A radar-based non-contact method for simultaneously extracting human breathing and heartbeat signals

Abstract

The present invention discloses a method for simultaneously extracting human respiration and heartbeat signals using radar without contact. To address this problem, the present invention establishes a model for simultaneously extracting respiration and heartbeat signals based on a characteristic weighted double cosine kernel decomposition method; utilizing the distribution characteristics of respiration and heartbeat signals in the frequency domain, the model weights and parameters are set, and through iterative solution, the frequency estimation values of the respiration and heartbeat signals are output. The present invention proposes a double cosine kernel decomposition method to simultaneously extract the respiration and heartbeat signals of the human body, with high accuracy and high speed, and can reliably and in real time separate the respiration and heartbeat waveforms of the human body, making the present invention highly innovative and competitive. This invention provides a feasible technical innovation solution for radar non-contact monitoring of human heartbeat and respiration status, and can provide convenient near-medical-grade physiological signal monitoring for non-contact monitoring during epidemics, health monitoring of special populations, sleep quality analysis, and real-time fatigue status of drivers.

Description

Technical Field

[0001] The present invention relates to a new generation of information technology, and is a radar non-contact method for simultaneously extracting human breathing and heartbeat signals. Background Art

[0002] Currently, the primary technology for monitoring vital signs like respiration and heartbeat relies on direct contact. Direct contact testing methods include electrocardiograms, finger-clip pulse oximeters, and electronic blood pressure monitors. However, these devices are expensive and often medical devices, making them difficult for non-professionals to operate. Furthermore, they require contact with the human body, making the test inconvenient.

[0003] Radar offers significant advantages for detecting breathing and heartbeats. First, radar has strong signal penetration, allowing it to be concealed in the surrounding environment for detection without contacting the human body. Second, radar has a wider detection range and is less susceptible to external environmental conditions, resulting in greater stability. Furthermore, radar can perform long-term detection, and utilizing this long-term detection data is very helpful in assessing human health. Contactless technology can continuously monitor a user's health, enabling timely warnings in emergencies such as fainting and cardiac arrest, reducing the risk of death or disability and thus alleviating the burden of medical care on society. Radar's advantages in detecting breathing and heartbeats have attracted widespread attention.

[0004] Although radar has great advantages in detecting respiration and heartbeat, the respiration and heartbeat signals are relatively weak. Under external interference, the accuracy of respiration and heartbeat separation directly affects the estimation effect. Therefore, an accurate respiration and heartbeat signal separation method is very important. Summary of the Invention

[0005] In view of this, the main purpose of the present invention is to propose a radar-based non-contact method for simultaneous extraction of human breathing and heartbeat signals, establish a feature-weighted double cosine kernel decomposition model, and use the frequency domain characteristics of breathing and heartbeat signals to solve the model, so as to achieve simultaneous and accurate separation of human breathing and heartbeat signals.

[0006] To achieve the above object, the method of the present invention comprises the following steps:

[0007] A radar non-contact method for simultaneously extracting human breathing and heartbeat signals, characterized by comprising the following steps:

[0008] Step 1: Based on the phase data obtained by radar measurement of the human body, a feature weighted double cosine kernel decomposition model is proposed:

[0009]

[0010] Where ⊙ represents element-wise multiplication, c1 and c2 are the frequency domain sparse estimates of heartbeat and respiration, w1 and w2 are the feature weights of c1 and c2, and y is the radar phase signal. is the IDCT (Inverse Discrete Cosine Transform) matrix; for the vector

[0011] Step 2: Initialize the model parameters according to the data length L of the radar phase signal y, c i is a real vector of length N, that is w i is a real vector of length N, that is is a complex matrix with L rows and N columns, that is D i Is an N-row L-column complex DCT matrix, that is Where N is L rounded up to the power of 2, i = 1, 2;

[0012] Step 3: Set the value of vector w1 to 100 and the value of vector w2 to 1000. According to the relationship between the sampling rate and the number of sampling points, set the value of w1 in the heartbeat range of 0.8-2Hz to 1, and the value of w2 in the breathing range of 0.1-0.5Hz to 1;

[0013] Step 4: Initialize the variables used to solve the model and set μ1=100,μ2=1,d i is an all-zero vector of length N, that is c i is the DCT value of y, i=1, 2; set the number of iterations n=30, and the threshold function thld(x, T)=sign(x)max(0,|x|-T);

[0014] Step 5: Set the parameter c i Enter the thld function and cut c i The influence of irrelevant coefficients in the i Subtract d i Get the intermediate variable v i , which can be expressed as follows:

[0015] v i ←thld(c i +d i , T i )-d i ,i=1,2

[0016] Step 6: Calculate the deviation k between the current time domain estimated value and the actual value y, which can be expressed as follows:

[0017]

[0018] Step 7: Perform DCT transformation on the deviation k to obtain the coefficient d in the frequency domain i , which can be expressed as follows:

[0019] d i ←C i D i k,i=1,2

[0020] Step 8: The coefficient v i With coefficient d i Add together to get the parameter c to be solved i , and adjust the value of n, which can be expressed as follows:

[0021] c i ←d i +v i ,i=1,2

[0022] n←n-1

[0023] Step 9: Determine whether n is zero. If n≠0, go to step 5. If n=0, terminate the iteration and obtain the heartbeat and respiratory frequency domain coefficient c. i , where c1 is the heartbeat frequency domain coefficient, c2 is the respiratory frequency domain coefficient, i IDCT can be used to obtain the time domain waveforms of heartbeat and respiratory signals.

[0024] The beneficial effects of this invention lie in that the proposed decomposition method enables simultaneous extraction of respiratory and heartbeat signals with high separation accuracy and speed, enabling real-time monitoring of respiratory and heartbeat signals, making the invention highly innovative and competitive. Further research on this research could facilitate the monitoring and analysis of human health conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 Schematic diagram of the workflow of a method for simultaneously extracting a radar breathing signal and a heartbeat signal in an embodiment of the present invention;

[0026] Figure 2 The spectrum of normal human breathing and heartbeat signals separated by a method for simultaneously extracting radar breathing signals and heartbeat signals in an embodiment of the present invention and the spectrum of breathing and heartbeat signals measured by medical reference equipment;

[0027] Figure 3 The time domain waveforms of normal human breathing and heartbeat signals separated by a method for simultaneously extracting radar breathing signals and heartbeat signals in an embodiment of the present invention;

[0028] Figure 4 The following is a comparison of the measurement results of the present invention and the medical-grade contact sensor in the normal human breathing scenario in the embodiment of the present invention;

[0029] Figure 5 The following is a comparison of the measurement results of the present invention and a medical-grade contact sensor in a scenario where human respiratory arrest occurs in an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0031] A method for simultaneously extracting radar breathing signals and heartbeat signals. The entire process is as follows: Figure 1 As shown in the figure, the steps for extracting the respiratory signal and the heartbeat signal simultaneously are as follows:

[0032] A radar non-contact method for simultaneously extracting human breathing and heartbeat signals, characterized by comprising the following steps:

[0033] Step 1: Based on the phase data obtained by radar measurement of the human body, a feature weighted double cosine kernel decomposition model is proposed:

[0034]

[0035] Where ⊙ represents element-wise multiplication, c1 and c2 are the frequency domain sparse estimates of heartbeat and respiration, w1 and w2 are the feature weights of c1 and c2, and y is the radar phase signal. is the IDCT (Inverse Discrete Cosine Transform) matrix; for the vector

[0036] Step 2: Initialize the model parameters according to the data length L of the radar phase signal y, c i is a real vector of length N, that is w i is a real vector of length N, that is is a complex matrix with L rows and N columns, that is D i Is an N-row L-column complex DCT matrix, that is Where N is L rounded up to the power of 2, i = 1, 2;

[0037] Step 3: Set the value of vector w1 to 100 and the value of vector w2 to 1000. According to the relationship between the sampling rate and the number of sampling points, set the value of w1 in the heartbeat range of 0.8-2Hz to 1, and the value of w2 in the breathing range of 0.1-0.5Hz to 1;

[0038] Step 4: Initialize the variables used to solve the model and set μ1=100,μ2=1,d i is an all-zero vector of length N, that is c i is the DCT value of y, i=1, 2; set the number of iterations n=30, and the threshold function thld(x, T)=sign(x)max(0,|x|-T);

[0039] Step 5: Set the parameter c i Enter the thld function and cut c i The influence of irrelevant coefficients in the i Subtract d i Get the intermediate variable v i , which can be expressed as follows:

[0040] v i ←thld(c i +d i , T i )-d i , i=1,2

[0041] Step 6: Calculate the deviation k between the current time domain estimated value and the actual value y, which can be expressed as follows:

[0042]

[0043] Step 7: Perform DCT transformation on the deviation k to obtain the coefficient d in the frequency domain i , which can be expressed as follows:

[0044] d i ←C i D i k, i = 1, 2

[0045] Step 8: The coefficient v i With coefficient d i Add together to get the parameter c to be solved i , and adjust the value of n, which can be expressed as follows:

[0046] c i ←d i +v i , i=1,2

[0047] n←n-1

[0048] Step 9: Determine whether n is zero. If n≠0, go to step 5. If n=0, terminate the iteration and obtain the heartbeat and respiratory frequency domain coefficient c. i , where c1 is the heartbeat frequency domain coefficient, c2 is the respiratory frequency domain coefficient, iIDCT can be used to obtain the time domain waveforms of heartbeat and respiratory signals.

[0049] In a further embodiment, the present invention uses radar phase signals to separate the breathing and heartbeat signals of a normal human body, and the separation results are as follows: Figure 2 As shown, Figure 2 (a) The figure shows the respiratory frequency domain after separation. Figure 2 (b) is the frequency domain of the medical respiratory sensor detection signal, Figure 2 (c) The figure shows the heartbeat frequency domain after separation. Figure 2 (d) is the frequency domain of the detection signal of medical ECG equipment.

[0050] Figure 3 Yes Figure 2 The time domain waveform change diagram of the respiratory and heartbeat signals obtained after IDCT transformation of the frequency domain signal, Figure 3 (a) is the waveform of the respiratory signal. Figure 3 (b) The figure shows the waveform of the heartbeat signal.

[0051] Figure 4 For comparison of the measurement results of the method of the present invention and the medical contact sensor under normal breathing conditions, Figure 4 (a) The solid line in the figure is the medical respiratory sensor signal, and the dotted line is the respiratory signal separated by the algorithm. Figure 4 (b) The solid line in the figure is the signal measured by the medical ECG device, and the dotted line is the heartbeat signal separated by the algorithm. The decomposed signal obtained using the method of the present invention is highly consistent with the actual human respiratory and heartbeat signals, and the separation effect is accurate.

[0052] In a further embodiment, the present invention uses radar phase signals to separate the respiratory and heartbeat signals of the human body in the scene of respiratory arrest, and the separation results are as follows: Figure 5 As shown, Figure 5 (a) The solid line is the medical respiratory sensor signal, and the dotted line is the respiratory signal separated by the present invention. Figure 4 (b) The solid line in the figure is the waveform measured by medical ECG contact, and the dotted line is the heartbeat signal separated by the present invention.

[0053] Depend on Figure 4 and Figure 5 Comparison of the measurement results of the medical contact sensor in the two scenarios shows that the present invention can separate the respiratory and heartbeat signals of the target in different scenarios, and the measurement results are close to those of the medical contact sensor.

[0054] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A radar non-contact method for simultaneously extracting human breathing and heartbeat signals, characterized in that: The following steps are involved: Step 1: Based on the phase data obtained by radar measurement of the human body, a feature weighted double cosine kernel decomposition model is proposed: Where ⊙ represents element-wise multiplication, c1 and c2 are the frequency domain sparse estimates of heartbeat and respiration, w1 and w2 are the feature weights of c1 and c2, and y is the radar phase signal. is the inverse discrete cosine transform IDCT matrix; for the vector Step 2: Initialize the model parameters according to the data length L of the radar phase signal y, c i is a real vector of length N, that is w i is a real vector of length N, that is is a complex matrix with L rows and N columns, that is D i Is an N-row L-column complex DCT matrix, that is Where N is L rounded up to the power of 2, i = 1, 2; Step 3: Set the value of vector w1 to 100 and the value of vector w2 to 1000. According to the relationship between the sampling rate and the number of sampling points, set the value of w1 in the heartbeat range of 0.8-2Hz to 1, and the value of w2 in the breathing range of 0.1-0.5Hz to 1; Step 4: Initialize the variables used to solve the model and set μ1=100,μ2=1,d i is an all-zero vector of length N, that is c i is the DCT value of y, i=1, 2; set the number of iterations n=30, and the threshold function thld(x, T)=sign(x)max(0,|x|-T); Step 5: Set the parameter c i Enter the thld function and cut c i The influence of irrelevant coefficients in the i Subtract d i Get the intermediate variable v i , which can be expressed as follows: v i ←thld(c i +d i ,T i )-d i ,i=1,2 Step 6: Calculate the deviation k between the current time domain estimated value and the actual value y, which can be expressed as follows: Step 7: Perform DCT transformation on the deviation k to obtain the coefficient d in the frequency domain i , which can be expressed as follows: d i ←C i D i k,i=1,2 Step 8: The coefficient v i With coefficient d i Add together to get the parameter c to be solved i , and adjust the value of n, which can be expressed as follows: c i ←d i +v i ,i=1,2 n←n-1 Step 9: Determine whether n is zero. If n≠0, go to step 5. If n=0, terminate the iteration and obtain the heartbeat and respiratory frequency domain coefficient c. i , where c1 is the heartbeat frequency domain coefficient, c2 is the respiratory frequency domain coefficient, i IDCT can be used to obtain the time domain waveforms of heartbeat and respiratory signals.

Images