Human body vital sign perception method based on VMD + ACMD joint decomposition
By adopting the combined VMD+ACMD decomposition and STFT time-frequency analysis methods in millimeter wave radar technology, the problem of noise interference in sign signal extraction is solved, and higher accuracy and higher quality signal extraction is achieved.
Patent Information
- Application Number
- CN202411624836.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-14
- Publication Date
- 2025-03-21
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing millimeter-wave radar technology is susceptible to noise interference in sign signal extraction, resulting in a decrease in signal quality and affecting the extraction accuracy of heart rate and respiration rate.
The human vital sign perception method based on VMD+ACMD joint decomposition is adopted, and the human body micro-movement signal is collected through millimeter wave radar, and the signal is decomposed using VMD to obtain multiple modal functions that do not overlap each other. It is further decomposed through ACMD to extract fine feature signals, and finally time-frequency analysis is performed through STFT to extract the heart rate signal in real time.
It significantly improves the separation accuracy of sign signals, enhances the anti-noise ability of the signal, ensures higher signal quality, and improves the extraction accuracy of heart rate and respiration rate.
Smart Images
Figure CN119669645A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of non-contact vital sign perception, and in particular relates to a human vital sign perception method based on VMD+ACMD joint decomposition. Background Art
[0002] With the continuous development of the economy and society and the acceleration of the aging process, people are paying more and more attention to health issues. As key indicators for measuring the health status of human life, long-term and stable monitoring of breathing and heartbeat can not only help people to grasp their own health status in a timely manner, but also provide early warning for certain cardiopulmonary diseases, so as to take effective preventive measures. In recent years, people have an increasing demand for non-contact vital sign perception, and vital sign detection technology based on millimeter-wave radar has attracted widespread attention in medical health monitoring in recent years. Millimeter-wave radar has the advantages of non-contact and long-distance detection, and can effectively detect the tiny movements and physiological signals of the human body, and then extract vital sign information such as heart rate and respiratory rate.
[0003] Existing technical means mainly rely on optical sensors (such as cameras), capacitive sensors and millimeter-wave radars to perceive vital signs information, among which optical and capacitive sensing technologies are widely used in physiological signal detection. Optical sensors (such as cameras) capture subtle changes in skin color to extract information such as heart rate and respiration, while capacitive sensors measure physiological signals by detecting changes in capacitance on the surface of human contact. However, optical sensing technology is easily affected by ambient light and is difficult to penetrate clothing or bedding for detection; capacitive sensors need to work close to the human body and are not suitable for contactless scenarios. Millimeter-wave radar technology detects human micro-movements (such as slight chest fluctuations) to achieve non-contact measurement of vital signs such as heart rate and respiration, which is not affected by ambient light or clothing, and can achieve effective monitoring at long distances or in barrier environments. However, traditional millimeter-wave radar technology usually relies on simple signal processing methods in vital sign signal extraction, which is easily affected by noise, especially in the extraction of pulse waves, which is very limited and has low accuracy.
[0004] Existing millimeter wave radar technology usually uses a single signal decomposition method, which is difficult to effectively separate complex vital signs signals, especially in an environment with large noise interference, the signal quality is significantly reduced, affecting the extraction accuracy of heart rate and respiratory rate. Therefore, how to solve the above technical problems and defects is the focus that needs to be solved urgently. Summary of the invention
[0005] The purpose of the present invention is to overcome the defects described in the background technology, so as to realize a human vital signs perception method based on VMD+ACMD joint decomposition, so as to solve the problems that millimeter wave radar technology in the prior art usually adopts a single signal decomposition method, it is difficult to effectively separate complex vital signs signals, the signal quality is significantly reduced in an environment with large noise interference, and the extraction accuracy of heart rate and respiratory rate is affected.
[0006] In order to achieve the above-mentioned object of the invention, the technical solution of the present invention is:
[0007] A human vital sign perception method based on VMD+ACMD joint decomposition includes the following steps:
[0008] Step 1: Collect data on the human body through millimeter wave radar and extract micro-motion signals in the chest area of the human body.
[0009] Step 2: Generate a distance-angle heat map. Perform FFT transformation on the sampling point of each chirp of the echo to generate a distance-angle heat map to determine the distance and angle of the target.
[0010] Step 3: DC offset removal: Remove the DC offset to improve the accuracy of pulse wave signal extraction.
[0011] Step 4: Expand DACM phase extraction. When the phase value exceeds the range of -π to π radians and wraps around, phase wraparound occurs. Apply the phase unwrapping algorithm to correct the phase wraparound and obtain the unwrapped phase value.
[0012] Step 5: Decompose the signal. After obtaining the radar micro-motion signal, use VMD to decompose the signal to obtain multiple non-overlapping modal functions.
[0013] Step 6: Signal reconstruction: For the decomposed modal signal, weighted reconstruction is performed according to the correlation coefficient RⅠ with the original signal to enhance the main frequency components of the heart rate and respiratory rate.
[0014] Step 7: Extract features. After signal reconstruction, the signal is further decomposed through ACMD to extract more refined feature signals.
[0015] Step 8: Time-frequency analysis. In the extracted micro-motion signal, STFT is used to perform time-frequency analysis to extract and dynamically track the heart rate signal in real time.
[0016] In the above-mentioned human vital sign sensing method based on VMD+ACMD joint decomposition, the human body stability signal collected by the data includes the micro-motion signal caused by the chest rise and fall.
[0017] In the above-mentioned human vital sign perception method based on VMD+ACMD joint decomposition, in the distance angle heat map generation step, the distance FFT transformation point number N and the angle FFT transformation point number N are determined according to the distance and angle requirements of the target.
[0018] In the above-mentioned human vital sign sensing method based on VMD+ACMD joint decomposition, in the DC offset removal step, the DC offset is removed so that the inverse tangent demodulation result is consistent with the phase caused by the actual physiological motion modulation echo signal. The DC offset calculation formula is:
[0019]
[0020] Where, DC Ⅰ Signal B Ⅰ (t) DC bias part; B Ⅰ (t) is the signal value at sampling point t; DC Q It is signal B Q The DC bias part of B Q (t) is signal B Q The signal value at sampling point t; M is the number of samples to be considered.
[0021] In the above-mentioned human vital sign perception method based on VMD+ACMD joint decomposition, in the extended DACM phase extraction step, a phase unwrapping algorithm is used to correct phase wrapping and obtain an unwrapped phase value to dynamically adapt to signal characteristics under different environments. The process formula for extracting phase information is:
[0022] I(t)=B I (t)-DC I ;
[0023] Q(t)=B Q (t)-DC Q ;
[0024]
[0025] Where I(t) and Q(t) represent the in-phase signal and quadrature signal after DC offset removal, respectively, and I(t)' and Q(t)' are the differential forms of I(t) and Q(t), respectively.
[0026] In the above-mentioned human vital sign perception method based on VMD+ACMD joint decomposition, in the signal decomposition step, the decomposition mode K and the penalty factor alpha are used to control the signal decomposition process. The decomposition mode number K determines the number of modes into which the signal is decomposed, and each mode represents a different frequency band; the penalty factor alpha controls the sparsity of the mode.
[0027] In the above-mentioned human vital sign perception method based on VMD+ACMD joint decomposition, in the time-frequency analysis step, STFT divides the signal into multiple time windows, and performs Fourier transform in each window to extract the heart rate and respiratory rate from the respiratory signal and pulse wave.
[0028] Compared with the prior art, the human vital signs perception method based on VMD+ACMD joint decomposition of the present invention has at least the following beneficial effects:
[0029] 1. In the human vital sign perception method based on VMD+ACMD joint decomposition of the present invention, the frequency resolution capability of VMD and the adaptive decomposition capability of ACMD are fully utilized by combining variational mode decomposition (VMD) with adaptive complex mode decomposition (ACMD). Through this joint decomposition, vital sign signals such as heart rate and respiratory rate can be accurately separated in multi-frequency and complex environments, the ability to analyze micro-motion signals is enhanced, and the accuracy of vital sign extraction is significantly improved. Compared with the traditional single decomposition method, this solution improves the separation accuracy of vital sign signals, reduces the problem of modal aliasing, and ensures higher signal quality.
[0030] 2. The present invention combines an extended differentiation and cross-multiply (DACM) algorithm to accurately extract phase information from millimeter-wave radars and dynamically adjust phase modulation to ensure that more subtle micro-motion signals are captured. This technology enhances the signal's anti-noise capability, especially in long distances or complex scenarios, and can ensure accurate signal extraction.
[0031] 3. Short-time Fourier transform (STFT) is introduced to realize real-time time-frequency analysis of millimeter-wave radar micro-motion signals. This method can dynamically track the changes in heart rate signals, provide more accurate heart rate monitoring, and improve the real-time processing capabilities of vital sign signals. STFT provides more accurate analysis in the time and frequency domains, allowing the system to quickly respond to changes in heart rate, which is suitable for real-time health monitoring and medical applications. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] Figure 1 It is a flow chart of the human vital signs perception method based on VMD+ACMD joint decomposition of the present invention. DETAILED DESCRIPTION
[0033] The human vital signs sensing method based on VMD+ACMD joint decomposition of the present invention is described in more detail below through specific implementation methods.
[0034] This embodiment discloses a method for sensing human vital signs based on VMD+ACMD joint decomposition, which is used to solve the problems in the prior art that millimeter wave radar technology usually adopts a single signal decomposition method, it is difficult to effectively separate complex vital sign signals, the signal quality is significantly reduced in an environment with large noise interference, and the extraction accuracy of heart rate and respiratory rate is affected.
[0035] The present invention adopts the following technical solutions to solve the technical problems:
[0036] A human vital sign perception method based on VMD+ACMD joint decomposition includes the following steps:
[0037] Step 1: Data collection: collect data on the human body through millimeter wave radar, and collect and extract micro-motion signals in the chest area of the human body.
[0038] In this embodiment, the human body stability signal collected by the data includes the micro-motion signal caused by the chest rise and fall, and the micro-motion signal includes but is not limited to pulse wave, breathing signal, etc. Millimeter wave radar has the ability of non-contact, long-distance, and penetrating obstacles such as clothing, and is suitable for physical sign detection in various complex environments.
[0039] Step 2: Generate a distance-angle heat map. Perform FFT transformation on the sampling point of each chirp of the echo to generate a distance-angle heat map to determine the distance and angle of the target.
[0040] In this embodiment, FFT is performed on the sampling point of each chirp of the echo, and the number of FFT transformation points N needs to be determined to determine the distance of the target.
[0041] At the same time, for vital sign extraction, only the horizontal direction angle is needed. FFT transformation is used along the horizontal direction. The number of FFT transformation points N needs to be determined to draw a distance angle heat map.
[0042] Step 3: DC offset removal: Remove the DC offset to improve the accuracy of pulse wave signal extraction.
[0043] The DC bias makes the inverse tangent demodulation result different from the phase caused by the actual physiological motion modulation echo signal. By removing the DC bias, the inverse tangent demodulation result is consistent with the phase caused by the actual physiological motion modulation echo signal, thereby improving the accuracy of pulse wave signal extraction.
[0044] In this step, the DC bias calculation formula is:
[0045]
[0046] Where, DC Ⅰ Signal B Ⅰ (t) DC bias part; B Ⅰ(t) is the signal value at sampling point t; DC Q It is signal B Q The DC bias part of B Q (t) is signal B Q The signal value at sampling point t; M is the number of samples to be considered.
[0047] Step 4: Expand DACM phase extraction. When the phase value exceeds the range of -π to π radians and wraps around, phase wraparound occurs. Apply the phase unwrapping algorithm to correct the phase wraparound and obtain the unwrapped phase value.
[0048] The phase component of the received signal carries information about the target's motion or vital signs. In order to extract the phase, a complex range image is usually used. In some cases, the extracted phase may be affected by phase wrapping. Phase wrapping occurs when the phase value exceeds the range of -π to π radians and wraps around. In order to correct the phase wrapping, a phase unwrapping algorithm needs to be applied to obtain the unwrapped phase value. In this embodiment, in order to dynamically adapt to the signal characteristics in different environments, the process formula for extracting the phase information is:
[0049] I(t)=B I (t)-DC I ;
[0050] Q(t)=B Q (t)-DC Q ;
[0051]
[0052] Where I(t) and Q(t) represent the in-phase signal and quadrature signal after DC offset removal, respectively, and I(t)' and Q(t)' are the differential forms of I(t) and Q(t), respectively.
[0053] Step 5: Decompose the signal. After obtaining the radar micro-motion signal, use VMD (variational mode decomposition) to decompose the signal to obtain multiple non-overlapping mode functions, each mode represents a different frequency component. This process can separate different frequency characteristics such as heart rate and breathing rate, effectively reduce noise interference, and retain key signal components.
[0054] In variational mode decomposition, the decomposition mode K and the penalty factor alpha need to be considered to control the decomposition process of the signal. The number of decomposition modes K determines how many modes the signal is decomposed into, and each mode represents a different frequency band. A K value that is too large will increase the computational complexity and may introduce unnecessary noise modes, while a K value that is too small may cause important frequency information to be lost. The penalty factor alpha controls the sparsity of the mode, that is, the width of the modal frequency band. A larger alpha value will make the mode more compact, but may ignore some frequency components; a smaller alpha value allows a wider frequency band, but may introduce noise components.
[0055] Step 6: Signal reconstruction, based on the correlation coefficient R with the original signal Ⅰ Weighted reconstruction is performed to enhance the main frequency components of heart rate and respiratory rate. This process can enhance the main frequency of vital sign signals, highlight the significant features of heart rate and respiratory signals, and improve signal quality. The reconstructed signal is more accurate, especially in noisy environments, and can effectively improve the signal-to-noise ratio.
[0056] In this embodiment, after VMD decomposition, in order to improve the reconstruction accuracy of the signal, the present invention adopts an envelope spectrum peak reconstruction method. This method calculates the envelope spectrum of each decomposed mode and calculates the peak value of the envelope spectrum according to the correlation coefficient R with the original signal. Ⅰ Weighted reconstruction is performed to enhance key vital signs such as heart rate and respiratory rate. The envelope spectrum peak reconstruction method can effectively suppress noise, highlight the main frequency components of the target signal, and improve the quality of the reconstructed signal. This step ensures that the reconstructed signal is more physically meaningful by combining the spectral characteristics of each mode, especially showing strong robustness in high noise environments.
[0057] Step 7: Feature extraction,After signal reconstruction, ACMD is used to further decompose the signal to extract more refined feature signals, especially pulse wave signals.,ACMD can capture subtle changes in vital sign signals and improve the extraction accuracy of heart rate signals by dynamically adjusting the decomposition process.
[0058] In this embodiment, in order to further optimize signal extraction, the present invention combines adaptive complex mode decomposition (ACMD) technology. ACMD is a mode decomposition method based on adaptive adjustment, which can automatically adjust the center frequency and bandwidth of each mode during the decomposition process. By extracting the pulse wave features in the signal through ACMD, the weak vital signs information in the millimeter wave radar signal can be better captured, especially the complex dynamic signals. ACMD effectively enhances the extraction capability of the pulse wave signal and improves the detection accuracy of the heart rate and breathing signals.
[0059] Step 8: Time-frequency analysis. In the extracted micro-motion signal, STFT is used for time-frequency analysis to extract and dynamically track the heart rate signal in real time. STFT divides the signal into multiple time windows and performs Fourier transform in each window to extract the heart rate and respiratory rate from the respiratory signal and pulse wave. By analyzing the frequency components of each time window, the system can track the changes in the heart rate signal in real time and output accurate heart rate values.
[0060] STFT is a commonly used time-frequency analysis method that can simultaneously analyze the changes of a signal in time and frequency by performing a local Fourier transform on the signal. Specifically, STFT performs a Fourier transform on the signal in different time windows to obtain a spectrum of frequency changes over time, which is used to capture the dynamic changes of the heart rate signal.
[0061] DACM phase extraction can dynamically adapt to the signal characteristics in different environments and enhance the anti-interference ability. Variational mode decomposition (VMD) decomposes the signal into multiple non-overlapping frequency bands, has good frequency resolution and noise resistance, and can effectively separate vital signs such as heart rate and breathing. Through envelope spectrum peak reconstruction, the reconstruction quality of the signal is further improved, and the main vital signs are highlighted. Combined with ACMD decomposition, the mode decomposition process is adaptively adjusted to accurately extract weak pulse wave signals. Finally, the short-time Fourier transform (STFT) is used to perform time-frequency analysis on the signal in the time window, and the heart rate signal is extracted and dynamically tracked in real time. The combined application significantly improves the accuracy, robustness and noise resistance of vital sign signal extraction, which is suitable for complex application scenarios such as smart medical care and health monitoring.
[0062] In addition, unless otherwise defined, technical or scientific terms used herein shall have the common meanings understood by one of ordinary skill in the art to which the present invention belongs.
[0063] The exemplary implementation modes of the present invention are described in detail above with reference to the preferred embodiments. The protection scope of the present invention is not limited thereto. It can be understood by those skilled in the art that, without departing from the concept of the present invention, various modifications and variations can be made to the above-mentioned specific embodiments, and various combinations of the various technical features proposed in the present invention can be made without exceeding the protection scope of the present invention.
Claims
1. A human vital sign perception method based on VMD+ACMD joint decomposition, characterized in that: The method comprises the following steps: Step 1: Data collection: using millimeter wave radar to collect data on the human body and extract micro-motion signals of the human chest area; Step 2: Generate the distance-angle heat map. Perform FFT transformation on the sampling point of each chirp of the echo to generate a distance-angle heat map to determine the distance and angle of the target. Step 3, removing the DC bias, removing the DC bias to improve the accuracy of pulse wave signal extraction; Step 4: Expand the DACM phase extraction. When the phase value exceeds the range of -π to π radians and wraps around, phase wrapping occurs. Apply the phase unwrapping algorithm to correct the phase wrapping and obtain the unwrapped phase value. Step 5: Decompose the signal. After obtaining the radar micro-motion signal, use VMD to decompose the signal to obtain multiple non-overlapping modal functions. Step 6: Signal reconstruction: for the decomposed modal signal, according to the correlation coefficient R with the original signal Ⅰ Perform weighted reconstruction to enhance the main frequency components of heart rate and respiratory rate; Step 7: Extract features. After signal reconstruction, further decompose the signal through ACMD to extract more refined feature signals. Step 8: Time-frequency analysis: STFT is used to perform time-frequency analysis on the extracted micro-motion signal to extract and dynamically track the heart rate signal in real time.
2. The human vital signs sensing method based on VMD+ACMD joint decomposition according to claim 1 is characterized by: The human body stability signal collected by the data includes a micro-motion signal caused by chest fluctuation.
3. The human vital sign perception method based on VMD+ACMD joint decomposition according to claim 2 is characterized by: In the distance angle heat map generation step, the distance FFT transformation point number N and the angle FFT transformation point number N are determined according to the distance and angle requirements of the target.
4. The human vital sign perception method based on VMD+ACMD joint decomposition according to claim 3 is characterized by: In the DC offset removal step, the DC offset is removed so that the inverse tangent demodulation result is consistent with the phase caused by the actual physiological motion modulation echo signal. The DC offset calculation formula is: Where, DC Ⅰ Signal B Ⅰ (t) DC bias part; B Ⅰ (t) is the signal value at sampling point t; DC Q It is signal B Q The DC bias part of B Q (t) is signal B Q The signal value at sampling point t; M is the number of samples to be considered.
5. The human vital sign sensing method based on VMD+ACMD joint decomposition according to claim 4 is characterized by: In the extended DACM phase extraction step, a phase unwrapping algorithm is used to correct phase wrapping and obtain an unwrapped phase value to dynamically adapt to signal characteristics in different environments. The process formula for extracting phase information is: I(t)=B I (t)-DC I ; Q(t)=B Q (t)-DC Q ; Where I(t) and Q(t) represent the in-phase signal and quadrature signal after DC offset removal, respectively, and I(t)' and Q(t)' are the differential forms of I(t) and Q(t), respectively.
6. The human vital sign sensing method based on VMD+ACMD joint decomposition according to claim 5 is characterized by: In the signal decomposition step, the decomposition mode K and the penalty factor alpha are used to control the decomposition process of the signal. The decomposition mode number K determines the number of modes into which the signal is decomposed, and each mode represents a different frequency band. The penalty factor alpha controls the sparsity of the modalities.
7. The human vital sign perception method based on VMD+ACMD joint decomposition according to claim 6 is characterized by: In the time-frequency analysis step, STFT divides the signal into multiple time windows, and performs Fourier transform in each window to extract the heart rate and respiratory rate from the respiratory signal and pulse wave.