Sign recognition method based on fmcw millimeter wave radar segmentation scanning algorithm

Abstract

The present application relates to the technical field of biomedical radar signal processing and artificial intelligence, and particularly relates to a sign recognition method based on a FMCW millimeter wave radar segmented scanning algorithm, wherein a monitoring space range of a FMCW millimeter wave radar is divided into multiple continuous and non-overlapping sub-regions, a body movement, a breathing and a heart rate feature vector are extracted for each sub-region respectively, a feature weight is calculated by using an entropy weight method, a best detection sub-region of each physiological signal is screened out by combining a weighted fuzzy comprehensive evaluation method, and finally parameter calculation is completed in the respective best region. The method effectively solves the problems of weak signal loss in non-peak regions, mutual interference of body movement, breathing and heart rate signals, difficulty in adapting to spatial distribution differences of different height / size populations and insufficient recognition accuracy of signal aliasing caused by traditional "global single region detection", and realizes high-precision, partitioned and robust detection of body movement, breathing and heart rate signals.

Description

Technical Field

[0001] This invention relates to the field of biomedical radar signal processing and artificial intelligence, and in particular to a vital sign recognition method based on the FMCW millimeter-wave radar segmented scanning algorithm. Background Technology

[0002] Respiratory rate, heart rate, and body movement status are core vital signs of the human body. Their non-contact, high-precision monitoring is of significant value in applications such as home monitoring, medical auxiliary diagnosis, and intelligent health management. FMCW (Frequency Modulated Continuous Wave) millimeter-wave radar, with its advantages of non-contact detection, strong resistance to environmental interference, and ability to penetrate clothing, has become the mainstream technology in this field. Traditional methods involve transmitting radar signals to cover the target area, receiving reflected signals, preprocessing them, and suppressing clutter to select the area with the largest signal amplitude as the detection target, thereby extracting physiological characteristics such as respiration and heart rate.

[0003] However, traditional FMCW millimeter-wave radar detection methods have significant limitations in practical applications. First, the method selects only a single range gate region with the largest amplitude, resulting in severe loss of weak signals in non-peak regions. For example, leg movement signals from shorter individuals or abdominal assisted breathing signals from thinner individuals cannot be captured due to their low amplitude, leading to missed body movement detection and incomplete respiratory characteristics. Second, body movement, respiratory, and heart rate signals are not detected separately. When body movement occurs, strong amplitude body movement signals can mask weak physiological phase changes in the same region, causing interruptions or distortions in respiratory and heart rate calculations. Furthermore, traditional methods cannot adapt to the differences in the spatial distribution of the chest, abdomen, and limbs among people of different heights and body types. For example, the difference in chest and abdominal distance between adults and children can be more than 0.5 meters, and the difference in respiratory signal radiation range between obese and thin individuals can be 3 to 5 range gates. A single peak region is difficult to dynamically adapt, leading to attenuation of chest and abdominal signals in shorter individuals or misjudgment of limb movements in taller individuals. Finally, the mixed processing of signals across the entire range can easily lead to signal aliasing, such as misinterpreting headboard reflections or slight movements of bedding as body movements, and the superposition of respiratory harmonics in the chest and abdominal regions with heart rate signals, which increases the frequency extraction error.

[0004] Therefore, existing technologies suffer from problems such as loss of information in non-peak areas, mutual interference between body movement and physiological signals, poor individual spatial adaptability, and insufficient accuracy in signal aliasing recognition. There is an urgent need for a technical solution that can subdivide the monitoring space, dynamically adapt to differences in human body position, and achieve accurate detection of body movement, respiration, and heart rate zones to improve the accuracy and robustness of vital sign monitoring. Summary of the Invention

[0005] The purpose of this invention is to provide a vital sign identification method based on the FMCW millimeter-wave radar segmented scanning algorithm, which solves problems such as information loss in non-peak areas, signal aliasing, and poor individual adaptability.

[0006] To achieve the above objectives, this invention provides a vital sign identification method based on the FMCW millimeter-wave radar segmented scanning algorithm, comprising the following steps:

[0007] The FMCW millimeter-wave radar transmits linear frequency modulated signals and receives echo signals, and preprocesses the echo signals.

[0008] The monitoring spatial range is mapped to a distance gate index range, and the distance gate index range is uniformly divided into multiple continuous non-overlapping sub-regions;

[0009] For the preprocessed echo signal, extract the signal of each sub-region according to the sub-region distance gate index;

[0010] For each sub-region, extract the body motion feature vector, respiratory feature vector, and heart rate feature vector respectively;

[0011] The entropy weight method was used to calculate the feature weights of the feature vectors of three types of signals: body movement, respiration, and heart rate.

[0012] Combining the aforementioned feature weights, a weighted fuzzy comprehensive evaluation method is used to score the fit of multiple sub-regions in terms of body movement, respiration, and heart rate.

[0013] Based on the fitness score, the optimal sub-regions corresponding to body movement, respiration, and heart rate are selected.

[0014] Body motion parameters are calculated within the optimal sub-region for body motion, respiratory rate is calculated within the optimal sub-region for respiration, and heart rate is calculated within the optimal sub-region for heart rate.

[0015] Specifically, for the preprocessed echo signal, the signal of each sub-region is extracted according to the sub-region distance gate index, including:

[0016] When extracting the signal of each sub-region, an adaptive weight based on dynamic calculation of signal-to-noise ratio is used to perform weighted summation of the echo complex signals of each distance gate in the sub-region.

[0017] This process involves extracting the signal from each sub-region of the preprocessed echo signal using a sub-region distance gate index, and prior to this, the following steps are also included:

[0018] Coherent accumulation of multi-channel echo signals is performed to suppress channel noise, and a spatiotemporal two-dimensional constant false alarm rate algorithm is used to eliminate static clutter.

[0019] For each sub-region, body motion feature vector, respiratory feature vector, and heart rate feature vector are extracted separately. This process also includes:

[0020] Phase unwrapping is performed on the sub-region signal to eliminate phase ambiguity.

[0021] Specifically, for each sub-region, body motion feature vector, respiratory feature vector, and heart rate feature vector are extracted, including:

[0022] The body motion feature vector includes the peak value of the phase change rate, the mean signal energy, and the variance of the body motion frequency band spectrum; the respiratory feature vector includes the peak energy of the wavelet packet of the respiratory frequency band, the bandwidth ratio of the main peak of the spectrum, and the autocorrelation coefficient of the respiratory cycle; the heart rate feature vector includes the energy of the intrinsic mode function components of the heart rate frequency band, the harmonic energy ratio, and the signal-to-noise ratio of the heart rate frequency band.

[0023] Specifically, the entropy weight method is used to calculate the feature weights of the feature vectors of the three types of signals: body movement, respiration, and heart rate. This includes:

[0024] The original feature values ​​of each feature are normalized.

[0025] Calculate the normalization percentage and entropy value for each feature;

[0026] The difference coefficients of the features are calculated based on the entropy values, and the entropy weights of each feature are finally determined.

[0027] Specifically, combining the aforementioned feature weights, a weighted fuzzy comprehensive evaluation method is used to score the fit of multiple sub-regions in terms of body movement, respiration, and heart rate, including:

[0028] A spatial continuity penalty coefficient is introduced in the fitness scoring process to penalize spatially discrete sub-regions.

[0029] Specifically, the calculation of body motion parameters is performed within the optimal sub-region for body motion, respiratory rate is performed within the optimal sub-region for respiration, and heart rate is performed within the optimal sub-region for heart rate. This includes:

[0030] The body movement parameters include body movement amplitude, average body movement frequency, and body movement state identifier; the respiratory rate is determined by spectral peak detection; and the heart rate is determined by time-frequency analysis using Hilbert-Huang transform.

[0031] This invention discloses a vital sign recognition method based on a segmented scanning algorithm of FMCW millimeter-wave radar. The method divides the monitoring space of the FMCW millimeter-wave radar into multiple continuous, non-overlapping sub-regions. For each sub-region, feature vectors for body movement, respiration, and heart rate are extracted. Entropy weighting is used to calculate feature weights, and a weighted fuzzy comprehensive evaluation method is combined to select the optimal detection sub-region for each type of physiological signal. Finally, parameter calculations are performed within their respective optimal regions. This method effectively solves the problems of traditional "global single-region detection," such as loss of weak signals in non-peak areas, interference between body movement and respiration / heart rate signals, difficulty in adapting to spatial distribution differences among people of different heights / body types, and insufficient accuracy in signal aliasing recognition. It achieves high-precision, regionalized, and robust detection of body movement, respiration, and heart rate signals. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0033] Figure 1 This is a flowchart of the steps of the vital sign recognition method based on the FMCW millimeter-wave radar segmented scanning algorithm of the present invention. Detailed Implementation

[0034] The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, but should not be construed as limiting the present invention.

[0035] Please see Figure 1 ,in, Figure 1 This is a flowchart of the steps of the vital sign recognition method based on the FMCW millimeter-wave radar segmented scanning algorithm of the present invention.

[0036] This invention provides a vital sign recognition method based on the FMCW millimeter-wave radar segmented scanning algorithm, comprising the following steps:

[0037] S101: Transmit linear frequency modulated signals and receive echo signals via FMCW millimeter-wave radar, and preprocess the echo signals;

[0038] S102: Map the monitoring spatial range to a distance gate index range, and evenly divide the distance gate index range into multiple continuous non-overlapping sub-regions;

[0039] S103: Extract the signal of each sub-region from the preprocessed echo signal according to the sub-region distance gate index;

[0040] S104: For each sub-region, extract the body motion feature vector, respiratory feature vector, and heart rate feature vector respectively;

[0041] S105: The entropy weight method is used to calculate the feature weights of the feature vectors of the three types of signals: body movement, respiration and heart rate;

[0042] S106: Combining the aforementioned feature weights, a weighted fuzzy comprehensive evaluation method is used to score the fit of multiple sub-regions in terms of body movement, respiration, and heart rate;

[0043] S107: Based on the fitness score, select the best sub-regions corresponding to body movement, respiration and heart rate respectively;

[0044] S108: Calculate body motion parameters within the optimal sub-region of body motion, calculate respiratory rate within the optimal sub-region of respiration, and calculate heart rate within the optimal sub-region of heart rate.

[0045] Specifically, the FMCW millimeter-wave radar acquires echo data of the covered area by transmitting a linear frequency modulated signal. The transmitted signal is:

[0046] ;

[0047] in, The FMCW linear frequency modulated signal transmitted by the radar. The amplitude of the transmitted signal; The imaginary unit is used to construct signals in the complex field. The radar carrier frequency; To adjust the frequency, ,in For bandwidth, This is the frequency sweep period; It is a time variable.

[0048] After the signal is reflected by the human body, the echo signal collected by the radar receiving antenna is mixed to obtain the difference frequency signal:

[0049] ;

[0050] in, This is the intermediate frequency difference signal after mixing; This represents the number of target points within the radar's detection range. For the first The amplitude of the echo signal from a target is related to the target's reflection coefficient and distance. For the first The instantaneous distance from the target to the radar antenna; For radar signal wavelength, , The speed of light; For the first The initial phase of the echo from each target; It is Gaussian white noise, which includes radar hardware noise, environmental electromagnetic interference, etc.

[0051] Monitoring spatial range Mapped to distance gate index range The actual distance corresponding to each distance gate is:

[0052] ;

[0053] in, For the first The actual distance corresponding to each distance gate; This is the minimum detection range for the radar; The distance gate index (positive integer). , This represents the total number of radar range gates. For radar range resolution, , This refers to the radar bandwidth.

[0054] Cover the range of distance gate indexes Evenly divided into The nth consecutive non-overlapping sub-regions, the nth The distance gate index range for each sub-region is:

[0055] ;

[0056] when Cannot be When dividing by integers, the index range of the last sub-region is adjusted as follows:

[0057] ;

[0058] in, For the first The set of distance gate indices corresponding to each sub-region ( ); The starting distance gate index (positive integer) corresponding to the coverage area; The termination distance gate index (a positive integer) corresponds to the coverage area, and ; The number of sub-regions (positive integer) is adaptively adjusted based on the monitoring length and distance resolution. To ensure that the number of distance gates in each sub-region is as uniform as possible, a floor function is used.

[0059] No. The actual spatial extent corresponding to each sub-region is:

[0060] ;

[0061] in, For the first The actual spatial distance range of each sub-region; For the first The minimum value of the distance gate index set for each sub-region; For the first The maximum value of the distance between each subregion and the gate index set; the meanings of the other symbols are the same as before.

[0062] For the preprocessed radar echo signal, index it by sub-region range gate. Extract the first Signals for each sub-region:

[0063] ;

[0064] in, For the first Echo complex signals of each sub-region; For distance gate index, ; Time of the first The first sub-region The adaptive weights of the distance gates are dynamically adjusted based on the signal-to-noise ratio; Time of the first The echo complex signal of the distance gate, , , These are the in-phase component and quadrature component of the distance gate, respectively.

[0065] Adaptive weights The formula for calculation is:

[0066] ;

[0067] in, Time of the first The first sub-region Signal-to-noise ratio of each distance gate; For the first Distance gate index within each sub-region, The denominator is the first... The sum of the signal-to-noise ratios of all distance gates within each sub-region is used to ensure that the sum of the weights is 1.

[0068] Signal-to-noise ratio The formula for calculation is:

[0069] ;

[0070] in, This is a time-domain expectation operator used to calculate the mean of signal power; For the first The power of the distance gate echo signal; Time of the first The variance of background noise in each sub-region.

[0071] Background noise variance The estimation formula is:

[0072] ;

[0073] in, The noise estimation window length (positive integer); Index of the sampling points within the window; The sampling interval for radar signals. , Sampling rate; This is a reference distance gate index outside the sub-region. This distance gate contains no human reflection signals and only background noise. This refers to a historical sampling moment.

[0074] Coherent accumulation of multi-channel echo signals to suppress channel noise.

[0075] ;

[0076] in, The first after coherent accumulation One distance gate echo signal; The number of radar receiving antennas (positive integer); For antenna channel index, ; For the first Channel, First Distance to door, The echo signal at a given moment; For the first Channel 1 Phase calibration coefficient of the distance gate.

[0077] A spatiotemporal two-dimensional CFAR constant false alarm rate algorithm is used to eliminate static noise from the bed frame, bedding, etc.

[0078] ;

[0079] in, Time of the first Clutter power estimates for each distance gate; The distance to the reference window (positive integer); The length of the time reference window (positive integer); This is a set of distance reference window indices, containing reference distance gates on both sides of the target distance gate; This is a set of time reference window indices, containing historical sampling points prior to the current time. This is the index offset within the reference window; This is the index offset within the time reference window.

[0080] The signal after clutter suppression is:

[0081] ;

[0082] in, The first clutter suppression One distance gate echo signal; CFAR threshold coefficient (positive real number); The denominator should be a very small positive number to avoid being zero. This is a rectangular window function.

[0083] Eliminate phase ambiguity in sub-region signals.

[0084] ;

[0085] in, For the first The unwrapping phase of signals in each sub-region; For the first The original phase of the signal in each sub-region, ; For phase sampling point index; This is a sign function; when the input is positive, zero, or negative, the output is 1, 0, or -1, respectively. , For two adjacent phase sampling times, phase continuous unwrapping is achieved by accumulating the results of the sign function.

[0086] For each sub-region Feature vectors corresponding to body movement, respiration, and heart rate are extracted respectively. , , The feature quantization formula is as follows:

[0087] Body motion feature vector ;

[0088] ;

[0089] in, For the first The peak value of the phase change rate of each sub-region reflects the intensity of the phase abrupt change caused by body motion; For feature extraction time window; This is the time derivative of the unwrapped phase, i.e., the rate of phase change; This is the maximum value operator, which extracts the peak value of the phase change rate within the time window.

[0090] ;

[0091] in, For the first The average energy of the signal in each sub-region reflects the overall intensity of the target reflected signal within the region; The integral operator within the time window calculates the total signal energy; the denominator is the length of the time window, which normalizes the total energy to the mean.

[0092] ;

[0093] in, For the first The spectral variance of each sub-region in the body motion frequency band reflects the frequency distribution dispersion of the body motion signal; The Fast Fourier Transform operator converts a time-domain signal into a frequency-domain spectrum. For frequency variables; This is the characteristic frequency band of body motion signals; A frequency band filtering operator that retains only the spectral components within a specified frequency band; This is the variance operator, used to calculate the variance of the filtered spectrum.

[0094] Respiratory feature vector ;

[0095] ;

[0096] in, For the first The peak energy of wavelet packets in the respiratory frequency band of each sub-region reflects the intensity of the respiratory signal; The wavelet packet decomposition operator is used to perform multi-scale decomposition of the unwrapped phase signal; This is a characteristic frequency band of respiratory signals; This is the peak energy extraction operator, used to calculate the maximum energy value within the decomposed breathing frequency band.

[0097] ;

[0098] in, For the first The bandwidth ratio of the main peak of the respiratory signal spectrum in each sub-region reflects the degree of frequency concentration of the respiratory signal. It is the main peak width calculation operator, which outputs the half-power bandwidth of the main peak of the spectrum; The peak frequency extraction operator outputs the frequency corresponding to the main peak of the spectrum.

[0099] ;

[0100] in, For the first The respiratory cycle autocorrelation coefficient of the unwrapped phase signal in each sub-region reflects the periodicity of the respiratory signal; This is a correlation coefficient operator that outputs the Pearson correlation coefficient between two signals, with a value range of... ; This is an estimate of the respiratory cycle. ; The delay signal is the unwrapped phase signal, and the delay time is the estimated value of the respiratory cycle.

[0101] Heart rate feature vector ;

[0102] ;

[0103] in, For the first The energy of the intrinsic mode function (IMF) component of each sub-region in the heart rate frequency band reflects the intensity of the heart rate signal; An operator for extracting the IMF components of the Hilbert-Huang transform is used to adaptively decompose the echo signal; This is the characteristic frequency band of the heart rate signal; This is an energy calculation operator that outputs the energy value of the IMF component.

[0104] ;

[0105] in, For the first The harmonic energy ratio of the heart rate signal in each sub-region reflects the harmonic richness of the heart rate signal; The harmonic order (a positive integer); For harmonic order index; For the first The amplitude and frequency of the second harmonic are... ; This is the fundamental frequency of the heart rate signal; The fundamental amplitude is the value of the fundamental wave; the molecule represents the amplitude of the preceding wave. The sum of squares of the amplitudes of the subharmonics, with the square of the amplitude of the fundamental wave in the denominator.

[0106] ;

[0107] in, For the first The signal-to-noise ratio of each sub-region after heart rate frequency band filtering reflects the purity of the heart rate signal; The Butterworth bandpass filter operator retains only the signal within the heart rate frequency band. The signal-to-noise ratio (SNR) calculation operator is used to calculate the SNR of the filtered signal.

[0108] The entropy weight method is used to calculate the feature weights, and the weighted fuzzy comprehensive evaluation method is combined with the method to evaluate the features. Each sub-region is scored for fit, and the optimal detection area for each physiological signal is selected.

[0109] For the Physiological signals ( ), No. Sub-region Normalized values ​​of each feature:

[0110] ;

[0111] in, For the first Class of signals Sub-region The normalized values ​​of each feature, with a range of values. ; For the first Class of signals Sub-region One original feature value; For sub-region indexing, ; For all sub-regions, the first Class of signals The minimum value of each feature; For all sub-regions, the first Class of signals The maximum value of each feature.

[0112] Feature entropy value:

[0113] ;

[0114] in, For the first Class of signals The entropy value of each feature reflects its discriminative power; It is the natural logarithm operator; For the first Class of signals Sub-region The normalized percentage of each feature.

[0115] Feature normalization percentage:

[0116] ;

[0117] in, For the first Class of signals Sub-region Normalized percentage of each feature (dimensionless), range of values The sum of the percentages of all sub-regions is 1; the rest conform to the same meaning as before;

[0118] Feature entropy weight:

[0119] ;

[0120] in, For the first Class of signals The weights of each feature, and their ranges. The sum of the weights of all features is 1; The coefficient of difference for each feature is denoted as 1; the denominator is the sum of the coefficients of difference for all features.

[0121] No. The sub-regions for the first Evaluation of the fit of physiological signals:

[0122] ;

[0123] in, For the first The sub-regions for the first The fit score for the signal type is as follows: the higher the score, the better the fit. This is a spatial continuity penalty coefficient used to penalize spatially discrete sub-regions; The sub-region index distance. ; For the first Sub-region mean index of the signal class .

[0124] Optimal choice:

[0125] ;

[0126] in, For the first The optimal sub-region index (positive integer) for the signal type; This is the maximum value index operator.

[0127] Spatial range of the optimal detection area:

[0128] ;

[0129] in, For the first The optimal detection area for this type of signal; The minimum distance gate index for the optimal sub-region; This is the maximum distance gate index for the best subregion; the meanings of the other symbols are the same as before.

[0130] Calculation of body motion parameters:

[0131] ;

[0132] in, The amplitude of body movement; The optimal sub-region index for body motion signals; The signal amplitude is the optimal range for motion.

[0133] ;

[0134] in, The numerator is the average frequency of the body movement; the denominator is the weighted sum of the frequency energies within the body movement frequency band; the denominator is the total energy within the body movement frequency band; the meanings of the other symbols are the same as before.

[0135] Body movement status indicators:

[0136] ;

[0137] in, This is a body movement status indicator; 1 indicates that there is body movement, and 0 indicates that there is no body movement. The threshold value for body movement amplitude is calibrated according to the actual application scenario.

[0138] Respiratory rate calculation:

[0139] ;

[0140] in, Respiratory rate; The optimal sub-region index for respiratory signals; This represents the unwrapping phase signal for the optimal breathing region; the meanings of the other symbols are the same as before.

[0141] Heart rate calculation:

[0142] ;

[0143] in, Heart rate; The optimal sub-region index for the heart rate signal; The symbol represents the Hilbert-Huang transform operator, used for time-frequency analysis of the signal in the optimal heart rate region; the meanings of the other symbols are the same as before.

[0144] By subdividing the radar coverage space into multiple sub-regions and employing entropy weighting to fuse multimodal features and weighted fuzzy comprehensive evaluation to select the optimal detection area, precise regional detection of body movement, respiration, and heart rate is achieved. This effectively solves core problems caused by traditional global single-region detection, such as severe loss of weak signals in non-peak areas, mutual interference between body movement and physiological signals, inability to adapt to individual differences in human body position, and insufficient recognition accuracy due to signal aliasing. Ultimately, this achieves high-precision recognition of body movement, respiration, and heart rate signals, and provides high spatiotemporal resolution quantitative biomarker data for related applications.

[0145] The above-disclosed embodiments are merely one or more preferred embodiments of this application and should not be construed as limiting the scope of this application. Those skilled in the art can understand that all or part of the processes for implementing the above embodiments and equivalent changes made in accordance with the claims of this application still fall within the scope of this application.

Claims

1. A sign recognition method based on FMCW millimeter wave radar segmentation scanning algorithm, characterized in that, Includes the following steps: The FMCW millimeter-wave radar transmits linear frequency modulated signals and receives echo signals, and preprocesses the echo signals. The monitoring spatial range is mapped to a distance gate index range, and the distance gate index range is uniformly divided into multiple continuous non-overlapping sub-regions; For the preprocessed echo signal, extract the signal of each sub-region according to the sub-region distance gate index; For each sub-region, extract the body motion feature vector, respiratory feature vector, and heart rate feature vector respectively; The entropy weight method was used to calculate the feature weights of the feature vectors of three types of signals: body movement, respiration, and heart rate. Combining the aforementioned feature weights, a weighted fuzzy comprehensive evaluation method is used to score the fit of multiple sub-regions in terms of body movement, respiration, and heart rate. Based on the fitness score, the optimal sub-regions corresponding to body movement, respiration, and heart rate are selected. Body motion parameters are calculated within the optimal sub-region for body motion, respiratory rate is calculated within the optimal sub-region for respiration, and heart rate is calculated within the optimal sub-region for heart rate. 2.The sign recognition method based on FMCW millimeter wave radar sector scan algorithm according to claim 1, wherein, For the preprocessed echo signal, the signal of each sub-region is extracted according to the sub-region distance gate index, specifically including: When extracting the signal of each sub-region, an adaptive weight based on dynamic calculation of signal-to-noise ratio is used to perform weighted summation of the echo complex signals of each distance gate in the sub-region. 3.The sign recognition method based on FMCW millimeter wave radar sector scan algorithm of claim 2, wherein, For the preprocessed echo signal, the signal of each sub-region is extracted according to the sub-region distance gate index. This process also includes: Coherent accumulation of multi-channel echo signals is performed to suppress channel noise, and a spatiotemporal two-dimensional constant false alarm rate algorithm is used to eliminate static clutter. 4.The sign recognition method based on FMCW millimeter wave radar sector scan algorithm of claim 3, wherein, For each sub-region, body motion feature vector, respiratory feature vector, and heart rate feature vector are extracted separately. This process also includes: Phase unwrapping is performed on the sub-region signal to eliminate phase ambiguity. 5.The sign recognition method based on FMCW millimeter wave radar sector scan algorithm according to claim 4, wherein, For each sub-region, body motion feature vector, respiratory feature vector, and heart rate feature vector are extracted, specifically including: The body motion feature vector includes the peak value of the phase change rate, the mean signal energy, and the variance of the body motion frequency band spectrum; the respiratory feature vector includes the peak energy of the wavelet packet of the respiratory frequency band, the bandwidth ratio of the main peak of the spectrum, and the autocorrelation coefficient of the respiratory cycle; the heart rate feature vector includes the energy of the intrinsic mode function components of the heart rate frequency band, the harmonic energy ratio, and the signal-to-noise ratio of the heart rate frequency band. 6.The sign recognition method based on FMCW millimeter wave radar sector scan algorithm according to claim 5, wherein, The entropy weight method is used to calculate the feature weights of the feature vectors of three types of signals: body movement, respiration, and heart rate. Specifically, this includes: The original feature values ​​of each feature are normalized. Calculate the normalization percentage and entropy value for each feature; The difference coefficients of the features are calculated based on the entropy values, and the entropy weights of each feature are finally determined.

7. The FMCW millimeter wave radar segment scan algorithm based sign recognition method according to claim 6, characterized in that, Based on the aforementioned feature weights, a weighted fuzzy comprehensive evaluation method is used to score the fit of multiple sub-regions in terms of body movement, respiration, and heart rate, specifically including: A spatial continuity penalty coefficient is introduced in the fitness scoring process to penalize spatially discrete sub-regions.

8. The vital sign recognition method based on the FMCW millimeter-wave radar segmented scanning algorithm as described in claim 7, characterized in that, Body motion parameters are calculated within the optimal sub-region for body motion, respiratory rate is calculated within the optimal sub-region for respiration, and heart rate is calculated within the optimal sub-region for heart rate, including: The body movement parameters include body movement amplitude, average body movement frequency, and body movement state identifier; the respiratory rate is determined by spectral peak detection; and the heart rate is determined by time-frequency analysis using Hilbert-Huang transform.

Images