Method and apparatus for extracting physiological information from radar signal
Patent Information
- Application Number
- PCT/CN2025/085725
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2025-03-28
- Publication Date
- 2026-10-01
Smart Images
Figure CN2025085725_01102026_PF_FP_ABST
Abstract
Description
A method and apparatus for extracting physiological information from radar signals Technical Field
[0001] This application relates to the field of living organism detection. Background Technology
[0002] Breathing and heartbeat are essential physiological activities for maintaining normal bodily functions and are vital signs of the human body. Monitoring breathing and heartbeat helps in diagnosing diseases and understanding a person's health status. Common breathing and heartbeat detection devices include professional medical equipment such as electrocardiogram (ECG) monitors and stethoscopes. In daily life, wearable devices such as smartwatches and smart bracelets can also be used to monitor changes in a person's physical indicators at all times. However, wearable devices have problems such as low wearing comfort and frequent charging.
[0003] Radar-based physiological detection is a non-contact detection method that can continuously monitor the respiratory and heart rate of the subject without requiring them to wear any devices. This method has a good user experience, high acceptance, and broad application prospects.
[0004] It should be noted that the above introduction to the technical background is only for the purpose of providing a clear and complete explanation of the technical solutions of this application and facilitating understanding by those skilled in the art. It should not be assumed that these technical solutions are known to those skilled in the art simply because they have been described in the background section of this application. Summary of the Invention
[0005] The inventors discovered that existing radar-based physiological detection methods mainly detect breathing and heart rate by detecting the displacement of the body surface caused by human breathing or heartbeat. However, because radar signals are very sensitive, they may be affected by various noises during physiological detection, affecting their detection accuracy. For example, when the body shakes, it may affect the physiological signals of breathing and heartbeat. In addition, the displacement of the body surface caused by heartbeat and the displacement of the body surface caused by breathing will also affect each other, affecting the accuracy of physiological detection.
[0006] To address at least one of the aforementioned problems or other similar issues, embodiments of this application provide a method and apparatus for extracting physiological information from radar signals. This method detects physiological activities such as heartbeat and respiration by measuring vibrations of the body surface caused by physiological activities. Since the vibration frequencies of the body surface caused by physiological activities such as respiration and heartbeat are relatively high, typically greater than 4Hz, while the frequency range of respiration is approximately 0.2–0.4Hz and the frequency range of heartbeat is approximately 0.8–2Hz, measuring the higher-frequency vibrations of the body surface can effectively filter out low-frequency noise. Furthermore, by processing the physiological signals extracted based on radar signals and the convolution kernel, thereby optimizing the convolution kernel, the accuracy of physiological information detection based on radar signals can be improved.
[0007] According to one aspect of the embodiments of this application, an apparatus for extracting physiological information from radar signals is provided, the apparatus comprising:
[0008] The extraction unit uses a convolution kernel to process the radar signal of the target in the first time period to obtain the first physiological signal;
[0009] The first processing unit processes the first physiological signal to obtain a first optimized parameter characterizing the energy features of the first physiological signal.
[0010] The second processing unit processes the convolution kernel to obtain a second optimized parameter characterizing the energy features of the convolution kernel;
[0011] A parameter optimization unit optimizes the convolution kernel based on at least one of the at least one first optimization parameter and at least one second optimization parameter to obtain an optimized convolution kernel;
[0012] The output unit processes the radar signal within the first time period using an optimized convolution kernel to obtain a first output result, which is then used as the physiological information of the target being measured.
[0013] According to another aspect of the embodiments of this application, a method for extracting physiological information from radar signals is provided, the method comprising:
[0014] The radar signal of the target under test in the first time period is processed using a convolution kernel to obtain the first physiological signal;
[0015] The first physiological signal is processed to obtain a first optimized parameter characterizing the energy properties of the first physiological signal;
[0016] The convolution kernel is processed to obtain a second optimized parameter characterizing the energy properties of the convolution kernel;
[0017] The convolution kernel is optimized according to at least one of the at least one first optimization parameter and at least one second optimization parameter to obtain an optimized convolution kernel;
[0018] The optimized convolutional kernel is used to process the radar signal within the first time period to obtain a first output result, which is then used as the physiological information of the target being measured.
[0019] According to another aspect of the embodiments of this application, a computer device is provided, the computer device including a memory and a processor, the memory storing a computer program, and the processor being configured to execute the computer program to implement the method as described above.
[0020] According to another aspect of the embodiments of this application, a storage medium storing a computer-readable program is provided, the computer-readable program causing a computer to perform the method described above.
[0021] One of the beneficial effects of this application embodiment is that, according to this application embodiment, physiological activities such as heartbeat and breathing are detected by measuring the vibration of the body surface caused by physiological activities. Since the vibration frequency of the body surface caused by physiological activities such as breathing and heartbeat is greater than its own frequency, by detecting the higher frequency body surface vibration, low-frequency noise can be effectively filtered out, and the accuracy of physiological information detection based on radar signals can be improved.
[0022] Specific embodiments of this application are disclosed in detail with reference to the following description and accompanying drawings, indicating how the principles of this application can be adopted. It should be understood that the embodiments of this application are not limited in scope. Within the scope of the appended claims, embodiments of this application include many changes, modifications, and equivalents.
[0023] Features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, combined with features in other embodiments, or substituted for features in other embodiments.
[0024] It should be emphasized that the term "including / comprises" as used herein refers to the presence of a feature, whole, step, or component, but does not exclude the presence or addition of one or more other features, wholes, steps, or components. Attached Figure Description
[0025] The elements and features described in one drawing or embodiment of this application may be combined with elements and features shown in one or more other drawings or embodiments. Furthermore, in the drawings, similar reference numerals denote corresponding parts in several drawings and can be used to indicate corresponding parts used in more than one embodiment.
[0026] The accompanying drawings, which form part of the specification, are used to provide a further understanding of the embodiments of this application and illustrate the implementation methods of this application, together with the textual description, to explain the principles of this application. Obviously, the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any creative effort. In the drawings:
[0027] Figure 1 is a schematic diagram of a method for extracting physiological signals from radar signals according to an embodiment of this application;
[0028] Figure 2 is a schematic diagram of the target under test and the radar according to an embodiment of this application;
[0029] Figure 3 is a schematic diagram comparing the IBI results obtained by the method of the present application for detecting the target under test with the actual IBI results;
[0030] Figure 4 is a schematic diagram of a device for extracting physiological information from radar signals according to an embodiment of this application;
[0031] Figure 5 is a schematic diagram of a computer device according to an embodiment of this application. Detailed Implementation
[0032] Referring to the accompanying drawings, the foregoing and other features of this application will become apparent from the following description. Specific embodiments of this application are specifically disclosed in the description and drawings, illustrating partial implementations in which the principles of this application may be employed. It should be understood that this application is not limited to the described embodiments; rather, it includes all modifications, variations, and equivalents falling within the scope of the appended claims.
[0033] In the embodiments of this application, the terms "first," "second," etc., are used to distinguish different elements by name, but do not indicate the spatial arrangement or chronological order of these elements, and these elements should not be limited by these terms. The term "and / or" includes any one or more of the terms listed in association and all combinations thereof. The terms "comprising," "including," "having," etc., refer to the presence of the stated features, elements, components, or assemblies, but do not exclude the presence or addition of one or more other features, elements, components, or assemblies.
[0034] In the embodiments of this application, the singular forms "a," "the," etc., including the plural forms, should be broadly understood as "a kind" or "a class" rather than limited to the meaning of "an." Furthermore, the term "the" should be understood to include both the singular and plural forms, unless the context explicitly indicates otherwise. Additionally, the term "according to" should be understood as "at least partially based on…," and the term "based on" should be understood as "at least partially based on…," unless the context explicitly indicates otherwise.
[0035] The various embodiments of this application will now be described with reference to the accompanying drawings.
[0036] First aspect of the embodiments
[0037] This application provides a method for extracting physiological information from radar signals.
[0038] Figure 1 is a schematic diagram of a method for extracting physiological information from radar signals according to an embodiment of this application. As shown in Figure 1, the method includes:
[0039] 110: The radar signal of the target under test in the first time period is processed using a convolution kernel to obtain the first physiological signal;
[0040] 120: Process the first physiological signal to obtain a first optimized parameter characterizing the energy properties of the first physiological signal;
[0041] 130: Process the convolution kernel to obtain a second optimized parameter characterizing the energy properties of the convolution kernel;
[0042] 140: Optimize the convolution kernel according to at least one of the at least one first optimization parameter and at least one second optimization parameter to obtain an optimized convolution kernel;
[0043] 150: The optimized convolutional kernel is used to process the radar signal in the first time period to obtain a first output result, and the first output result is used as the physiological information of the target being tested.
[0044] It is worth noting that Figure 1 above is only a schematic illustration of an embodiment of this application, but this application is not limited thereto. For example, other operations may be added or some operations may be removed. Those skilled in the art can make appropriate modifications based on the above content, and are not limited to the description in Figure 1 above.
[0045] According to the embodiments of this application, by measuring the vibration of the body surface at a higher frequency, low-frequency noise can be effectively filtered out. Furthermore, by processing the physiological signals extracted based on radar signals and the convolution kernel, thereby optimizing the convolution kernel, the accuracy of physiological information detection based on radar signals can be improved.
[0046] In this embodiment of the application, the radar periodically transmits wireless signals to its surrounding environment when it is working, and then receives and processes the wireless signals reflected by the target being measured, thereby enabling it to perceive the target being measured in the environment.
[0047] In this application embodiment, the type of radar is not limited. The radar can be a Doppler radar, a Frequency Modulated Continuous Wave (FMCW) radar, a Stepped Frequency Continuous Wave (SFCW) radar, etc.
[0048] In this embodiment of the application, the signal received by the radar includes not only the signal reflected from the target being tested, but also the signal reflected from other objects in the environment. In order to improve the detection accuracy and avoid the influence of the signal reflected from other objects in the environment on the detection result, in this embodiment of the application, the signal received by the radar (the original radar signal) can be processed to filter out the signal reflected from other objects in the environment, improve the signal-to-noise ratio of the radar signal reflected by the human target, and obtain better detection performance.
[0049] Taking FMCW radar as an example, after obtaining the original radar signal, FMCW radar performs a Fast Fourier Transform (FFT) operation on the original radar signal to obtain radar signals reflected from different distances. It then performs an angle FFT operation on the radar signal, thereby obtaining radar signals reflected from different distances and angles. Compared with the original radar signal (which includes both signals reflected from the target and signals reflected from other objects in the environment), using only the radar signals reflected from the distance and angle of the target can achieve better detection results.
[0050] In the above embodiments, the signal processing method may further include time-domain filtering, frequency-domain filtering, Doppler processing, etc. The signal processing method can be implemented by other devices or can be directly processed by radar. For details, please refer to the relevant technologies. This application does not impose any restrictions.
[0051] In the embodiments of this application, the target to be measured can be a person. Further, the target to be measured can be a specific person, or all people in the environment surrounding the radar, or a specific group of people. For example, the radar can detect all elderly people within its detection range, or detect all disabled people within its detection range. For the identification of a specific group of people or a specific person, they can be marked in advance, or determined in advance through other detection methods. For specific details, please refer to relevant technologies. This application does not impose any restrictions.
[0052] In this embodiment of the application, during operation 110, radar signals reflected from the target under test over a period of time can be processed. This period of time is at least longer than the cycle of a physiological signal, for example, at least longer than the cycle of one breath and / or one heartbeat. This allows for the complete extraction of physiological information from the radar signal, representing one breath and / or heartbeat, facilitating subsequent analysis of the target under test. The radar signal here can be the radar signal after the aforementioned signal processing, or, in other words, the radar signal after filtering out signals reflected from other objects in the environment.
[0053] The radar signal can be expressed by the following formula: S={s t ,1≤t≤NS}
[0054] Where, N S N is the duration of the radar signal. S At least longer than the cycle of a physiological signal, s t This represents the radar signal reflected by the target at time t. The radar signal at time t can be a complex signal. Taking a complex signal as an example, for s... t It can be calculated using the following formula: s t =P t expjφ t
[0055] Among them, P t and φ t These represent the amplitude and phase of the radar signal, respectively. In this embodiment, the aforementioned complex signal s can be used. t The aforementioned radar signal (complex signal s) can also be used as input to extract physiological information. t The amplitude or phase of the radar signal is used as input to extract physiological information. This application does not impose any limitations. In the following embodiments, the embodiments of this application are illustrated by taking the input radar signal as a complex signal.
[0056] In operation 110, in some embodiments, a convolution kernel is used to perform a convolution operation on the radar signal of the target being measured within a first time period to obtain a first convolution result. The amplitude of the first convolution result at each moment within the first time period is calculated to obtain a first physiological signal.
[0057] The first time period is at least longer than the period of a physiological signal. When extracting the physiological signal from the radar signal of the target within the first time period, a convolution operation is first performed on the input radar signal. The convolution kernel can be represented by the following formula: G={g j ,-(N G -1) / 2≤j≤(N G -1) / 2}
[0058] Where, N G Let G represent the length of the convolution kernel. The input radar signal is convolved using kernel G to obtain the convolved signal, i.e., the first convolution result. This first convolution result can be expressed as: C = {c t ,1≤t≤N S}
[0059] Where, N S c represents the length of the input radar signal. t c is the result of convolving the radar signal at time t. t The following formula can be used to calculate: c t =(S*G)t =∑ 1≤t≤k s t g k-t
[0060] In the example above, when the input radar signal S is a complex number, the output result c after convolution is... t It can also be a complex number, or, when the parameters of the convolution kernel are complex, the output result c. t It is also a plural number.
[0061] In the above embodiment, the amplitude A of the signal at each time point in the first convolution result is calculated, and this amplitude A is used as the physiological signal extracted from the radar signal of the target in the first time period, that is, the first physiological signal, which can be expressed as: A={a t ,1≤t≤N S}
[0062] Among them, a t a represents the amplitude of the radar signal convolution result at time t. t =|c t That is, at time t, c can be calculated. t The modulus is used to calculate the physiological signal corresponding to the radar signal at time t.
[0063] In this embodiment of the application, after obtaining the first physiological signal of the target within a first time period, the first physiological signal can be further processed to obtain a first optimized parameter characterizing the energy characteristics of the first physiological signal. The calculation of the first optimized parameter is illustrated below.
[0064] In some embodiments, during operation 120, an FFT operation is performed on the first physiological signal to obtain a first energy value, and a first optimization parameter is obtained based on the first energy value. The first energy value represents the energy of the first physiological signal at various frequencies within a first frequency range, where the first frequency range is the frequency range covered by the first physiological signal.
[0065] Taking the amplitude A of the first physiological signal obtained through the above convolution as an example, in the above embodiment, the first physiological signal A = {a t ,1≤t≤N S Performing an FFT operation yields the energy of the first physiological signal at various frequencies, i.e., the first energy value. The first energy value can be expressed by the following formula:
[0066] in, This represents the energy of the first physiological signal A at frequency f. This indicates the first frequency range, which is the frequency range covered by the first physiological signal A.
[0067] In the above embodiments, in some possible implementations, obtaining the first optimization parameter based on the first energy value may include:
[0068] The first optimization parameter is obtained by calculating the ratio of the peak energy of the first energy value within the second frequency range to the total energy of the first energy value within the first frequency range. Here, the second frequency range refers to the frequency range within the first frequency range that represents human respiration or heartbeat.
[0069] In the above embodiments, the first optimization parameter can be calculated using the following formula:
[0070] in, This indicates the first energy value within the second frequency range. The peak energy within, that is, The first optimization parameter r0 is determined by... The ratio of the first energy value to the total energy within the first frequency range is determined. It can be seen that the first optimized parameter characterizes the energy characteristics of the first physiological signal. In addition, the first optimized parameter also characterizes the spectral characteristics of the first physiological signal.
[0071] In some embodiments, the second frequency range In the frequency range of the first physiological signal The frequency range of human breathing or heartbeat, that is, the frequency range of human physiological activities.
[0072] For example, the frequency range of the first physiological signal The intersection of this frequency range with the frequency range representing human respiration or heartbeat is used as the second frequency range. For example, if the frequency range of the first physiological signal is 0 to 20 Hz, the frequency range of human respiration is 0.2 to 0.4 Hz, and the frequency range of human heartbeat is 0.8 to 2 Hz, then when using the method of this application embodiment to detect human respiration, the second frequency range can be 0.2 to 0.4 Hz, and when using the method of this application embodiment to detect human heartbeat, the second frequency range can be 0.8 to 2 Hz. The frequency range representing human respiration or heartbeat can be preset.
[0073] In the above embodiments, in some other possible implementations, obtaining the first optimization parameter based on the first energy value may further include:
[0074] The first optimization parameter is obtained by calculating the ratio of the peak energy of the first energy value in the second frequency range to the total energy of the first energy value in the third frequency range. Here, the second frequency range is the frequency range within the first frequency range representing human respiration or heartbeat, and the third frequency range is the frequency range within the first frequency range that is less than or equal to the second frequency range.
[0075] In the above embodiments, the first optimization parameter can be calculated using the following formula:
[0076] That is, the first optimization parameter r1 is determined by the peak energy of the first energy value within the second frequency range. With the first energy value in the third frequency range Total energy within The ratio is determined, where, This indicates the first energy value within the second frequency range. Peak energy within, for peak energy Second frequency range Please refer to the above explanation of the first optimization parameter r0, which will not be repeated here.
[0077] In the above embodiments, the third frequency range is the first frequency range. Less than or equal to the second frequency range frequency range For example, the frequency range of the first physiological signal is 0 to 20 Hz, the second frequency range representing human respiration is 0.2 to 0.4 Hz, the second frequency range representing human heartbeat is 0.8 to 2 Hz, and when the method of the embodiment of this application is used to detect human respiration, the third frequency range can be 0 to 0.4 Hz, and when the method of the embodiment of this application is used to detect human heartbeat, the third frequency range can be 0 to 2 Hz.
[0078] In the above embodiments, by calculating the first optimization parameter based on the first physiological signal, the convolution kernel G used in the physiological signal extraction process can be optimized using the first optimization parameter to improve the accuracy of radar physiological detection. In some examples, when optimizing the convolution kernel G, the first optimization parameter r0 and the first optimization parameter r1 can be used, or any one of the two first optimization parameters can be used. This application does not impose any restrictions.
[0079] In the embodiments of this application, in addition to optimizing the convolution kernel according to the first optimization parameter, in some examples, in order to further improve the accuracy of radar physiological detection, the convolution kernel itself can also be processed to obtain the second optimization parameter, and the convolution kernel can be further optimized using the second optimization parameter. The calculation of the second optimization parameter is illustrated below.
[0080] In some embodiments, calculating the second optimization parameter includes:
[0081] Perform an FFT operation on the convolution kernel to obtain a second energy value. Calculate the ratio of the total energy of the second energy value in the fifth frequency range to the total energy of the convolution kernel in the fourth frequency range to obtain the second optimized parameters. Here, the second energy value represents the energy of the convolution kernel at each frequency in the fourth frequency range; the fourth frequency range is the frequency range covered by the convolution kernel; and the fifth frequency range is a pre-set frequency range that includes the frequency range of vibrations on the body surface caused by human physiological activities.
[0082] In the above embodiment, the convolution kernel is the convolution kernel G used in the physiological signal extraction operation. An FFT operation is performed on the convolution kernel G to obtain the energy of the convolution kernel G at each frequency within its covered frequency range, that is, the second energy value. The second energy value can be expressed by the following formula:
[0083] in, Let G be the energy of the convolution kernel at frequency f. This represents the frequency range covered by the convolution kernel G, that is, the fourth frequency range, and the second energy value E. G The spectral characteristics of the convolution kernel G were characterized based on the second energy value E. G The second optimization parameter is calculated using the following formula:
[0084] in, This indicates that the convolution kernel G is in the fourth frequency range. The total energy within, that is, the second energy value E G The sum of energy at all frequencies within its frequency range. Indicates the second energy value E G In the fifth frequency range That is, the convolution kernel G is in the fifth frequency range Total energy within.
[0085] In the above embodiments, the fifth frequency range represents the frequency range that includes vibrations of the body surface caused by human physiological activities, such as heartbeat and / or breathing.
[0086] In the above embodiments, the frequency range of body surface vibrations caused by heartbeat and / or breathing can be determined by prior knowledge or by experimental measurement. For example, the frequency range of body surface vibrations caused by heartbeat can be preset to 8 to 20 Hz. That is, when the method of the present application embodiment is used to detect heartbeat, the fifth frequency range can be preset to 8 to 20 Hz.
[0087] According to the above embodiments, a second optimization parameter is calculated, and the convolution kernel is optimized according to the second optimization parameter, which can further improve the accuracy of physiological information detection based on radar signals.
[0088] In this application embodiment, there is no restriction on the calculation order of the first optimization parameter and the second optimization parameter. The first optimization parameter can be calculated first, or the second optimization parameter can be calculated first. In addition, there is no restriction on the calculation order of the first optimization parameters r0 and r1. The optimization parameters can be arbitrarily combined in their calculation order as needed. This application does not impose any restrictions.
[0089] In the embodiments of this application, the convolution kernel can be optimized using only the first optimization parameter (e.g., the aforementioned r0 and / or r1), or only the second optimization parameter (e.g., the aforementioned r2), or both the first and second optimization parameters can be used simultaneously. The optimization of the convolution kernel will be described exemplarily below.
[0090] In some embodiments, in operation 140, optimizing the convolution kernel using the first optimization parameter and the second optimization parameter may involve weighting at least one of the first optimization parameter and at least one of the second optimization parameters to obtain an optimization target, modifying the parameters in the convolution kernel to maximize the optimization target, and obtaining the optimized convolution kernel when the optimization target is maximized.
[0091] The optimization objective can be obtained based on only the first optimization parameter and the second optimization parameter, or based on only one of the first optimization parameter and the second optimization parameter. The first optimization parameter may include the first optimization parameter r0 and the first optimization parameter r1, or one of the first optimization parameter r0 and the first optimization parameter r1. Furthermore, this application does not limit the number of the first optimization parameter and the second optimization parameter, and multiple optimization parameters can be selected for weighted processing.
[0092] For example, using the first optimization parameter r0, the first optimization parameter r1, and the second optimization parameter r2 for weighted processing, the weighted processing process can be expressed by the following formula: L = a0r0 + a1r1 + a2r2
[0093] Where a0, a1, and a2 are all non-negative weighting coefficients, that is, the first optimization parameter and the second optimization parameter are weighted to obtain the optimization objective L. Based on the optimization objective L, the parameters of the convolution kernel are modified to maximize the result of the optimization objective L, thereby obtaining the optimal convolution kernel parameters, that is, the optimized convolution kernel.
[0094] In the above embodiments, by modifying the parameters in the convolution kernel, both the first and second optimization parameters will change, and the optimization objective L will also change accordingly. By continuously modifying the parameters in the convolution kernel, the optimization objective L is maximized. At this point, the parameters in the convolution kernel are saved, and the convolution kernel optimization is completed, resulting in an optimized convolution kernel. The parameters corresponding to this convolution kernel can maximize the optimization objective L. Taking the above formula as an example, modifying the parameters in the convolution kernel G changes the first optimization parameters r0 and r1 and the second optimization parameter r2, thereby changing the optimization objective L. Thus, by continuously modifying the parameters in the convolution kernel G, the optimization objective L is maximized, completing the optimization of the convolution kernel.
[0095] The maximum value of the optimization objective can be a pre-set number. For example, if the optimization objective is 10, the parameters in the convolution kernel can be modified to make the optimization objective 10. Alternatively, the first optimization parameter and the second optimization parameter can be modified to make the optimization objective reach the minimum value to complete the optimization of the convolution kernel. This application does not restrict the setting of the optimization objective. For details, please refer to relevant technologies.
[0096] In the above embodiments, this application does not limit the optimization method of the convolution kernel. For example, batch gradient descent, stochastic gradient descent, or genetic algorithm can be used to modify the parameters in the convolution kernel, thereby optimizing the convolution kernel.
[0097] In the above embodiments, the same first optimization parameters and / or second optimization parameters are used to optimize the convolution kernel for radar signals within the first time period. This application is not limited to this. For radar signals within the first time period, different optimization parameters can be set at different time nodes according to the changes in the target position or movement within the first time period. For example, in time period a within the first time period, the first optimization parameters in time period a are used to optimize the convolution kernel; in time period b within the first time period, the first optimization parameters and second optimization parameters in time period b are used to optimize the convolution kernel. Thus, different optimization parameters can be set to optimize the convolution kernel according to the different positions or movements of the target, resulting in more accurate physiological detection results.
[0098] In the above embodiments, the target being measured can be all people. In some examples, the convolution kernel is optimized using radar signals reflected from all people within the radar detection range to obtain an optimized convolution kernel. Further, physiological signals of specific groups or individuals within this population are detected, and the optimized convolution kernel is further optimized using radar signals reflected from these specific groups or individuals. This further optimizes the convolution kernel and improves detection accuracy. For example, after optimizing the convolution kernel based on radar signals reflected from all people within the radar range, convolution kernel a is obtained. To detect physiological information of individuals with abnormalities within the population, the radar signals reflected from these individuals are used to optimize convolution kernel a, resulting in convolution kernel b. Convolution kernel b has higher detection accuracy. These individuals with abnormalities can be people with heart rhythm abnormalities, such as the elderly or those with a history of heart disease.
[0099] In some examples, for radar signals within the first time period, after optimizing the convolution kernel using the radar signal, when the optimization target is close to the maximum, the LOSS value of the radar signal can be checked, and the radar signal data corresponding to outliers can be removed. Then, the checked radar signal can be used to further train the convolution kernel. In this way, noise in the convolution kernel radar signal can be filtered out. For example, when detecting the physiological information of adults in the target, the above method can be used to remove the data of minors. In addition, in other examples, when it is necessary to detect the data of a few people or abnormal people in the population (e.g., minors or people with a history of heart disease), the radar signal can be checked again, the radar signal corresponding to the outliers can be retained, and the retained radar signal can be used to further train the convolution kernel. In this way, the physiological information of the population can also be obtained.
[0100] The above describes the optimization of the convolution kernel based on the first optimization parameter and / or the second optimization parameter. In some examples, after optimizing the convolution kernel, the optimized convolution kernel is directly used to process the radar signal within a first time period, and the first output result is used as the physiological information of the target. In some embodiments, the first output result can be further regularized, and the result after regularization (referred to as the second output result) is used as the physiological information of the target. The regularization process is described below with examples.
[0101] In some embodiments, the first output result is regularized to obtain a second output result, which is then used as the physiological information of the target being tested.
[0102] In some embodiments, regularization of the first output result includes:
[0103] Calculate the period of the first output result and the range of the neighborhood signal of the first output result;
[0104] The second output result is obtained by calculating the ratio of the energy of the first output result to the average energy of the neighborhood signal at each time point within the first time period.
[0105] In the above embodiments, the convolution kernel G is optimized using at least one of at least a first optimization parameter and at least one second optimization parameter to obtain an optimized convolution kernel G1. The optimized convolution kernel G1 is then used to process the radar signal within a first time period to obtain a first output result A1. The period Δ of the first output result A1 is then calculated. t And the range λΔ of the neighborhood signal of the first output result A1 t λ is a positive number, where, in some examples, λ is a positive number less than 1, for example, λ is 0.8.
[0106] In some examples, the energy of the first output A1 at various times is calculated, a 1t This represents the first output result at time t. Let A1 represent the energy of the first output result A1 at time t; calculate the average energy of the neighborhood signal at time t, which can be done using the following formula:
[0107] in, Let λΔ be the range of the signal in this neighborhood. t The number of internal signals, therefore, the ratio of the energy of the first output at time t to the average energy of its neighboring signals can be expressed by the following formula:
[0108] Where, x t The second output result at time t represents the radar signal during the first time interval. The second output result can be expressed by the following formula: X = {x} t ,1≤t≤N 1S}
[0109] Where, N 1S This indicates the length of the optimized convolution kernel G1.
[0110] In some embodiments, the period of the first output result can be calculated using the following two methods:
[0111] Perform an FFT operation on the first output result to obtain the first calculation result. Use the frequency period with the highest energy in the first calculation result as the period of the first output result, or...
[0112] Calculate the time interval between the local maxima of the first data result, and use this time interval as the period of the first output result.
[0113] For example, the period of the first output result is calculated using the frequency method, the radar signal in the first time period is processed using the optimized convolution kernel G1 to obtain the first output result A1, the first output signal A1 is subjected to FFT operation to obtain the spectrum of the first output signal A1, and the frequency with the highest energy in the frequency range of the first output signal A1 is calculated based on the spectrum. The period of this frequency is the period of the first output signal A1.
[0114] For example, the period of the first output result can be calculated using a time-domain method. The first output result A1 is obtained by processing the radar signal in the first time period using the optimized convolution kernel G1. The first output result A1 is divided into multiple local signals with equal time intervals. The maximum point of each local signal is calculated. The average time interval between the multiple maximum points is taken to obtain the period of the first output signal A1.
[0115] The above method for calculating the period of the first output result is only an example. In addition, other methods can be used to calculate the period of the first output result, such as using power spectral density or autocorrelation function. This application does not impose any restrictions, as long as the period of the first output result can be calculated.
[0116] In summary, by performing regularization on the first output result, the accuracy of physiological information detection can be further improved, and the stability of the convolution kernel can be enhanced.
[0117] The following describes some effects of the embodiments of this application.
[0118] Figure 2 is a schematic diagram of the positional relationship between the target under test and the radar according to an embodiment of this application. Figure 3 is a schematic diagram comparing the IBI (Inter-beat interval) result obtained by detecting the target under test according to the method of an embodiment of this application with the actual IBI result.
[0119] In the example in Figure 3, taking the number of targets to be tested as 36, and the distances d between these 36 targets and the radar as 0.2m, 0.4m, 0.6m, 0.8m, and 1m respectively as an example, that is, each target to be tested is tested 5 times at a position 0.2m, 0.4m, 0.6m, 0.8m, and 1m away from the radar.
[0120] Figure 3(a) shows the actual heartbeat signal and the heartbeat signal extracted according to the method of the embodiment of this application. As shown in Figure 3(a), curve (a1) represents the actual heartbeat signal, and curve (a2) represents the heartbeat signal extracted according to the method of the embodiment of this application. Figure 3(b) shows the actual IBI result of the target and the IBI result obtained by the method of the embodiment of this application. As shown in Figure 3(b), curve (b1) represents the actual IBI result, and curve (b2) represents the IBI result of the embodiment of this application. Taking Figure 3 as an example, the actual heartbeat signal and the actual IBI result of the target can be extracted using the PPG (Photoplethysmography) method, or other methods can also be used.
[0121] As can be seen from Figure 3, the heartbeat signal and IBI result extracted using the method of this application embodiment have a very small error compared with the actual heartbeat signal and IBI result of the target being tested. The method of this application embodiment can accurately extract the physiological information of the target being tested.
[0122] Table 1 illustrates, for example, the IBI results obtained by extracting heartbeat signals from the target using the method of this application embodiment and existing radar heartbeat detection methods.
[0123] Table 1:
[0124] As shown in Table 1, the evaluation target is the proportion of the IBI of the 36 tested targets described above, whose error is less than 100ms compared to the true IBI of the tested targets. Method 1 is the method for extracting physiological information from radar signals according to an embodiment of this application, where the physiological information is heartbeat information. Method 2 is an existing radar heartbeat detection method. The proportion of IBI errors less than 100ms obtained using the existing radar heartbeat detection method is 79.75%, while the proportion obtained using the method of this application is 88.29%. That is, compared to the existing radar heartbeat detection method, the method of this application improves the accuracy by 8.54%. It is evident that using the method described in this application significantly increases the proportion of Inter-Beat Interval (IBI) errors less than 100ms, and the method described in this application can improve the accuracy of physiological information detection.
[0125] In the above example, existing radar heartbeat information detection methods can be spectrum-based Doppler radar heartbeat information detection methods or deep learning-based radar heartbeat information detection methods. For specific details, please refer to relevant technologies. This application does not impose any restrictions.
[0126] The above embodiments use the physiological information of the target being tested, such as heartbeat or respiration, as an example. This application is not limited to this. The machine learning model can output other detection information, such as the pulse signal of the target being tested, according to the specific needs of actual operation.
[0127] The above embodiments are merely illustrative examples of the methods described in this application, but this application is not limited thereto, and appropriate modifications can be made based on the above embodiments. For example, the above embodiments can be used alone, or one or more of the above embodiments can be combined.
[0128] According to the embodiments of this application, by measuring the vibration of the body surface at a higher frequency, low-frequency noise can be effectively filtered out. Furthermore, by processing the physiological signals extracted based on radar signals and the convolution kernel, thereby optimizing the convolution kernel, the accuracy of physiological information detection based on radar signals can be improved.
[0129] Second aspect of the embodiments
[0130] This application provides an apparatus for extracting physiological information from radar signals. Since the principle by which this apparatus solves the problem is similar to the method of the first aspect embodiment, its specific implementation can be referred to the implementation of the method of the first aspect embodiment, and the similarities will not be repeated.
[0131] Figure 4 is a schematic diagram of a device for extracting physiological information from radar signals according to an embodiment of this application. As shown in Figure 4, the device 400 for extracting physiological information from radar signals according to an embodiment of this application includes:
[0132] The physiological signal extraction unit 401 uses a convolution kernel to process the radar signal of the target under test within the first time period to obtain the first physiological signal.
[0133] The first processing unit 402 processes the first physiological signal to obtain a first optimized parameter characterizing the energy characteristics of the first physiological signal.
[0134] The second processing unit 403 processes the convolution kernel to obtain a second optimized parameter characterizing the energy properties of the convolution kernel;
[0135] The optimization unit 404 optimizes the convolution kernel according to at least one of the at least one first optimization parameter and at least one second optimization parameter to obtain the optimized convolution kernel;
[0136] The first output unit 405 uses the optimized convolution kernel to process the radar signal within the first time period to obtain a first output result, and uses the first output result as the physiological information of the target being measured.
[0137] In some embodiments, the device 400 further includes:
[0138] The second output unit 406 performs regularization processing on the first output result to obtain a second output result, and uses the second output result as the physiological information of the target being tested.
[0139] In some embodiments, the physiological signal extraction unit 401 uses the convolution kernel to perform a convolution operation on the radar signal within the first time period to obtain a first convolution result, calculates the amplitude of the first convolution result at each moment within the first time period, and obtains the first physiological signal.
[0140] In some embodiments, the first processing unit 402 performs an FFT operation on the first physiological signal to obtain a first energy value, and obtains the first optimization parameter based on the first energy value, wherein the first energy value represents the energy of the first physiological signal at each frequency within a first frequency range, and the first frequency range is the frequency range covered by the first physiological signal.
[0141] In some embodiments, the first processing unit 402 calculates the ratio of the peak energy of the first energy value in the second frequency range to the total energy of the first energy value in the first frequency range, to obtain the first optimization parameter.
[0142] The second frequency range is the frequency range within the first frequency range that represents human breathing or heartbeat.
[0143] In some embodiments, the first processing unit 402 calculates the ratio of the peak energy of the first energy value in the second frequency range to the total energy of the first energy value in the third frequency range to obtain the first optimization parameter.
[0144] Wherein, the second frequency range is the frequency range in the first frequency range that represents human breathing or heartbeat; the third frequency range is the frequency range in the first frequency range that is less than or equal to the second frequency range.
[0145] In some embodiments, the second processing unit 403 performs an FFT operation on the convolution kernel to obtain a second energy value, calculates the ratio of the total energy of the second energy value in a fifth frequency range to the total energy of the convolution kernel in a fourth frequency range, and obtains the second optimized parameter. The second energy value represents the energy of the convolution kernel at each frequency in the fourth frequency range. The fourth frequency range is the frequency range covered by the convolution kernel, and the fifth frequency range is a preset frequency range that includes the frequency range of body surface vibrations caused by human physiological activities.
[0146] In some embodiments, the human physiological activities include human respiration and / or heartbeat.
[0147] In some embodiments, the optimization unit 404 performs weighted processing on at least one of the at least one first optimization parameter and at least one second optimization parameter to obtain an optimization target, modifies the parameters in the convolution kernel to maximize the optimization target, and obtains the optimized convolution kernel when the optimization target is maximized.
[0148] In some embodiments, the second output unit 406 performs regularization processing on the first output result, including:
[0149] Calculate the period of the first output result and the range of the neighborhood signal of the first output result;
[0150] Based on the period and the range of the neighborhood signal, the ratio of the energy of the first output result to the average energy of the neighborhood signal at each moment within the first time period is calculated to obtain the second output result.
[0151] In some embodiments, the second output unit 406 performs an FFT operation on the first output result to obtain a first calculation result, and uses the frequency period with the highest energy in the first calculation result as the period of the first output result; or, the second output unit 406 calculates the time interval between the local maxima of the first output result and uses the time interval as the period of the first output result.
[0152] It is worth noting that the above description only covers the components or modules relevant to this application, but this application is not limited thereto. The apparatus 400 for extracting physiological information from radar signals may also include other components or modules, and for details regarding these components or modules, please refer to relevant technologies.
[0153] For simplicity, Figure 4 only illustrates the connection relationships or signal flow between the various components or modules, but those skilled in the art should understand that various related technologies such as bus connections can be used. The aforementioned components or modules can be implemented using hardware facilities such as processors and memory; this application does not limit this.
[0154] The above embodiments are merely illustrative examples of embodiments of this application, but this application is not limited thereto, and appropriate modifications can be made based on the above embodiments. For example, the above embodiments can be used alone, or one or more of the above embodiments can be combined.
[0155] According to the embodiments of this application, by measuring the vibration of the body surface at a higher frequency, low-frequency noise can be effectively filtered out. Furthermore, by processing the physiological signals extracted based on radar signals and the convolution kernel, thereby optimizing the convolution kernel, the accuracy of radar physiological detection can be improved.
[0156] Third aspect of the embodiments
[0157] This application provides a computer device including a means 400 for extracting physiological information from radar signals as described in the second aspect of the embodiment, the contents of which are incorporated herein by reference. This computer device may be, for example, a computer, server, workstation, laptop computer, smartphone, etc.; however, this application is not limited thereto.
[0158] Figure 5 is a schematic diagram of a computer device according to an embodiment of this application. As shown in Figure 5, the computer device 500 may include a processor (e.g., a central processing unit, CPU) 510 and a memory 520; the memory 520 is coupled to the central processing unit 510. The memory 520 can store various data; in addition, it also stores an information processing program 521, and executes the program 521 under the control of the processor 510.
[0159] In some embodiments, the function of the apparatus 400 for extracting physiological information from radar signals is integrated into the processor 510. The processor 510 is configured to implement the vital information detection method as described in the first aspect of the embodiment.
[0160] In some embodiments, the device 400 for extracting physiological information from radar signals is configured separately from the processor 510. For example, the device 400 for extracting physiological information from radar signals can be configured as a chip connected to the processor 510, and the function of the device 400 for extracting physiological information from radar signals can be realized through the control of the processor 510.
[0161] In addition, as shown in Figure 5, the computer device 500 may also include: input / output (I / O) devices 530 and a display 540, etc.; the functions of the above components are similar to those in the prior art, and will not be described in detail here. It is worth noting that the computer device 500 does not necessarily include all the components shown in Figure 5; furthermore, the computer device 500 may also include components not shown in Figure 5, which can be referred to in related technologies.
[0162] This application also provides a computer-readable program, wherein when the program is executed in a computer device, the program causes the computer device to perform the method described in the first embodiment.
[0163] This application provides a storage medium storing a computer-readable program, wherein the computer-readable program causes a computer device to perform the method described in the first aspect of the embodiment.
[0164] The apparatus and methods described above in this application can be implemented in hardware or in combination with software. This application relates to a computer-readable program that, when executed by a logic component, enables the logic component to implement the apparatus or components described above, or to implement the various methods or steps described above. This application also relates to storage media for storing the above programs, such as hard disks, magnetic disks, optical disks, DVDs, flash memory, etc.
[0165] The methods / apparatus described in conjunction with the embodiments of this application can be directly embodied in hardware, software modules executed by a processor, or a combination of both. For example, one or more and / or combinations of one or more functional block diagrams shown in the figures can correspond to various software modules in a computer program flow, or to various hardware modules. These software modules can correspond to the various steps shown in the figures, respectively. These hardware modules can be implemented, for example, using a field-programmable gate array (FPGA) to embed these software modules.
[0166] The software module can reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art. A storage medium can be coupled to the processor, enabling the processor to read information from and write information to the storage medium; or the storage medium can be an integral part of the processor. The processor and storage medium can reside in an ASIC. The software module can be stored in the memory of a mobile terminal or in a memory card that can be inserted into the mobile terminal. For example, if the device (such as a mobile terminal) uses a high-capacity MEGA-SIM card or a high-capacity flash memory device, the software module can be stored in the MEGA-SIM card or the high-capacity flash memory device.
[0167] One or more and / or one or more combinations of functional blocks described in the accompanying drawings can be implemented as a general-purpose processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic device, discrete hardware component, or any suitable combination thereof for performing the functions described herein. One or more and / or one or more combinations of functional blocks described in the accompanying drawings can also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors in communication with a DSP, or any other such configuration.
[0168] The present application has been described above with reference to specific embodiments. However, those skilled in the art should understand that these descriptions are exemplary and not intended to limit the scope of protection of the present application. Those skilled in the art can make various modifications and variations to the present application based on its spirit and principles, and these modifications and variations are also within the scope of the present application.
Claims
1. A device for extracting physiological information from radar signals, wherein, The device includes: The physiological signal extraction unit uses a convolution kernel to process the radar signal of the target under test within the first time period to obtain the first physiological signal. The first processing unit processes the first physiological signal to obtain a first optimized parameter characterizing the energy properties of the first physiological signal. The second processing unit processes the convolution kernel to obtain a second optimized parameter characterizing the energy properties of the convolution kernel; An optimization unit optimizes the convolution kernel based on at least one of the at least one first optimization parameter and at least one second optimization parameter to obtain an optimized convolution kernel; The first output unit processes the radar signal within the first time period using the optimized convolution kernel to obtain a first output result, which is then used as the physiological information of the target being measured.
2. The apparatus according to claim 1, wherein, The device further includes: The second output unit performs regularization processing on the first output result to obtain a second output result, and uses the second output result as the physiological information of the target being tested.
3. The apparatus according to claim 1, wherein, The physiological signal extraction unit uses the convolution kernel to perform a convolution operation on the radar signal within the first time period to obtain a first convolution result, calculates the amplitude of the first convolution result at each moment within the first time period, and obtains the first physiological signal.
4. The apparatus according to claim 1, wherein, The first processing unit performs an FFT operation on the first physiological signal to obtain a first energy value, and obtains the first optimization parameter based on the first energy value. The first energy value represents the energy of the first physiological signal at each frequency within a first frequency range, and the first frequency range is the frequency range covered by the first physiological signal.
5. The apparatus according to claim 4, wherein, The first processing unit calculates the ratio of the peak energy of the first energy value within the second frequency range to the total energy of the first energy value within the first frequency range, and obtains the first optimization parameter. The second frequency range is the frequency range within the first frequency range that represents human breathing or heartbeat.
6. The apparatus according to claim 4, wherein, The first processing unit calculates the ratio of the peak energy of the first energy value within the second frequency range to the total energy of the first energy value within the third frequency range, and obtains the first optimization parameter. Wherein, the second frequency range is the frequency range in the first frequency range that represents human breathing or heartbeat; the third frequency range is the frequency range in the first frequency range that is less than or equal to the second frequency range.
7. The apparatus according to claim 1, wherein, The second processing unit performs an FFT operation on the convolution kernel to obtain a second energy value, and calculates the ratio of the total energy of the second energy value in the fifth frequency range to the total energy of the convolution kernel in the fourth frequency range to obtain the second optimized parameters. Wherein, the second energy value represents the energy of the convolution kernel at each frequency within the fourth frequency range; The fourth frequency range is the frequency range covered by the convolution kernel, and the fifth frequency range is a preset frequency range that includes the frequency range of body surface vibrations caused by human physiological activities.
8. The apparatus according to claim 7, wherein, The human physiological activities include human respiration and / or heartbeat.
9. The apparatus according to claim 1, wherein, The optimization unit performs weighted processing on at least one of the at least one first optimization parameter and at least one second optimization parameter to obtain an optimization target, modifies the parameters in the convolution kernel to maximize the optimization target, and obtains the optimized convolution kernel when the optimization target is maximized.
10. The apparatus according to claim 2, wherein, The second output unit performs regularization processing on the first output result, including: Calculate the period of the first output result and the range of the neighborhood signal of the first output result; Based on the period and the range of the neighborhood signal, the ratio of the energy of the first output result to the average energy of the neighborhood signal at each moment within the first time period is calculated to obtain the second output result.
11. The apparatus according to claim 10, wherein, The second output unit performs an FFT operation on the first output result to obtain a first calculation result, and uses the frequency period with the highest energy in the first calculation result as the period of the first output result, or... The second output unit calculates the time interval between the local maximum points of the first output result, and uses the time interval as the period of the first output result.
12. A method for extracting physiological information from radar signals, wherein, The method includes: The radar signal of the target under test in the first time period is processed using a convolution kernel to obtain the first physiological signal; The first physiological signal is processed to obtain a first optimized parameter characterizing the energy properties of the first physiological signal; The convolution kernel is processed to obtain a second optimized parameter characterizing the energy properties of the convolution kernel; The convolution kernel is optimized according to at least one of the at least one first optimization parameter and at least one second optimization parameter to obtain an optimized convolution kernel; The optimized convolutional kernel is used to process the radar signal within the first time period to obtain a first output result, which is then used as the physiological information of the target being measured.
13. The method according to claim 12, wherein, The method further includes: The first output result is regularized to obtain a second output result, which is then used as the physiological information of the target being tested.
14. A computer device, the computer device comprising a processor and a memory coupled to the processor, the memory storing a computer program, wherein, The processor is configured to execute the computer program to implement the method for extracting physiological information from radar signals as described below: The radar signal of the target under test in the first time period is processed using a convolution kernel to obtain the first physiological signal; The first physiological signal is processed to obtain a first optimized parameter characterizing the energy properties of the first physiological signal; The convolution kernel is processed to obtain a second optimized parameter characterizing the energy properties of the convolution kernel; The convolution kernel is optimized according to at least one of the at least one first optimization parameter and at least one second optimization parameter to obtain an optimized convolution kernel; The optimized convolutional kernel is used to process the radar signal within the first time period to obtain a first output result, which is then used as the physiological information of the target being measured.