Radar-based vital sign monitoring method, system, and readable storage medium

Abstract

The application provides a radar-based vital sign monitoring method, system and readable storage medium, uses a radar to emit electromagnetic waves and capture reflected waves caused by micro-movement of a human chest cavity and containing micro phase changes, extracts feature signals corresponding to each reflection point based on radar echo signals, and performs interference elimination processing on the radar echo signals to eliminate overall motion interference, so as to extract accurate breathing signals and heartbeat signals. Meanwhile, since the radar point cloud data used can only reflect displacement and distance information of a part, image information of a monitored person is not collected, privacy information of the monitored person is protected to a certain extent, actual monitoring requirements are met, and the method has certain popularization value.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, and in particular to a radar-based method, system, and readable storage medium for monitoring vital signs. Background Technology

[0002] Non-contact monitoring technologies have developed rapidly in recent years, with image sensor-based methods being particularly widely used. This technology uses a camera to capture video images of the human chest cavity and employs image processing algorithms to analyze the periodic movements of the contours, thereby extracting respiratory features. However, image monitoring relies on good lighting conditions and an unobstructed field of view. When the monitored individual is obstructed by other objects or in a dimly lit environment, the subtle movements of the chest cavity are difficult to capture effectively, leading to a significant decrease in monitoring accuracy. Furthermore, image sensors continuously collect image information including the monitored individual's face, body contours, and even the surrounding environment, which can easily expose personal privacy. These factors limit its widespread application and make it difficult to meet monitoring needs. Summary of the Invention

[0003] To overcome the problems existing in related technologies, this specification provides a radar-based method, system, and readable storage medium for monitoring vital signs.

[0004] Firstly, a radar-based method for monitoring vital signs is provided, the method comprising: Acquire radar echo signals for the target thoracic cavity region, wherein the radar echo signals carry characteristic information related to the micro-movement of the target thoracic cavity; The radar echo signal is subjected to interference removal processing, and characteristic signals representing thoracic cavity micro-movements are extracted from the processed signal. The characteristic signals representing chest cavity micro-movements are separated and processed to obtain respiratory and heartbeat signals that reflect the target's vital signs.

[0005] According to the radar-based vital sign monitoring method provided in this application, the step of performing interference removal processing on the radar echo signal and extracting characteristic signals representing thoracic cavity micro-movements from the processed signal includes: extracting displacement time-series data corresponding to multiple reflection points on the surface of the target thoracic cavity based on the radar echo signal to generate multiple sets of point cloud data. The multiple sets of point cloud data are registered to obtain corrected point cloud data after removing the overall motion interference of the target; The phase-time signal characterizing thoracic cavity micromovement is extracted from the corrected point cloud data and used as the feature signal characterizing thoracic cavity micromovement.

[0006] According to the radar-based vital sign monitoring method provided in this application, the multiple sets of point cloud data include point cloud data of the current frame and point cloud data of the previous frame. The registration process of the multiple sets of point cloud data to obtain corrected point cloud data after removing interference from the overall motion of the target includes: Register the point cloud data of the current frame with the point cloud data of the previous frame to obtain the optimal rotation matrix and translation vector of the current frame relative to the previous frame. Based on the optimal rotation matrix and translation vector, the point cloud data of the current frame is calibrated to the unified reference coordinate system of the point cloud of the previous frame, so as to obtain the corrected point cloud data after removing the overall motion interference of the target.

[0007] According to the radar-based vital sign monitoring method provided in this application, after registering the point cloud data of the current frame with the point cloud data of the previous frame, the method further includes: Determine whether the registration process meets the preset reliability conditions; If the conditions are not met, the point cloud data of the current frame is registered with the point cloud data of a preset reference frame, and the point cloud data of the current frame is corrected according to the registration result.

[0008] According to the radar-based vital sign monitoring method provided in this application, the step of registering the point cloud data of the current frame with the point cloud data of the previous frame to obtain the optimal rotation matrix and translation vector of the current frame relative to the previous frame includes: The point cloud data of the current frame and the point cloud data of the previous frame are decentralized to obtain a decentralized coordinate matrix. Construct a weighted covariance matrix based on the decentralized coordinate matrix; Singular value decomposition is performed on the weighted covariance matrix to obtain the optimal rotation matrix and optimal translation vector between point clouds of different frames.

[0009] According to the radar-based vital sign monitoring method provided in this application, the step of acquiring radar echo signals targeting the target chest cavity region includes: The captured raw echo signal is preprocessed to obtain a radar echo signal targeting the chest cavity region of the target; wherein the raw echo signal contains emission components from stationary and moving objects, and the corresponding preprocessing includes static clutter filtering.

[0010] According to the radar-based vital sign monitoring method provided in this application, the step of extracting phase-time signals characterizing thoracic cavity micro-movements from the corrected point cloud data includes: Determine the residual vector between the corrected point cloud data and the previous frame's point cloud data; Based on the component of the residual vector in the radar radial direction, the instantaneous micro-displacement at the current moment is generated; The instantaneous micro-displacements at multiple consecutive moments are combined in chronological order to form the phase-time signal reflecting the micro-movements of the thoracic cavity.

[0011] According to the radar-based vital sign monitoring method provided in this application, after separating and processing the characteristic signals representing chest cavity micro-movements to obtain respiratory and heartbeat signals reflecting the target's vital signs, the method further includes: Periodic feature detection is performed on the respiratory signal and / or the heartbeat signal to determine the relevant signal features of the respiratory signal, the relevant signal features including amplitude features and / or temporal regularity features; Based on the relevant signal characteristics of the respiratory signal and / or the heartbeat signal, and the preset event type, identify abnormal vital signs events.

[0012] According to the radar-based vital sign monitoring method provided in this application, the method further includes: The phase timing signal, respiratory signal and / or heartbeat signal, or features extracted from them, are input into a pre-trained deep learning model; Obtain the output of the deep learning model and generate at least one of the following results: Denoising or enhancement processing of vital signs signals; Results of respiratory pattern or abnormal event identification; The results of sleep stage segmentation.

[0013] Secondly, a radar-based vital signs monitoring system is provided, the system comprising: A radar transceiver array includes a radar transmitting antenna array and a receiving antenna array, wherein the radar transmitting antenna array is configured to periodically transmit frequency-modulated continuous waves with linearly increasing frequency, and the receiving antenna array is connected to the radar transmitting antenna array and is configured to acquire radar echo signals containing the chest cavity region of a target. The processor, coupled to the radar transceiver array, is used to implement the radar-based vital sign monitoring method as described in any one of the first aspects above.

[0014] Thirdly, a computer-readable storage medium is provided, on which a radar-based vital signs monitoring program is stored, wherein the radar-based vital signs monitoring program, when executed, implements the steps of any of the radar-based vital signs monitoring methods described above.

[0015] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the radar-based vital sign monitoring method as described in any of the above.

[0016] This application provides a radar-based method, system, and readable storage medium for monitoring vital signs, which has the following advantages compared to the low accuracy and security of current image monitoring methods: This method uses radar to emit electromagnetic waves and capture reflected waves containing minute phase changes caused by slight movements of the human chest cavity. Based on the radar echo signals, characteristic signals corresponding to each reflection point are extracted. Interference removal processing is then applied to the radar echo signals to eliminate overall motion interference, thereby extracting accurate respiratory and heartbeat signals. Furthermore, since the radar point cloud data used only reflects the displacement and distance information of the monitored area and does not collect image information of the monitored person, it protects the privacy of the monitored person to a certain extent, meets practical monitoring needs, and has certain promotional value.

[0017] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this specification. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this specification and, together with the description, serve to explain the principles of this specification.

[0019] Figure 1 This is a schematic flowchart illustrating a radar-based vital signs monitoring method according to an exemplary embodiment of this specification; Figure 2 This is another schematic flowchart illustrating a radar-based vital signs monitoring method according to an exemplary embodiment of this specification; Figure 3 This is a schematic block diagram of a radar-based vital signs monitoring device illustrated in this specification according to an exemplary embodiment. Detailed Implementation

[0020] The technical solutions in the embodiments (or "implementations") of this application will be clearly and completely described herein with reference to the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements.

[0021] If the embodiments of this application contain terms relating to directional indications or positional relationships (such as up, down, left, right, front, back, inside, outside, top, bottom, center, vertical, horizontal, longitudinal, transverse, length, width, counterclockwise, clockwise, axial, radial, circumferential, etc.), such terms are only used to explain the relative positional relationships and movement of the components in a specific posture (as shown in the attached figures); if the specific posture changes, the directional indications or positional relationships will also change accordingly. Furthermore, the terms "first" and "second" used in the embodiments of this application are only for descriptive convenience and should not be construed as indicating or implying relative importance.

[0022] This application provides a radar-based method, system, and readable storage medium for monitoring vital signs. The following detailed description, in conjunction with the accompanying drawings, illustrates this application. The features described in the embodiments and implementations can be combined with each other.

[0023] To address the aforementioned technical problems, this specification provides a radar-based method for monitoring vital signs.

[0024] The aim is to use radar to emit electromagnetic waves and capture reflected waves containing minute phase changes caused by subtle movements of the human chest cavity (breathing and / or heartbeat). Singular Value Decomposition (SVD) is used to register the radar point cloud to eliminate overall motion interference, thereby extracting accurate respiratory and heartbeat signals. Furthermore, since the radar point cloud data used only reflects the displacement and distance information of the monitored area and does not collect image information of the monitored person, it protects the privacy of the monitored person to a certain extent, and has certain promotional value.

[0025] The radar-based vital sign monitoring method described in this article features non-contact operation, privacy protection, and interference resistance, making it suitable for scenarios such as medical and health monitoring centers, home environments, rescue operations, and vehicles. The following explanation uses a vehicle scenario as an example, integrating radar sensors into the car seat or cockpit to monitor the driver's respiratory rate and heart rate changes, determining whether fatigue driving, sudden illness, or other conditions are present. It should be noted that specific implementation methods for other scenarios are generally similar and are not specifically limited here.

[0026] This application provides an embodiment of a radar-based vital sign monitoring method, referring to... Figure 1 , Figure 1 This is a schematic flowchart of a radar-based vital sign monitoring method provided in the embodiments of this specification.

[0027] In some embodiments, the radar-based vital signs monitoring method of this application is applied to the processor of a radar-based vital signs monitoring system.

[0028] Radar-based vital sign monitoring systems include radar transceiver arrays and processors.

[0029] A radar transceiver array includes a radar transmitting antenna array and a receiving antenna array. The radar transmitting antenna array is configured to periodically transmit frequency-modulated continuous waves with linearly increasing frequency. The receiving antenna array is connected to the radar transmitting antenna array and is configured to acquire radar echo signals containing the chest cavity region of a target.

[0030] The processor, coupled to the radar transceiver array, is used to implement the radar-based vital sign monitoring method.

[0031] It is understandable that radar equipment is installed in a specific location within a vehicle. This radar equipment has a built-in dedicated radar chip, which is connected to a transmitting antenna array. Through the transmitting antenna array, frequency-modulated continuous electromagnetic waves (called chirps) with linearly increasing frequencies are periodically emitted. The radar chip controls the electromagnetic waves emitted by the transmitting antenna array, with a fixed linear frequency growth slope, and the emission period is synchronized with the echo signal acquisition period to ensure the temporal continuity of point cloud data acquisition.

[0032] As an example, the transmitting antenna array employs a multi-channel synchronous transmission mode, with each channel synchronously transmitting electromagnetic waves at linearly increasing frequencies to simultaneously capture echo signals from multiple reflection points in the human chest cavity, generating corresponding sets of point cloud data. This frequency modulation mechanism is crucial for simultaneously and accurately measuring the target's distance and relative velocity (radial velocity).

[0033] As an example, the operating frequency band scans from 60.0 GHz to 64 GHz, with a configurable frame rate of 60 Hz-100 Hz, before quickly jumping back to the starting frequency to begin the next cycle.

[0034] The transmitting antenna array emits radio waves (FMCW waves) whose frequency varies linearly with time. These waves illuminate the human chest cavity. Because breathing and heartbeat cause extremely small periodic undulations on the surface of the chest cavity, the path of the reflected waves also changes constantly. This minute change in distance is reflected in the phase change of the reflected waves. By receiving and analyzing these phase changes in the reflected waves, radar can extract the human's breathing and heart rate.

[0035] Subsequently, these extremely weak echo signals containing the aforementioned variations are captured by a receiving antenna array. The echo signals are then processed by a processor to execute a radar-based vital sign monitoring method, thereby enabling the monitoring of vital signs of targets inside vehicle seats or cockpits.

[0036] The strongest energy reflections originate primarily from the surface of the chest cavity, due to the micro-movements (tiny displacements) of the chest cavity caused by respiration and heartbeat. The echo signal exhibits two key changes compared to the transmitted signal: time delay (τ), determined by the constant speed of light and proportional to the absolute distance of the radar reaching the chest cavity; and phase shift (Δφ), caused by the micro-Doppler effect and proportional to the instantaneous displacement velocity of the chest cavity. These two key parameters are crucial for capturing micrometer-level motion.

[0037] In some embodiments, the radar transceiver array is disposed inside the vehicle, and the detection direction of the radar transceiver array is directed toward the chest cavity area of ​​the human body on the vehicle seat.

[0038] It should be noted that the transmitting antenna array and receiving antenna array in this embodiment are installed inside the vehicle according to the actual location of the monitored target, and may be, but are not limited to: The steering wheel is positioned close to the driver's chest cavity, providing strong micro-motion signals. The driver's seat backrest is fitted to the backrest, and the micro-motion signal is stable. The sunlight is directed downwards onto the chest cavity from the overhead canopy of the driver's seat (near the A-pillar or B-pillar), without any obstruction or human contact. The dashboard and center console are positioned facing the driver or front passenger, making them suitable for monitoring multiple people simultaneously. The front passenger seat and the backrests of the rear seats are used for monitoring passengers' vital signs.

[0039] The radar-based vital sign monitoring method specifically includes the following steps 101 to 105: In step 101, radar echo signals for the target thoracic cavity region are acquired, and the radar echo signals carry characteristic information related to the micro-movement of the target thoracic cavity.

[0040] By periodically transmitting electromagnetic waves with linearly varying frequencies through a transmitting antenna array, these waves illuminate the chest cavity region of the target human body. Multiple spatially independent effective reflection points are formed by tissues at different locations on the chest cavity surface. Each effective reflection point reflects the incident electromagnetic waves, generating a corresponding echo signal component. Therefore, the radar echo signal received by the receiving antenna array is a superimposed signal containing echo components from multiple reflection points, including feature information related to micro-motions on the chest cavity surface. This feature information includes, but is not limited to, distance information and micro-displacement information generated by respiration and / or heartbeat.

[0041] In some embodiments, acquiring radar echo signals targeting the thoracic cavity region includes: The captured raw echo signal is preprocessed to obtain a radar echo signal targeting the chest cavity region of the target; wherein the raw echo signal contains emission components from stationary and moving objects, and the corresponding preprocessing includes static clutter filtering.

[0042] During operation, radar equipment captures raw echo signals covering the monitored area. These raw echo signals include not only the echo components reflected from the chest cavity region of the target human body, but also the reflection components generated by stationary objects and the interference reflection components from non-target areas. These stationary objects include, but are not limited to, vehicle interiors, seats, doors, and center consoles.

[0043] To extract the effective signal that reflects only the subtle movements of the target's chest cavity, the captured raw echo signal needs to be preprocessed. As an example, preprocessing includes at least static clutter filtering. By filtering out the fixed reflection components corresponding to stationary objects in the raw echo signal, static environmental interference is suppressed, while the dynamic effective echo components caused by moving objects such as the target human body's chest cavity breathing and heartbeat are retained. This is then used for target localization or detection, resulting in a radar echo signal with a higher signal-to-noise ratio, specifically targeting the target's chest cavity region. This provides clean raw data for subsequent point cloud data generation, point cloud registration, and vital sign signal extraction.

[0044] As an example, algorithms such as high-pass filters or moving target indicators (MTI) can be used to filter out stationary echo signals generated by stationary objects such as walls and furniture.

[0045] The signal containing target motion information is obtained by static clutter filtering. The clutter-suppressed echo signal is then used for target detection to determine the range cell or spatial region where the target's chest cavity is located. Subsequently, the radar echo signal for the target's chest cavity region is extracted from the range cell or spatial region.

[0046] As an example, the distance information of each reflection point is first obtained based on FMCW (Frequency Modulated Continuous Wave) radar. The reflection points are then roughly classified according to the preset seat distance range. Then, combined with the spatial coordinate distribution of point cloud data, reflection points that are close in distance and spatially clustered are grouped into the same target, thereby distinguishing different targets such as the driver, front passenger, and rear passenger.

[0047] This embodiment greatly improves the signal-to-noise ratio (SNR) of dynamic micro-motion signals, which is a prerequisite for detecting weak vital signs.

[0048] In step 102, interference removal processing is performed on the radar echo signal, and characteristic signals representing thoracic cavity micro-movements are extracted from the processed signal.

[0049] In this embodiment, the radar echo signal is subjected to interference removal processing to eliminate irrelevant interference factors such as overall target motion and static clutter. Then, feature signals that can characterize the micro-motion state of the target's chest cavity are extracted from the signal after interference removal processing.

[0050] Specifically, refer to Figure 2 , Figure 2 This is another schematic flowchart of a radar-based vital sign monitoring method provided in the embodiments of this specification. In some embodiments, the interference removal processing of the radar echo signal and the extraction of characteristic signals representing thoracic cavity micromovements from the processed signal include the following steps 1021 to 1023: Step 1021: Based on the radar echo signal, extract the displacement time series data corresponding to multiple reflection points on the surface of the target chest cavity, and generate multiple sets of point cloud data.

[0051] The radar receiving antenna array synchronously acquires and analyzes multiple echo signals, obtaining distance, azimuth, and phase change information corresponding to multiple effective reflection points on the thoracic surface. Based on the position changes of each reflection point in multiple consecutive frames of radar echo signals, the displacement time series data of each reflection point over time is calculated. This displacement time series data reflects the micro-motion trajectory of the corresponding reflection point within the monitoring period.

[0052] The spatial coordinates and displacement information of all effective reflection points at the same time are integrated to form multiple sets of point cloud data. Each frame of point cloud data corresponds to the multi-point motion state of the chest cavity surface at a sampling time. Continuous frames of point cloud data constitute the original data basis for subsequent registration correction and micro-motion extraction.

[0053] In step 1022, the multiple sets of point cloud data are registered to obtain corrected point cloud data after removing interference from the overall motion of the target.

[0054] In some embodiments, the multiple sets of point cloud data include point cloud data of the current frame and point cloud data of the previous frame. The registration process of the multiple sets of point cloud data to obtain corrected point cloud data after removing overall target motion interference includes: Register the point cloud data of the current frame with the point cloud data of the previous frame to obtain the optimal rotation matrix and translation vector of the current frame relative to the previous frame. Based on the optimal rotation matrix and translation vector, the point cloud data of the current frame is calibrated to the unified reference coordinate system of the point cloud of the previous frame, so as to obtain the corrected point cloud data after removing the overall motion interference of the target.

[0055] As an example, the point cloud data of the current frame P ={ p1, p2, p3... pn},in, pi = ( xi , yi , zi (Point cloud data from the previous frame) Q ={ q1,q2,q3...qn}, where q i = ( xi' , yi' , zi' ), each point in the point cloud data of the current frame pi The point q corresponding to the point cloud data in the previous frame i pair.

[0056] As an example, the step of registering the point cloud data of the current frame with the point cloud data of the previous frame to obtain the optimal rotation matrix and translation vector of the current frame relative to the previous frame includes: The point cloud data of the current frame and the point cloud data of the previous frame are decentralized to obtain a decentralized coordinate matrix. Construct a weighted covariance matrix based on the decentralized coordinate matrix; Singular value decomposition is performed on the weighted covariance matrix to obtain the optimal rotation matrix and optimal translation vector between point clouds of different frames.

[0057] The specific process is as follows: Step 1: Data preparation and decentralization.

[0058] a. Calculate the centroid of the two point sets based on their weights: The best-fit transformation is usually performed around the centroid of the two point sets, which is the average position of the point sets.

[0059] Centroid of the current frame point cloud dataset:

[0060] The calculation method for the centroid of the previous frame's point cloud dataset is the same, so it will not be repeated here.

[0061] b. Calculate the decentralized coordinates of all points.

[0062] Translate the two point sets (i.e., the point cloud data of the previous frame and the point cloud data of the current frame) so that their centroids coincide with the origin of the coordinate system.

[0063] Centralized current frame point cloud data :

[0064] Centralized previous frame point cloud data :

[0065] Step 2: Construct the weighted covariance matrix.

[0066] a. Construct a 3×3 weighted covariance matrix H:

[0067] Singular value decomposition (SVD) is performed on matrix H. SVD decomposes matrix H into the product of three matrices:

[0068] U and V are both 3×3 orthogonal matrices (i.e., their columns are unit orthogonal), and Σ is a 3×3 diagonal matrix. The elements on the diagonal are called singular values, which are usually arranged in descending order.

[0069] The purpose of this step is that any linear transformation (matrix H) can be viewed as first rotating, then stretching along the coordinate axes (Σ), and finally rotating again (U).

[0070] Step 3: Calculate the optimal rotation and translation.

[0071] a. Calculate the optimal rotation matrix R: The optimal rotation matrix can be directly obtained from the SVD results:

[0072] Special handling: If det(R) = -1 (indicating that R is a reflection), then invert the third column of V and then calculate R to ensure that it is a pure rotation rather than a reflection.

[0073] b. Calculate the optimal translation vector T: Once the rotation matrix R is obtained, the translation vector can be directly obtained through the centroid relation:

[0074] That is, first rotate the center of gravity of the source point cloud to the orientation of the target point cloud, and then translate them to make them coincide.

[0075] Step 4, Logical Output: Transformation result: Transform each point in the current frame point cloud data pi By applying the calculated R and T values, a new position aligned with the previous frame's point cloud data is obtained, eliminating the overall rigid body motion displacement of the human body. This transformation represents the optimal fitting position that minimizes the overall deviation.

[0076] Fit Report: The system calculates and reports statistics such as the maximum deviation, average deviation, and standard deviation after transformation, which are used to accurately quantify the fit quality.

[0077] In some embodiments, after registering the point cloud data of the current frame with the point cloud data of the previous frame, the method further includes: Determine whether the registration process meets the preset reliability conditions; If the conditions are not met, the point cloud data of the current frame is registered with the point cloud data of a preset reference frame, and the point cloud data of the current frame is corrected according to the registration result.

[0078] As an example, the preset reference frame point cloud data is the first frame point cloud data collected at the initial moment, or the point cloud data collected at a certain historical moment that meets the reliability conditions.

[0079] As an example, the reliability conditions include, but are not limited to: registration error less than a preset threshold, point cloud overlap meeting a preset ratio, registration iteration number within a reasonable range, and the change range of rotation matrix and translation vector not exceeding the reasonable range of normal rigid body motion of the human body.

[0080] The validity of the point cloud registration result between the current frame and the previous frame is verified to determine whether the current registration process meets the preset reliability conditions. If the result is that the preset reliability conditions are not met, it means that there may be point cloud loss, excessive interference, sudden change in human posture, or tracking failure between the current frame and the previous frame. In this case, the previous frame is no longer used as the registration reference. Instead, the point cloud data of the current frame is re-registered with the point cloud data of the pre-saved preset reference frame. The point cloud data of the current frame is then corrected based on the rotation matrix and translation vector obtained from this registration to avoid errors in the extraction of chest cavity micro-motion signals due to inter-frame registration failure, thereby improving the stability and robustness of vital sign detection in complex scenarios.

[0081] Through the above embodiments, point cloud data is registered to eliminate displacement caused by overall human movement. That is, point cloud registration is used to compensate for overall movement by matrix decomposition, thereby extracting pure local micro-motion signals (such as chest rise and fall caused by heartbeat and / or breathing), and thus obtaining vital signs signals.

[0082] In step 1023, a phase-time signal characterizing thoracic cavity micromovement is extracted from the corrected point cloud data and used as the feature signal characterizing thoracic cavity micromovement.

[0083] In some embodiments, extracting the phase-time signal characterizing thoracic cavity micromovements from the corrected point cloud data includes: Determine the residual vector between the corrected point cloud data and the previous frame's point cloud data; Based on the component of the residual vector in the radar radial direction, the instantaneous micro-displacement at the current moment is generated; The instantaneous micro-displacements at multiple consecutive moments are combined in chronological order to form the phase-time signal reflecting the micro-movements of the thoracic cavity.

[0084] For the same reflection point in the current frame point cloud data after registration and correction, the coordinate difference between the two points is calculated to obtain the residual vector. This corrected residual vector eliminates the rigid body motion component of the human body as a whole, retaining only the minute chest cavity motion component caused by breathing and heartbeat. Then, the residual vector is projected onto the radar detection radial direction to obtain the projection component of the residual vector in the radar radial direction. The instantaneous micro-motion displacement corresponding to the current moment is determined based on this projection component. Finally, according to the sampling time sequence of the radar signal, the instantaneous micro-motion displacements obtained from multiple consecutive sampling moments are arranged and combined sequentially to form a time-varying time sequence. This time sequence is the phase time sequence signal reflecting the chest cavity micro-motion, which is used for subsequent separation of respiratory and heartbeat signals.

[0085] As an example, we extract the complex phase information Φ(t) of the signal. The core principle is: ΔΦ(t)≈(4π / λ)*Δd(t) Where λ is the radar wavelength (~3.9 mm, 77 GHz), and Δd(t) is the instantaneous minute displacement of the thoracic cavity relative to the radar.

[0086] A phase-time signal Φ(t) is obtained that is completely synchronized with the change of chest cavity displacement over time, which directly modulates the movement of breathing and heartbeat.

[0087] In step 105, the phase timing signal is separated to obtain respiratory and heartbeat signals that reflect the target's vital signs.

[0088] In some embodiments, the phase timing signal is subjected to a first bandpass filtering process to extract a signal in a first frequency range as a breathing signal; The phase timing signal is subjected to a second bandpass filter to extract the signal in the second frequency range as the heartbeat signal.

[0089] For example, bandpass filtering is performed on the phase timing signal Φ(t) to extract the signal using the normal frequency range corresponding to the respiratory signal and heartbeat signal.

[0090] Respiratory signals: extracted using a bandpass filter of 0.1-0.5 Hz (i.e., 6-30 breaths / minute).

[0091] Heartbeat signal: Extracted using a bandpass filter of 0.8-2.5Hz (i.e., 48-150 beats / minute). Due to the weaker signal, extremely high requirements are placed on the signal-to-noise ratio and the algorithm.

[0092] Through the above process, pure, time-domain respiratory and heartbeat waveforms are obtained.

[0093] In some embodiments, after separating and processing the characteristic signals representing thoracic cavity micromovements to obtain respiratory and heartbeat signals reflecting the target's vital signs, the method further includes: Periodic feature detection is performed on the respiratory signal and / or the heartbeat signal to determine the relevant signal features of the respiratory signal, the relevant signal features including amplitude features and / or temporal regularity features; Based on the relevant signal characteristics of the respiratory signal and / or the heartbeat signal, and the preset event type, identify abnormal vital signs events.

[0094] Time-domain analysis of respiratory and / or heartbeat signals, including at least one of the following: Peak / trough detection: Identify the peak (end of inspiration) or trough (end of expiration) of each respiratory cycle on the filtered respiratory waveform.

[0095] Calculate the interval: Calculate the time interval T between consecutive peaks.

[0096] To calculate the respiratory rate: Instantaneous respiratory rate (breaths / minute) = 60 / T, then calculate the average respiratory rate.

[0097] Amplitude analysis: Calculates the difference between peaks and troughs, reflecting the depth of breathing.

[0098] Based on the relevant signal features obtained from the above analysis, abnormal events are identified.

[0099] As an example, Apnea: detects whether the amplitude of the respiratory signal remains below a preset threshold for a period of time (usually >10 seconds).

[0100] Hypopnea: Detects whether there is a significant (e.g., a decrease of 50%) and a sustained (>10 seconds) reduction in the amplitude of the respiratory signal.

[0101] Rhythm abnormalities: Analyze the regularity of waveforms to identify periodic, gradually increasing and decreasing abnormal patterns such as Cheyne-Stokes respiration.

[0102] In some embodiments, by introducing deep learning models into the non-contact vital sign detection process, the performance bottleneck of traditional signal processing algorithms in complex vehicle environments can be overcome, achieving an upgrade from simple signal extraction to intelligent perception, accurate identification, and automatic analysis. This enables the radar vital sign monitoring system to have stronger anti-interference capabilities, higher detection accuracy, and richer physiological analysis capabilities, making it particularly suitable for scenarios such as vehicle sleep monitoring and driving health and safety monitoring where human body movements are weak, environmental interference is complex, and breathing patterns are atypical.

[0103] As an example, the method also includes: The phase timing signal, respiratory signal and / or heartbeat signal, or features extracted from them, are input into a pre-trained deep learning model; Obtain the output of the deep learning model and generate at least one of the following results: Denoising or enhancement processing of vital signs signals; Results of respiratory pattern or abnormal event identification; The results of sleep stage segmentation.

[0104] The processed time-series waveforms, spectrograms, or original features are input into a pre-trained deep learning model (such as 1D-CNN, LSTM, or Transformer).

[0105] a. Signal enhancement: More accurately separate weak vital signs from noise.

[0106] b. Intelligent recognition: Automatically identifies and classifies complex and atypical breathing patterns or sleep events.

[0107] c. Sleep stages: By combining body movement, breathing and heartbeat patterns, the sleep stages are automatically divided (wakefulness, light sleep, deep sleep, REM sleep).

[0108] By deeply integrating deep learning with radar vital sign detection, the system achieves more accurate signal recognition and more intelligent analysis, enabling it to maintain high precision and robustness in complex in-vehicle environments. This provides reliable technical support for applications such as vehicle driving safety, in-vehicle health monitoring, and sleep monitoring.

[0109] This application provides a radar-based method, system, and readable storage medium for monitoring vital signs, which has the following advantages compared to the low accuracy and security of current image monitoring methods: This method uses radar to emit electromagnetic waves and capture reflected waves containing minute phase changes caused by subtle movements of the human chest cavity (breathing and / or heartbeat). Based on the radar echo signal, characteristic signals corresponding to each reflection point are extracted. Interference removal processing is then applied to the radar echo signal to eliminate overall motion interference, thereby extracting accurate respiratory and heartbeat signals. Furthermore, since the radar point cloud data used only reflects the displacement and distance information of the monitored area and does not collect image information of the monitored person, it protects the privacy of the monitored person to a certain extent, meets practical monitoring needs, and has certain promotional value.

[0110] Based on the same concept as the above method, this application also proposes a radar-based vital signs monitoring system.

[0111] The radar-based vital signs monitoring system includes a radar transceiver array and a processor coupled to the radar transceiver array.

[0112] A radar transceiver array includes a radar transmitting antenna array and a receiving antenna array, wherein the radar transmitting antenna array is configured to periodically transmit frequency-modulated continuous waves with linearly increasing frequency, and the receiving antenna array is connected to the radar transmitting antenna array and is configured to acquire radar echo signals containing the chest cavity region of a target. A processor for implementing the radar-based vital sign monitoring method as described in any of the above embodiments.

[0113] The specific implementation process of the functions and roles of each module in the above device can be found in the implementation process of the corresponding steps in the above method, which can achieve the same technical effect, and will not be repeated here.

[0114] Figure 3 An example is a schematic diagram of the physical structure of a radar-based vital sign monitoring device, such as... Figure 3 As shown, the radar-based vital sign monitoring device may include: a processor 810, a communication interface 820, a memory 830, and a communication bus 840. The processor 810, communication interface 820, and memory 830 communicate with each other via the communication bus 840. The processor 810 can call logical instructions from the memory 830 to execute the radar-based vital sign monitoring method.

[0115] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion 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 this application. 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.

[0116] On the other hand, this application also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the radar-based vital sign monitoring methods provided by the above methods.

[0117] In another aspect, this application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the radar-based vital sign monitoring methods provided by the above methods.

[0118] It should be noted that the technical solutions or features described in the above embodiments can be combined or supplemented with each other without conflict. The scope of protection of this application is not limited to the precise structures described in the above embodiments and shown in the accompanying drawings; all modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A radar-based method for monitoring vital signs, characterized in that, The method includes: Acquire radar echo signals for the target thoracic cavity region, wherein the radar echo signals carry characteristic information related to the micro-movement of the target thoracic cavity; The radar echo signal is subjected to interference removal processing, and characteristic signals representing thoracic cavity micro-movements are extracted from the processed signal. The characteristic signals representing chest cavity micro-movements are separated and processed to obtain respiratory and heartbeat signals that reflect the target's vital signs.

2. The radar-based vital sign monitoring method as described in claim 1, characterized in that, The interference removal process on the radar echo signal, and the extraction of characteristic signals representing thoracic cavity micromovements from the processed signal, includes: Based on the radar echo signal, displacement time-series data corresponding to multiple reflection points on the surface of the target thoracic cavity are extracted to generate multiple sets of point cloud data. The multiple sets of point cloud data are registered to obtain corrected point cloud data after removing the overall motion interference of the target; The phase-time signal characterizing thoracic cavity micromovement is extracted from the corrected point cloud data and used as the feature signal characterizing thoracic cavity micromovement.

3. The radar-based vital sign monitoring method as described in claim 1, characterized in that, The multiple sets of point cloud data include point cloud data from the current frame and point cloud data from the previous frame. The registration process performed on the multiple sets of point cloud data to obtain corrected point cloud data after removing overall target motion interference includes: Register the point cloud data of the current frame with the point cloud data of the previous frame to obtain the optimal rotation matrix and translation vector of the current frame relative to the previous frame. Based on the optimal rotation matrix and translation vector, the point cloud data of the current frame is calibrated to the unified reference coordinate system of the point cloud of the previous frame, so as to obtain the corrected point cloud data after removing the overall motion interference of the target.

4. The radar-based vital sign monitoring method as described in claim 3, characterized in that, After registering the point cloud data of the current frame with the point cloud data of the previous frame, the method further includes: Determine whether the registration process meets the preset reliability conditions; If the conditions are not met, the point cloud data of the current frame is registered with the point cloud data of a preset reference frame, and the point cloud data of the current frame is corrected according to the registration result.

5. The radar-based vital sign monitoring method as described in claim 3, characterized in that, The step of registering the point cloud data of the current frame with the point cloud data of the previous frame to obtain the optimal rotation matrix and translation vector of the current frame relative to the previous frame includes: The point cloud data of the current frame and the point cloud data of the previous frame are decentralized to obtain a decentralized coordinate matrix. Construct a weighted covariance matrix based on the decentralized coordinate matrix; Singular value decomposition is performed on the weighted covariance matrix to obtain the optimal rotation matrix and optimal translation vector between point clouds of different frames.

6. The radar-based vital sign monitoring method as described in claim 1, characterized in that, The acquisition of radar echo signals targeting the chest cavity region includes: The captured raw echo signal is preprocessed to obtain a radar echo signal targeting the chest cavity region of the target; wherein the raw echo signal contains emission components from stationary and moving objects, and the corresponding preprocessing includes static clutter filtering.

7. The radar-based vital sign monitoring method as described in claim 2, characterized in that, The extraction of phase-time signals characterizing thoracic cavity micromovements from the corrected point cloud data includes: Determine the residual vector between the corrected point cloud data and the previous frame's point cloud data; Based on the component of the residual vector in the radar radial direction, the instantaneous micro-displacement at the current moment is generated; By combining the instantaneous micro-displacements at multiple consecutive moments in chronological order, a phase-time signal reflecting the micro-movements of the thoracic cavity is formed.

8. The radar-based vital sign monitoring method as described in claim 2, characterized in that, After separating and processing the characteristic signals representing thoracic cavity micromovements to obtain respiratory and heartbeat signals reflecting the target's vital signs, the method further includes: Periodic feature detection is performed on the respiratory signal and / or the heartbeat signal to determine the relevant signal features of the respiratory signal, the relevant signal features including amplitude features and / or temporal regularity features; Based on the relevant signal characteristics of the respiratory signal and / or the heartbeat signal, and the preset event type, identify abnormal vital signs events.

9. The radar-based vital sign monitoring method as described in claim 1, characterized in that, The method further includes: Phase timing signals, respiratory signals and / or heartbeat signals, or features extracted from them, are input into a pre-trained deep learning model; Obtain the output of the deep learning model and generate at least one of the following results: Denoising or enhancement processing of vital signs signals; Results of respiratory pattern or abnormal event identification; The results of sleep stage segmentation.

10. A radar-based vital sign monitoring system, characterized in that, The system includes: A radar transceiver array includes a radar transmitting antenna array and a receiving antenna array, wherein the radar transmitting antenna array is configured to periodically transmit frequency-modulated continuous waves with linearly increasing frequency, and the receiving antenna array is connected to the radar transmitting antenna array and is configured to acquire radar echo signals containing the chest cavity region of a target. The processor, coupled to the radar transceiver array, is used to implement the radar-based vital sign monitoring method as described in any one of claims 1 to 9.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a radar-based vital signs monitoring program, which, when executed, implements the steps of the radar-based vital signs monitoring method as described in any one of claims 1 to 9.

Images