Far infrared radar non-contact sign detection method and device

Abstract

The application relates to the technical field of sign detection, and discloses a far-infrared radar non-contact sign detection method and device, which comprises the following steps: collecting radar echo data and body temperature data of multiple target objects in a detection area through a far-infrared sensor and a millimeter wave radar; performing heartbeat and respiration signal analysis on the radar echo data to generate first heartbeat parameters and first respiration parameters, and simultaneously performing body temperature distribution analysis on the body temperature data to generate body temperature distribution parameters; calculating second heartbeat parameters and second respiration parameters of each target object based on the first heartbeat parameters and the first respiration parameters; and generating a sign detection result of each target object according to the second heartbeat parameters, the second respiration parameters and the body temperature distribution parameters. The application can automatically identify heartbeat, respiration and body temperature abnormal states and generate a comprehensive health evaluation result, thereby improving the accuracy of non-contact sign detection.

Description

Technical Field

[0001] This invention relates to the field of vital sign detection technology, and in particular to a non-contact vital sign detection method and device using far-infrared radar. Background Technology

[0002] Existing non-contact vital sign detection technologies mainly rely on single sensing technologies. For example, pure infrared thermal imaging technology can only detect body temperature information, while pure radar technology can detect heartbeat and respiration but cannot acquire body temperature data. Furthermore, signal separation is difficult in multi-target scenarios. Current signal processing methods mostly employ fixed-parameter filtering algorithms, lacking adaptive optimization capabilities based on physiological signal characteristics. They cannot dynamically adjust according to the differences in physiological features among individuals, resulting in low accuracy in vital sign detection using current technologies. Summary of the Invention

[0003] This invention provides a non-contact vital sign detection method and device using far-infrared radar. This invention can automatically identify abnormal states of heartbeat, respiration, and body temperature and generate comprehensive health assessment results, thereby improving the accuracy of non-contact vital sign detection.

[0004] In a first aspect, the present invention provides a non-contact vital sign detection method using far-infrared radar, the far-infrared radar non-contact vital sign detection method comprising:

[0005] The radar echo data and body temperature data of multiple target objects in the detection area are collected using far-infrared sensors and millimeter-wave radar.

[0006] The radar echo data is analyzed to generate heartbeat and respiratory signals, and first heartbeat parameters and first respiratory parameters are generated. At the same time, the body temperature data is analyzed to generate body temperature distribution parameters.

[0007] Calculate the second heartbeat parameter and the second respiratory parameter for each target object based on the first heartbeat parameter and the first respiratory parameter;

[0008] The vital signs detection results for each target object are generated based on the second heart rate parameter, the second respiratory parameter, and the body temperature distribution parameter.

[0009] In conjunction with the first aspect, in a first implementation of the first aspect of the present invention, the step of collecting radar echo data and body temperature data of multiple target objects in the detection area using a far-infrared sensor and a millimeter-wave radar includes:

[0010] The original echo signal is obtained by transmitting a preset electromagnetic wave signal to the detection area using millimeter-wave radar and receiving the Doppler frequency shift echo generated by the chest cavity movement of the target object.

[0011] The original echo signal is converted from analog to digital to obtain radar echo data;

[0012] The body temperature data of the target object's body surface within the detection area is collected using a far-infrared sensor.

[0013] In conjunction with the first aspect, in a second implementation of the first aspect of the present invention, the step of analyzing the radar echo data for heart and respiratory signals to generate first heartbeat parameters and first respiratory parameters, and simultaneously analyzing the body temperature data for body temperature distribution to generate body temperature distribution parameters, includes:

[0014] The radar echo data is analyzed for heart rate signals to obtain heart rate characteristic data, and the radar echo data is also processed for respiratory fundamental frequency isolation to obtain respiratory characteristic data.

[0015] The Bio-RCS values ​​of the heartbeat feature data and the respiratory feature data are calculated respectively, and the RR interval variability and respiratory interval variability of each target object are extracted to obtain the first heartbeat parameter and the first respiratory parameter.

[0016] Based on the body temperature data, multi-target temperature field reconstruction and spatial temperature distribution calculation are performed to identify the body temperature distribution parameters of each target object.

[0017] In conjunction with the first aspect, in a third implementation of the first aspect of the present invention, the step of performing heartbeat frequency signal analysis on the radar echo data to obtain heartbeat feature data, and simultaneously performing respiratory fundamental frequency isolation processing on the radar echo data to obtain respiratory feature data, includes:

[0018] The radar echo data is separated according to the preset heartbeat frequency band and respiratory frequency band to obtain heartbeat frequency domain signal and respiratory frequency domain signal;

[0019] Electromagnetic scattering calculations are performed on the heartbeat frequency domain signal to obtain the heartbeat scattering component, and electromagnetic scattering calculations are performed on the respiratory frequency domain signal to obtain the respiratory scattering component;

[0020] The heart rate harmonic interference in the heartbeat scattering component is iteratively optimized and eliminated to obtain an interference-free heartbeat signal. At the same time, the respiratory scattering component is subjected to fundamental frequency isolation filtering to obtain a pure respiratory signal.

[0021] Wavelet transform and cross-coupling correction are performed on the de-interference heartbeat signal to obtain heartbeat feature data, and wavelet transform is performed on the pure respiratory signal to obtain respiratory feature data.

[0022] In conjunction with the first aspect, in the fourth implementation of the first aspect of the present invention, the step of calculating the Bio-RCS values ​​of the heartbeat feature data and the respiratory feature data respectively, and extracting the RR interval variability and respiratory interval variability of each target object to obtain the first heartbeat parameter and the first respiratory parameter includes:

[0023] Based on the heartbeat feature data, the heartbeat bio-radar cross section is calculated to obtain the first heartbeat Bio-RCS value. At the same time, based on the respiratory feature data, the respiratory bio-radar cross section is calculated to obtain the first respiratory Bio-RCS value.

[0024] Peak detection is performed on the heartbeat feature data to identify heartbeat peak sequences, and respiratory cycle detection is performed on the respiratory feature data to identify respiratory peak sequences.

[0025] The RR interval sequence is calculated based on the time difference between adjacent R wave peaks in the heart rate peak sequence, and the respiratory interval sequence is calculated based on the time difference between adjacent respiratory peaks in the respiratory peak sequence.

[0026] Statistical analysis of variance was performed on the RR interval sequences to obtain the first heart rate variability parameter for each target subject, and statistical analysis of variance was performed on the respiratory interval sequences to obtain the first respiratory variability parameter for each target subject.

[0027] The first heartbeat parameter is generated by combining the first heartbeat Bio-RCS value and the first heart rate variability parameter, and the first respiratory parameter is generated by combining the first respiratory Bio-RCS value and the first respiratory variability parameter.

[0028] In conjunction with the first aspect, in a fifth implementation of the first aspect of the present invention, the step of calculating the second heartbeat parameter and the second respiratory parameter for each target object based on the first heartbeat parameter and the first respiratory parameter includes:

[0029] Spatial positioning calculations are performed on the radar echo data to obtain the spatial azimuth information of each target object;

[0030] Based on the spatial azimuth information, the preset controller is driven to calculate the beamforming weight vector, and the directional enhancement weight matrix of each target object is generated according to the beamforming weight vector.

[0031] Based on the targeted enhancement weight matrix, target separation is performed on the first heartbeat Bio-RCS value and the first heart rate variability parameter in the first heartbeat parameter to obtain the second heartbeat Bio-RCS value and the second heart rate variability parameter for each target object.

[0032] Based on the targeted enhancement weight matrix, target separation is performed on the first respiratory Bio-RCS value and the first respiratory variability parameter in the first respiratory parameter to obtain the second respiratory Bio-RCS value and the second respiratory variability parameter for each target object.

[0033] The second heart rate Bio-RCS value and the second heart rate variability parameter of each target object are combined into a second heart rate parameter, and the second respiratory Bio-RCS value and the second respiratory variability parameter of each target object are combined into a second respiratory parameter.

[0034] In conjunction with the first aspect, in the sixth implementation of the first aspect of the present invention, the step of performing spatial positioning calculations on the radar echo data to obtain the spatial azimuth information of each target object includes:

[0035] Calculate the physiological signal correlation matrix based on the radar echo data;

[0036] Matrix decomposition is performed on the physiological signal correlation matrix to obtain a sequence of feature values ​​reflecting each target object;

[0037] The number of targets is determined based on the magnitude of each feature value in the feature value sequence, and the azimuth matrix of each target object is calculated based on the relationship between the feature values, the array element spacing, and the working wavelength.

[0038] The azimuth matrix is ​​corrected to obtain the spatial azimuth information of each target object.

[0039] In conjunction with the first aspect, in the seventh implementation of the first aspect of the present invention, the step of driving a preset controller to calculate a beamforming weight vector based on the spatial azimuth information, and generating a directional enhancement weight matrix for each target object according to the beamforming weight vector, includes:

[0040] The spatial azimuth information is used as input to drive the preset controller to calculate the ratio of the target signal power to the interference signal power, obtain the power ratio, and determine the beam pointing vector of each target object based on the power ratio.

[0041] The phase delay difference between the M array elements in the millimeter-wave radar and each target object is calculated based on the beam pointing vector, and the array steering vector of each target object is generated based on the phase delay difference.

[0042] Based on the array steering vector, minimum variance distortionless response calculation is performed to obtain the beamforming weight vector of each target object;

[0043] The beamforming weight vector is reorganized into a matrix according to the target object number to obtain the directional enhancement weight matrix for each target object.

[0044] In conjunction with the first aspect, in the eighth implementation of the first aspect of the present invention, the step of generating the vital sign detection results for each target object based on the second heart rate parameter, the second respiratory parameter, and the body temperature distribution parameter includes:

[0045] Based on the comparison between the second heartbeat Bio-RCS value and the second heart rate variability parameter in the second heartbeat parameter and the preset normal heartbeat range, the heartbeat detection status of each target object is identified.

[0046] Based on the comparison between the second respiratory Bio-RCS value and the second respiratory variability parameter in the second respiratory parameter and the preset normal respiratory range, the respiratory detection status of each target object is identified.

[0047] Based on the comparison between the body temperature distribution parameters and the preset normal body temperature range, the body temperature detection status of each target object is identified.

[0048] Based on the heart rate, respiration, and body temperature of each target object, and combined with the spatial orientation information of the target object, the vital signs detection results of each target object are obtained.

[0049] Secondly, the present invention provides a far-infrared radar non-contact vital sign detection device, the far-infrared radar non-contact vital sign detection device comprising:

[0050] The acquisition module is used to acquire radar echo data and body temperature data of multiple target objects in the detection area through far-infrared sensors and millimeter-wave radar;

[0051] The analysis module is used to analyze the radar echo data for heartbeat and respiratory signals, generate first heartbeat parameters and first respiratory parameters, and simultaneously analyze the body temperature data for body temperature distribution, generating body temperature distribution parameters.

[0052] The calculation module is used to calculate the second heartbeat parameter and the second respiratory parameter for each target object based on the first heartbeat parameter and the first respiratory parameter;

[0053] The generation module is used to generate the vital sign detection results for each target object based on the second heartbeat parameter, the second respiratory parameter, and the body temperature distribution parameter.

[0054] The technical solution provided by this invention integrates far-infrared sensors and millimeter-wave radar technology to achieve simultaneous non-contact detection of three physiological parameters: body temperature, heart rate, and respiratory rate, overcoming the limitations of incomplete information acquisition by existing single technologies. By establishing a far-infrared electromagnetic propagation physiological signal dynamic model and full-wave analysis of Maxwell's equations, it effectively separates multi-target cross-coupling interference, solving the technical problem of physiological signal aliasing in multi-target environments. A physiological signal-guided Harris Hawks optimization algorithm is used to achieve adaptive parameter adjustment, specifically optimized for the 0.8-2.0Hz heart rate and 0.1-0.5Hz respiratory frequency bands, significantly improving detection accuracy. Combined with physiological signal spatial angle estimation and a multi-target reinforcement learning controller, it achieves accurate target localization and identification, automatically identifying abnormal heart rate, respiration, and body temperature states and generating comprehensive health assessment results. Attached Figure Description

[0055] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 This is a schematic diagram of the steps of the far-infrared radar non-contact vital sign detection method in an embodiment of the present invention;

[0057] Figure 2 This is a schematic diagram of the structure of the far-infrared radar non-contact vital sign detection device in an embodiment of the present invention. Detailed Implementation

[0058] This invention provides a method and apparatus for non-contact vital sign detection using far-infrared radar. The terms "first," "second," "third," "fourth," etc. (if present)," in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0059] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1One embodiment of the far-infrared radar non-contact vital sign detection method of the present invention includes:

[0060] Step S1: Collect radar echo data and body temperature data of multiple target objects in the detection area using far-infrared sensors and millimeter-wave radar;

[0061] Specifically, a high-precision millimeter-wave radar transmitter is deployed in the detection area. This transmitter continuously emits electromagnetic wave signals of specific waveforms according to a preset frequency and modulation scheme. When these signals encounter the small, periodic movements of parts of the human body, such as the chest cavity, due to breathing or heartbeat rhythms, they form echo signals containing Doppler frequency shift characteristics. These echo signals are spatially distributed and received by a 64-element receiver array, acquiring the original radar reflection signals covering multiple target objects. The system's integrated high-resolution analog-to-digital converter performs real-time analog-to-digital conversion on the original analog radar echo signals, converting the analog signals into radar echo data at a sampling frequency of 1000Hz while preserving the phase difference and amplitude variations between multiple channels. Simultaneously, a far-infrared sensor module, without interfering with the millimeter-wave radar signal, synchronously collects the surface thermal radiation data of all target objects in the detection area based on the 8–14 micrometer far-infrared band. Two-dimensional thermal imaging is used to generate thermal image data with a resolution of 0.1℃ per pixel, and the correspondence between each heat source area and the target in the radar echo signal is automatically calibrated, thus achieving the fusion acquisition of body temperature and radar motion data in the same spatial domain.

[0062] Step S2: Analyze the radar echo data for heartbeat and respiratory signals to generate the first heartbeat parameter and the first respiratory parameter. At the same time, analyze the body temperature data for body temperature distribution to generate body temperature distribution parameters.

[0063] Specifically, the raw radar echo data undergoes frequency decomposition based on wavelet transform. High-frequency components corresponding to the heartbeat frequency (between 0.8Hz and 2.0Hz) are selected for target-guided filtering and time-frequency domain enhancement to extract signal components highly correlated with the heartbeat rhythm and construct heartbeat feature data. In the lower frequency domain (0.1Hz to 0.5Hz), fundamental frequency isolation and bandpass enhancement are performed on the respiratory main frequency components to filter out stray interference and cross-coupling components, extracting respiratory feature data formed by respiratory rhythm fluctuations. The Bio-RCS value of the physiological signal radar cross section for each target under the current radar view is calculated. The Bio-RCS value characterizes the variation of an individual's physiological vibration scattering ability in radar waves. Based on this, time-series variability analysis is performed on the RR interval of the heartbeat signal and the respiratory cycle interval of the respiratory signal. By extracting key statistical characteristic parameters such as fluctuation range, fluctuation frequency, and rhythm dispersion, the first heartbeat parameters and first respiratory parameters of the target individual are constructed. Meanwhile, when analyzing far-infrared body temperature data, multiple target objects are segmented by spatial thermal radiation image registration and heat source target region division technology. Temperature field inversion and isothermal region fitting are performed by combining thermal pixel intensity gradient field to form temperature spatial modeling results of multiple heat source targets. On this basis, the average body temperature, local temperature maximum, temperature distribution variance and thermal center coordinates of each target are calculated to obtain body temperature distribution parameters.

[0064] Step S3: Calculate the second heartbeat parameter and second respiratory parameter for each target object based on the first heartbeat parameter and the first respiratory parameter;

[0065] Specifically, spatial positioning calculations are performed on the radar echo data. By analyzing the phase difference, amplitude response, and time delay information between each channel in the 64-element receiver array, the independent spatial azimuth coordinates of each target individual in the radar field of view are extracted. Combined with the array structure and the radar operating wavelength, the spatial positioning information of each target object is determined. Based on the spatial azimuth information, a preset multi-target feature separation controller is driven, and the beamforming weight vector solution operation is performed by calling the minimum variance distortion-free response criterion. A set of complex weight coefficients with the strongest response to a specific direction and the least interference to non-target directions is dynamically constructed, generating a set of directional enhancement weight matrices covering all target objects. Each matrix corresponds to a beam direction enhancement channel for one target object. Using a directional enhancement weight matrix, target attribution unmixing and beam projection are performed on the first heartbeat parameter in the radar data. The enhanced heartbeat radar cross-section (Bio-RCS) value for each target individual in a specific direction is extracted. Temporal feature reconstruction and discrete rhythm identification are then performed on the original heart rate variability index to obtain the second heartbeat Bio-RCS value and second heart rate variability parameter after removing multi-target cross-interference. The controller separates the unique respiratory Bio-RCS value and respiratory rhythm variability index corresponding to each target object from the first respiratory parameter based on the amplitude fluctuation and periodic sequence of the respiratory dominant frequency in the time domain, obtaining independently identified second respiratory Bio-RCS value and second respiratory variability parameter. The second heartbeat Bio-RCS value of each target object and its corresponding second heart rate variability parameter are jointly modeled and encapsulated into a structured second heartbeat parameter. Simultaneously, the second respiratory Bio-RCS value and the second respiratory variability parameter are integrated to form the second respiratory parameter.

[0066] Step S4: Generate the vital signs detection results for each target object based on the second heartbeat parameter, the second respiratory parameter, and the body temperature distribution parameter.

[0067] Specifically, the second heartbeat Bio-RCS value and the second heart rate variability parameter are matched against physiological thresholds. These parameters are compared item by item with the preset normal human heartbeat physiological range, where the normal heartbeat range is set as a specific dynamic interval based on age group, gender, and clinical heart rate benchmark. By comparing the amplitude stability of the Bio-RCS value and the periodic regularity of the variability index, it is determined whether the individual's heartbeat status is in a normal rhythm pattern, thereby identifying the target subject's heartbeat detection status as "normal," "high," "low," or "abnormal fluctuation." The respiratory Bio-RCS value and respiratory variability index in the second respiratory parameter are processed in the same way, comparing whether their respiratory rhythm stability and signal intensity meet the definition criteria of the normal respiratory rate range, determining whether the target subject has a state of slow breathing, rapid breathing, or irregular breathing, and outputting a respiratory detection status label accordingly. Simultaneously, the body temperature distribution parameters are compared and analyzed against the normal body temperature range. Based on the average temperature, distribution variance, and temperature rise trend of multiple heat source pixels, it is determined whether the body surface temperature is at a normal level. If the temperature is higher than the fever threshold, it is marked as "high temperature abnormality," and if it is lower than the low temperature threshold, it is identified as "low temperature alarm." The system also identifies local hot spots or uneven temperature areas within the temperature field, forming a body temperature detection status identifier. The three types of detection status results for each target object are jointly analyzed with its spatial orientation information within the detection area. An orientation matching algorithm is used to establish a one-to-one correspondence between the detection results and the target individual. Based on this, a comprehensive vital sign detection result containing the target ID, heart rate status, respiratory status, body temperature status, and location information is generated.

[0068] In one specific embodiment, the process of performing step S1 may specifically include the following steps:

[0069] The original echo signal is obtained by transmitting a preset electromagnetic wave signal to the detection area using millimeter-wave radar and receiving the Doppler frequency shift echo generated by the chest cavity movement of the target object.

[0070] The original echo signal is converted from analog to digital to obtain radar echo data;

[0071] The body temperature data of the target object's body surface within the detection area is collected using a far-infrared sensor.

[0072] Specifically, a millimeter-wave radar transmitting array with high frequency stability and directional consistency is constructed. The radar transmitting array operates at a center frequency of 24.125 GHz and is modulated using continuous wave modulation. It periodically transmits preset electromagnetic wave signals into the detection area according to the array signal transmission protocol. When the electromagnetic waves encounter a human target during propagation, they are modulated by the minute mechanical vibrations of the chest cavity caused by heartbeat and respiration, resulting in time-varying phase and frequency shifts in the reflected wave. This shift phenomenon is the Doppler frequency shift signal caused by physiological micro-movements. A millimeter-wave antenna array consisting of 64 receiving channels is used for spatially distributed reception of the reflected signal, synchronously acquiring the target's reflected echo in real time. This forms a raw radar echo signal containing azimuth angle, range, and physiological modulation components. The raw radar echo signal exists in the form of analog voltage and changes continuously. An embedded analog-to-digital conversion module performs high-precision analog-to-digital conversion on the raw analog radar echo signal, with a sampling frequency set to 1000 Hz and an analog-to-digital conversion accuracy of 16 bits. This converts the analog signal into a radar echo data stream with timestamps, preserving the fine-grained temporal characteristics of the target object's motion information. Meanwhile, the far-infrared sensor module is activated synchronously during the millimeter-wave radar data acquisition process. The far-infrared detector operates in the thermal radiation sensitive band of 8 to 14 micrometers. It uses non-contact infrared sensing technology to capture high-resolution imaging of the heat naturally released by the human body surface in the detection area. It uses a thermoelectric matrix sensor array to independently track the heat source of each target individual and calculates the temperature value of its body surface through pixel temperature decoding.

[0073] In one specific embodiment, the process of performing step S2 may specifically include the following steps:

[0074] Heart rate signal analysis was performed on the radar echo data to obtain heart rate characteristic data. At the same time, respiratory fundamental frequency isolation processing was performed on the radar echo data to obtain respiratory characteristic data.

[0075] The Bio-RCS values ​​of the heartbeat feature data and respiratory feature data were calculated separately, and the RR interval variability and respiratory interval variability of each target object were extracted to obtain the first heartbeat parameter and the first respiratory parameter.

[0076] Based on body temperature data, multi-target temperature field reconstruction and spatial temperature distribution calculation are performed to identify the body temperature distribution parameters of each target object.

[0077] Specifically, a multi-channel receiving structure is deployed in the millimeter-wave radar system. A spatial array of 64 elements acquires high spatiotemporal resolution radar echo data covering the detection area. The radar echo data contains high-frequency micro-motion signals caused by heartbeats and low-frequency fluctuations caused by respiration. After receiving the radar echo data, a heart rate-band-guided wavelet transform method is used to perform frequency decomposition on the signal. A scale interval of 0.8Hz to 2.0Hz is selected to extract the time-frequency distribution energy band matching the rhythm of chest cavity contraction and expansion, thus separating the heartbeat signal with physiological rhythm characteristics and generating heartbeat feature data. Furthermore, to avoid overlap and interference between the respiratory frequency component and the heartbeat signal, bandpass filtering and wavelet reconstruction methods are applied to select scale components in the frequency range of 0.1Hz to 0.5Hz. Fundamental frequency isolation processing is performed on the relatively slow displacement changes in the radar signal to extract the periodic low-frequency respiratory feature signal caused by chest cavity expansion. Heartbeat and respiratory feature data are mapped back to the radar reflection model. Based on the target's incident angle and reflection amplitude in the radar array, the heartbeat Bio-RCS value and respiratory Bio-RCS value for each target are calculated to reflect the target's physiological rhythm's ability to scatter radar signals. The temporal variation of the RR interval sequence and respiratory cycle sequence is extracted through periodic maximum amplitude point spacing analysis. On this basis, variability measurement operations are performed on the heartbeat interval and respiratory interval using a sliding time window to calculate their standard deviation, variance, and rhythm stationarity index, generating first heartbeat parameters and first respiratory parameters with individual differences. Simultaneously, in the far-infrared body temperature data acquisition module, thermal imaging analysis algorithms are called to reconstruct the temperature field of multiple targets. The heat source contour of each target is identified by the image segmentation algorithm, and a pixel temperature matrix is ​​established. Then, combined with spatial matching coordinate mapping, the heat source regions of each target are separated from the thermal image, and temperature field modeling is performed within the region. By calculating the region's average temperature, maximum temperature, temperature gradient, and center offset, a body temperature distribution parameter set containing the local temperature characteristics and global thermal distribution characteristics of each target is formed.

[0078] In one specific embodiment, the process of performing heart rate signal analysis on the radar echo data to obtain heart rate characteristic data, and simultaneously performing respiratory fundamental frequency isolation processing on the radar echo data to obtain respiratory characteristic data, can specifically include the following steps:

[0079] The radar echo data is separated according to the preset heartbeat frequency band and respiratory frequency band to obtain the heartbeat frequency domain signal and the respiratory frequency domain signal;

[0080] Electromagnetic scattering calculations were performed on the heartbeat frequency domain signal to obtain the heartbeat scattering component, and electromagnetic scattering calculations were performed on the respiratory frequency domain signal to obtain the respiratory scattering component.

[0081] The heart rate harmonic interference in the heartbeat scattering component is iteratively optimized and eliminated to obtain an interference-free heartbeat signal. At the same time, the respiratory scattering component is subjected to fundamental frequency isolation filtering to obtain a pure respiratory signal.

[0082] Wavelet transform and cross-coupling correction are performed on the interference-free heartbeat signal to obtain heartbeat feature data, and wavelet transform is performed on the pure respiratory signal to obtain respiratory feature data.

[0083] Specifically, a frequency domain identification channel is constructed. After receiving radar echo data collected by a 64-element millimeter-wave receiving array, the frequency spectrum distribution of the radar signal in different time periods is obtained through short-time Fourier transform or continuous wavelet transform. Using the heartbeat frequency band (0.8Hz to 2.0Hz) and the respiratory frequency band (0.1Hz to 0.5Hz) as separation criteria, two bandpass intervals are extracted from the radar echo data in the frequency domain, forming heartbeat and respiratory frequency domain signals with clear frequency band boundaries. The separated heartbeat and respiratory frequency domain signals are then input into an analysis module based on an electromagnetic wave scattering model. The far-infrared radar scattering dynamics model, constructed using Maxwell's equations, calculates the change in the scattering cross-section of the radar echo in each frequency band. Based on the relationship between the array receiving angle and the target's micro-motion direction, the heartbeat and respiratory scattering components are solved using a radiation pattern function. This enables the system to identify and quantify the signal scattering behavior of the target caused by physiological activities in complex reflection environments. To address harmonic interference in the heartbeat scattering component, an iterative optimization method combined with a harmonic cancellation function is employed. In the frequency domain, integer multiples of the dominant heart rate frequency are sequentially removed. By setting an interference estimation residual threshold and a maximum number of iterations, the accuracy and computational efficiency of the elimination process are ensured. After several iterations, the heartbeat spectrum free of harmonic contamination is obtained, and the de-interference heartbeat signal is recovered through inverse transformation. Simultaneously, to improve the clarity of the respiratory signal, a fundamental frequency isolation filter is performed on the respiratory scattering component. A broadband notch filter is used to shield signals in non-fundamental frequency regions, and a low-pass filter is employed to retain the dominant respiratory frequency component, resulting in a clean respiratory signal centered on the dominant respiratory frequency with effectively suppressed sideband interference. High-precision wavelet transforms were performed on both the de-interferenced heartbeat signal and the clean respiratory signal to obtain time-frequency feature maps. The heartbeat signal processing channel used the Morlet mother wavelet function for multi-scale decomposition, extracting the heartbeat amplitude distribution at multiple scales to ensure that aperiodic abnormal fluctuations were significantly expressed. A cross-coupling correction mechanism for multi-target environments was introduced, using the mapping matrix between historical target trajectories and channel interference to demix and compensate for the coupling parts of adjacent targets in the heartbeat features, generating heartbeat feature data. Simultaneously, the clean respiratory signal was input into the wavelet decomposition path, and the rhythmicity and amplitude changes of the respiratory cycle were modeled using scale parameters that matched the respiratory frequency structure, outputting respiratory feature data including respiratory frequency, respiratory rhythm amplitude, and low-frequency trend change characteristics.

[0084] In one specific embodiment, the process of calculating the Bio-RCS values ​​of heartbeat feature data and respiratory feature data respectively, and extracting the RR interval variability and respiratory interval variability of each target object to obtain the first heartbeat parameter and the first respiratory parameter can specifically include the following steps:

[0085] Based on heartbeat feature data, the heartbeat bio-radar cross section is calculated to obtain the first heartbeat Bio-RCS value. At the same time, based on respiratory feature data, the respiratory bio-radar cross section is calculated to obtain the first respiratory Bio-RCS value.

[0086] Peak detection is performed on heartbeat feature data to identify heartbeat peak sequences, and respiratory cycle detection is performed on respiratory feature data to identify respiratory peak sequences.

[0087] The RR interval sequence is calculated based on the time difference between adjacent R wave peaks in the heart rate peak sequence, and the respiratory interval sequence is calculated based on the time difference between adjacent respiratory peaks in the respiratory peak sequence.

[0088] Statistical analysis of variance was performed on the RR interval sequences to obtain the first heart rate variability parameter for each target subject, and statistical analysis of variance was performed on the respiratory interval sequences to obtain the first respiratory variability parameter for each target subject.

[0089] The first heartbeat parameter is generated by combining the first heartbeat Bio-RCS value and the first heart rate variability parameter, and the first respiratory parameter is generated by combining the first respiratory Bio-RCS value and the first respiratory variability parameter.

[0090] Specifically, based on heartbeat characteristic data, and according to the radiation pattern response model and target azimuth information in the array radar system, a physiological micro-motion scattering modeling method is used to calculate the heartbeat bio-radar cross section. The calculation is based on the scattering intensity of the target's heartbeat amplitude in the millimeter-wave band. The reflected energy after directional enhancement is extracted by integrating the signal power within each time period and normalized in the space-time domain to obtain the first heartbeat Bio-RCS value for each target, reflecting the effective electromagnetic scattering capability caused by the target's heartbeat. Simultaneously, the same electromagnetic scattering modeling process is performed on respiratory characteristic data, but its scattering rhythm is slower than that of the heartbeat signal. Therefore, a long-period integration window and a fundamental frequency enhancement weight parameter are introduced during the calculation to ensure that the small but periodically stable reflection behavior of respiratory amplitude is fully captured, yielding the corresponding first respiratory Bio-RCS value. Peak detection is performed on heartbeat characteristic data. This involves identifying local maxima along the time axis of the heartbeat signal and using a heart rate scale window to correct and filter out abnormal peaks, outputting a continuous sequence of heartbeat peaks. Each peak is considered the corresponding moment of an R wave. Based on this heartbeat peak sequence, the time difference between adjacent R wave peaks is calculated sequentially to construct the RR interval sequence for the target individual over a continuous time period. For respiratory characteristic data, respiratory cycle detection is performed simultaneously. This involves identifying periodic peaks in the respiratory signal—the peaks formed by each inhalation or exhalation—and constructing a respiratory peak sequence. The respiratory interval sequence is then established using the time difference between adjacent respiratory peaks. Statistical variance analysis is performed on both heartbeat and respiratory intervals, including calculating the variance, standard deviation, and coefficient of variation of the interval sequences. This reflects the heart rate variability and respiratory rhythm variability of the target individual during the detection period. Higher heart rate variability indicates stronger cardiac autoregulation, while respiratory variability relates to the respiratory system's responsiveness to changes in internal and external loads. The first heartbeat Bio-RCS value is combined with the corresponding first heart rate variability parameter to generate a structured first heartbeat parameter. At the same time, the first breath Bio-RCS value is integrated with the first breath variability parameter to generate the corresponding first breath parameter.

[0091] In one specific embodiment, the process of performing step S3 may specifically include the following steps:

[0092] Spatial positioning calculations are performed on radar echo data to obtain the spatial azimuth information of each target object;

[0093] Based on spatial azimuth information, a preset controller is driven to calculate the beamforming weight vector, and a directional enhancement weight matrix for each target object is generated according to the beamforming weight vector.

[0094] Based on the targeted enhancement weight matrix, target separation is performed on the first heartbeat Bio-RCS value and the first heart rate variability parameter in the first heartbeat parameter to obtain the second heartbeat Bio-RCS value and the second heart rate variability parameter for each target object.

[0095] Based on the targeted enhancement weight matrix, target separation is performed on the first respiratory Bio-RCS value and the first respiratory variability parameter in the first respiratory parameter to obtain the second respiratory Bio-RCS value and the second respiratory variability parameter for each target object;

[0096] The second heart rate Bio-RCS value and the second heart rate variability parameter of each target object are combined into a second heart rate parameter, and the second respiratory Bio-RCS value and the second respiratory variability parameter of each target object are combined into a second respiratory parameter.

[0097] Specifically, an array signal direction estimation algorithm is executed on the radar echo data acquired by the 64-element millimeter-wave radar receiving array. This algorithm analyzes the phase difference and cross-correlation matrix between each element, and calculates the arrival direction of each target in two-dimensional angular space based on the energy distribution of the physiological rhythm signal. Improved spatial spectrum estimation methods, such as the MUSIC or SAGE algorithms, are used to extract the incident azimuth and elevation angles of multiple signal sources, thereby calibrating the spatial azimuth information of each target. Based on this spatial azimuth information, the built-in multi-target beamforming control module is activated. This module calls the minimum variance distortionless response criterion or the adaptive linear constraint minimum variance algorithm to calculate a corresponding beamforming weight vector for each azimuth angle. The beamforming weight vector represents the phase offset and gain factor that needs to be applied to each element in the 64-element array, in order to construct a receiving channel with maximum signal gain in a specific direction and minimal interference in other directions. By performing parallel weight calculations on the azimuth angles of all targets, multiple directional enhancement channels are constructed, and a set of directional enhancement weight matrices indexed by target numbers is generated. Each column in the matrix corresponds to the weight allocation of array elements, and each row corresponds to the target direction. Using the directional enhancement weight matrices, signal separation operations are performed on the heartbeat Bio-RCS value and heart rate variability parameter contained in the first heartbeat parameter. The multi-target mixed signal in the original radar echo is directionally demixed through array weighted combination, significantly enhancing the signal matching the specified target direction while suppressing signals from other directions. Directional gain can be used to suppress cross-coupling effects and improve the accuracy of heartbeat scattering energy estimation for each target, thereby extracting the second heartbeat Bio-RCS value for each target object in its own direction. Simultaneously, the heart rate variability sequence is also subjected to time-domain filtering and channel demixing using the same beam processing method, ensuring that the periodic data used for variability analysis comes from a clean single-target signal channel, thus extracting the second heart rate variability parameter. Using the same procedure, beamforming and signal separation were performed on the respiratory Bio-RCS values ​​and respiratory variability parameters included in the first respiratory parameters. The mixed respiratory signals were projected onto subspaces corresponding to the azimuths of each target using a directional weight matrix. This removed multi-target superposition interference and extracted the independent second respiratory Bio-RCS values ​​for each target. Simultaneously, respiratory variability was recalculated based on the cleaned respiratory interval sequence to obtain the second respiratory variability parameters. The second heartbeat Bio-RCS values ​​and second heart rate variability parameters for each target were combined according to time sequence and data structure to form the second heartbeat parameters; and the second respiratory Bio-RCS values ​​and second respiratory variability parameters were paired and encapsulated to form the second respiratory parameters.

[0098] Before performing spatial positioning calculations on radar echo data to obtain the spatial azimuth information of each target object, the process includes a step of enhancing the physiological features of multiple targets through a physiological signal radar fusion network: Radar echo data is input into a heartbeat radar echo encoder and a respiratory radar echo encoder at frequencies of 0.8-2.0Hz in the heartbeat frequency domain and 0.1-0.5Hz in the respiratory frequency domain, respectively, for parallel feature extraction. The heartbeat encoder uses a one-dimensional convolutional neural network with a kernel size of 25 to match the heartbeat signal cycle, and the respiratory encoder uses a one-dimensional convolutional neural network with a kernel size of 100 to match the respiratory signal cycle, resulting in heartbeat convolutional feature maps and respiratory convolutional feature maps. Learnable physiological signal feature labels are integrated into each processing layer of the heartbeat and respiratory convolutional feature maps. An attention mechanism is used to adaptively fuse the physiological signal feature labels with the convolutional features, and the feature label parameters are dynamically adjusted based on the physiological signal differences of each target object, resulting in label-enhanced heartbeat features and label-enhanced respiratory features. Label-enhanced heartbeat and respiratory features are input into a multi-channel radar echo fusion module. This module integrates a dedicated physiological signal adaptive filter, which adjusts the filter parameters in real time using a physiological signal radar scattering coefficient correction algorithm. The scattering coefficient correction factor is calculated based on the ratio of the bio-radar scattering cross-section of each target object to a standard reference value, resulting in adaptively corrected fusion features. A physiological feature extraction matrix is ​​constructed based on the adaptively corrected fusion features. This matrix captures the unique physiological scattering patterns and frequency domain response characteristics of each target object by encoding the physiological feature differences of different individuals in the far-infrared band. Combined with spatial correlation analysis of physiological signals between targets, an enhanced multi-target physiological feature representation vector is obtained. The enhanced multi-target physiological feature representation vector is used as improved radar echo data and input into a subsequent spatial positioning calculation module. Through physiological feature enhancement processing, the signal separation and target recognition accuracy in multi-target environments are significantly improved, resulting in physiologically enhanced radar data for precise spatial positioning.

[0099] In one specific embodiment, the process of performing spatial positioning calculations on radar echo data to obtain the spatial azimuth information of each target object can specifically include the following steps:

[0100] Calculate the physiological signal correlation matrix based on radar echo data;

[0101] Matrix decomposition is performed on the physiological signal correlation matrix to obtain the feature value sequence reflecting each target object;

[0102] The number of targets is determined based on the magnitude of each eigenvalue in the eigenvalue sequence, and the azimuth matrix of each target object is calculated based on the relationship between the eigenvalues, the array element spacing, and the working wavelength.

[0103] The azimuth matrix is ​​corrected to obtain the spatial azimuth information of each target object.

[0104] Specifically, the multi-channel radar echo data is preprocessed. The radar echo data is collected by 64 array elements simultaneously within a time period, forming 64 channels of time-series signals. Each channel contains weak modulation signals from multiple target objects caused by chest cavity micro-movements due to heartbeat and respiration. Based on frequency separation or filtering, the effective frequency band components related to physiological signals in the radar echo data are extracted and constructed into a unified time-window radar data matrix, the dimension of which is the number of array elements multiplied by the sampling duration. Based on the time-window radar data matrix, the spatial correlation between signals is calculated. The conjugate product between every two array element channels is statistically averaged using the expected value form to construct a physiological signal correlation matrix containing the interrelationships between all array elements. The physiological signal correlation matrix is ​​a 64×64 complex-valued covariance matrix, expressing the spatial structural relationship of the physiological micro-movement information collected by each channel of the radar array within the monitoring period. Eigenvalue decomposition is performed on the physiological signal correlation matrix. The Hermitian matrix eigenvalue decomposition method from linear algebra is invoked to decompose the correlation matrix into a product of an eigenvector matrix and a diagonal eigenvalue matrix, yielding an eigenvalue sequence associated with the number of targets. In this eigenvalue sequence, each non-zero or significantly non-attenuated eigenvalue is considered a target object with an independent physiological signal source in the current space. All eigenvalues ​​in the sequence are sorted and analyzed from largest to smallest amplitude. Based on a set signal-to-noise ratio threshold, eigenvalues ​​are filtered to remove minor features caused by noise, retaining only the principal eigenvalues ​​corresponding to valid physiological signal targets, thus achieving automatic target quantity determination and detection. After determining the number of targets, the azimuth angle of each target is calculated using the eigenvector corresponding to each significant eigenvalue and the array structure, combined with the element spacing d and the working wavelength λ, through the eigenspectral method. This process employs a spatial spectrum estimation algorithm, such as the MUSIC algorithm in subspace projection methods. The MUSIC algorithm utilizes the orthogonality between the noise subspace and the signal subspace, projecting the array steering vector and the feature subspace at different angles to obtain the response intensity of the spatial spectrum function. It then identifies the peak positions within the response function, and the angle corresponding to each peak is the estimated azimuth angle of the target object within the radar array's view. By projecting and scanning all detected principal eigenvalues, a preliminary azimuth matrix of multiple targets is formed. Error correction is then performed on the initially obtained azimuth matrix. The correction process combines system calibration data with historical observation data to perform adaptive fitting correction, or performs multi-angle joint optimization based on maximum likelihood estimation, to obtain the spatial azimuth information of each target object.

[0105] In one specific embodiment, the process of executing the steps of driving a preset controller to calculate the beamforming weight vector based on spatial azimuth information and generating the directional enhancement weight matrix for each target object according to the beamforming weight vector can specifically include the following steps:

[0106] The spatial azimuth information is used as input state to drive the preset controller to calculate the ratio of target signal power to interference signal power, obtain the power ratio, and determine the beam pointing vector of each target object based on the power ratio.

[0107] The phase delay difference between the M array elements in the millimeter-wave radar and each target object is calculated based on the beam pointing vector, and the array steering vector of each target object is generated based on the phase delay difference.

[0108] Based on the array steering vector, minimum variance distortionless response calculation is performed to obtain the beamforming weight vector of each target object;

[0109] The beamforming weight vector is reorganized into a matrix according to the target object number to obtain the directional enhancement weight matrix for each target object.

[0110] Specifically, the spatial azimuth information of each target object is input as a state variable into the intelligent beam control module. The intelligent beam control module has a pre-built action decision-maker based on deep reinforcement learning or traditional heuristic algorithms, and dynamically evaluates the control strategy with the objective function of maximizing the target signal power while minimizing the interference signal power. During control, the array received signal power in the direction corresponding to each target is estimated. The power ratio between the target and interference is calculated by combining the signal energy in the corresponding direction from the radar echo data with the distribution of non-target energy in the background signal. This power ratio is used as a reference indicator for the current beam pointing decision. By analyzing the gradient distribution of the power ratio at different azimuth angles, a set of beam pointing vectors that maximize the power ratio is determined. Each vector contains the optimal receiving direction angle for a target, representing the spatial direction in which the antenna array focuses energy on that target. Phase delay difference calculation is performed based on each beam direction and array structure parameters. For multiple elements in the radar array, the phase delay difference of each element in the spatial direction is calculated based on the geometric relationship between the target azimuth angle, element spacing, and millimeter-wave operating wavelength, establishing a phase difference vector representing the array response delay law. The phase differences of all array elements are uniformly mapped to complex numbers to construct the array steering vector for each target object in its corresponding direction, reflecting the receiving gain and phase coupling relationship of all array elements in that direction. Based on the array steering vector, a minimum variance distortionless response beamforming algorithm is used to calculate the weight vector for each target direction. This minimizes the array output's response to signals in other directions while ensuring no distortion in the signal amplitude in the desired direction, achieving spatially selective interference suppression. In the specific calculation process, a noise covariance matrix is ​​constructed based on the array's received signal, and a series of matrix transformations are performed in conjunction with the target steering vector to obtain the optimal beamforming weight vector. The optimal beamforming weight vector defines the complex weight coefficient of each antenna element in the array in the target direction, containing both amplitude adjustment and phase modulation information. After calculating the beamforming weight vector for all targets, a matrix-level recombination operation is performed on all weight vectors according to the target number. The weight vector of each target is concatenated into a weight matrix in column vector form, where each column corresponds to a target and each row corresponds to the complex weight of an array element in all target directions, thus forming a directional enhancement weight matrix.

[0111] In one specific embodiment, the process of performing step S4 may specifically include the following steps:

[0112] Based on the comparison between the second heartbeat Bio-RCS value and the second heart rate variability parameter in the second heartbeat parameters and the preset normal heartbeat range, the heartbeat detection status of each target object is identified.

[0113] The respiratory detection status of each target object is identified by comparing the second respiratory Bio-RCS value and the second respiratory variability parameter with the preset normal respiratory range.

[0114] The body temperature distribution parameters are compared with the preset normal body temperature range to determine the body temperature detection status of each target object.

[0115] Based on the heart rate, respiration, and body temperature of each target object, and combined with the spatial orientation information of the target object, the vital signs detection results of each target object are obtained.

[0116] Specifically, the second heartbeat Bio-RCS value and the second heart rate variability parameter are used as basic input variables. By comparing them with the preset normal physiological range of heartbeat, the heartbeat status is identified and judged. Among them, the second heartbeat Bio-RCS value mainly reflects the radar scattering intensity caused by the heartbeat of the target individual. If the second heartbeat Bio-RCS value is significantly lower than the reflection threshold caused by normal physiological movement, there are abnormal signals such as circulatory system inhibition, weak heartbeat, or insufficient physiological activity. At the same time, the rhythm stability analysis of the second heart rate variability parameter is performed. When the standard deviation, variance, or coefficient of variation of the RR interval is abnormally high, it indicates that the target has arrhythmia, abnormal sympathetic nerve tone, or autonomic nervous system regulatory imbalance. Conversely, if the variability is too low, there is a risk of bradycardia, sinus arrest, or restricted electrical conduction. The absolute amplitude and variability of the Bio-RCS are compared with the temporal distribution characteristics of a preset heart rate range model, and heart rate detection status labels such as "normal," "too fast," "too slow," or "abnormal" are output according to the degree of matching. The same logic is used to compare and judge the second respiratory Bio-RCS value and the second respiratory variability parameter in the second respiratory parameters. The respiratory Bio-RCS value mainly reflects the low-frequency scattering behavior caused by the periodic fluctuations of the chest wall. If its energy amplitude is consistently low, it indicates shallow and weak breathing; if it suddenly increases or fluctuates frequently, it is associated with problems such as rapid breathing, drastic changes in intrathoracic pressure, or respiratory system restriction. Simultaneously, by detecting the time series changes of the respiratory intervals in continuous respiratory cycles and performing variability analysis, when obvious periodicity, apnea, or over-rate breathing are observed, it is determined whether the target subject's breathing has functional impairment or irregular patterns. Combined with the scattering structure characteristics of the Bio-RCS, respiratory detection statuses such as "normal breathing," "shallow and weak," "rapid," and "abnormal rhythm" are output. Simultaneously, temperature spatial modeling and static feature matching analysis are performed on body temperature distribution parameters. Based on temperature field images acquired by far-infrared sensors, information such as the target body surface average temperature, local extrema, and temperature gradient distribution are extracted. These temperature features are then compared item by item with a preset normal body temperature range (e.g., 36.1°C to 37.2°C). If the detection result is higher than the preset normal body temperature range, it is marked as "high temperature"; if it is lower than the lower limit, it is identified as "low temperature." Obvious temperature unevenness or abnormal hot spots indicate inflammation or circulatory abnormalities. Each target object is assigned a structured body temperature detection status result. The heart rate, respiration, and body temperature detection status of each target object are structurally fused, and combined with spatial orientation information, spatial labels are bound to each detection result to ensure accurate matching of their attribution relationships in a multi-target environment.This is used to construct a vital sign detection result recording unit. Each unit contains three types of vital sign judgment information: target spatial location identifier (e.g., azimuth coordinates), heart rate status, respiratory status, and body temperature status. This constitutes the core health status output module of the multi-target non-contact physiological sensing system.

[0117] Before obtaining the vital sign detection results for each target object based on its heartbeat, respiration, and body temperature detection status, combined with its spatial location information, the process includes a step of identifying individual physiological rhythm features through a physiological rhythm difference attention module. This involves dividing the second heartbeat and second respiration parameters into continuous physiological signal time-series segments according to time windows, and calculating the physiological signal time-series difference equations for the heartbeat radar echo embedding and respiration radar echo embedding within adjacent time windows to obtain a time-series difference vector reflecting individual physiological rhythm changes. Based on this time-series difference vector, a multi-target physiological rhythm difference matrix is ​​constructed. By calculating the time-dependent characteristics of the RR interval variability and respiratory interval variability of each target object within different time windows, and combining this with the time-varying characteristics of the individual physiological scattering coefficient, a rhythm feature matrix characterizing the unique physiological rhythm pattern of each target object is obtained. Finally, the rhythm feature matrix is ​​input into the difference-level attention weights. The redistribution algorithm quantifies individual rhythm differences, normalizes the physiological rhythm difference coefficients of each target object using the Softmax function, and assigns different attention weights based on the significance of rhythm differences to obtain an individualized physiological rhythm attention weight distribution. Based on this physiological rhythm attention weight distribution, the second heart rate Bio-RCS value, second heart rate variability parameter, second respiratory Bio-RCS value, and second respiratory variability parameter of each target object are weighted and fused to highlight the individual physiological rhythm characteristics of each target object, resulting in enhanced physiological feature parameters. Based on these enhanced physiological feature parameters, an individual physiological rhythm recognition model is established for each target object. By encoding the time-dependent information related to individual physiological rhythms in the heartbeat and respiratory signals detected by far-infrared radar, accurate differentiation of target identity and personalized assessment of physiological state are achieved, resulting in enhanced vital sign recognition results containing individual rhythm characteristics.

[0118] The above describes the non-contact vital sign detection method using far-infrared radar in the embodiments of the present invention. The following describes the non-contact vital sign detection device using far-infrared radar in the embodiments of the present invention. Please refer to [link / reference]. Figure 2 One embodiment of the far-infrared radar non-contact vital sign detection device of the present invention includes:

[0119] The acquisition module 210 is used to acquire radar echo data and body temperature data of multiple target objects in the detection area through far-infrared sensors and millimeter-wave radar;

[0120] The analysis module 220 is used to analyze the radar echo data for heartbeat and respiratory signals, generate the first heartbeat parameter and the first respiratory parameter, and at the same time analyze the body temperature data for body temperature distribution, generate body temperature distribution parameters.

[0121] Calculation module 230 is used to calculate the second heartbeat parameter and the second respiratory parameter for each target object based on the first heartbeat parameter and the first respiratory parameter;

[0122] The generation module 240 is used to generate the vital sign detection results for each target object based on the second heartbeat parameter, the second respiratory parameter, and the body temperature distribution parameter.

[0123] Through the synergistic cooperation of the aforementioned components, far-infrared sensors and millimeter-wave radar technologies are combined. The far-infrared sensor is responsible for non-contact body temperature detection, while the millimeter-wave radar is responsible for acquiring heartbeat and respiratory signals. This enables simultaneous monitoring of three key physiological parameters: body temperature, heart rate, and respiratory rate. This overcomes the limitations of existing single technologies that can only detect some physiological parameters. By establishing a dynamic model of far-infrared electromagnetic propagation of physiological signals and employing the Maxwell equations full-wave analytical method to separate the heartbeat scattering component, respiratory scattering component, and multi-target cross-coupling interference component, the technical challenge of physiological signal aliasing in multi-target environments is effectively solved, enabling simultaneous detection of physiological signals from multiple targets without interference. By using wavelet transform of heart rate and respiratory characteristics combined with a physiological signal-guided Harris Hawks optimization algorithm, the detection parameters can be adaptively adjusted according to the differences in physiological characteristics of different individuals, overcoming the limitations of traditional fixed-parameter methods. Specific optimizations are performed for the 0.8-2.0Hz frequency band of heartbeat signals and the 0.1-0.5Hz frequency band of respiratory signals, significantly improving the accuracy of physiological parameter detection. By employing a physiological signal spatial angle estimation algorithm and a multi-target reinforcement learning controller for collaborative computation, a complete target localization and identity recognition mechanism was established. This mechanism can accurately determine the spatial location of each target object and associate it with corresponding physiological parameters, solving the key problem that existing technologies cannot distinguish between detection results from different individuals. A multi-level anomaly judgment system based on Bio-RCS values ​​and variability parameters was also established, which can automatically identify abnormal states in three dimensions: heart rate, respiration, and body temperature, and generate comprehensive detection results including anomaly warnings and health status assessments.

[0124] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0125] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0126] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A non-contact vital sign detection method using far-infrared radar, characterized in that, include: The radar echo data and body temperature data of multiple target objects in the detection area are collected using far-infrared sensors and millimeter-wave radar. The radar echo data is analyzed to generate heartbeat and respiratory signals, and first heartbeat parameters and first respiratory parameters are generated. At the same time, the body temperature data is analyzed to generate body temperature distribution parameters. Specifically, this includes: analyzing the radar echo data to obtain heart rate signal to obtain heart rate characteristic data; simultaneously performing respiratory fundamental frequency isolation processing on the radar echo data to obtain respiratory characteristic data; calculating the heart rate bio-radar cross section based on the heart rate characteristic data to obtain a first heart rate Bio-RCS value; simultaneously calculating the respiratory bio-radar cross section based on the respiratory characteristic data to obtain a first respiratory Bio-RCS value; performing peak detection on the heart rate characteristic data to identify heart rate peak sequences; simultaneously performing respiratory cycle detection on the respiratory characteristic data to identify respiratory peak sequences; and calculating the R wave peak value based on the time difference between adjacent R wave peak points in the heart rate peak sequence. The R-interval sequence is used to calculate the respiratory interval sequence based on the time difference between adjacent respiratory peak points in the respiratory peak sequence. Statistical variance analysis is performed on the R-interval sequence to obtain the first heart rate variability parameter for each target object. Statistical variance analysis is also performed on the respiratory interval sequence to obtain the first respiratory variability parameter for each target object. The first heartbeat Bio-RCS value and the first heart rate variability parameter are combined to generate the first heartbeat parameter, and the first respiratory Bio-RCS value and the first respiratory variability parameter are combined to generate the first respiratory parameter. Based on the body temperature data, multi-target temperature field reconstruction and spatial temperature distribution calculation are performed to identify the body temperature distribution parameters of each target object. The calculation of second heartbeat and second respiratory parameters for each target object is based on the first heartbeat and first respiratory parameters. Specifically, this includes: calculating a physiological signal correlation matrix based on the radar echo data; performing matrix decomposition on the physiological signal correlation matrix to obtain a sequence of characteristic values ​​reflecting each target object; determining the number of targets based on the amplitude of each characteristic value in the characteristic value sequence, and calculating the azimuth matrix of each target object based on the relationship between the characteristic values, the array element spacing, and the working wavelength; correcting the azimuth matrix to obtain the spatial azimuth information of each target object; using the spatial azimuth information as input to drive a preset controller to calculate the ratio of target signal power to interference signal power, obtaining a power ratio, and determining the beam pointing vector of each target object based on the power ratio; calculating the phase delay difference from M array elements in the millimeter-wave radar to each target object based on the beam pointing vector, and generating the array steering vector of each target object based on the phase delay difference; and based on the... The array steering vector performs minimum variance distortion-free response calculation to obtain beamforming weight vectors for each target object; the beamforming weight vectors are then matrix-reorganized according to the target object number to obtain the directional enhancement weight matrix for each target object; based on the directional enhancement weight matrix, the first heartbeat Bio-RCS value and the first heart rate variability parameter in the first heartbeat parameter are separated to obtain the second heartbeat Bio-RCS value and the second heart rate variability parameter for each target object; based on the directional enhancement weight matrix, the first respiratory Bio-RCS value and the first respiratory variability parameter in the first respiratory parameter are separated to obtain the second respiratory Bio-RCS value and the second respiratory variability parameter for each target object; the second heartbeat Bio-RCS value and the second heart rate variability parameter for each target object are combined to form the second heartbeat parameter, and the second respiratory Bio-RCS value and the second respiratory variability parameter for each target object are combined to form the second respiratory parameter; The vital signs detection results for each target object are generated based on the second heart rate parameter, the second respiratory parameter, and the body temperature distribution parameter.

2. The far-infrared radar non-contact vital sign detection method according to claim 1, characterized in that, The method involves collecting radar echo data and body temperature data of multiple target objects in the detection area using far-infrared sensors and millimeter-wave radar, including: The original echo signal is obtained by transmitting a preset electromagnetic wave signal to the detection area using millimeter-wave radar and receiving the Doppler frequency shift echo generated by the chest cavity movement of the target object. The original echo signal is converted from analog to digital to obtain radar echo data; The body temperature data of the target object's body surface within the detection area is collected using a far-infrared sensor.

3. The far-infrared radar non-contact vital sign detection method according to claim 1, characterized in that, The process involves analyzing the radar echo data to obtain heart rate signal data, and simultaneously performing respiratory fundamental frequency isolation processing on the radar echo data to obtain respiratory characteristic data, including: The radar echo data is separated according to the preset heartbeat frequency band and respiratory frequency band to obtain the heartbeat frequency domain signal and the respiratory frequency domain signal; Electromagnetic scattering calculations are performed on the heartbeat frequency domain signal to obtain the heartbeat scattering component, and electromagnetic scattering calculations are performed on the respiratory frequency domain signal to obtain the respiratory scattering component; The heart rate harmonic interference in the heartbeat scattering component is iteratively optimized and eliminated to obtain an interference-free heartbeat signal. At the same time, the respiratory scattering component is subjected to fundamental frequency isolation filtering to obtain a pure respiratory signal. Wavelet transform and cross-coupling correction are performed on the de-interference heartbeat signal to obtain heartbeat feature data, and wavelet transform is performed on the pure respiratory signal to obtain respiratory feature data.

4. The far-infrared radar non-contact vital sign detection method according to claim 1, characterized in that, The step of generating vital sign detection results for each target object based on the second heart rate parameter, the second respiratory parameter, and the body temperature distribution parameter includes: Based on the comparison between the second heartbeat Bio-RCS value and the second heart rate variability parameter in the second heartbeat parameter and the preset normal heartbeat range, the heartbeat detection status of each target object is identified. Based on the comparison between the second respiratory Bio-RCS value and the second respiratory variability parameter in the second respiratory parameter and the preset normal respiratory range, the respiratory detection status of each target object is identified. Based on the comparison between the body temperature distribution parameters and the preset normal body temperature range, the body temperature detection status of each target object is identified. Based on the heart rate, respiration, and body temperature of each target object, and combined with the spatial orientation information of the target object, the vital signs detection results of each target object are obtained.

5. A far-infrared radar non-contact vital sign detection device, characterized in that, For performing the far-infrared radar non-contact vital sign detection method as described in any one of claims 1-4, the far-infrared radar non-contact vital sign detection device comprises: The acquisition module is used to acquire radar echo data and body temperature data of multiple target objects in the detection area through far-infrared sensors and millimeter-wave radar; The analysis module is used to analyze the radar echo data for heartbeat and respiratory signals, generate first heartbeat parameters and first respiratory parameters, and simultaneously analyze the body temperature data for body temperature distribution, generating body temperature distribution parameters. The calculation module is used to calculate the second heartbeat parameter and the second respiratory parameter for each target object based on the first heartbeat parameter and the first respiratory parameter; The generation module is used to generate the vital sign detection results for each target object based on the second heartbeat parameter, the second respiratory parameter, and the body temperature distribution parameter.

Images