CN110596705A - Human body target identity recognition method and system based on vital sign SAR imaging - Google Patents

Abstract

The invention discloses a human body target identity recognition method and a system based on vital sign SAR imaging, wherein the method comprises the following steps: collecting the respiratory waveform of a person to be identified, extracting the characteristic parameters of the respiratory waveform, inputting the respiratory waveform into a classifier for training, and obtaining an identity identification model; determining the accurate position of a human body target in a scene by utilizing a vital sign SAR imaging method based on phase demodulation; moving the radar to the position right in front of the determined human body target to acquire the respiratory waveform of the human body target; and inputting the respiratory waveform to be recognized into the identity recognition model, so as to accurately obtain the identity of the human target. The system comprises a respiratory waveform extraction module, a feature extraction module, a classification model establishing module, an imaging module, a human body target positioning module and an identity recognition module. The method has the advantages of high identity identification accuracy, good robustness and wider applicability, and can accurately obtain the position and the identity information of the human body target in the scene.

Description

Human body target identity recognition method and system based on vital sign SAR imaging

Technical Field

The invention belongs to the technical field of radars, and particularly relates to a human body target identity recognition method and system based on vital sign SAR imaging.

Background

The mobile radar platform can obtain larger synthetic aperture and better spatial resolution by the azimuth movement of the antenna, and meanwhile, the antenna used by the radar platform is much smaller than that used by the traditional beam scanning radar, so that the deployment and loading are convenient. Nowadays, the frequency modulation continuous wave radar technology is developed more and more mature, and modern frequency modulation continuous wave radar has the advantages of light weight, low cost, high resolution and the like, and has great potential in the application of the fields of earth science, security protection, vital sign monitoring and the like.

With the continuous development of radars, the demands for positioning and identity recognition of human targets in scenes are increasing. Because the method has great application prospect in the aspects of home security, user authentication, health monitoring and the like. The existing FMCW radar system is limited by precision, the specific position of a human body target cannot be directly obtained from imaging, the micro Doppler information of the target needs to be measured by combining the CW radar system, and the target position of a living body can be obtained, so that the complexity and the cost of the system are increased; at present, the research of determining the identity of a human target according to a breathing mode is not mature, and the problems of low identification accuracy, insufficient data set and the like mainly exist. Therefore, the method has a great application prospect in rapidly and accurately acquiring the position of the human target and further identifying the identity of the human target.

Disclosure of Invention

The invention aims to provide a human body target identity recognition method and system based on vital sign SAR imaging.

The technical solution for realizing the purpose of the invention is as follows: a human body target identity recognition method based on vital sign SAR imaging comprises the following steps:

step 1, statically placing an FMCW radar right ahead of a person to be identified by L meters, collecting radar echo signals, preprocessing the echo signals to obtain a respiratory waveform Re (tau)_a)；

Step 2, extracting each respiration waveform Re (tau) obtained in the step 1_a) The characteristic parameters of (1);

step 3, establishing an identity recognition model according to the characteristic parameters of the respiratory waveforms extracted in the step 2;

step 4, randomly distributing the personnel to be identified in a certain scene, operating an FMCW radar mobile platform in the scene according to a straight track to acquire echo signals, preprocessing the echo signals, and then carrying out SAR imaging to obtain an imaging result I;

step 5, in the imaging result I of the step 4, positioning the human body target position L_x；

Step 6, after the human body target position is determined, the FMCW radar is moved to the human body target azimuth position L_xCollecting radar echo signals, and preprocessing the echo signals to obtain a human target respiration waveform Rex (tau)_a)；

Step 7, the human target respiration waveform Rex (tau) to be identified in the step 6_a) And (4) inputting the identity recognition model obtained in the step (3), and recognizing the identity of the human body target to be recognized.

The system for realizing the human body target identity recognition method based on the vital sign SAR imaging comprises the following steps:

the respiratory waveform extraction module is used for collecting target respiratory waveforms of all the people to be identified;

the characteristic extraction module is used for extracting characteristic parameters of the human target respiration waveform;

the classification model establishing module is used for establishing an identity recognition model according to the characteristic parameters of the respiration waveform;

the imaging module is used for carrying out SAR imaging on a scene;

the human body target positioning module is used for acquiring the position of a human body target in the SAR imaging result;

and the identity recognition module is used for inputting the respiratory waveform of the human target collected in the scene into the identity recognition model to obtain the identity of the human target.

Compared with the prior art, the invention has the following remarkable advantages: 1) according to the invention, the scene area is subjected to the vital sign SAR imaging through the FMCW mobile platform, so that the position of the human body target is determined, the position of the human body target does not need to be known in advance, the limitation is overcome, and the applicable scene is wider; 2) the invention utilizes the radar sensor to obtain the breathing signal of the human target, compared with the traditional contact type monitoring, the device is simple and easy to operate, and can reduce the uncomfortable feeling of the human body, thereby not influencing the breathing of experimenters in the aspect of experiments, reducing the measurement error, and overcoming a plurality of limitations, such as the contact piece can not be directly contacted due to the burning or other factors on the contact part required; 3) according to the invention, proper characteristic parameters are selected to represent different personal identities, a model capable of identifying the human target identity is trained by a machine learning method, the final identification result is better, and the accuracy is higher; 4) the method is simple and effective, the equipment is simple and easy to realize, the cost is low, the operation is easy, and the performance is reliable.

The invention is further described below with reference to the accompanying drawings.

Drawings

Fig. 1 is a flow chart of a human body target identity recognition method based on vital sign SAR imaging.

FIG. 2 is a diagram illustrating dynamic fractional area ratio characteristics of a respiratory waveform in an embodiment of the present invention.

FIG. 3 is a schematic diagram of the breathing characteristics of an embodiment of the present invention when the lungs are full.

FIG. 4 is a diagram illustrating an SVM confusion matrix according to an embodiment of the present invention.

FIG. 5 is a design drawing of an experimental scenario in an embodiment of the present invention.

FIG. 6 is a BP algorithm imaging graph in an embodiment of the present invention.

Fig. 7 is a schematic diagram of a phase signal after unwrapping according to an embodiment of the present invention.

Fig. 8 is a schematic diagram of the unwrapped phase signal after 10 th order linear fit in an embodiment of the present invention.

FIG. 9 is a diagram of a raw respiration signal in an embodiment of the invention.

FIG. 10 is a schematic diagram of a smoothed respiration signal according to an embodiment of the invention.

Fig. 11 is a diagram of the final positioning and recognition result in the embodiment of the present invention.

Detailed Description

With reference to fig. 1, the human body target identity recognition method based on vital sign SAR imaging of the present invention comprises the following steps:

step 1, statically placing an FMCW radar right ahead of a person to be identified by L meters, collecting radar echo signals, preprocessing the echo signals to obtain a respiratory waveform Re (tau)_a)；

Step 2, extracting each respiration waveform Re (tau) obtained in the step 1_a) The characteristic parameters of (1);

step 3, establishing an identity recognition model according to the characteristic parameters of the respiratory waveforms extracted in the step 2;

step 4, randomly distributing the personnel to be identified in a certain scene, operating an FMCW radar mobile platform in the scene according to a straight track to acquire echo signals, preprocessing the echo signals, and then carrying out SAR imaging to obtain an imaging result I;

step 5, in the imaging result I of the step 4, positioning the human body target position L_x；

Step 6, after the human body target position is determined, the FMCW radar is moved to the human body target azimuth position L_xCollecting radar echo signals, and preprocessing the echo signals to obtain a human target respiration waveform Rex (tau)_a)；

Step 7, the human target respiration waveform Rex (tau) to be identified in the step 6_a) And (4) inputting the identity recognition model obtained in the step (3), and recognizing the identity of the human body target to be recognized.

Further, step 1, statically placing the FMCW radar right ahead of the person to be identified by L meters, collecting radar echo signals, preprocessing the echo signals to obtain a respiration waveform Re (tau)_a) The method specifically comprises the following steps:

step 1-1, statically placing an FMCW radar at a position L meters right ahead of a person to be identified, and collecting a radar echo signal s (t);

step 1-2, preprocessing the echo signal s (t), specifically:

the echo signal s (t) is:

in the formula, s_TIs a signal to be transmitted by the radar,being complex conjugates of the received signal of the radar, A₁Is a signal s_TAmplitude of (A)₂Is a signalK represents the chirp rate, λ represents the wavelength, R represents the distance of the target from the radar, t represents the "fast time" of wave propagation, c represents the speed of wave propagation;

fourier transforming the echo signal s (t):

in the formula, A₁A₂T represents a waveform period as the amplitude of the signal; f represents the range-wise frequency;

step 1-3, constructing a radar echo matrix according to echo signals after Fourier transform, specifically: according to Fourier transformEcho signal construction M_p×N_pThe radar echo matrix is in a row distance direction, each row stores echo signals obtained in a pulse repetition period, and the rows are in an azimuth direction; wherein M is_pRepresents M_pOne pulse repetition period, N_pIndicating N within a pulse repetition period_pA sampling point N_p；

Step 1-4, extracting echo segments SH in a range gate at a distance L meters from a radar in a radar echo matrix₁For the echo segment SH₁Performing phase unwrapping to obtain the distance R between the target and the radar₁：

Wherein ang is the phase angle after the phase angle alpha is unwound,

wherein the phase anglea is the echo segment SH₁The real part of the medium complex number element, b is the echo segment SH₁The imaginary part of the medium complex number element;

step 1-5, obtaining the respiration waveform Re (tau) of the human body target_a) The method specifically comprises the following steps: distance R between target and radar₁Slow time tau corresponding to each azimuth sample_aObtaining R₁(τ_a)，R₁(τ_a) Including the direct distance R of the radar from the target₀And the respiration waveform Re (tau) of the human target_a) Thus:

Re(τ_a)＝R₁(τ_a)-R₀。

further, step 2 extracts each respiration waveform Re (τ) obtained in step 1_a) The characteristic parameters are as follows:

step 2-1, calculating the respiratory frequency Bf, which specifically comprises the following steps: for respiration wave Re (tau)_a) Performing FFT to obtain the respiratory frequency Bf;

step 2-2, calculating average expiration starting time period T_exThe method specifically comprises the following steps:

step 2-2-1, obtaining respiration waveform Re (tau)_a) Set of displacement peaks of (2):

P_ex＝[p_x1 p_x2 p_x3 p_x4 … … p_xN]

in the formula, p_xiIs the peak value of the ith respiration waveform, and N represents N respiration waveforms;

step 2-2-2, calculating the expiration starting time period T according to the time index corresponding to the displacement peak value_exiComprises the following steps:

T_exi＝tp_x(i+1)-tp_x(i)

in the formula, tp_xiIs the peak value p of the ith respiration waveform_xiA corresponding time index;

step 2-2-3, calculating average expiration start time period T_exComprises the following steps:

step 2-3, calculating the average breathing depth, which specifically comprises the following steps:

step 2-3-1, obtaining a respiration waveform Re (tau)_a) Set of nadirs of (a):

N_in＝[n_x1 n_x2 n_x3 n_x4 ... n_xN]

in the formula, n_xiIs the lowest point of the ith respiratory waveform, and N represents N respiratory waveforms;

step 2-3-2, calculating the depth of breath d_xiComprises the following steps:

d_xi＝p_xi-n_xi

step 2-3-3, solving the average depth of breath depth as:

step 2-4, calculating average expiration starting time period T_exThe standard deviation of (a) is specifically:

step 2-5, calculating an average inspiration starting time period Tin, specifically:

step 2-5-1, calculating an inspiration starting time period T according to a time index corresponding to the lowest point of the respiration waveform_iniComprises the following steps:

T_ini＝tn_x(i+1)-tn_xi

in the formula, tn_xiIs the lowest point n of the ith respiration waveform_xiA corresponding time index;

step 2-5-2, calculating the average inspiration starting time period Tin as:

step 2-6, calculating the standard deviation of the average inspiration starting time period TinThe method specifically comprises the following steps:

step 2-7, calculating average expiratory velocity v_exiThe method specifically comprises the following steps:

step 2-8, calculating average air suction speed v_iniThe method specifically comprises the following steps:

step 2-9, calculating the average area ratio, specifically:

step 2-9-1, extracting points with displacement value of q% of peak value in the respiratory segment as follows:

p_0.qi＝0.q*[p_x1 p_x2 p_x3 p_x4 ... ... p_xN]

extracting points in the respiratory segment with a displacement value s% higher than the lowest point:

n_1.si＝1.s*[n_x1 n_x2 n_x3 n_x4 … n_xN]

step 2-9-2, obtaining the expiratory-inspiratory area A-ex according to the step 2-9-1_iThe four boundary points of (2) are: [ p ]_0.qi n_1.sip_0.qi n_1.si+1](ii) a Inspiratory-expiratory area A-in_iThe four boundary points of (2) are: [ p ]_0.qi n_1.si+1 p_0.qi+1 n_1.si+1]；

Step 2-9-3, calculating the area ratio, and solving the mean value r1 as follows:

step 2-10, solving the mean r2 of Euclidean distance ratio of 1 second before and after each respiratory waveform peak value as:

where da represents the euclidean distance from data point to vertex one second after the vertex appears and dp represents the euclidean distance from data point to vertex one second before the vertex appears.

Exemplary preferably, q% ═ 70% in step 2-9-1; and s% ═ 30%.

Further, step 3 establishes an identity recognition model according to the characteristic parameters of the respiration waveform extracted in step 2, specifically: will respiration wave form Re (tau)_a) And (3) as samples, inputting the characteristics of each sample and the corresponding identity label into a classifier for training to obtain an identity recognition model.

Further, in step 4, the FMCW radar mobile platform operates in the scene according to a straight track to acquire an echo signal, and performs SAR imaging after preprocessing the echo signal to obtain an imaging result I, which specifically is as follows:

step 4-1, the FMCW radar runs along a straight track to collect an original echo signal s' (t), the running direction is the azimuth direction, the irradiation direction of the FMCW radar is perpendicular to the running track, and the irradiation direction is the distance direction;

step 4-2, performing Fourier transform on the original echo signal s' (t) and then constructing a radar echo matrix, wherein the process is the same as that of the steps 1-2 and 1-3;

and 4-3, carrying out SAR imaging on the radar echo matrix in the step 4-2 by utilizing a back projection BP algorithm to obtain an imaging result I.

Further, step 5 in the imaging result I of step 4, the human target position L is located_xThe method specifically comprises the following steps:

step 5-1, acquiring all points with the maximum local energy, namely all targets, in an imaging graph I;

step 5-2, positioning the azimuth position L of each target_nThe method specifically comprises the following steps: aiming at each target, the row coordinate value and the azimuth resolution rho of the corresponding local energy maximum point are determined_aAs the product of the azimuth position L of the target_n(ii) a Wherein the azimuthal resolution ρ_aComprises the following steps:

in the formula, L_aIs the true aperture of the antenna;

step 5-3, locating the distance position H of each target_nI.e. the distance door H where the target is located_nThe method specifically comprises the following steps: aiming at each target, the column coordinate value and the range resolution rho of the corresponding local energy maximum point are used_rAs the product of the distance to the target position H_n(ii) a Wherein the distance resolution ρ_rComprises the following steps:

wherein c is the speed of light, B_pIs the FMCW radar signal bandwidth;

step 5-4, extracting a distance gate H where the target is located_nAnd (3) carrying out phase unwrapping on the echo segment SHn after internal preprocessing to obtain the distance R between the target and the radar:

wherein ang is the phase angle after the phase angle alpha is unwound,

wherein the phase anglea is the real part of the complex element in the echo segment SHn, and b is the imaginary part of the complex element in the echo segment SHn;

step 5-5, performing fitting removal processing on the distance R between the target and the radar in the step 5-4 to obtain micro Doppler information of the target, specifically:

step 5-5-1, corresponding the distance R between the target and the radar to the slow time tau of each azimuth sampling_aObtaining R (tau)_a)，R(τ_a) Involving a distance R generated by radar motion_x(τ_a) And micro-Doppler information Re' (τ) of the target_a) (ii) a Wherein:

in the formula, R₀Indicates the distance length, tau, of the radar and the target when the radar is right in front of the target_aRepresents the slow time of the azimuth motion, v represents the velocity of the radar motion;

step 5-5-2, all R (tau) corresponding to the distance R between the target and the radar_a) Obtaining a linear fit using the principle of least squares of deviationCurve function fit (τ)_a) And is composed of fit (τ)_a) In place of R_x(τ_a)；

Step 5-5-3, based on R (τ)_a) And fit (τ)_a) Obtaining target micro Doppler information Re' (tau)_a) Comprises the following steps:

Re'(τ_a)＝R(τ_a)-fit(τ_a)

step 5-6, micro Doppler information Re' (tau) to the target_a) And carrying out smoothing treatment by using a smoothing formula as follows:

in the formula, Res (τ)_a)_mM is the target micro Doppler information Re' (τ) for the smoothed target micro Doppler waveform_a) The mth sampling point of the middle azimuth slow time, M ∈ [1, M >_n]；M_nCounting the number of sampling points; re' (τ)_a)_mIs target micro Doppler information Re' (tau)_a) The m-th sampling point of (a); re' (τ)_a)_m+kIs target micro Doppler information Re' (tau)_a) The m + k th sampling point of (a); k is the weighted point number of each sampling point, and the smooth point number can be set by setting the value of K;

step 5-7, assuming that the radar moves to the azimuth position L where the target is located_nAt a time t_aAccording to t_aTime Res (τ)_a)_mAnd (3) judging and positioning a human body target by using a waveform: if at t_aRes (τ) around time_a)_mIf regular waveforms exist in the waveforms, the target is indicated to be a human target, and the azimuth position L corresponding to the target is recorded_x(ii) a If at t_aRes (τ) around time_a)_mThe waveform does not have great difference with the waveforms at other moments, which indicates that the target is a non-living object.

Further, after the human target position is determined in the step 6, the FMCW radar is moved to the human target azimuth position L_xCollecting radar echo signals, preprocessing the echo signals to obtain respiration waveforms Rex (tau)_a) The method specifically comprises the following steps:

step 6-1, moving the FMCW radar to the azimuth position L of the target_xAcquiring echo data;

6-2, after Fourier transformation is carried out on the echo data collected in the step 6-1, a radar echo matrix is constructed, and the specific process is the same as that in the step 1-3;

6-3, positioning the distance direction position Hx of the human body target, namely the distance door Hx where the human body target is located, specifically: the column coordinate value of the energy maximum point in the radar echo matrix in the step 6-2 and the distance direction resolution rho_rThe product of the distance position Hx and the distance position Hx of the human target;

step 6-4, extracting the preprocessed echo segment SHx in the range gate Hx where the human body target is located, performing phase unwrapping on the echo segment SHx, and obtaining the distance R between the human body target and the radar_x：

Step 6-5, obtaining the respiratory waveform Rex (tau) of the human body target_a) The method specifically comprises the following steps: distance R between target and radar_xSlow time tau corresponding to each azimuth sample_aObtaining R_x(τ_a)，R_x(τ_a) Including the direct distance R of the radar from the target₀And the respiratory waveform Rex (tau) of the human target_a) And then:

Rex(τ_a)＝R_x(τ_a)-R₀。

the system for realizing the human body target identity recognition method based on the vital sign SAR imaging comprises the following steps:

the respiratory waveform extraction module is used for collecting target respiratory waveforms of all the people to be identified;

the characteristic extraction module is used for extracting characteristic parameters of the human target respiration waveform;

the classification model establishing module is used for establishing an identity recognition model according to the characteristic parameters of the respiration waveform;

the imaging module is used for carrying out SAR imaging on a scene;

the human body target positioning module is used for acquiring the position of a human body target in the SAR imaging result;

and the identity recognition module is used for inputting the respiratory waveform of the human target collected in the scene into the identity recognition model to obtain the identity of the human target.

The present invention will be described in further detail with reference to examples.

Examples

With reference to fig. 1, the human body target identity recognition method based on vital sign SAR imaging of the present invention includes the following contents:

1. in this embodiment, an FMCW radar system based on a mobile platform is adopted, the carrier frequency of the radar system is 5.8GHz, the bandwidth of the transmitted signal is 320MHz, the gain of the antenna is 11.3dB, the half-power angle is 46 °, and the sampling frequency is 192 kHz.

2. The FMCW radar is placed at a position 1 meter in front of an experimenter statically, respiratory waveform data of 10 experimenters are collected, each pack of respiratory data is 30 seconds, and each experimenter collects 50 packs of 500 packs of data.

3. And extracting characteristic parameters of the respiration waveform, wherein the dynamic segmentation area ratio characteristic of the respiration waveform is shown in figure 2, and the respiration characteristic of one second before and after the lung full-sucking is shown in figure 3.

4. Labeling 1-10 labels on 10 experimenters respectively, then sending the extracted characteristic parameters and the corresponding labels into a classifier for training to obtain an identity recognition model, wherein a classification result confusion matrix is shown in figure 4.

5. Setting a scene as a rectangular area with the azimuth length of 3.6m and the distance length of 5m, and placing two targets in the scene area, wherein the target L₁As a reflector, an object L₂The target is an experimenter with the identity of 1, the target is 1.2m and 2.4m away from the left edge of the scene, the distance from the lower edge of the scene is 1m, and the specific experimental scene design is shown in fig. 5.

6. Let the edge (azimuth) under the scene that the radar prolongs 3.6m long move with speed v ═ 0.2m/s from left to right, use B after preprocessing echo dataThe P algorithm performs SAR imaging on the echo matrix, and the imaging result I is shown in FIG. 6, in which two targets L are shown₁And L₂。

7. Firstly, because the distance direction distances of two targets are the same, the distance door H where any target is located is extracted₁The echo segment SH1 is processed to obtain the distance R (tau) between the radar and the target_a) As shown in fig. 7; secondly, to eliminate the distance information R generated by the radar motion_x(τ_a) Calculating a fitting function fit (tau) of 10 th order_a) The results are shown in FIG. 8; finally, with R (τ)_a) Subtract the fitting function fit (τ)_a) Thus obtaining the micro Doppler information Re' (tau) of the target_a) As shown in fig. 9; to Re' (tau)_a) Performing 250-point smoothing to obtain a smoothed target micro Doppler waveform Res (tau)_a) As shown in fig. 10; from fig. 10 it can be seen that the regular vibration signal is clearly seen in the dashed line box 12 to 27 seconds, with an amplitude 10 times greater than the first 12 seconds, and the time 12 to 27 seconds, it is the radar that moves to the target L₂Time of the neighborhood, that is, target L₂Is the human target in fig. 6.

8. Moving FMCW to a human target L₂The respiratory waveform of a stationary measurement target.

9. The measured human body target L₂The respiratory waveform is input into an identity recognition model, and a human body target L can be obtained₂The identity of (c). The final recognition result of this embodiment is shown in FIG. 11, where the human target L₂Is an experimenter with a label of 1.

The method comprises the steps of collecting physiological signals through a radar sensor, extracting characteristic parameters of a human target respiration waveform, training an identity recognition model, positioning the human target by using a vital sign SAR imaging method based on phase demodulation, extracting the characteristic parameters of the human target respiration waveform, sending the characteristic parameters into the identity recognition model, and recognizing the identity of the human target. The human body target positioning method has the advantages of good human body target positioning effect, high identity identification accuracy, good robustness and wider applicability.

Claims (9)

1. A human body target identity recognition method based on vital sign SAR imaging is characterized by comprising the following steps:

step 1, statically placing an FMCW radar right ahead of a person to be identified by L meters, collecting radar echo signals, preprocessing the echo signals to obtain a respiratory waveform Re (tau)_a)；

Step 2, extracting each respiration waveform Re (tau) obtained in the step 1_a) The characteristic parameters of (1);

step 3, establishing an identity recognition model according to the characteristic parameters of the respiratory waveforms extracted in the step 2;

step 4, randomly distributing the personnel to be identified in a certain scene, operating an FMCW radar mobile platform in the scene according to a straight track to acquire echo signals, preprocessing the echo signals, and then carrying out SAR imaging to obtain an imaging result I;

step 5, in the imaging result I of the step 4, positioning the human body target position L_x；

Step 6, after the human body target position is determined, the FMCW radar is moved to the human body target azimuth position L_xCollecting radar echo signals, and preprocessing the echo signals to obtain a human target respiration waveform Rex (tau)_a)；

Step 7, the human target respiration waveform Rex (tau) to be identified in the step 6_a) And (4) inputting the identity recognition model obtained in the step (3), and recognizing the identity of the human body target to be recognized.

2. The method for identifying the identity of a human target based on the SAR imaging of vital signs according to claim 1, wherein step 1 comprises statically placing the FMCW radar right in front of the person to be identified by L meters, collecting radar echo signals, and preprocessing the echo signals to obtain the respiration waveform Re (τ) (I)_a) The method specifically comprises the following steps:

step 1-1, statically placing an FMCW radar at a position L meters right ahead of a person to be identified, and collecting a radar echo signal s (t);

step 1-2, preprocessing the echo signal s (t), specifically:

the echo signal s (t) is:

in the formula, s_TIs a signal to be transmitted by the radar,being complex conjugates of the received signal of the radar, A₁Is a signal s_TAmplitude of (A)₂Is a signalK represents the chirp rate, λ represents the wavelength, R represents the distance of the target from the radar, t represents the "fast time" of wave propagation, c represents the speed of wave propagation;

fourier transforming the echo signal s (t):

in the formula, A₁A₂T represents a waveform period as the amplitude of the signal; f represents the range-wise frequency;

step 1-3, constructing a radar echo matrix according to echo signals after Fourier transform, specifically: construction of M from Fourier transformed echo signals_p×N_pThe radar echo matrix is in a row distance direction, each row stores echo signals obtained in a pulse repetition period, and the rows are in an azimuth direction; wherein M is_pRepresents M_pOne pulse repetition period, N_pIndicating N within a pulse repetition period_pA sampling point N_p；

Step 1-4, extracting echo segments SH in a range gate at a distance L meters from a radar in a radar echo matrix₁For the echo segment SH₁Performing phase unwrapping to obtain the distance R between the target and the radar₁：

Wherein ang is the phase angle after the phase angle alpha is unwound,

wherein the phase anglea is the echo segment SH₁The real part of the medium complex number element, b is the echo segment SH₁The imaginary part of the medium complex number element;

step 1-5, obtaining the respiration waveform Re (tau) of the human body target_a) The method specifically comprises the following steps: distance R between target and radar₁Slow time tau corresponding to each azimuth sample_aObtaining R₁(τ_a)，R₁(τ_a) Including the direct distance R of the radar from the target₀And the respiration waveform Re (tau) of the human target_a) Thus:

Re(τ_a)＝R₁(τ_a)-R₀。

3. the method for identifying the identity of a human target based on SAR imaging of vital signs according to claim 1, wherein step 2 extracts each respiration waveform Re (τ) obtained in step 1_a) The characteristic parameters are as follows:

step 2-1, calculating the respiratory frequency Bf, which specifically comprises the following steps: for respiration wave Re (tau)_a) Performing FFT to obtain the respiratory frequency Bf;

step 2-2, calculating average expiration starting time period T_exThe method specifically comprises the following steps:

step 2-2-1, obtaining respiration waveform Re (tau)_a) Set of displacement peaks of (2):

P_ex＝[p_x1 p_x2 p_x3 p_x4 … … p_xN]

in the formula, p_xiFor the ith callThe peak value of the inhalation waveform, N represents N respiration waveforms;

step 2-2-2, calculating the expiration starting time period T according to the time index corresponding to the displacement peak value_exiComprises the following steps:

T_exi＝tp_x(i+1)-tp_x(i)

in the formula, tp_xiIs the peak value p of the ith respiration waveform_xiA corresponding time index;

step 2-2-3, calculating average expiration start time period T_exComprises the following steps:

step 2-3, calculating the average breathing depth, which specifically comprises the following steps:

step 2-3-1, obtaining a respiration waveform Re (tau)_a) Set of nadirs of (a):

N_in＝[n_x1 n_x2 n_x3 n_x4 … n_xN]

in the formula, n_xiIs the lowest point of the ith respiratory waveform, and N represents N respiratory waveforms;

step 2-3-2, calculating the depth of breath d_xiComprises the following steps:

d_xi＝p_xi-n_xi

step 2-3-3, solving the average depth of breath depth as:

step 2-4, calculating average expiration starting time period T_exThe standard deviation of (a) is specifically:

step 2-5, calculating an average inspiration starting time period Tin, specifically:

step 2-5-1, corresponding to the lowest point of the respiration waveformTime indexing the inspiration onset time period T_iniComprises the following steps:

T_ini＝tn_x(i+1)-tn_xi

in the formula, tn_xiIs the lowest point n of the ith respiration waveform_xiA corresponding time index;

step 2-5-2, calculating the average inspiration starting time period Tin as:

step 2-6, calculating the standard deviation of the average inspiration starting time period TinThe method specifically comprises the following steps:

step 2-7, calculating average expiratory velocity v_exiThe method specifically comprises the following steps:

step 2-8, calculating average air suction speed v_iniThe method specifically comprises the following steps:

step 2-9, calculating the average area ratio, specifically:

step 2-9-1, extracting points with displacement value of q% of peak value in the respiratory segment as follows:

p_0.qi＝0.q*[p_x1 p_x2 p_x3 p_x4 … … p_xN]

extracting points in the respiratory segment with a displacement value s% higher than the lowest point:

n_1.si＝1.s*[n_x1 n_x2 n_x3 n_x4 … n_xN]

step 2-9-2, obtaining the expiration-inspiration area A _ ex according to the step 2-9-1_iThe four boundary points of (2) are: [ p ]_0.qi n_1.si p_0.qin_1.si+1](ii) a Inspiratory-expiratory area A _ in_iThe four boundary points of (2) are: [ p ]_0.qi n_1.si+1 p_0.qi+1 n_1.si+1]；

Step 2-9-3, calculating the area ratio, and solving the mean value r1 as follows:

step 2-10, solving the mean r2 of Euclidean distance ratio of 1 second before and after each respiratory waveform peak value as:

where da represents the euclidean distance from data point to vertex one second after the vertex appears and dp represents the euclidean distance from data point to vertex one second before the vertex appears.

4. The human target identification method based on vital sign SAR imaging according to claim 3, characterized in that, in step 2-9-1, the q% ═ 70%; and s% ═ 30%.

5. The method for identifying the identity of the human target based on the vital sign SAR imaging according to claim 2 or 3, wherein the step 3 of establishing the identity identification model according to the characteristic parameters of the respiration waveform extracted in the step 2 specifically comprises: will respiration wave form Re (tau)_a) And (3) as samples, inputting the characteristics of each sample and the corresponding identity label into a classifier for training to obtain an identity recognition model.

6. The method for identifying the identity of a human target based on the vital sign SAR imaging according to claim 1 or 2, wherein the FMCW radar mobile platform in step 4 operates in the scene according to a straight track to acquire an echo signal, and performs SAR imaging after preprocessing the echo signal to obtain an imaging result I, specifically:

step 4-1, the FMCW radar runs along a straight track to collect an original echo signal s' (t), the running direction is the azimuth direction, the irradiation direction of the FMCW radar is perpendicular to the running track, and the irradiation direction is the distance direction;

step 4-2, performing Fourier transform on the original echo signal s' (t) and then constructing a radar echo matrix, wherein the process is the same as that of the steps 1-2 and 1-3;

and 4-3, carrying out SAR imaging on the radar echo matrix in the step 4-2 by utilizing a back projection BP algorithm to obtain an imaging result I.

7. The method for identifying the identity of a human target based on SAR imaging of vital signs according to claim 6, wherein the step 5 locates the target position L of the human body in the imaging result I of the step 4_xThe method specifically comprises the following steps:

step 5-1, acquiring all points with the maximum local energy, namely all targets, in an imaging graph I;

step 5-2, positioning the azimuth position L of each target_nThe method specifically comprises the following steps: aiming at each target, the row coordinate value and the azimuth resolution rho of the corresponding local energy maximum point are determined_aAs the product of the azimuth position L of the target_n(ii) a Wherein the azimuthal resolution ρ_aComprises the following steps:

in the formula, L_aIs the true aperture of the antenna;

step 5-3, locating the distance position H of each target_nI.e. the distance door H where the target is located_nThe method specifically comprises the following steps: aiming at each target, the column coordinate value and the range resolution rho of the corresponding local energy maximum point are used_rAs the product of the distance to the target position H_n(ii) a Wherein the distance directionResolution ρ_rComprises the following steps:

wherein c is the speed of light, B_pIs the FMCW radar signal bandwidth;

step 5-4, extracting a distance gate H where the target is located_nAnd (3) carrying out phase unwrapping on the echo segment SHn after internal preprocessing to obtain the distance R between the target and the radar:

wherein ang is the phase angle after the phase angle alpha is unwound,

wherein the phase anglea is the real part of the complex element in the echo segment SHn, and b is the imaginary part of the complex element in the echo segment SHn;

step 5-5, performing fitting removal processing on the distance R between the target and the radar in the step 5-4 to obtain micro Doppler information of the target, specifically:

step 5-5-1, corresponding the distance R between the target and the radar to the slow time tau of each azimuth sampling_aObtaining R (tau)_a)，R(τ_a) Involving a distance R generated by radar motion_x(τ_a) And micro-Doppler information Re' (τ) of the target_a) (ii) a Wherein:

in the formula, R₀Indicates the distance length, tau, of the radar and the target when the radar is right in front of the target_aIndicating the slow time of the azimuthal movement,v represents the velocity of the radar motion;

step 5-5-2, all R (tau) corresponding to the distance R between the target and the radar_a) Obtaining a linear fitting curve function fit (tau) by using the principle of minimum deviation sum of squares_a) And is composed of fit (τ)_a) In place of R_x(τ_a)；

Step 5-5-3, based on R (τ)_a) And fit (τ)_a) Obtaining target micro Doppler information Re' (tau)_a) Comprises the following steps:

Re'(τ_a)＝R(τ_a)-fit(τ_a)

step 5-6, micro Doppler information Re' (tau) to the target_a) And carrying out smoothing treatment by using a smoothing formula as follows:

in the formula, Res (τ)_a)_mM is the target micro Doppler information Re' (τ) for the smoothed target micro Doppler waveform_a) The mth sampling point of the middle azimuth slow time, M ∈ [1, M >_n]；M_nCounting the number of sampling points; re' (τ)_a)_mIs target micro Doppler information Re' (tau)_a) The m-th sampling point of (a); re' (τ)_a)_m+kIs target micro Doppler information Re' (tau)_a) The m + k th sampling point of (a); k is the weighted point number of each sampling point, and the smooth point number can be set by setting the value of K;

step 5-7, assuming that the radar moves to the azimuth position L where the target is located_nAt a time t_aAccording to t_aTime Res (τ)_a)_mAnd (3) judging and positioning a human body target by using a waveform: if at t_aRes (τ) around time_a)_mIf regular waveforms exist in the waveforms, the target is indicated to be a human target, and the azimuth position L corresponding to the target is recorded_x(ii) a If at t_aRes (τ) around time_a)_mThe waveform does not have great difference with the waveforms at other moments, which indicates that the target is a non-living object.

8. The method for identifying the identity of a human target based on SAR imaging of vital signs according to claim 7, wherein the FMCW radar is moved to the target direction L after the target position of the human body is determined in step 6_xCollecting radar echo signals, preprocessing the echo signals to obtain respiration waveforms Rex (tau)_a) The method specifically comprises the following steps:

step 6-1, moving the FMCW radar to the azimuth position L of the target_xAcquiring echo data;

6-2, after Fourier transformation is carried out on the echo data collected in the step 6-1, a radar echo matrix is constructed, and the specific process is the same as that in the step 1-3;

6-3, positioning the distance direction position Hx of the human body target, namely the distance door Hx where the human body target is located, specifically: the column coordinate value of the energy maximum point in the radar echo matrix in the step 6-2 and the distance direction resolution rho_rThe product of the distance position Hx and the distance position Hx of the human target;

step 6-4, extracting the preprocessed echo segment SHx in the range gate Hx where the human body target is located, performing phase unwrapping on the echo segment SHx, and obtaining the distance R between the human body target and the radar_x：

Step 6-5, obtaining the respiratory waveform Rex (tau) of the human body target_a) The method specifically comprises the following steps: distance R between target and radar_xSlow time tau corresponding to each azimuth sample_aObtaining R_x(τ_a)，R_x(τ_a) Including the direct distance R of the radar from the target₀And the respiratory waveform Rex (tau) of the human target_a) And then:

Rex(τ_a)＝R_x(τ_a)-R₀。

9. a system for implementing the method for identifying human target identity based on vital sign SAR imaging according to any one of claims 1 to 8, comprising:

the respiratory waveform extraction module is used for collecting target respiratory waveforms of all the people to be identified;

the characteristic extraction module is used for extracting characteristic parameters of the human target respiration waveform;

the classification model establishing module is used for establishing an identity recognition model according to the characteristic parameters of the respiration waveform;

the imaging module is used for carrying out SAR imaging on a scene;

the human body target positioning module is used for acquiring the position of a human body target in the SAR imaging result;

and the identity recognition module is used for inputting the respiratory waveform of the human target collected in the scene into the identity recognition model to obtain the identity of the human target.