Multi-person sleep detection system based on non-contact physical sign monitoring

Abstract

The invention discloses a multi-person sleep detection system based on non-contact type physical sign monitoring, belongs to the technical field of sleep detection systems, and aims to solve the problem that the accuracy is greatly reduced due to the fact that an existing non-contact type sleep monitoring system is influenced by human body shaking. The system does not need to wear equipment, improves comfort and naturalness, is suitable for long-term monitoring, simultaneously monitors respiration and heartbeat data of a plurality of individuals, improves efficiency and coverage, provides high-precision real-time monitoring by using a millimeter wave radar and a laser radar in combination with an advanced signal processing algorithm, and realizes real-time monitoring through denoising and feature extraction technologies. The system ensures accuracy and stability of monitoring data, displays and stores the monitoring data in real time, supports subsequent analysis, is beneficial to health management and illness state tracking, is suitable for a plurality of scenes such as families, hospitals and nursing homes, and provides support for personal health monitoring and medical treatment.

Description

Technical Field

[0001] The present invention relates to the technical field of sleep detection systems, and in particular to a multi-person sleep detection system based on non-contact vital sign monitoring. Background Art

[0002] As people pay more and more attention to the health level of life, vital signs detection technology has important value. Respiratory rate and heart rate are the most basic vital signs of the human body. At present, the conventional detection methods of these two frequencies require direct contact with the monitored objects, which not only causes discomfort to the human body, but also has great restrictions on application occasions. Therefore, non-contact detection methods have broad development prospects and application value.

[0003] Doppler radar has the advantages of all-day, all-weather, and strong penetration. In recent years, non-contact vital signs detection technology based on millimeter-wave radar has gradually become a hot research direction; although the technology has been initially applied at this stage, there are still some huge challenges, the most typical of which is the unconscious human body shaking during the detection process; this randomly generated body movement is comparable to or even greater than the amplitude of human chest wall displacement, and is an extremely strong noise source, which will greatly reduce the accuracy of vital signs estimation. Summary of the invention

[0004] The purpose of the present invention is to provide a multi-person sleep detection system based on non-contact vital sign monitoring, which solves the problem that the existing non-contact sleep monitoring system in the background technology is greatly affected by human body shaking, resulting in greatly reduced accuracy.

[0005] To achieve the above object, the present invention provides the following technical solution: a multi-person sleep detection system based on non-contact vital sign monitoring, comprising a data monitoring unit and a data processing unit, wherein the data monitoring unit comprises:

[0006] Millimeter-wave radar: The millimeter-wave radar is selected as a 77GHz millimeter-wave radar, which is used to simultaneously identify and track multiple targets and collect the breathing and heartbeat data of the monitored person in real time;

[0007] LiDAR: LiDAR obtains detailed point cloud data of the monitoring area through 3D scanning, providing the system with accurate spatial information. Based on the point cloud data, it uses advanced algorithms to identify and extract human targets, continuously track them, record their positions and movement trajectories, and provide human target jitter information as a supplement and verification of millimeter wave radar data.

[0008] The data processing unit comprises:

[0009] Data acquisition and processing module: Start the millimeter-wave radar and lidar modules to collect breathing and heartbeat signals of multiple people in real time, receive the raw data from the millimeter-wave radar and lidar, and apply the denoising algorithm to remove high-frequency noise and background interference in the raw signals;

[0010] Feature extraction module: extracts frequency, amplitude and phase difference characteristic parameters from the preprocessed signal, and uses signal separation technology to distinguish the breathing and heartbeat signals of different people;

[0011] Algorithm processing module: through the EKF algorithm, accurately estimate each person's breathing and heartbeat status in real time;

[0012] Parameter calculation module: Calculates each person's breathing and heart rate based on the data processed by the EKF algorithm;

[0013] Display storage module: used to display the calculated respiratory and heart rate in real time on the monitoring interface, providing medical staff or users with immediate physiological status feedback, and also used to store the monitoring data, including raw data and processed results, in a local or cloud database for subsequent data analysis, disease tracking and health management;

[0014] The processing steps of the EKF algorithm in the algorithm processing module are:

[0015] S11. Initial state estimate and covariance matrix, setting the initial state estimate and initial covariance matrix of the system, the initial value is based on the prior knowledge of the system and obtained through preliminary measurement;

[0016] S12. Prediction step: predict the state estimation value and covariance matrix of the previous moment to obtain the predicted state and predicted covariance of the current moment;

[0017] S13. Update step: receiving observations from radar and lidar, and updating state estimates and covariance matrices using observation models and Kalman gains;

[0018] S14. Repeated process: Through continuous prediction and updating, the EKF algorithm can continuously track and update each person's breathing and heartbeat status;

[0019] The state vector x k It is expressed as: Where N is the number of people, f r ,i(k) and f h ,i(k) respectively represent the breathing rate and heart rate of the i-th person at time k;

[0020] The observation vector is defined as: Where M is the number of sensors, z f ,i(k) and zΦ ,i(k) respectively represent the frequency and phase observed by the i-th sensor at time k;

[0021] The observation model is: k =h(x k )+v k , where h(·) is the nonlinear observation function, v k is zero-mean Gaussian observation noise;

[0022] The formula for state prediction in S12 is: k|k-1 =F k x k-1|k-1 +w k , where F k is the state transfer matrix, w k is the process noise; the formula for covariance prediction is: Where Q k is the covariance matrix of the process noise;

[0023] The formula of the Kalman gain in S13 is: Among them, H k is the Jacobian matrix of the observation model, R k is the covariance matrix of the observation noise; the state estimation update formula is: k|k =x k|k-1 +K k (z k -h(x k|k-1 )), the covariance matrix update formula is: P k|k =(IK k H k ) k|k-1 .

[0024] Furthermore, the millimeter wave radar is also used to extract distance information when identifying the breathing and heartbeat data of the detected person, and the extraction steps are:

[0025] S21. Generate FMCW signal: generate linear frequency modulation signal;

[0026] S22. Transmit signal: transmit FMCW signal through antenna;

[0027] S23. Receive signal: receive the signal reflected from the target;

[0028] S24. Extracting the difference frequency signal: Mixing the received signal with the transmitted signal through a mixer to obtain a difference frequency signal;

[0029] S25. Calculate distance information: Calculate the distance of the target using the difference frequency signal.

[0030] Furthermore, the distance calculation formula is:

[0031] Where R is the distance and c is the speed of light 3*10 8 m / s, Δf is the frequency difference, and B is the frequency modulation bandwidth.

[0032] Furthermore, the data acquisition and processing module uses a discrete wavelet transform algorithm to process the signal, remove high-frequency noise, smooth the signal, and extract characteristic parameters related to breathing and heartbeat.

[0033] Furthermore, the formula of the discrete wavelet transform is: Among them, x(n) is the input signal discrete time series, ψ(·) is the wavelet function, a is the scale parameter that controls the expansion and contraction of the wavelet function, and b is the translation parameter that controls the position of the wavelet function on the time axis.

[0034] Furthermore, the algorithm processing module further includes a minimum mean square error in addition to the EKF algorithm, and the minimum mean square error algorithm is used to further process the signal or optimize the performance of the EKF.

[0035] Furthermore, the weight update formula of the minimum mean square error is: k+1 =w k +2μe k x k (n), the error signal calculation formula is: where w k is the filter weight vector at the kth iteration, μ is the learning rate or step size, e k is the error signal at the kth iteration, d(n) is the expected output, i.e., the target signal, x k (n) is the input signal vector at time n during the kth iteration, Indicates w k The transpose of .

[0036] Furthermore, the calculation steps of the parameter calculation module are:

[0037] S81. Data extraction: Extract the state estimates related to each person from the output of the EKF algorithm. The state estimates usually include estimates of each person's breathing rate and heart rate. Ensure that these data are organized according to each person's identity or index so that they can be calculated separately later;

[0038] S82. State stability check: Before directly calculating the frequency, the state estimate is checked for stability to ensure the reliability of the calculated frequency;

[0039] S83. Frequency calculation: Once the steady-state estimate is determined, the breathing and heart rate of each person can be calculated. The frequency calculation is usually based on the periodic changes within a time interval. For periodic physiological signals such as breathing and heart rate, the frequency can be calculated by measuring the time difference between adjacent peaks or troughs.

[0040] Furthermore, in S83, the specific steps of frequency calculation are:

[0041] S831. Determine the cycle: For each person's breathing and heartbeat signals, find adjacent peaks or troughs and calculate the time difference between them. This time difference is the cycle of the signal;

[0042] S832. Calculate frequency: Convert the period to frequency, that is, frequency = 1 / period, the period is in seconds and the frequency is in Hertz.

[0043] Furthermore, in S82, since the EKF is an iterative process based on prediction and update, the estimated values ​​at certain moments may fluctuate greatly due to noise or model inaccuracy. Therefore, by checking whether the state estimate values ​​for several consecutive time steps remain relatively stable, a sliding window averaging method is used to ensure that the calculated frequency is reliable.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] The present invention provides a multi-person sleep detection system based on non-contact vital sign monitoring. It does not require wearing equipment, improves comfort and naturalness, is suitable for long-term monitoring, and simultaneously monitors the breathing and heartbeat data of multiple individuals to improve efficiency and coverage. It uses millimeter-wave radar and lidar, combined with advanced signal processing algorithms, to provide high-precision real-time monitoring. Through denoising and feature extraction technology, it ensures the accuracy and stability of monitoring data, displays and stores monitoring data in real time, supports subsequent analysis, and is helpful for health management and disease tracking. It is suitable for multiple scenarios such as homes, hospitals, and nursing homes, and provides support for personal health monitoring and medical treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a module diagram of the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] In order to solve the technical problem that the existing non-contact sleep monitoring system is greatly affected by human body shaking, such as Figure 1 As shown, the following preferred technical solutions are provided:

[0049] A multi-person sleep detection system based on non-contact vital sign monitoring includes a data monitoring unit and a data processing unit, wherein the data monitoring unit includes:

[0050] Millimeter-wave radar: The millimeter-wave radar is selected as a 77GHz millimeter-wave radar, which is used to simultaneously identify and track multiple targets and collect the breathing and heartbeat data of the monitored person in real time;

[0051] LiDAR: LiDAR obtains detailed point cloud data of the monitoring area through 3D scanning, providing the system with accurate spatial information. Based on the point cloud data, it uses advanced algorithms to identify and extract human targets, continuously track them, record their positions and movement trajectories, and provide human target jitter information as a supplement and verification of millimeter wave radar data.

[0052] The data processing unit comprises:

[0053] Data acquisition and processing module: Start the millimeter-wave radar and lidar modules to collect breathing and heartbeat signals of multiple people in real time, receive the raw data from the millimeter-wave radar and lidar, and apply the denoising algorithm to remove high-frequency noise and background interference in the raw signals;

[0054] Feature extraction module: extracts frequency, amplitude and phase difference characteristic parameters from the preprocessed signal, and uses signal separation technology to distinguish the breathing and heartbeat signals of different people;

[0055] Algorithm processing module: through the EKF algorithm, accurately estimate each person's breathing and heartbeat status in real time;

[0056] Parameter calculation module: Calculates each person's breathing and heart rate based on the data processed by the EKF algorithm;

[0057] Display storage module: used to display the calculated respiratory and heart rate in real time on the monitoring interface, providing medical staff or users with immediate physiological status feedback, and also used to store the monitoring data, including raw data and processed results, in a local or cloud database for subsequent data analysis, disease tracking and health management;

[0058] The processing steps of the EKF algorithm in the algorithm processing module are:

[0059] S11. Initial state estimate and covariance matrix, setting the initial state estimate and initial covariance matrix of the system, the initial value is based on the prior knowledge of the system and obtained through preliminary measurement;

[0060] S12. Prediction step: predict the state estimation value and covariance matrix of the previous moment to obtain the predicted state and predicted covariance of the current moment;

[0061] S13. Update step: receiving observations from radar and lidar, and updating state estimates and covariance matrices using observation models and Kalman gains;

[0062] S14. Repeated process: Through continuous prediction and updating, the EKF algorithm can continuously track and update each person's breathing and heartbeat status;

[0063] Where N is the number of people, f r ,i(k) and f h ,i(k) respectively represent the breathing rate and heart rate of the i-th person at time k;

[0064] The observation vector is defined as: Where M is the number of sensors, z f ,i(k) and z Φ ,i(k) respectively represent the frequency and phase observed by the i-th sensor at time k;

[0065] The observation model is: k =h(x k )+v k , where h(·) is the nonlinear observation function, v k is zero-mean Gaussian observation noise;

[0066] The formula for state prediction in S12 is: kk-1 =F k x k-1k-1 +w k , where F k is the state transfer matrix, w k is the process noise; the formula for covariance prediction is: Where Q k is the covariance matrix of the process noise;

[0067] The formula of the Kalman gain in S13 is: Among them, H k is the Jacobian matrix of the observation model, R k is the covariance matrix of the observation noise; the state estimation update formula is: kk =x kk-1 +K k (z k -h(x kk-1 )), the covariance matrix update formula is: P kk =(IKk H k ) kk-1 .

[0068] The millimeter wave radar is also used to extract distance information when identifying the breathing and heartbeat data of the detected person, and the extraction steps are as follows:

[0069] S21. Generate FMCW signal: generate linear frequency modulation signal;

[0070] S22. Transmit signal: transmit FMCW signal through antenna;

[0071] S23. Receive signal: receive the signal reflected from the target;

[0072] S24. Extracting the difference frequency signal: Mixing the received signal with the transmitted signal through a mixer to obtain a difference frequency signal;

[0073] S25. Calculate distance information: Calculate the distance of the target using the difference frequency signal.

[0074] The distance calculation formula is:

[0075] Where R is the distance and c is the speed of light 3*10 8 m / s, Δf is the frequency difference, and B is the frequency modulation bandwidth.

[0076] The data acquisition and processing module uses a discrete wavelet transform algorithm to process the signal, remove high-frequency noise, smooth the signal, and extract characteristic parameters related to breathing and heartbeat.

[0077] The formula of the discrete wavelet transform is: Among them, x(n) is the input signal discrete time series, ψ(·) is the wavelet function, a is the scale parameter that controls the expansion and contraction of the wavelet function, and b is the translation parameter that controls the position of the wavelet function on the time axis.

[0078] The algorithm processing module includes a minimum mean square error in addition to the EKF algorithm. The minimum mean square error algorithm is used to further process the signal or optimize the performance of the EKF.

[0079] The weight update formula of the minimum mean square error is: k+1 =w k +2μe k x k (n), the error signal calculation formula is: where w k is the filter weight vector at the kth iteration, μ is the learning rate or step size, e k is the error signal at the kth iteration, d(n) is the expected output, i.e., the target signal, xk (n) is the input signal vector at time n during the kth iteration, Indicates w k The transpose of .

[0080] The calculation steps of the parameter calculation module are:

[0081] S81. Data extraction: Extract the state estimates related to each person from the output of the EKF algorithm. The state estimates usually include estimates of each person's breathing rate and heart rate. Ensure that these data are organized according to each person's identity or index so that they can be calculated separately later;

[0082] S82. State stability check: Before directly calculating the frequency, the state estimate is checked for stability to ensure the reliability of the calculated frequency;

[0083] S83. Frequency calculation: Once the steady-state estimate is determined, the breathing and heart rate of each person can be calculated. The frequency calculation is usually based on the periodic changes within a time interval. For periodic physiological signals such as breathing and heart rate, the frequency can be calculated by measuring the time difference between adjacent peaks or troughs.

[0084] In S83, the specific steps of frequency calculation are:

[0085] S831. Determine the cycle: For each person's breathing and heartbeat signals, find adjacent peaks or troughs and calculate the time difference between them. This time difference is the cycle of the signal;

[0086] S832. Calculate frequency: Convert the period to frequency, that is, frequency = 1 / period, the period is in seconds and the frequency is in Hertz.

[0087] In S82, since EKF is an iterative process based on prediction and update, the estimated values ​​at certain moments may fluctuate greatly due to noise or model inaccuracy. Therefore, a sliding window averaging method is used to ensure that the calculated frequency is reliable by checking whether the state estimate values ​​for several consecutive time steps remain relatively stable.

[0088] Specifically, the millimeter-wave radar and lidar sensors work together to monitor the breathing and heartbeat data of multiple people in real time. The millimeter-wave radar is responsible for simultaneously identifying and tracking multiple targets and collecting breathing and heartbeat signals; the lidar obtains detailed point cloud data through three-dimensional scanning, provides accurate spatial information, identifies human targets and tracks their positions and movement trajectories, and supplements and verifies the data of the millimeter-wave radar. The data processing unit is responsible for collecting and processing the data of these sensors;

[0089] First, the system starts the radar module to collect signals in real time and uses a denoising algorithm to remove high-frequency noise. Then, it extracts characteristic parameters from the preprocessed signals and uses signal separation technology to distinguish the breathing and heartbeat signals of different people.

[0090] In the algorithm processing part, the system applies the extended Kalman filter algorithm to predict and update each person's breathing and heartbeat status to ensure real-time and accurate state estimation. Subsequently, the system calculates each person's breathing and heartbeat frequency based on these estimated data, and displays the results in real time on the monitoring interface through the display storage module. At the same time, the data is stored locally or in the cloud for subsequent analysis and health management.

[0091] The entire system accurately measures the distance to the target through the FMCW signal generation and reception, difference frequency signal extraction and distance calculation of the millimeter wave radar, and combines the point cloud data of the lidar with discrete wavelet transform and minimum mean square error algorithm to further optimize signal processing and feature extraction, ultimately providing accurate and reliable physiological status monitoring.

[0092] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0093] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A multi-person sleep detection system based on non-contact vital sign monitoring, comprising a data monitoring unit and a data processing unit, characterized in that: The data monitoring unit comprises: Millimeter-wave radar: The millimeter-wave radar is selected as a 77GHz millimeter-wave radar, which is used to simultaneously identify and track multiple targets and collect the breathing and heartbeat data of the monitored person in real time; LiDAR: LiDAR obtains detailed point cloud data of the monitoring area through 3D scanning, providing the system with accurate spatial information. Based on the point cloud data, it uses advanced algorithms to identify and extract human targets, continuously track them, record their positions and movement trajectories, and provide human target jitter information as a supplement and verification of millimeter wave radar data. The data processing unit comprises: Data acquisition and processing module: Start the millimeter-wave radar and lidar modules to collect breathing and heartbeat signals of multiple people in real time, receive the raw data from the millimeter-wave radar and lidar, and apply the denoising algorithm to remove high-frequency noise and background interference in the raw signals; Feature extraction module: extracts frequency, amplitude and phase difference characteristic parameters from the preprocessed signal, and uses signal separation technology to distinguish the breathing and heartbeat signals of different people; Algorithm processing module: through the EKF algorithm, accurately estimate each person's breathing and heartbeat status in real time; Parameter calculation module: Calculates each person's breathing and heart rate based on the data processed by the EKF algorithm; Display storage module: used to display the calculated respiratory and heart rate in real time on the monitoring interface, providing medical staff or users with immediate physiological status feedback, and also used to store the monitoring data, including raw data and processed results, in a local or cloud database for subsequent data analysis, disease tracking and health management; The processing steps of the EKF algorithm in the algorithm processing module are: S11. Initial state estimate and covariance matrix, setting the initial state estimate and initial covariance matrix of the system, the initial value is based on the prior knowledge of the system and obtained through preliminary measurement; S12. Prediction step: predict the state estimation value and covariance matrix of the previous moment to obtain the predicted state and predicted covariance of the current moment; S13. Update step: receiving observations from radar and lidar, and updating state estimates and covariance matrices using observation models and Kalman gains; S14. Repeated process: Through continuous prediction and updating, the EKF algorithm can continuously track and update each person's breathing and heartbeat status; The state vector x k It is expressed as: Where N is the number of people, f r ,i(k) and f h ,i(k) respectively represent the breathing rate and heart rate of the i-th person at time k; The observation vector is defined as: Where M is the number of sensors, z f ,i(k) and z Φ ,i(k) respectively represent the frequency and phase observed by the i-th sensor at time k; The observation model is: k =h(x k )+v k , where h(·) is the nonlinear observation function, v k is zero-mean Gaussian observation noise; The formula for state prediction in S12 is: k|k-1 =F k x k-1|k-1 +w k , where F k is the state transfer matrix, w k is the process noise; the formula for covariance prediction is: Where Q k is the covariance matrix of the process noise; The formula of the Kalman gain in S13 is: Among them, H k is the Jacobian matrix of the observation model, R k is the covariance matrix of the observation noise; the state estimation update formula is: k|k =x k|k-1 +K k (z k -h(x k|k-1 )), the covariance matrix update formula is: P k|k =(IK k H k ) k|k-1 .

2. A multi-person sleep detection system based on non-contact vital sign monitoring as claimed in claim 1, characterized in that: The millimeter wave radar is also used to extract distance information when identifying the breathing and heartbeat data of the detected person, and the extraction steps are as follows: S21. Generate FMCW signal: generate linear frequency modulation signal; S22. Transmit signal: transmit FMCW signal through antenna; S23. Receive signal: receive the signal reflected from the target; S24. Extracting the difference frequency signal: Mixing the received signal with the transmitted signal through a mixer to obtain a difference frequency signal; S25. Calculate distance information: Calculate the distance of the target using the difference frequency signal.

3. A multi-person sleep detection system based on non-contact vital sign monitoring as claimed in claim 2, characterized in that: The distance calculation formula is: Where R is the distance and c is the speed of light 3*10 8 m / s, Δf is the frequency difference, and B is the frequency modulation bandwidth.

4. A multi-person sleep detection system based on non-contact vital sign monitoring as claimed in claim 1, characterized in that: The data acquisition and processing module uses a discrete wavelet transform algorithm to process the signal, remove high-frequency noise, smooth the signal, and extract characteristic parameters related to breathing and heartbeat.

5. A multi-person sleep detection system based on non-contact vital sign monitoring as claimed in claim 4, characterized in that: The formula of the discrete wavelet transform is: Among them, x(n) is the input signal discrete time series, ψ(·) is the wavelet function, a is the scale parameter that controls the expansion and contraction of the wavelet function, and b is the translation parameter that controls the position of the wavelet function on the time axis.

6. A multi-person sleep detection system based on non-contact vital sign monitoring as claimed in claim 1, characterized in that: The algorithm processing module includes a minimum mean square error in addition to the EKF algorithm. The minimum mean square error algorithm is used to further process the signal or optimize the performance of the EKF.

7. A multi-person sleep detection system based on non-contact vital sign monitoring as claimed in claim 6, characterized in that: The weight update formula of the minimum mean square error is: k+1 =w k +2μe k x k (n), the error signal calculation formula is: where w k is the filter weight vector at the kth iteration, μ is the learning rate or step size, e k is the error signal at the kth iteration, d(n) is the expected output, i.e., the target signal, x k (n) is the input signal vector at time n during the kth iteration, Indicates w k The transpose of .

8. The multi-person sleep detection system based on non-contact vital sign monitoring as claimed in claim 1, characterized in that: The calculation steps of the parameter calculation module are: S81. Data extraction: Extract the state estimates related to each person from the output of the EKF algorithm. The state estimates usually include estimates of each person's breathing rate and heart rate. Ensure that these data are organized according to each person's identity or index so that they can be calculated separately later; S82. State stability check: Before directly calculating the frequency, the state estimate is checked for stability to ensure the reliability of the calculated frequency; S83. Frequency calculation: Once the steady-state estimate is determined, the breathing and heart rate of each person can be calculated. The frequency calculation is based on the periodic changes within a time interval. For periodic physiological signals such as breathing and heart rate, the frequency is calculated by measuring the time difference between adjacent peaks or troughs.

9. A multi-person sleep detection system based on non-contact vital sign monitoring as claimed in claim 8, characterized in that: In S83, the specific steps of frequency calculation are: S831. Determine the cycle: For each person's breathing and heartbeat signals, find adjacent peaks or troughs and calculate the time difference between them. This time difference is the cycle of the signal; S832. Calculate frequency: Convert the period to frequency, that is, frequency = 1 / period, the period is in seconds and the frequency is in Hertz.

10. A multi-person sleep detection system based on non-contact vital sign monitoring as claimed in claim 9, characterized in that: In S82, since EKF is an iterative process based on prediction and update, the estimated values ​​at certain moments may fluctuate greatly due to noise or model inaccuracy. Therefore, a sliding window averaging method is used to ensure that the calculated frequency is reliable by checking whether the state estimate values ​​for several consecutive time steps remain relatively stable.

Images