A method, system and computer device for recognizing vital signs of multiple human targets

Abstract

The present application proposes a method, system and computer device for identifying the vital signs of multiple human targets, the method comprising: using an ultra-wideband radar to detect the area to be detected to obtain a radar echo matrix corresponding to the area, and filtering out static clutter in the radar echo matrix; calculating the permutation entropy value of the slow time series corresponding to each range gate in the radar echo matrix; determining the initial range gate area where the human target is located according to each permutation entropy value; determining the number of human targets and the position of each human target according to the initial range gate area where the human target is located; according to the position of each human target, extracting the body surface vibration signal of each human target from the radar echo matrix from which static clutter is filtered out and identifying each human target respectively. The technical solution provided by the present invention uses the KMeans++ clustering algorithm and the HOC algorithm to identify the vital signs of multiple human targets, thereby improving the efficiency, accuracy and intelligence of identification, thereby providing a guarantee for timely and effective rescue.

Description

Technical Field

[0001] The present application relates to the technical field of human life target recognition, and in particular to a method, system and computer equipment for recognizing multiple human life signs. Background Art

[0002] Life detection radar is a high-tech life-saving device developed by combining micropower ultra-wideband radar technology with biomedical engineering techniques. It is specifically designed to locate survivors in emergency rescue missions such as earthquakes and landslides. It utilizes the wide spectrum, strong penetration, high resolution, excellent anti-interference performance, and low power consumption of nanosecond electromagnetic pulses. It can penetrate non-metallic media and transmit nanosecond pulsed electromagnetic waves over long distances. The radar uses the time-domain Doppler effect caused by human motion on the radar echo to analyze the presence of human targets within the detection range, as well as vital signs and specific location information.

[0003] Existing research mainly focuses on the positioning of single human targets and the recognition algorithms of breathing / heartbeat characteristics. It is unable to identify vital signs when there are multiple human targets in the detection range, which in turn affects the timeliness and accuracy of rescue. Summary of the Invention

[0004] The present application provides a method, system and computer device for recognizing the vital signs of multiple human targets, so as to at least solve the technical problems in the related art of being unable to recognize multiple human targets and inaccurate recognition.

[0005] The first embodiment of the present application provides a method for identifying vital signs of multiple human targets, including:

[0006] Using an ultra-wideband radar to detect an area to be detected to obtain a radar echo matrix corresponding to the area, and filtering out static clutter in the radar echo matrix;

[0007] Calculate the permutation entropy value of the slow time series corresponding to each range gate in the radar echo matrix;

[0008] Determine an initial range gate area where a human target is located in the area to be detected according to each of the permutation entropy values;

[0009] Determine the number of human targets and the position of each human target according to the initial range gate area where each human target is located in the detected area;

[0010] According to the position of each human target, the body surface vibration signal of each human target is extracted from the radar echo matrix after filtering out static clutter, and each human target is identified respectively.

[0011] The second embodiment of the present application provides a multi-human target vital sign recognition system, comprising:

[0012] A filtering module is used to detect the area to be detected using an ultra-wideband radar to obtain a radar echo matrix corresponding to the area, and to filter out static clutter in the radar echo matrix;

[0013] The calculation module is used to calculate the permutation entropy value of the slow time series corresponding to each range gate in the radar echo matrix;

[0014] A first determining module is used to determine an initial range gate area where a human target is located in the detection area according to each of the permutation entropy values;

[0015] A second determining module is used to determine the number of human targets and the position of each human target according to the initial range gate area where each human target is located in the detected area;

[0016] The identification module is used to extract the body surface vibration signal of each human target from the radar echo matrix after filtering out static clutter according to the position of each human target and identify each human target respectively.

[0017] The third aspect of the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the identification method of the first aspect of the present application is implemented.

[0018] The technical solutions provided by the embodiments of this application bring at least the following beneficial effects:

[0019] The present application proposes a method, system, and computer device for identifying the vital signs of multiple human targets. The method includes: using an ultra-wideband radar to detect an area to be detected to obtain a radar echo matrix corresponding to the area, and filtering out static clutter in the radar echo matrix; calculating the permutation entropy value of the slow time series corresponding to each range gate in the radar echo matrix; determining the initial range gate area where the human targets in the area to be detected are located based on each permutation entropy value; determining the number of human targets and the position of each human target based on the initial range gate area where each human target in the area to be detected is located; and extracting the surface vibration signal of each human target from the radar echo matrix from which static clutter has been filtered out, and identifying each human target separately, based on the position of each human target. The technical solution provided by the present invention can identify the vital signs of multiple human targets, improve the efficiency and accuracy of identification, and thus provide a guarantee for timely and effective rescue.

[0020] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0022] Figure 1 A flowchart of a method for identifying vital signs of multiple human targets provided according to one embodiment of the present application is shown;

[0023] Figure 2 The figure is a structural diagram of a multi-human target vital sign recognition system provided according to one embodiment of the present application. DETAILED DESCRIPTION

[0024] The following describes in detail embodiments of the present application, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0025] The following describes a method, system, and computer device for recognizing vital signs of multiple human targets according to an embodiment of the present application with reference to the accompanying drawings.

[0026] Example 1

[0027] Figure 1 FIG. 1 is a flow chart of a method for identifying vital signs of multiple human targets according to an embodiment of the present application, as shown in FIG. Figure 1 As shown, this may include:

[0028] Step 1: Use ultra-wideband radar to detect the area to be detected to obtain a radar echo matrix corresponding to the area, and filter out static clutter in the radar echo matrix;

[0029] It should be noted that, assuming that the pulse signal transmitted by the ultra-wideband pulse radar is s(τ), the echo signal received by the antenna is: Where n(τ,t) is the random noise signal of the receiver. Then, the echo signal is discretized to obtain: Where R(m,n) is the radar echo matrix, δ s and T s Represent the sampling interval of fast time and slow time respectively, M and N represent the total number of sampling points of fast time and slow time respectively, τ v (t) is the time delay of the human body micro-motion echo, t and τ are slow time and fast time respectively, a v is the amplitude of the human body micro-motion echo, a i is the amplitude of the echo from the stationary object in the environment, τ i (t) is the time delay of the echo from a stationary object in the environment, and v is the propagation speed of the electromagnetic wave, which is 3×108 m / s.

[0030] It should be noted that filtering out static clutter in the radar echo matrix includes:

[0031] The static clutter in the radar echo matrix is ​​filtered out using a subtraction averaging method.

[0032] The specific implementation process of the subtraction averaging method is as follows: first, the stationary object echo signal of a certain sampling period is found within the period, and then the target echo signal is obtained by subtracting the reference received pulse from each beam of received pulse. The expression of the reference received pulse is: Where M is the number of sampling points in each sampling period, i is the number of the current sampling period, and N is the total number of reference pulse sampling periods. The processed signal is: R[m,n] = R[m,n] - C[m].

[0033] Step 2: Calculate the permutation entropy value of the slow time series corresponding to each range gate in the radar echo matrix;

[0034] In the embodiment of the present disclosure, the calculation of the permutation entropy value of the slow time series corresponding to each range gate in the radar echo matrix includes:

[0035] A fast time point is set in the fast time direction of each radar echo in the radar echo matrix, and a slow time sequence obtained by sampling the corresponding radar receiving pulse is determined;

[0036] According to each of the slow time series, respectively generate overlapping column vector matrices corresponding to each of the slow time series;

[0037] Reconstructing the overlapping column vector matrices corresponding to the slow time series respectively;

[0038] Taking each reconstructed component in each reconstructed overlapping column vector matrix as a sequence, determining the sorting probability of each sequence;

[0039] Determining the permutation entropy value corresponding to each of the slow time series according to the sorting probability of each sequence;

[0040] Each row in the overlapping column vector matrix serves as a reconstruction component, and the elements in each reconstruction component after reconstruction are arranged in ascending order.

[0041] For example, let the fast time point be m1 and the length of the slow time series be N. Then the slow time series can be expressed as X[m1,i]={y(1),y(2),...,y(i)},i=1,2,3...,N;Y∈R M×N , reconstruct the phase space and obtain the reconstruction matrix Y:

[0042] Where, τ d is the delay time, q is the embedding dimension, K=N-(q-1)τ d In this embodiment, the parameters of permutation entropy are set as follows: q is 7, τ d is 1. Each row of the phase space matrix is ​​called a reconstruction component. Arrange the elements in the j-th reconstruction component in ascending order to obtain a new reconstruction component: X j ={x(j+(j1-1)τ d )≤x(j+(j2-1)τ d )≤...≤x(j+(j q -1)τ d )}where j1,j2,...,j q Represents X j The position of each element in the dimension. If the elements are equal, they are sorted by the size of j value. For any reconstruction component of the phase space matrix, a set of sequences can be obtained: S(l)={j1,j2,...,j q}, where l = {1, 2, ..., K}, and K ≤ q! . A q-dimensional sequence produces q! sorts of S(l) sorts. Calculate the probability of each sort of S(l) sequence, and the sum of all the sorting probabilities is 1. Then the permutation entropy of X[m1,i] can be defined in terms of Shannon entropy as: H p The size of H represents the complexity of the slow time series X[m1,i], p The smaller the value, the more regular the time series changes. On the contrary, the more random the time series changes. M×N In the , the slow time direction data of the human target position range changes relatively regularly, so the permutation entropy value corresponding to the human target area, i.e., the PE value, is low.

[0043] Step 3: determining an initial range gate area where the human target in the area to be detected is located according to each of the permutation entropy values;

[0044] In the embodiment of the present disclosure, determining the initial range gate area where the human target in the area to be detected is located according to each of the permutation entropy values ​​includes:

[0045] Obtaining the maximum value among the permutation entropy values;

[0046] Binarization processing is used to determine whether the absolute value of the difference between each permutation entropy value and the maximum value is greater than a preset difference threshold. If so, there is a human target in the range gate area corresponding to the permutation entropy value and the binarized value corresponding to the range gate is 1, and the range gate area corresponding to the permutation entropy value is used as the initial range gate area where the human target is located.

[0047] For example, the formula for the binarization process is Where, T r is the preset difference threshold. In this embodiment, T r The calculation formula is: M represents the total number of fast-time sampling points. If the Q(m) values ​​after binarization are all 0, there are no human targets in the environment of the test area. If the number of Q(m) values ​​after binarization is 1 or greater, there are human targets in the environment of the test area.

[0048] Step 4: determining the number of human targets and the position of each human target according to the initial range gate area where each human target in the area to be measured is located;

[0049] In the embodiment of the present disclosure, determining the number of human targets and the position of each human target according to the initial range gate area where each human target in the area to be detected is located includes:

[0050] Determine whether the number of human targets in the initial range gate area where each human target in the area to be measured is equal to 1. If so, extract the range gate corresponding to the minimum value in the permutation entropy curve, and determine the position of the human target based on the range gate. Otherwise, use the KMeans++ clustering algorithm to process the initial range gate area where each human target in the area to be measured is located to obtain the number of human targets and the position of each human target.

[0051] The permutation entropy curve is a curve drawn with the permutation entropy value corresponding to each of the slow time series as the vertical coordinate and the range gate as the horizontal coordinate.

[0052] It should be noted that the determining whether the number of human targets in the initial range gate area where each human target in the area to be detected is equal to 1 includes:

[0053] The range gates with a binarized value of 1 are numbered to form a one-dimensional data set. The differences between adjacent numbers are calculated and the maximum value is obtained. If the maximum value is less than or equal to the number of sampling points occupied by the chest thickness in the radar receiving pulse, the number of human targets in the test area is considered to be equal to 1.

[0054] The calculation formula for the number of sampling points P occupied by the chest thickness in the radar receiving pulse can be: Where D tho is the thickness of the human chest cavity, v is the propagation speed of electromagnetic waves, which can be 3×10 8 m / s, δ s is the fast time sampling interval.

[0055] It should be noted that the KMeans++ clustering algorithm is used to process the initial range gate area where each human target in the area to be measured is located to obtain the number of human targets and the position of each human target, including:

[0056] Setting different cluster numbers K and using the KMeans++ clustering algorithm to cluster the one-dimensional data set;

[0057] Determine the silhouette coefficients corresponding to different clustering numbers K, and use the clustering number K corresponding to the maximum value of the silhouette coefficient as the number of human targets in the initial range gate area where each human target is located;

[0058] Extracting the range gate where the permutation entropy value is minimum in each of the clusters corresponding to the maximum value of the silhouette coefficient, and determining the position of each human target based on each of the range gates;

[0059] The number of groups K is equal to Y-1, where Y is the total number of people in the area to be tested.

[0060] It should be noted that the energy / variance / standard deviation / kurtosis, etc. of the slow time series corresponding to each range gate in the radar echo matrix can also be obtained, and then the KMeans++ clustering algorithm is used to perform clustering to determine the number of human targets and the position of each human target.

[0061] For example, first, the K value is set to 2, 3, 4....10, where the maximum value is not fixed, that is, the maximum number of detected targets;

[0062] Secondly, perform KMeans++ clustering. The specific steps of each clustering are as follows:

[0063] Step F1: Randomly select a sample (number) from the one-dimensional data set as the initial cluster center;

[0064] Step F2: Calculate the shortest distance between each sample and the selected cluster center, then calculate the probability of each sample being selected as the next cluster center, and finally select the next cluster center according to the roulette wheel method;

[0065] Step F3: Repeat step 2 until all cluster centers are selected;

[0066] Step F4: Calculate the distance between all sample points and the cluster center, and assign each sample point to the cluster with the closest distance to the cluster center;

[0067] Step F5: Obtain the centroid of the cluster as the new cluster center and repeat the above process until the cluster assignment remains unchanged or the maximum number of iterations is reached.

[0068] Then, the silhouette coefficient method is applied to evaluate the clustering effect when the K value is 2, 3, 4...10. The specific steps of the silhouette coefficient method are as follows:

[0069] For each sample point in the cluster, calculate their silhouette coefficient separately:

[0070]

[0071] Among them, a(i) = average (the distance from sample point i to all other points in the cluster to which it belongs), b(i) = min (the average distance from sample point i to all points in a cluster that does not contain it).

[0072] The average of the silhouette coefficients of all sample points is the total silhouette coefficient of the clustering result.

[0073] Finally, according to the distance gate P where the human target is located pos , using the formula The distance of the human target is estimated, where Range is the distance between the human target and the radar antenna. v is the propagation speed of electromagnetic waves, which is equal to 3×10 8 m / s, δ s is the fast time sampling interval.

[0074] Step 5: According to the position of each human target, the surface vibration signal of each human target is extracted from the radar echo matrix after filtering out static clutter, and each human target is identified respectively.

[0075] In the embodiment of the present disclosure, extracting the body surface vibration signal of each human target from the radar echo matrix after filtering out static clutter and identifying each human target respectively according to the position of each human target includes:

[0076] Extracting the body surface vibration signal of each human target from the radar echo matrix after filtering out static clutter based on the position of each human target;

[0077] According to the human respiratory frequency band and heartbeat frequency band, the Butterworth filter is used to reconstruct the respiratory and heartbeat time domain signals of each human target respectively.

[0078] It should be noted that the square amplitude function of the Butterworth filter is defined as: N is the order of the Butterworth filter, ωc is the 3dB cutoff frequency, and ε is the parameter that controls the bandpass ripple amplitude.

[0079] Calculate the high-order cumulative amount of each reconstructed respiratory and heartbeat signal, and perform fast Fourier transform on each high-order cumulative amount to identify the respiratory and heartbeat frequencies of each human target;

[0080] The human body's respiratory frequency band ranges from 0.1 Hz to 0.8 Hz, and the heartbeat frequency band ranges from 0.8 Hz to 2.8 Hz.

[0081] It should be noted that using the higher order cumulant (HOC) algorithm to process the signal and then perform frequency domain analysis can significantly reduce the influence of surrounding clutter and thus improve the signal-to-noise ratio compared to the traditional FFT algorithm. In this embodiment, the fourth order cumulant is used, and its calculation formula is: in, Where {x(n)} is the original respiratory / heartbeat signal. When n≤0 or n≥N, x(n)=0. It can be seen that the fourth-order cumulant of Gaussian noise is always zero. Let τ1=τ, τ2=τ3=0, and obtain a one-dimensional slice signal for fast Fourier transform to read the respiratory / heartbeat frequency.

[0082] In summary, the present application proposes a method for identifying the vital signs of multiple human targets, which uses the KMeans++ clustering algorithm and the HOC algorithm to identify the vital signs of multiple human targets. It can quickly and accurately obtain the number of human targets in the test area, their specific locations and vital signs information, thereby improving the efficiency and accuracy of identification, and providing a guarantee for timely and effective rescue.

[0083] Example 2

[0084] Furthermore, Figure 2 This is a structural diagram of a multi-human target vital sign recognition system provided according to one embodiment of the present application, such as Figure 2 As shown, this may include:

[0085] The filtering module 100 is configured to detect an area to be detected using an ultra-wideband radar to obtain a radar echo matrix corresponding to the area, and to filter out static clutter in the radar echo matrix;

[0086] The calculation module 200 is used to calculate the permutation entropy value of the slow time series corresponding to each range gate in the radar echo matrix;

[0087] A first determining module 300 is configured to determine an initial range gate region where a human target is located in the detection area according to each of the permutation entropy values;

[0088] The second determining module 400 is configured to determine the number of human targets and the position of each human target according to the initial range gate area where each human target is located in the detected area;

[0089] The identification module 500 is used to extract the body surface vibration signal of each human target from the radar echo matrix after filtering out static clutter according to the position of each human target and identify each human target respectively.

[0090] In an embodiment of the present disclosure, filtering out static clutter in the radar echo matrix includes:

[0091] The static clutter in the radar echo matrix is ​​filtered out using a subtraction averaging method.

[0092] In the embodiment of the present disclosure, the calculation module 200 includes:

[0093] The first determining unit 201 is configured to determine each slow time sequence obtained by sampling the radar receiving pulse corresponding to each fast time point in the radar echo matrix;

[0094] A generating unit 202 is configured to generate overlapping column vector matrices corresponding to the slow time series according to the slow time series;

[0095] A first reconstruction unit 203 is configured to reconstruct the overlapping column vector matrices corresponding to the slow time series respectively;

[0096] A second determining unit 204 is configured to treat each reconstructed component in each reconstructed overlapping column vector matrix as a sequence and determine a sorting probability of each sequence;

[0097] The third determining unit 205 is configured to determine the permutation entropy value corresponding to each of the slow time series according to the sorting probability of each sequence;

[0098] Each row in the overlapping column vector matrix serves as a reconstruction component, and the elements in each reconstruction component after reconstruction are arranged in ascending order.

[0099] In the embodiment of the present disclosure, the first determining module 300 includes:

[0100] An acquiring unit 301 is configured to acquire a maximum value among the permutation entropy values;

[0101] The judgment unit 302 is used to use binarization processing to determine whether the absolute value of the difference between each of the permutation entropy values ​​and the maximum value is greater than a preset difference threshold. If so, there is a human target in the range gate area corresponding to the permutation entropy value and the binarized value corresponding to the range gate is 1, and the range gate area corresponding to the permutation entropy value is used as the initial range gate area where the human target is located.

[0102] In the embodiment of the present disclosure, the second determining module 400 is specifically configured to:

[0103] Determine whether the number of human targets in the initial range gate area where each human target in the area to be measured is equal to 1. If so, extract the range gate corresponding to the minimum value in the permutation entropy curve, and determine the position of the human target based on the range gate. Otherwise, use the KMeans++ clustering algorithm to process the initial range gate area where each human target in the area to be measured is located to obtain the number of human targets and the position of each human target.

[0104] The permutation entropy curve is a curve drawn with the permutation entropy value corresponding to each of the slow time series as the vertical coordinate and the range gate as the horizontal coordinate.

[0105] Furthermore, the determining whether the number of human targets in the initial range gate area where each human target in the area to be detected is equal to 1 includes:

[0106] The range gates with a binarized value of 1 are numbered to form a one-dimensional data set. The differences between adjacent numbers are calculated and the maximum value is obtained. If the maximum value is less than or equal to the number of sampling points occupied by the chest thickness in the radar receiving pulse, the number of human targets in the test area is considered to be equal to 1.

[0107] Furthermore, the KMeans++ clustering algorithm is used to process the initial range gate area where each human target in the area to be measured is located to obtain the number of the human targets and the position of each human target, including:

[0108] Determine the silhouette coefficients corresponding to different grouping numbers K, and use the grouping number K corresponding to the maximum value of the silhouette coefficient as the number of human targets in the initial range gate area where each human target in the area to be measured is located;

[0109] Extracting the range gate where the permutation entropy value is minimum in each of the clusters corresponding to the maximum value of the silhouette coefficient, and determining the position of each human target based on each of the range gates;

[0110] The number of groups K is equal to Y-1, where Y is the total number of people in the area to be tested.

[0111] The number of groups k is equal to 2, 3, 4, ..., Y in sequence.

[0112] In the embodiment of the present disclosure, the identification module 500 includes:

[0113] An extraction unit 501 is configured to extract a body surface vibration signal of each human target from the radar echo matrix after filtering out static clutter based on the position of each human target;

[0114] The second reconstruction unit 502 is used to reconstruct the respiratory and heartbeat time domain signals of each human target using a Butterworth filter according to the respiratory frequency band range and the heartbeat frequency band range of the human body;

[0115] The identification unit 503 is used to calculate the high-order cumulative amount of each reconstructed respiratory and heartbeat signal, and perform fast Fourier transform on each high-order cumulative amount to identify the respiratory and heartbeat frequencies of each human target;

[0116] The human body's respiratory frequency band may range from 0.1 Hz to 0.8 Hz, and the heartbeat frequency band may range from 0.8 Hz to 2.8 Hz.

[0117] In summary, the multi-human target vital signs recognition system proposed in this application uses the KMeans++ clustering algorithm and the HOC algorithm to recognize the vital signs of multiple human targets, which can quickly and accurately obtain the number of human targets in the test area, the specific location and vital signs information, thereby improving the efficiency and accuracy of recognition, and providing a guarantee for timely and effective rescue.

[0118] Example 3

[0119] In order to implement the above embodiments, the embodiments of the present application also propose a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method described in Embodiment 1 of the present application is implemented.

[0120] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0121] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may be performed out of the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present application belong.

[0122] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A method for identifying vital signs of multiple human targets, characterized in that: The method comprises: Using an ultra-wideband radar to detect an area to be detected to obtain a radar echo matrix corresponding to the area, and filtering out static clutter in the radar echo matrix; Calculate the permutation entropy value of the slow time series corresponding to each range gate in the radar echo matrix; determine the initial range gate area where the human target in the area to be detected is located according to each permutation entropy value; Determine the number of human targets and the position of each human target according to the initial range gate area where each human target in the area to be detected is located; According to the position of each human target, extract the body surface vibration signal of each human target from the radar echo matrix after filtering out static clutter and identify each human target respectively; Determining the initial range gate area where the human target in the area to be detected is located according to each of the arrangement entropy values ​​includes: Obtaining the maximum value among the permutation entropy values; Using binarization processing to determine whether the absolute value of the difference between each permutation entropy value and the maximum value is greater than a preset difference threshold, if so, there is a human target in the range gate area corresponding to the permutation entropy value and the binarized value corresponding to the range gate is 1, and the range gate area corresponding to the permutation entropy value is used as the initial range gate area where the human target is located; Determining the number of human targets and the position of each human target according to the initial range gate area where each human target in the area to be detected is located includes: Determine whether the number of human targets in the initial range gate area where each human target in the area to be detected is equal to 1. If so, extract the range gate corresponding to the minimum value in the permutation entropy curve, and determine the position of the human target based on the range gate. Otherwise, use the KMeans++ clustering algorithm to process the initial range gate area where each human target in the area to be detected is located to obtain the number of human targets and the position of each human target. The permutation entropy curve is a curve drawn with the permutation entropy value corresponding to each of the slow time series as the ordinate and the range gate as the abscissa.

2. The method according to claim 1, characterized in that The filtering out static clutter in the radar echo matrix comprises: The static clutter in the radar echo matrix is ​​filtered out by using the subtraction average method.

3. The method according to claim 1, characterized in that The step of calculating the permutation entropy value of the slow time series corresponding to each range gate in the radar echo matrix includes: Determine each slow time series obtained by sampling the radar receiving pulse corresponding to each fast time point in the radar echo matrix; According to each of the slow time series, respectively generate overlapping column vector matrices corresponding to each of the slow time series; Reconstructing the overlapping column vector matrices corresponding to the slow time series respectively; Taking each reconstructed component in each reconstructed overlapping column vector matrix as a sequence, determining the sorting probability of each sequence; Determining the permutation entropy value corresponding to each of the slow time series according to the sorting probability of each sequence; Each row in the overlapping column vector matrix serves as a reconstruction component, and the elements in each reconstruction component after reconstruction are arranged in ascending order.

4. The method according to claim 1, characterized in that The step of judging whether the number of human targets in the initial range gate area where each human target in the area to be detected is equal to 1 comprises: The range gates with a value of 1 after binarization are organized into a one-dimensional data set according to the numbers, the differences between adjacent numbers are calculated and the maximum value is obtained. If the maximum value is less than or equal to the number of sampling points occupied by the chest thickness in the radar receiving pulse, the number of human targets in the detection area is considered to be equal to 1.

5. The method according to claim 4, characterized in that The KMeans++ clustering algorithm is used to process the initial range gate area where each human target in the to-be-detected area is located to obtain the number of human targets and the position of each human target, including: Setting different clustering numbers K, and using KMeans++ clustering algorithm to cluster the one-dimensional data set; Determine the silhouette coefficients corresponding to different grouping numbers K, and use the grouping number K corresponding to the maximum value of the silhouette coefficient as the number of human targets in the initial range gate area where each human target is located; Extracting the distance gate where the minimum value of permutation entropy is located in each group among the number of groups corresponding to the maximum value of the silhouette coefficient, and determining the position of each human target based on each of the distance gates; The number of groups K is equal to Y-1, where Y is the total number of people in the area to be detected.

6. The method according to claim 5, characterized in that The method of extracting the body surface vibration signal of each human target from the radar echo matrix after filtering out static clutter according to the position of each human target and identifying each human target respectively includes: Extracting the body surface vibration signal of each human target from the radar echo matrix after filtering out static clutter based on the position of each human target; According to the human respiratory frequency band and heartbeat frequency band, the Butterworth filter is used to reconstruct the respiratory and heartbeat time domain signals of each human target respectively. The high-order cumulants of each reconstructed breathing and heartbeat signal are calculated, and each high-order cumulant is subjected to fast Fourier transform to identify the breathing and heartbeat frequencies of each human target.

7. A multi-human target vital sign recognition system, characterized in that: The system comprises: A filtering module, used to detect the area to be detected using an ultra-wideband radar to obtain a radar echo matrix corresponding to the area, and to filter out static clutter in the radar echo matrix; A calculation module is used to calculate the permutation entropy value of the slow time series corresponding to each range gate in the radar echo matrix; A first determination module is used to determine an initial range gate area where a human target is located in the area to be detected according to each of the permutation entropy values; A second determination module is used to determine the number of human targets and the position of each human target according to the initial range gate area where each human target is located in the area to be detected; An identification module, used for extracting the body surface vibration signal of each human target from the radar echo matrix after filtering out static clutter according to the position of each human target and identifying each human target respectively; The first determining module is further used to obtain the maximum value among the permutation entropy values; Using binarization processing to determine whether the absolute value of the difference between each permutation entropy value and the maximum value is greater than a preset difference threshold, if so, there is a human target in the range gate area corresponding to the permutation entropy value and the binarized value corresponding to the range gate is 1, and the range gate area corresponding to the permutation entropy value is used as the initial range gate area where the human target is located; The second determination module is further used to determine whether the number of human targets in the initial range gate area where each human target in the area to be detected is equal to 1, if so, extract the range gate corresponding to the minimum value in the permutation entropy curve, and determine the position of the human target based on the range gate, otherwise, use the KMeans++ clustering algorithm to process the initial range gate area where each human target in the area to be detected is located, and obtain the number of the human targets and the position of each human target; The permutation entropy curve is a curve drawn with the permutation entropy value corresponding to each of the slow time series as the ordinate and the range gate as the abscissa.

8. A computer device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method according to any one of claims 1 to 6 is implemented.

Images