizones, K _yElement value K in the image _y(i j) includes in said 7 minizones, that is: successively

The 1st interval range is: $\left[K_y{(i,j)}_{\mathrm{Min},}\frac{K_y{(i,j)}_{\mathrm{Max}}+6K_y{(i,j)}_{\mathrm{Min}}}{7}\right),$

The 2nd interval range is: $[\frac{K_y{(i,j)}_{\mathrm{Max}}+6K_y{(i,j)}_{\mathrm{Min}}}{7},\frac{2K_y{(i,j)}_{\mathrm{Max}}+5K_y{(i,j)}_{\mathrm{Min}}}{7}),$

The 3rd interval range is: $[\frac{2K_y{(i,j)}_{\mathrm{Max}}+5K_y{(i,j)}_{\mathrm{Min}}}{7},\frac{3K_y{(i,j)}_{\mathrm{Max}}+4K_y{(i,j)}_{\mathrm{Min}}}{7}),$

The 4th interval range is: $[\frac{3K_y{(i,j)}_{\mathrm{Max}}+4K_y{(i,j)}_{\mathrm{Min}}}{7},\frac{4K_y{(i,j)}_{\mathrm{Max}}+3K_y{(i,j)}_{\mathrm{Min}}}{7}),$

The 5th interval range is: $[\frac{4K_y{(i,j)}_{\mathrm{Max}}+3K_y{(i,j)}_{\mathrm{Min}}}{7},\frac{5K_y{(i,j)}_{\mathrm{Max}}+2K_y{(i,j)}_{\mathrm{Min}}}{7}),$

The 6th interval range is: $[\frac{5K_y{(i,j)}_{\mathrm{Max}}+2K_y{(i,j)}_{\mathrm{Min}}}{7},\frac{6K_y{(i,j)}_{\mathrm{Max}}+K_y{(i,j)}_{\mathrm{Min}}}{7}),$

The 7th interval range is: $[\frac{6K_y{(i,j)}_{\mathrm{Max}}+K_y{(i,j)}_{\mathrm{Min}}}{7},K_y{(i,j)}_{\mathrm{Max}}];$

Add up the arithmetic mean of instantaneous value of all elements in the 2nd to the 6th interval range, obtain new image K _yThe middle eigenvalue that changes that extracts is designated as

New image K in the obtaining step described in the step 8 seven _yThe middle eigenvalue that changes that extracts obtains

Cycle T _rMethod be:

Step C1, get first point in

as fiducial value, assignment is in

Step C2, establish

H ∈ [2, n-1], from h=2 begin with h one by one substitution F (h) calculate, when for the first time F (h)＜0 occurring, the h of note this moment is h ₁Continue substitution h and calculate, occur at F (h)＜0 o'clock for the second time, note h at this moment is h ₂(h then ₂-1) T _f=T _R1, T _R1Be the cycle of asking;

Step C3, get the 4th point

With the 7th point The process of repeating step C2 obtains T respectively _R2And T _R3

Step C4, when there not being corresponding h ₁, h ₂The time, then corresponding T _R1, T _R2, T _R3Be taken as 0; T _r=median (T _R1+ T _R2+ T _R3), be:

Cycle.

The duration T of the video described in the step 2 _vBe 3s.

Beneficial effect: the present invention is based on image capturing; Propose the method for measuring heart rate that a kind of combining image is handled, carry out IMAQ by photographic head in the processing, avoided contact method for measuring heart rate in the past; Have lower noise and higher reliability, accuracy of measurement is high; Simultaneously, the cost of photographic head is lower, is convenient to use widely.This heart rate measurement mode has more practical value in reality.

Description of drawings

Fig. 1 is image K new among the present invention _yThe middle eigenvalue that extracts variation gets access to

Cycle T _rSketch map.

The specific embodiment

The specific embodiment one, a kind of method for measuring heart rate based on image capturing, it is realized by following steps:

Step 1, initialization: set and detect spatial characteristic vector T, said characteristic vector T is 3 * 1 vector; Setting is to the threshold epsilon of bianry image conversion;

Step 2, employing photographic head are taken the video of one section heart rate detection, and like the video image of human body face, hand blood vessel, this photographic head gets final product with common photographic head, satisfies FPS and gets final product greater than 15; And from video, extract each two field picture, and be labeled as F successively ₁, F ₂..., F _x, x ∈ [1, n], n is the frame number of image in this section video, F _xBe the expression of x two field picture under rgb space, that is: F _xIt is the matrix of M * N * 3;

If F _x ¹Be that x opens the gray value of image on the R space, F _x ²Be that x opens the gray value of image on the G space, F _x ³Be that x opens the gray value of image on the B space;

Use F _x ¹(i, j), F _x ²(i, j), F _x ³(i j) representes that respectively x opens the element value of image at the capable j row of the i on the R space, on the G space, on the B space, i ∈ [1, M], j ∈ [1, N];

Interval between adjacent two two field pictures does

T _vDuration for video;

Step 3, with image F _xProject to and detect in the space, obtain the gray-scale map G under the described detection of step 1 space _x, G _xMatrix for M * N; G _x(i j) is G _xIn the element value of the capable j of i row, get:

$G_x(i,j)={F_x}^1(i,j)\&times;\frac{T_{11}}{\sqrt{{T_{11}}^2+{T_{21}}^2+{T_{31}}^2}}+{F_x}^2(i,j)\&times;\frac{T_{21}}{\sqrt{{T_{11}}^2+{T_{21}}^2+{T_{31}}^2}}$

$+{F_x}^3(i,j)\&times;\frac{T_{31}}{\sqrt{{T_{11}}^2+{T_{21}}^2+{T_{31}}^2}}$

Wherein, T ₁₁, T ₂₁, T ₃₁Be respectively vector T first to the third line element;

Step 4, with every gray-scale map G of gained in the step 3 _xRespectively with first gray-scale map G ₁It is poor to do, and obtains n-1 after taking absolute value and open new difference gray-scale map R _y, y ∈ [1, n-1], R _y(i j) is R _yIn the element value of the capable j of i row, that is: R _y(i, j)=| G _y(i+1, j+1)-G ₁(i, j) |;

Step 5, extract the region of variation M of each two field picture medium vessels image _y

Step 6, extract the region of variation M of each two field picture medium vessels image in the step 5 _yImage K _y, detailed process does; Region of variation M with each the two field picture medium vessels image that obtains in the step 5 _yOne by one with step 4 in the gray-scale map R that obtains _yCarry out NAND operation, obtain K _y

K _y(i j) is K _yIn the element value of the capable j of i row, i.e. K _yIn the value of each element be: K _y(i, j)=M _y(i, j) * R _y(i, j);

Step 7, at image K _yThe middle eigenvalue that changes that extracts is designated as

Image K in step 8, the obtaining step seven _yThe middle eigenvalue that changes that extracts

Cycle T _r

Step 9, according to formula:

$\mathrm{HR}=\frac{60}{T_r}$

Obtain heart rate value HR.

Extract the region of variation M of each two field picture medium vessels image described in the step 5 _yMethod be:

Steps A 1, with the difference gray-scale map R that obtains in the step 4 ₁, R ₂..., R _N-1Being converted into ε is the binary image encoder B that Threshold Segmentation forms ₁, B ₂..., B _N-1, in the formula: B _y(i, j) expression B _yThe element value of the capable j row of i among the figure, that is:

$B_y(i,j)=\left\{\begin{array}{c}1,R_y(i,j)\&GreaterEqual;\mathrm{\&epsiv;}\\ 0,R_y(i,j)<\mathrm{\&epsiv;}\end{array}\right.;$

Steps A 2, to B in the steps A 1 _yExecution is the opening operation of structural element (diamond structure element) with Q, to remove the influence of noise of outlier, obtains C ₁, C ₂..., C _N-1, wherein:

$Q=\left(\begin{array}{ccccccc}0& 0& 0& 1& 0& 0& 0\\ 0& 0& 1& 1& 1& 0& 0\\ 0& 1& 1& 1& 1& 1& 0\\ 1& 1& 1& 1& 1& 1& 1\\ 0& 1& 1& 1& 1& 1& 0\\ 0& 0& 1& 1& 1& 0& 0\\ 0& 0& 0& 1& 0& 0& 0\end{array}\right)$

C _y＝B _yοQ；

Steps A 3: to the Elements C in the steps A 2 _yExecution is the closed operation of structural element (diamond structure element) with Q, so that with the internal image reparation of blood vessel with fill completely, obtains the region of variation M of each two field picture medium vessels image _y, that is: M _y=B _yQ.

Described in the step 7 from new image K _yThe middle method of extracting the eigenvalue that changes is:

Gained image K in step B1, the calculation procedure six _yElement in minima and maximum, be designated as K respectively _y(i, j) _MinAnd K _y(i, j) _MaxAnd with [K _y(i, j) _Min, K _y(i, j) _Max] interval on average is divided into 7 minizones, K _yElement value K in the image _y(i j) includes in said 7 minizones, that is: successively

The 1st interval range is: $\left[K_y{(i,j)}_{\mathrm{Min},}\frac{K_y{(i,j)}_{\mathrm{Max}}+6K_y{(i,j)}_{\mathrm{Min}}}{7}\right),$

The 2nd interval range is: $[\frac{K_y{(i,j)}_{\mathrm{Max}}+6K_y{(i,j)}_{\mathrm{Min}}}{7},\frac{2K_y{(i,j)}_{\mathrm{Max}}+5K_y{(i,j)}_{\mathrm{Min}}}{7}),$

The 3rd interval range is: $[\frac{2K_y{(i,j)}_{\mathrm{Max}}+5K_y{(i,j)}_{\mathrm{Min}}}{7},\frac{3K_y{(i,j)}_{\mathrm{Max}}+4K_y{(i,j)}_{\mathrm{Min}}}{7}),$

The 4th interval range is: $[\frac{3K_y{(i,j)}_{\mathrm{Max}}+4K_y{(i,j)}_{\mathrm{Min}}}{7},\frac{4K_y{(i,j)}_{\mathrm{Max}}+3K_y{(i,j)}_{\mathrm{Min}}}{7}),$

The 5th interval range is: $[\frac{4K_y{(i,j)}_{\mathrm{Max}}+3K_y{(i,j)}_{\mathrm{Min}}}{7},\frac{5K_y{(i,j)}_{\mathrm{Max}}+2K_y{(i,j)}_{\mathrm{Min}}}{7}),$

The 6th interval range is: $[\frac{5K_y{(i,j)}_{\mathrm{Max}}+2K_y{(i,j)}_{\mathrm{Min}}}{7},\frac{6K_y{(i,j)}_{\mathrm{Max}}+K_y{(i,j)}_{\mathrm{Min}}}{7}),$

The 7th interval range is: $[\frac{6K_y{(i,j)}_{\mathrm{Max}}+K_y{(i,j)}_{\mathrm{Min}}}{7},K_y{(i,j)}_{\mathrm{Max}}];$

Add up the arithmetic mean of instantaneous value of all elements in the 2nd to the 6th interval range, obtain new image K _yThe middle eigenvalue that changes that extracts is designated as

New image K in the obtaining step described in the step 8 seven _yThe middle eigenvalue that changes that extracts obtains

Cycle T _rMethod be:

Step C1, get first point in

as fiducial value, assignment is in

Step C2, establish

H ∈ [2, n-1], from h=2 begin with h one by one substitution F (h) calculate, when for the first time F (h)＜0 occurring, the h of note this moment is h ₁Continue substitution h and calculate, occur at F (h)＜0 o'clock for the second time, note h at this moment is h ₂(h then ₂-1) T _f=T _R1, T _R1Be the cycle of asking;

Step C3, get the 4th point With the 7th point

The process of repeating step C2 obtains T respectively _R2And T _R3

Step C4, when there not being corresponding h ₁, h ₂The time, then corresponding T _R1, T _R2, T _R3Be taken as 0; T _r=median (T _R1+ T _R2+ T _R3), be:

Cycle.

The duration T of the video described in the step 2 _vBe 3s.

Claims (5)

1. heart rate measurement ten thousand methods based on image capturing, it is characterized in that: it is realized by following steps:

Step 1, initialization: set and detect spatial characteristic vector T, said characteristic vector T is 3 * 1 vector; Setting is to the threshold epsilon of bianry image conversion;

Step 2, employing photographic head are taken the video of one section heart rate detection; And from video, extract each two field picture, and be labeled as F successively ₁, F ₂..., F _x, x ∈ [1, n], n is the frame number of image in this section video, F _xBe the expression of x two field picture under rgb space, that is: F _xIt is the matrix of M * N * 3;

If F _x ¹Be that x opens the gray value of image on the R space, F _x ²Be that x opens the gray value of image on the G space, F _x ³Be that x opens the gray value of image on the B space;

Use F _x ¹(i, j), F _x ²(i, j), F _x ³(i j) representes that respectively x opens the element value of image at the capable j row of the i on the R space, on the G space, on the B space, i ∈ [1, M], j ∈ [1, N];

Interval between adjacent two two field pictures does

T _vDuration for video;

Step 3, with image F _xProject to and detect in the space, obtain the gray-scale map G under the described detection of step 1 space _x, G _xMatrix for M * N; G _x(i j) is G _xIn the element value of the capable j of i row, get:

$G_x(i,j)={F_x}^1(i,j)\&times;\frac{T_{11}}{\sqrt{{T_{11}}^2+{T_{21}}^2+{T_{31}}^2}}+{F_x}^2(i,j)\&times;\frac{T_{21}}{\sqrt{{T_{11}}^2+{T_{21}}^2+{T_{31}}^2}}$

$+{F_x}^3(i,j)\&times;\frac{T_{31}}{\sqrt{{T_{11}}^2+{T_{21}}^2+{T_{31}}^2}}$

Wherein, T ₁₁, T ₂₁, T ₃₁Be respectively vector T first to the third line element;

Step 4, with every gray-scale map G of gained in the step 3 _xRespectively with first gray-scale map G ₁It is poor to do, and obtains n-1 after taking absolute value and open new difference gray-scale map R _y, y ∈ [1, n-1], R _y(i j) is R _yIn the element value of the capable j of i row, that is: R _y(i, j)=| G _y(i+1, j+1)-G ₁(i, j) |;

Step 5, extract the region of variation M of each two field picture medium vessels image _y

Step 6, extract the region of variation M of each two field picture medium vessels image in the step 5 _yImage K _y, detailed process does; Region of variation M with each the two field picture medium vessels image that obtains in the step 5 _yOne by one with step 4 in the gray-scale map R that obtains _yCarry out NAND operation, obtain K _y

K _y(i j) is K _yIn the element value of the capable j of i row, i.e. K _yIn the value of each element be: K _y(i, j)=M _y(i, j) * R _y(i, j);

Step 7, at image K _yThe middle eigenvalue that changes that extracts is designated as

Image K in step 8, the obtaining step seven _yThe middle eigenvalue that changes that extracts

Cycle T _r

Step 9, according to formula:

$\mathrm{HR}=\frac{60}{T_r}$

Obtain heart rate value HR.

2. a kind of method for measuring heart rate based on image capturing according to claim 1 is characterized in that, extracts the region of variation M of each two field picture medium vessels image described in the step 5 _yMethod be:

Steps A 1, with the difference gray-scale map R that obtains in the step 4 ₁, R ₂..., R _N-1Being converted into ε is the binary image encoder B that Threshold Segmentation forms ₁, B ₂..., B _N-1, in the formula: B _y(i, j) expression B _yThe element value of the capable j row of i among the figure, that is: $B_y(i,j)=\left\{\begin{array}{c}1,R_y(i,j)\&GreaterEqual;\mathrm{\&epsiv;}\\ 0,R_y(i,j)<\mathrm{\&epsiv;}\end{array}\right.;$

Steps A 2, to B in the steps A 1 _yExecution is the opening operation of structural element with Q, obtains C ₁, C ₂..., C _N-1, wherein:

$Q=\left(\begin{array}{ccccccc}0& 0& 0& 1& 0& 0& 0\\ 0& 0& 1& 1& 1& 0& 0\\ 0& 1& 1& 1& 1& 1& 0\\ 1& 1& 1& 1& 1& 1& 1\\ 0& 1& 1& 1& 1& 1& 0\\ 0& 0& 1& 1& 1& 0& 0\\ 0& 0& 0& 1& 0& 0& 0\end{array}\right)$

C _y＝B _yοQ；

Steps A 3: to the Elements C in the steps A 2 _yExecution is the closed operation of structural element with Q, obtains the region of variation M of each two field picture medium vessels image _y, that is: M _y=B _yQ.

3. a kind of method for measuring heart rate based on image capturing according to claim 1 is characterized in that, described in the step 7 from new image K _yThe middle method of extracting the eigenvalue that changes is:

Gained image K in step B1, the calculation procedure six _yElement in minima and maximum, be designated as K respectively _y(i, j) _MinAnd K _y(i, j) _MaxAnd with [K _y(i, j) _Min, K _y(i, j) _Max] interval on average is divided into 7 minizones, K _yElement value K in the image _y(i j) includes in said 7 minizones, that is: successively

The 1st interval range is: $[K_y{(i,j)}_{\mathrm{Min}},\frac{K_y{(i,j)}_{\mathrm{Max}}+6K_y{(i,j)}_{\mathrm{Min}}}{7}),$

The 2nd interval range is: $[\frac{K_y{(i,j)}_{\mathrm{Max}}+6K_y{(i,j)}_{\mathrm{Min}}}{7},\frac{2K_y{(i,j)}_{\mathrm{Max}}+5K_y{(i,j)}_{\mathrm{Min}}}{7}),$

The 3rd interval range is: $[\frac{2K_y{(i,j)}_{\mathrm{Max}}+5K_y{(i,j)}_{\mathrm{Min}}}{7},\frac{3K_y{(i,j)}_{\mathrm{Max}}+4K_y{(i,j)}_{\mathrm{Min}}}{7}),$

The 4th interval range is: $[\frac{3K_y{(i,j)}_{\mathrm{Max}}+4K_y{(i,j)}_{\mathrm{Min}}}{7},\frac{4K_y{(i,j)}_{\mathrm{Max}}+3K_y{(i,j)}_{\mathrm{Min}}}{7}),$

The 5th interval range is: $[\frac{4K_y{(i,j)}_{\mathrm{Max}}+3K_y{(i,j)}_{\mathrm{Min}}}{7},\frac{5K_y{(i,j)}_{\mathrm{Max}}+2K_y{(i,j)}_{\mathrm{Min}}}{7}),$

The 6th interval range is: $[\frac{5K_y{(i,j)}_{\mathrm{Max}}+2K_y{(i,j)}_{\mathrm{Min}}}{7},\frac{6K_y{(i,j)}_{\mathrm{Max}}+K_y{(i,j)}_{\mathrm{Min}}}{7}),$

The 7th interval range is: $[\frac{6K_y{(i,j)}_{\mathrm{Max}}+K_y{(i,j)}_{\mathrm{Min}}}{7},K_y{(i,j)}_{\mathrm{Max}}];$

Add up the arithmetic mean of instantaneous value of all elements in the 2nd to the 6th interval range, obtain new image K _yThe middle eigenvalue that changes that extracts is designated as

4. a kind of method for measuring heart rate based on image capturing according to claim 1 is characterized in that, new image K in the obtaining step described in the step 8 seven _yThe middle eigenvalue that changes that extracts obtains

Cycle T _rMethod be:

Step C1, get first point in

as fiducial value, assignment is in

Step C2, establish

H ∈ [2, n-1], from h=2 begin with h one by one substitution F (h) calculate, when for the first time F (h)＜0 occurring, the h of note this moment is h ₁Continue substitution h and calculate, occur at F (h)＜0 o'clock for the second time, note h at this moment is h ₂(h then ₂-1) T _f=T _R1, T _R1Be the cycle of asking;

Step C3, get the 4th point

With the 7th point

The process of repeating step C2 obtains T respectively _R2And T _R3

Step C4, when there not being corresponding h ₁, h ₂The time, then corresponding T _R1, T _R2, T _R3Be taken as 0; T _r=median (T _R1+ T _R2+ T _R3), be:

Cycle.

5. a kind of method for measuring heart rate based on image capturing according to claim 1 is characterized in that, the duration T of the video described in the step 2 _vBe 3s.

Images