Personnel vital sign detection method and system based on body motion compensation

Abstract

The invention discloses a human vital sign detection method and system based on body motion compensation, and the method comprises the steps: enabling echo data collected by a millimeter wave radar to generate a three-dimensional data matrix, and carrying out the phase compensation through a CBF method, so as to enhance a target signal; background noise is filtered by adopting a static clutter removal technology, and a dynamic reflection signal is extracted; and eliminating random occlusion artifacts by using a voting mechanism based on a CFAR result, and accurately positioning a target position. For random motion interference, a two-stage body motion compensation method is adopted to correct large-amplitude motion offset and suppress small-amplitude motion harmonic interference. And finally, screening an effective signal range through an autocorrelation coefficient, separating respiration and heartbeat signals by using an improved multi-channel variational mode decomposition algorithm, and extracting life signals. The method can effectively eliminate environmental noise, random shielding and motion interference, realizes high-precision monitoring of respiration and heartbeat frequencies in a complex scene, and is suitable for various application scenes such as an intelligent cabin.

Description

Technical Field

[0001] The present invention belongs to the technical field of non-contact vital sign monitoring, and particularly relates to a method and system for detecting a person's vital signs based on body movement compensation. Background Art

[0002] Vital signs are important physiological parameters reflecting the health status of the human body. Among them, respiration and heartbeat, as the most direct monitoring indicators, have extensive application values in fields such as medical health monitoring, home health management, and driver status monitoring. Currently, most traditional vital sign monitoring devices adopt contact measurement methods or video-based non-contact processing solutions, bringing additional burdens or privacy issues to the measured subjects. Therefore, millimeter-wave radar, which is not affected by visible light and has a wide range of applications, has gradually become a research hotspot in non-contact vital sign signal monitoring technology.

[0003] Currently, many studies at home and abroad have explored the technology of vital sign monitoring based on radar. For example, Texas Instruments in the United States uses a 77GHz frequency-modulated continuous wave (FMCW) radar combined with range gating and beamforming technologies to successfully achieve accurate estimation of vital signs in a single-target scenario and separation of multi-target signals; Yonsei University in South Korea uses a 24GHz FMCW Doppler radar combined with the fast Fourier transform (FFT) and multiple signal classification (MUSIC) algorithms to achieve separation and estimation of multi-target signals at relatively short distances. These methods perform well in single-target or multi-target scenarios at different distances, but often ignore the interference problem caused by the random movement of the measured subject. Usually, a fixed filter scheme is adopted, lacking adaptability in different scenarios and being difficult to meet the application requirements in complex scenarios.

[0004] Aiming at the technical difficulty of random movement interference, researching a method for detecting a person's vital signs based on body movement compensation to achieve high-precision signal extraction has important research value for the field of non-contact vital sign monitoring. Summary of the Invention

[0005] To solve the problem of inaccurate measurement for randomly moving objects in the prior art, the present invention provides a method and system for detecting a person's vital signs based on body movement compensation, which realizes real-time vital sign monitoring in a complex environment by combining movement compensation and multi-channel signal processing technologies.

[0006] According to one aspect of the specification of the present invention, a method for detecting a person's vital signs based on body movement compensation is provided, including:

[0007] Obtaining echo data of a target individual based on a millimeter-wave radar and generating a three-dimensional data matrix;

[0008] Perform beamforming processing, static clutter removal, and target detection on the three-dimensional data matrix in sequence to obtain a target area;

[0009] Perform two-stage body motion compensation on the target area, including: calculating the displacement amount between adjacent frames using the two-dimensional cross-correlation method to correct the position offset of the target area; downsampling the corrected signal, and removing the low-frequency and large-amplitude motion trends through B-spline curve fitting to suppress the harmonic interference of small random motions;

[0010] Extract vital signs from the signal after body motion compensation.

[0011] As a further technical solution, correcting the position offset of the target area includes:

[0012] Set a time window, calculate the constant false alarm rate results for each frame within this time window to obtain the possible positions where the target exists;

[0013] Regard the constant false alarm rate result of each frame as a vote, and then accumulate the voting results of all frames within the time window;

[0014] When the cumulative vote count in a certain range bin reaches half of the time window length, it is considered that this range bin is the position of the target;

[0015] Calculate the average Euclidean distance between the corresponding front and rear frames of the time window, and select the window with the smallest Euclidean distance as the window where the human body motion is the most stable, thereby obtaining the subscript of the target stable frame;

[0016] Calculate the centroid positions of all connected components, and start from the time window adjacent to the stable time window to perform connected component matching forward and backward;

[0017] Match each connected component in the time window where the target position has been determined in the adjacent time window with each connected component in the time window where the target position has not been determined one by one. The matching score is defined as the reciprocal of the Euclidean distance between the centroids of the two connected components;

[0018] Retain the connected component with the highest matching score in the time window where the target position has not been determined. If the highest score is not obtained in the matching with all the connected components of the previous windows, it is considered that this connected component is a connected component of motion artifacts, and mask elimination processing is performed in the constant false alarm rate result cumulative matrix;

[0019] After determining the new target position, calculate the offset amount compensated towards the stable window to obtain the compensation offset amount.

[0020] As a further technical solution, downsampling the corrected signal and removing the low-frequency and large-amplitude motion trends through B-spline curve fitting to suppress the harmonic interference of small random motions includes:

[0021] Downsample the original signal to avoid overfitting of the respiratory phase.

[0022] Use B-spline curves to fit the low-frequency and large-amplitude target motion trends and eliminate them from the original phase sequence;

[0023] Subtract the original artifacts from the original phase to obtain a cleaner signal with richer respiratory and heartbeat signals.

[0024] As a further technical solution, extract vital signs from the signal after body motion compensation, including:

[0025] Separate respiratory and heartbeat signals based on an improved multi-channel variational mode decomposition algorithm.

[0026] As a further technical solution, perform beamforming processing on the three-dimensional data matrix, including:

[0027] Set multiple groups of phase compensation weights for the received signal according to the target direction angle;

[0028] Enumerate each group of weights and perform weighted superposition on the signal, select the group of weights with the strongest target signal energy for phase compensation to obtain an enhanced target reflection signal.

[0029] As a further technical solution, remove static clutter from the target reflection signal, including:

[0030] Set a time window, calculate the average value of the target reflection signal along the time dimension within the window to estimate the background noise;

[0031] Extract the dynamic reflection signal according to the target reflection signal and the estimated value of the background noise.

[0032] As a further technical solution, perform target detection on the dynamic reflection signal and locate the target area based on a voting mechanism, including:

[0033] In the multi-frame constant false alarm rate detection results, accumulate the number of times each target area is marked into a matrix;

[0034] Set a threshold, screen the accumulated results, and retain the target areas with accumulated scores greater than the threshold.

[0035] According to one aspect of the specification of the present invention, there is provided a personnel vital sign detection system based on body motion compensation, including:

[0036] A first main module for obtaining echo data of a target individual based on a millimeter-wave radar and generating a three-dimensional data matrix;

[0037] A second main module, which is configured to perform beamforming processing, static clutter removal, and target detection on the three-dimensional data matrix in sequence to obtain a target area;

[0038] A third main module, which is configured to perform two-stage body movement compensation on the target area, including: calculating the displacement amount between adjacent frames by using a two-dimensional cross-correlation method to correct the position offset of the target area; downsampling the corrected signal, and removing the low-frequency and large-amplitude movement trends through B-spline curve fitting to suppress the harmonic interference of small random movements;

[0039] A fourth main module, which is configured to extract vital signs from the signal after body movement compensation.

[0040] According to one aspect of the specification of the present invention, an electronic device is provided, including a memory and a processor. The memory stores program instructions executed by the processor, and the processor calls the program instructions to execute the steps of the method for detecting a person's vital signs based on body movement compensation.

[0041] According to one aspect of the specification of the present invention, a non-transitory computer-readable storage medium is provided. The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions cause the computer to execute the steps of the method for detecting a person's vital signs based on body movement compensation.

[0042] One or more of the above technical solutions of the present invention have at least one or more of the following technical effects:

[0043] 1. In the method provided by the present invention, high-precision extraction of vital sign signals can be achieved based on the motion compensation and signal separation technologies of the target area, so as to realize the accurate monitoring of multi-target vital signs. After locating the chest area through constant false alarm rate (CFAR) detection and voting mechanism, a two-stage body movement compensation method is adopted to correct the displacement of the target area and remove motion interference. On this basis, an improved multi-channel variational mode decomposition (BM-VMD) algorithm is used to separate and extract respiration and heartbeat signals, realizing the efficient extraction and processing of vital sign signals.

[0044] 2. Since the determination of the chest area is obtained by multi-frame cumulative voting based on the CFAR detection result, the position offset of the target area is corrected by two-dimensional cross-correlation, and small-amplitude motion interference is removed by B-spline curve fitting. Therefore, the extracted signal mainly reflects the vital sign characteristics of the target and has high credibility. In addition, the improved multi-channel BM-VMD algorithm combines the information of multiple signal channels, decomposes the signal modes through adaptively optimized parameters, effectively distinguishes respiration and heartbeat signals, and further enhances the accuracy and robustness of signal extraction.

[0045] 3. The method of the present invention combines motion compensation and multi-channel signal processing technologies to achieve real-time vital sign monitoring in complex environments. It can adapt to scenarios where multiple targets are at the same distance, effectively solving the problem in the prior art that it is difficult to handle target breath-holding or motion artifacts. Compared with traditional methods, the present invention does not require additional equipment to assist in data acquisition. Through multi-channel processing of radar signals, accurate extraction of vital sign signals can be completed, greatly improving the monitoring efficiency, ensuring real-time performance and high precision, and significantly reducing the complexity and cost of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings used in the description of the embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0047] Figure 1 It is a flowchart of the method for detecting human vital signs based on body motion compensation in the embodiments of the present invention;

[0048] Figure 2 It is a schematic diagram of the experimental scenario and equipment used in the embodiments of the present invention;

[0049] Figure 3 For Figure 1 It is the CBF processing flowchart in the method shown in

[0050] Figure 4 It is a flowchart of the body motion compensation algorithm in the embodiments of the present invention;

[0051] Figure 5 It is a flowchart of the optimized particle swarm algorithm in step S5 in the embodiments of the present invention

[0052] Figure 6 It is a thermal image after random occlusion cancellation processing in the embodiments of the present invention;

[0053] Figure 7 It is a data comparison diagram before and after phase unwrapping processing in the embodiments of the present invention;

[0054] Figure 8 It is the target motion trend obtained in step S5 in the embodiments of the present invention;

[0055] Figure 9 It is the signal obtained after motion elimination in step S6 in the embodiments of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0056] The present invention provides a method for detecting a person's vital signs based on body movement compensation, which is used to solve the technical problem in the prior art that it is difficult to handle target breath-holding or motion artifacts. The method of the present invention can effectively eliminate environmental noise, random occlusion, and motion interference, and achieve high-precision monitoring of respiration and heart rate in complex scenarios, and is applicable to various application scenarios such as intelligent cockpits.

[0057] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention. In addition, the technical features in each embodiment or a single embodiment provided by the present invention can be combined with each other arbitrarily to form a new technical solution. This combination is not restricted by the order of steps and / or the mode of structural composition, but must be based on the fact that those of ordinary skill in the art can implement it. When the combination of technical solutions conflicts with each other or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the protection scope required by the present invention.

[0058] The experiment of the embodiment of the present invention was carried out in a simulated cockpit environment. The experimental equipment includes an IWR6843ISK millimeter-wave radar module and a DCA1000 data acquisition card. The radar was installed at the same height as the target's chest cavity, covering the activity range of the target. In the experimental scenario of this experiment, the distance between the target person and the millimeter-wave radar was about 1.0 m. In the experiment, the target person maintained a natural sitting posture and simulated the actual operations during driving, including arm movements and slight body tilts. The target person needed to wear a polar h10 heart rate test belt to record the ECG reference value during the detection period. The data was collected by listening to the UDP port through a Python program and stored as a binary format file. The frequency modulation slope of the signal was set to 70 MHz / μs, and the single data acquisition duration was 50 seconds.

[0059] Based on this experimental scenario, this embodiment provides a method for detecting a person's vital signs based on body movement compensation. Refer to Figure 1 , and the method includes:

[0060] Step S1: Data acquisition and matrix generation

[0061] The millimeter-wave radar transmits a linearly frequency-modulated signal, receives the target reflection signal, and stores the received echo data as a three-dimensional data matrix through an analog-to-digital converter (ADC); where the row is the received antenna signal; the column is the time sampling point; and the depth is the number of channels, which is determined by the combination of the transmitting and receiving antennas.

[0062] Step S2: CBF (Conventional Beamforming) processing

[0063] Specifically, its core idea is to regard the signals received by the antenna array of the millimeter-wave radar as frequency-domain signals, calculate their spatial spectra through FFT to obtain the power distributions in different directions; on this basis, phase compensation is performed through beamforming technology. After phase compensation for each signal, signals in a certain direction are obtained by accumulation.

[0064] As an alternative implementation, please refer to Figure 3 , which is the CBF processing flowchart. Step S2 specifically includes:

[0065] Step S2.1: Set multiple groups of phase compensation weights for the received signals according to the target direction angle

[0066] Among them, L represents the number of virtual antennas, d r represents the distance between two receiving antennas, and θ represents the direction of the signal to be enhanced. After phase compensation for each signal, signals in a certain direction are obtained by accumulation Among them, w θ (l) represents the weight corresponding to the l-th virtual antenna, and x l (t) represents the signal received by the l-th virtual antenna at time t.

[0067] Step S2.2: Enumerate each group of weights and perform weighted superposition on the signals, select the group of weights with the strongest target signal energy for phase compensation to obtain the enhanced target reflection signal. Specifically, it includes:

[0068] Step S2.2.1: Enumerate the angles of arrival (AoA) of multiple groups of reflected signals, calculate the corresponding phase compensation weights, and perform weighted superposition on the received signals;

[0069] Step S2.2.2: Select the group of weights for which the phase difference of the signal is completely compensated, that is, the group of weights with the maximum signal energy, for phase compensation, so as to enhance the signal in the target direction and suppress the interference in non-target directions, and achieve higher angular resolution.

[0070] Step S3: Remove static clutter from the signals processed by CBF to suppress the influence of background noise and roughly locate the range bin corresponding to the driver, including:

[0071] Step S3.1: Set a time window, calculate the average value of the signals along the time dimension within the window, and estimate the background noise Among them, W represents half of the window length for clutter removal, and p θ (t,m) is the power of the original signal, is the signal power obtained by filtering the background noise from the original signal. m represents the range bin index, that is, different distance points within the radar detection range.

[0072] Step S3.2: Subtract the background noise from the current signal to extract the dynamic reflection signal of the target person.

[0073] Step S4: Use the CFAR algorithm to detect the target in the dynamic reflection signal, and locate the target chest area based on the voting mechanism to counteract the influence of random occlusion. Specifically, it includes:

[0074] Step S4.1: Use the CFAR algorithm to detect the target in the dynamic reflection signal. In the multi-frame CFAR detection results, accumulate the number of times each target area is marked into a matrix;

[0075] Step S4.2: Set the threshold as threshold = λ × (frames - half_window), compare the CFAR statistical results in each range bin with this threshold, and retain the bins higher than the threshold for subsequent extraction of vital sign information.

[0076] Among them, λ is the coefficient used when determining the threshold, satisfying 0 < λ < 1. The smaller λ is, the easier it is to select the range bins containing vital signs, but the more interference may be introduced. Frames refers to the total number of collected data frames, and half_window refers to the window size determined during clutter removal.

[0077] Furthermore, obtain the matrix after random occlusion cancellation:

[0078]

[0079] Among them, A i，j is the element at position i, j in the matrix storing the total accumulation result, and A′ i，j is the corresponding element at the position in the matrix obtained after random occlusion cancellation. The non-zero positions in this matrix represent the rough positions of the target chest.

[0080] Step S5: Perform two-stage body movement compensation on the target area.

[0081] Specifically, according to whether the amplitude of the target person's movement exceeds the range-angle resolution, the random movement is divided into two categories. Set two-stage movement compensation to solve the problems brought by these two types of random movements. It mainly includes two-dimensional cross-correlation movement compensation based on adjacent sliding windows and movement artifact elimination based on B-spline curves.

[0082] Preferably, please refer to Figure 4 , which is a flowchart for body movement compensation. Step S5 specifically includes:

[0083] Step S5.1: Calculate the displacement amount between adjacent frames using the two-dimensional cross-correlation method to correct the position offset of the target area. Specifically, it includes:

[0084] Step S5.1.1: Set a time window, calculate the CFAR result of each frame within the time window, and obtain the possible positions where the target exists;

[0085] Step S5.1.2: Treat the CFAR result of each frame as a vote, and then accumulate the voting results of all frames within the time window;

[0086] Step S5.1.3: When the cumulative vote count within a certain range bin reaches half of the time window length, it is considered that this range bin is the position of the target, that is:

[0087]

[0088] where n represents the time window length, V i,j refers to the value of the cumulative matrix of the original data of each frame within the time window after CFAR at positions i and j, and V' i,j is the result after being screened by the voting rule and is the matrix used for final target detection.

[0089] Step S5.1.4: Calculate the average Euclidean distance between the corresponding previous and next frames of the time window:

[0090]

[0091] where A is the matrix data of the frame corresponding to the current time window, B and C are the matrix data of the previous and next frames of the current frame respectively, a ij , b ij , c ij are the elements at positions (i, j) in matrices A, B, and C respectively, representing the signal detection results of the millimeter-wave radar. And select the window with the smallest Euclidean distance as the window where the human body movement is the most stable. The subscript of the target stable frame is:

[0092]

[0093] where k * is the index of the optimal frame, k is the index of the current frame, and d Euclidean (A k , A k-1 ), d Euclidean (A k , A k+1 ) respectively represent the current frame AK The Euclidean distance from the previous frame A k-1 and the next frame A k+1 therebetween.

[0094] Step S5.1.5, calculate the centroid positions of all connected components

[0095]

[0096] and perform connected component matching forward and backward starting from the time windows adjacent to the stable time window. Among them, represent the horizontal and vertical coordinates of the centroid respectively, and i k and j k represent the horizontal and vertical coordinates of all points in the connected component.

[0097] Step S5.1.6, match each connected component in the time window with a determined target position in the adjacent time window with each connected component in the time window without a determined target position one by one, and the matching score x and y represent the horizontal and vertical coordinates in two connected components respectively.

[0098] Step S5.1.7, retain the connected component with the highest matching score in the time window without a determined target position. If the highest score is not obtained in the matching with all the connected components in the previous windows, it is considered that this connected component is a connected component of motion artifacts, and mask elimination processing is performed in the CFAR result accumulation matrix;

[0099] Step S5.1.8, after determining the new target position, calculate the offset compensated to the stable window. Among them, the two-dimensional cross-correlation calculation method is:

[0100]

[0101] In the formula, A(i, j) is the CFAR accumulation result matrix of the previous window, B(i, j) is the CFAR accumulation result matrix of the next window, and (u, v) represents the offset.

[0102] Furthermore, obtain the compensation offset offset:

[0103]

[0104] In the formula, B.shape[.] represents the dimension information in the CFAR accumulation result matrix of the next window.

[0105] Step S5.1.9, ignore the signals of other targets, and adopt a per-target alignment strategy for signal alignment to ensure that the alignment process is only performed for the current target.

[0106] Step S5.2: Downsample the corrected signal, and remove the low-frequency and large-amplitude motion trends by B-spline curve fitting to suppress the harmonic interference of small random motions. Specifically, it includes:

[0107] Step S5.2.1: Downsample the original signal to avoid overfitting of the respiratory phase. The downsampling factor is set to 12, and the optimization problem is where p(n) represents the observed data value corresponding to the nth sampling point in the sequence, is the control point, and λ is the smoothing factor used to control the smoothness of the curve representing the motion trend obtained by fitting. is the value of the motion trend at time t.

[0108] Step S5.2.2: Use the B-spline curve to fit the low-frequency and large-amplitude target motion trend, and the human body motion trend is and eliminate it from the original phase sequence, where represents the transpose of the control point vector, and f(t) represents the vector of the values of the B-spline basis functions, specifically the calculation results of each basis function at time point t;

[0109] Step S5.2.3: Subtract the original artifact from the original phase to obtain a cleaner signal with richer respiratory and heartbeat signals.

[0110] Step S6: Extract the vital signs from the compensated signal.

[0111] Preferably, Step S6 specifically includes:

[0112] Step S6.1: Use the autocorrelation coefficient to screen the effective range containing the respiratory and heartbeat signals in the chest area. The autocorrelation coefficient ρ k is calculated as:

[0113]

[0114] where ρ k represents the autocorrelation coefficient at lag k, which is used to measure the periodicity of the phase sequence; Var(p b，t ) represents the variance of the phase change p b (t) caused by respiration at time t.

[0115] Since the phase change caused by respiration is much larger than the phase changes caused by heartbeat and background noise, the autocorrelation coefficient will show a peak at the position corresponding to the respiratory cycle.

[0116] Step S6.2: Separate the respiratory and heartbeat signals based on the improved multi-channel variational mode decomposition (BM-VMD) algorithm.

[0117] Specifically, it includes:

[0118] Step S6.2.1: Update the decomposition mode. The calculation method for updating the K-th mode in the b-th range bin is as follows:

[0119]

[0120] where, represents the representation of the k-th mode in the frequency domain in the b-th range bin; represents the representation of the observed signal in the b-th range bin in the frequency domain; α is the regularization parameter used to control the smoothness of the model; ω k represents the center frequency of the k-th mode

[0121] Step S6.2.2: Update the center frequency. The calculation method for updating the center frequency is as follows:

[0122]

[0123] Step S6.2.3: Calculate the modulus of the difference between the previous and current iterations of each mode IMF. If it exceeds the threshold range, repeat Steps S6.2.1 - S6.2.2; if it is within the threshold range, take the decomposed mode and center frequency at this moment as the final result;

[0124] Step S6.2.4: Calculate the channel weights. Introduce a set of adaptive weights to balance multiple range bins to minimize the total variance after signal fusion. The optimization objective of the adaptive weights is where w is the weight vector and s represents the matrix containing the phase sequences of different range bins;

[0125] Step S6.2.5: Select parameters based on the principle of minimum envelope entropy. Take the envelope entropy as the fitness function for optimizing the BM-VMD parameters. The optimization objective of the optimization algorithm is to minimize the envelope entropy of K modes, that is:

[0126]

[0127] Specifically, please refer to Figure 5 , from Step S6.2.5, the optimized algorithm is obtained as follows:

[0128] Step S6.2.5.1: Decompose the signal with the optimized parameters to obtain a series of modal components;

[0129] Step S6.2.5.2: Initialize the parameters of the particle swarm, including the size of the particle swarm, velocity limit, inertia weight, and learning factor;

[0130] Step S6.2.5.3: Initialize the fitness of the particles (i.e., the minimum envelope entropy of the K modes obtained by decomposition) to evaluate the quality of the current position of the particles;

[0131] Step S6.2.5.4: Update the positions and velocities of the particles. The velocity and position of each particle are updated according to the following formulas:

[0132]

[0133] where represents the velocity of the i-th particle in the t-th generation, represents the position of the i-th particle in the t-th generation, w is the inertia weight, c1 and c2 are acceleration factors that respectively control the degree to which the particles approach the individual optimum and the global optimum, rand is a random number with a value range of 0 to 1, and gbest is the global optimum position among all particles.

[0134] Step S6.2.5.5: Evaluate and update the individual and global optimum solutions;

[0135] Step S6.2.5.6: Check the end condition. When the iteration reaches the upper limit of iteration or the change in fitness is less than the set threshold, the steps are ended, and the global optimum position is returned as the decomposition mode number and penalty factor of this set of data. Otherwise, repeat steps S6.2.5.3 - S6.2.5.6.

[0136] In this experimental scenario, the reference value of the heart rate of the target person measured by the polar h10 heart rate belt is 70 - 72 beats per minute, and the reference value of the respiratory rate is 14 breaths per minute. Using the method for detecting human vital signs based on body movement compensation proposed in this embodiment, the detection results of the heart rate and respiratory rate are as Figure 9 shown. The heart rate of the target person is 71.15 beats per minute, and the respiratory rate of the target person is 14 breaths per minute. The results are all within the reference value range.

[0137] The implementation basis of each embodiment of the present invention is achieved through programmed processing by a device with a processor function. Therefore, in engineering practice, the technical solutions and their functions of each embodiment of the present invention are encapsulated into various modules. Based on this actual situation, on the basis of the above embodiments, an embodiment of the present invention provides a system for detecting human vital signs based on body movement compensation, and this system is used to execute a method for detecting human vital signs based on body movement compensation in the above method embodiments.

[0138] The system includes: a first main module for obtaining echo data of a target individual based on a millimeter-wave radar and generating a three-dimensional data matrix; a second main module for successively performing beamforming processing, static clutter removal, and target detection on the three-dimensional data matrix to obtain a target area; a third main module for performing two-stage body movement compensation on the target area, including: calculating the displacement amount between adjacent frames using a two-dimensional cross-correlation method to correct the position offset of the target area; downsampling the corrected signal, and removing low-frequency and large-amplitude movement trends through B-spline curve fitting to suppress harmonic interference of small random movements; a fourth main module for extracting vital signs from the signal after body movement compensation.

[0139] An apparatus for detecting human vital signs based on body movement compensation provided by an embodiment of the present invention addresses the problem of inaccurate measurement of randomly moving objects in the prior art. By using the foregoing several modules and combining motion compensation and multi-channel signal processing technologies, real-time vital sign monitoring in a complex environment is achieved.

[0140] It should be noted that the system embodiment provided by the present invention is used not only to implement the method in the above method embodiment, but also to implement the methods in other method embodiments provided by the present invention. The difference lies only in setting corresponding functional modules. The principle is basically the same as that of the above system embodiment provided by the present invention. As long as those skilled in the art, based on the above system embodiment, refer to the specific technical solutions in other method embodiments, obtain corresponding technical means by combining technical features, and the technical solutions constituted by these technical means, and on the premise of ensuring the practicability of the technical solutions, improve the modules in the above system embodiment to obtain corresponding system-like embodiments for implementing the methods in other method-like embodiments.

[0141] Based on the same inventive concept as the foregoing embodiment, an embodiment of the present invention further provides an electronic device, including a memory and a processor. The memory stores program instructions executed by the processor, and the processor invokes the program instructions to execute the steps of the method for detecting human vital signs based on body movement compensation.

[0142] Based on the same inventive concept as the foregoing embodiment, an embodiment of the present invention further provides a non-transitory computer-readable storage medium. The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions cause the computer to execute the steps of the method for detecting human vital signs based on body movement compensation.

[0143] In summary, the present invention has the following advantages or beneficial technical effects:

[0144] 1. High-precision vital sign signal extraction. Traditional non-contact vital sign detection methods are easily affected by background clutter and target movement interference, resulting in low signal extraction accuracy. The present invention uses a two-stage body movement compensation method, combined with two-dimensional cross-correlation method and B-spline curve fitting, to effectively correct target offset and remove low-frequency large-amplitude motion artifacts, significantly improving the extraction accuracy of vital sign signals.

[0145] 2. Strong adaptability to complex scenarios. In the prior art, under the condition of multiple targets at the same distance, it is often necessary for the target to hold its breath or use complex equipment to avoid interference. The present invention uses a CFAR detection and multi-frame voting mechanism, combined with an improved multi-channel variational mode decomposition (BM-VMD) algorithm, to accurately separate respiratory and heartbeat signals in complex scenarios.

[0146] 3. Both real-time and efficient. Traditional monitoring methods have slow response and lag problems when the target moves and the signal changes dynamically. The present invention adopts multi-channel signal fusion and adaptive optimization technology to update the target position and signal separation parameters in real time, ensuring the efficient output of monitoring results and meeting the requirements of real-time vital sign monitoring.

[0147] 4. Simple, efficient and widely applicable. The present invention requires no additional auxiliary equipment except for necessary millimeter-wave radar equipment, and can achieve high-precision vital sign monitoring using millimeter-wave radar. This method has a simple structure and strong robustness, and is applicable to various application scenarios such as intelligent driving, home health monitoring and public health screening, with broad promotion prospects and practical value.

[0148] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting a person's vital signs based on body movement compensation, characterized in that, Including: Obtaining echo data of a target individual based on a millimeter-wave radar and generating a three-dimensional data matrix; Successively performing beamforming processing, static clutter removal, and target detection on the three-dimensional data matrix to obtain a target region; Performing two-stage body movement compensation on the target region, including: calculating the displacement amount between adjacent frames using a two-dimensional cross-correlation method to correct the position offset of the target region; downsampling the corrected signal, and removing low-frequency and large-amplitude movement trends through B-spline curve fitting to suppress harmonic interference of small-amplitude random movements; Extracting vital signs from the signal after body movement compensation.

2. The method for detecting a person's vital signs based on body movement compensation according to claim 1, wherein, Correcting the position offset of the target region, including: Setting a time window, calculating the constant false alarm rate results for each frame within the time window to obtain the possible positions where the target exists; Regarding the constant false alarm rate results of each frame as a vote, and then accumulating the voting results of all frames within the time window; When the cumulative vote count in a certain range bin reaches half of the time window length, it is considered that the range bin is the position of the target; Calculating the average Euclidean distance between the corresponding front and rear frames of the time window, and selecting the window with the smallest Euclidean distance as the window with the most stable human movement, thereby obtaining the subscript of the target stable frame; Calculating the centroid positions of all connected components, and performing connected component matching forward and backward starting from the time window adjacent to the stable time window; Matching each connected component in the time window where the target position has been determined in the adjacent time window with each connected component in the time window where the target position has not been determined one by one, and defining the matching score as the reciprocal of the Euclidean distance between the centroids of the two connected components; Retaining the connected component with the highest matching score in the time window where the target position has not been determined. If the highest score is not obtained in the matching with all previous window connected components, it is considered that the connected component is a connected component of motion artifacts, and mask elimination processing is performed in the constant false alarm rate result cumulative matrix; After determining the new target position, calculating the offset amount compensated towards the stable window to obtain the compensation offset amount.

3. The method for detecting a person's vital signs based on body movement compensation according to claim 2, wherein, Downsampling the corrected signal, and removing low-frequency and large-amplitude movement trends through B-spline curve fitting to suppress harmonic interference of small-amplitude random movements, including: Downsampling the original signal to avoid overfitting of the respiratory phase. Using a B-spline curve to fit the low-frequency and large-amplitude target movement trends and eliminating them from the original phase sequence; Subtracting the original artifacts from the original phase to obtain a cleaner signal with richer respiratory and heartbeat signals.

4. The method for detecting a person's vital signs based on body movement compensation according to claim 3, wherein, Extracting vital signs from the signal after body movement compensation, including: Separating respiratory and heartbeat signals based on an improved multi-channel variational mode decomposition algorithm.

5. The method for detecting a person's vital signs based on body movement compensation according to claim 1, wherein, Performing beamforming processing on the three-dimensional data matrix, including: Setting multiple groups of phase compensation weights for the received signal according to the target direction angle; Enumerating each group of weights and performing weighted superposition on the signal, and selecting the group of weights with the strongest target signal energy for phase compensation to obtain an enhanced target reflection signal.

6. The method for detecting a person's vital signs based on body movement compensation according to claim 5, characterized in that, Performing static clutter removal on the target reflection signal, including: Setting a time window, calculating the average value of the target reflection signal along the time dimension within the window to estimate the background noise; Extracting the dynamic reflection signal according to the target reflection signal and the estimated value of the background noise.

7. The method for detecting a person's vital signs based on body movement compensation according to claim 6, wherein Perform target detection on the dynamic reflection signal and locate the target area based on a voting mechanism, including: In the multi-frame constant false alarm rate detection results, accumulate the number of times each target area is marked into a matrix; Set a threshold, screen the accumulated results, and retain the target areas with accumulated scores greater than the threshold.

8. A personnel vital sign detection system based on body movement compensation, characterized in that, Including: The first main module is used to obtain echo data of a target individual based on a millimeter-wave radar and generate a three-dimensional data matrix; The second main module is used to perform beamforming processing, static clutter removal, and target detection on the three-dimensional data matrix in sequence to obtain a target area; The third main module is used to perform two-stage body motion compensation on the target area, including: calculating the displacement between adjacent frames using a two-dimensional cross-correlation method to correct the position offset of the target area; downsampling the corrected signal, and removing the low-frequency and large-amplitude motion trends through B-spline curve fitting to suppress the harmonic interference of small random motions; The fourth main module is used to extract vital signs from the signal after body motion compensation.

9. An electronic device, characterized in that, It includes a memory and a processor. The memory stores program instructions executed by the processor, and the processor calls the program instructions to execute the steps of a method for detecting human vital signs based on body motion compensation according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions cause the computer to execute the steps of a method for detecting human vital signs based on body motion compensation according to any one of claims 1 to 7.

Images