JP2023035888A - Device and method for extracting heart beat data on the basis of radio radar signal - Google Patents

Abstract

To provide a device and a method for extracting heart beat data on the basis of a radio radar signal.SOLUTION: A method includes: collecting a radio radar signal obtained by sensing a detection target using radar; performing band-pass filtering in a first frequency band on radio radar data; calculating frequency spectrum energy at each time accumulated in a time window; acquiring band-pass filtering feature data by performing feature extraction on data after the band-pass filtering and/or acquiring frequency spectrum energy feature data by performing feature extraction on frequency spectrum energy at each time in a second frequency band; inputting the band-pass filtering feature data and/or the frequency spectrum energy feature data to a neural network based detection model; and obtaining a heart beat interval and the number of heart beats of the detection target on the basis of waveform data.SELECTED DRAWING: Figure 2

Description

The present invention relates to the technical field of radar detection.

With the improvement of living standards, people's health awareness is gradually increasing, and it is necessary to regularly evaluate heartbeats to monitor health conditions. Generally speaking, there are two metrics to assess heartbeat, one is resting heart rate and the other is heart rate variability (HRV). Resting heart rate refers to the number of heart beats per minute in a normal person's resting state, and heart rate variability refers to the change in the difference between heartbeat cycles. An index for evaluating heartbeat is a very important guideline for disease diagnosis and health management, and heart rate variability can also be used to evaluate stress index, identify (specify) emotions, and the like.

In medicine, an electrocardiogram (ECG) is usually used to monitor a patient's resting heart rate and heart rate variability. Although ECG monitoring is highly accurate, it requires a sheet to be attached to a specified position on the skin during the test. , it is inconvenient to use, feels uncomfortable, and is inconvenient to monitor using an electrocardiogram at any time.

With the popularization of smart wearable devices, heart rate monitoring is integrated into most smart bracelets/watches, but most of these devices can only provide heart rate indicators, making it difficult to accurately assess heartbeat. In addition, since it is necessary to wear it for a long time, it is easy for the elderly to forget to wear it, and it is troublesome to wear it. On the other hand, contactless monitoring of the heartbeat has been proposed so far, for example by analyzing the radar signal reflected by the human body to extract the signal related to the heartbeat.

It should be noted that the above introduction of the background art is to clearly and completely describe the technical solution of the present invention and facilitate the understanding of those skilled in the art. These technical proposals are described in the background art of the present invention and should not be construed as being well known to those skilled in the art.

However, the inventor discovered the following. That is, since the physiological signal includes not only the heartbeat signal but also the respiratory signal, the extracted heartbeat signal is susceptible to the respiratory signal. At present, heartbeat contactless monitoring schemes can generally detect the presence or absence of heartbeat signals, but the detection accuracy may be low and accurate heartbeat data may not be extracted.

In view of at least one of the above technical problems, embodiments of the present invention provide an apparatus and method for extracting heartbeat data based on wireless radar signals. Accurate contactless acquisition of heartbeat data enables monitoring of resting heart rate and heart rate variability.

According to one aspect of an embodiment of the present invention, there is provided an apparatus for extracting heartbeat data based on wireless radar signals, the apparatus comprising:
an acquisition unit for acquiring radio radar signals detected by the radar;
a filtering unit for bandpass filtering an acquired wireless radar signal in a first frequency band, said first frequency band being a predetermined frequency band for said heartbeat to be detected;
a computing unit for computing the frequency spectrum energy at each instant accumulated by the time window of the data after bandpass filtering;
performing feature extraction on the data after the bandpass filtering to obtain bandpass filtering feature data, and/or performing feature extraction on the frequency spectrum energy at each time point in the second frequency band to obtain the frequency spectrum a feature extraction unit for obtaining energy feature data;
a detection unit for inputting the bandpass filtering feature data in the time domain and/or the frequency spectrum energy feature data in the frequency domain into a neural network-based detection model and using the detection model to obtain detected waveform data; and an acquisition unit for obtaining the heartbeat interval and heart rate of the detection target based on the waveform data.

According to another aspect of an embodiment of the invention, there is provided a method of extracting heartbeat data based on wireless radar signals, the method comprising:
Collecting radio radar signals from detection targets detected by the radar;
performing bandpass filtering on the acquired wireless radar signal in a first frequency band, the first frequency band being a predetermined frequency band associated with the heartbeat to be detected;
calculating the frequency spectrum energy at each instant accumulated by the time window of the data after bandpass filtering;
performing feature extraction on the data after the bandpass filtering to obtain bandpass filtering feature data, and/or performing feature extraction on the frequency spectrum energy at each time point in the second frequency band to obtain the frequency spectrum obtaining energy signature data;
inputting the bandpass filtering feature data in the time domain and/or the frequency spectrum energy feature data in the frequency domain into a neural network-based detection model to obtain post-detection waveform data using the detection model; and the waveform. Obtaining a heartbeat interval and a heart rate of the detection target based on the data.

Advantageous effects of embodiments of the present invention are at least the following. That is, obtaining bandpass filtering feature data and/or frequency spectrum energy feature data based on the wireless radar signal, and neural network-based detection of the time domain bandpass filtering feature data and/or frequency domain frequency spectrum energy feature data. input into a model to obtain post-detection waveform data using the detection model; and heartbeat interval and heart rate to be detected based on the waveform data. It can accurately acquire heartbeat data without contact and realize resting heart rate and heart rate variability monitoring, so heartbeat monitoring is not only convenient at any time, but also provides a comfortable user experience. and high accuracy.

Also, features described and/or shown for one embodiment may be used in the same or similar manner in one or more of the other embodiments, combined with features in other embodiments, or described in other embodiments. You can also replace features in

It should be noted that terms such as "including/having," as used herein, refer to the presence of a feature, element, step, or assembly, but not one or more other features, elements, steps, or does not exclude the presence or addition of assemblies.

Elements and features shown in one drawing or embodiment of the invention can be combined with elements and features shown in one or more other drawings or embodiments. Also, in the figures, like reference numerals are used to indicate corresponding parts in several figures and to indicate corresponding parts used in several embodiments.
1 is a diagram showing an electrocardiographic signal acquired by an electrocardiogram (ECG); FIG. FIG. 4 illustrates a method for extracting heartbeat data based on wireless radar signals in an embodiment of the present invention; It is a figure which shows an example of the radio radar data in the Example of this invention. It is a figure which shows an example of the data after band-pass filtering in the Example of this invention. FIG. 3 is a diagram showing an example of frequency spectrum energy at multiple times in an embodiment of the present invention; FIG. 4 is a diagram showing an example of an electrocardiographic signal acquired by ECG in the embodiment of the present invention; It is a figure which shows an example of the bandpass filtering feature data in the Example of this invention. FIG. 4 illustrates another method of extracting heartbeat data based on wireless radar signals in an embodiment of the present invention; FIG. 4 is a diagram showing an example of comparison between an electrocardiographic signal acquired by ECG and triangular wave data in the embodiment of the present invention; FIG. 3 shows an example of heart rate and heartbeat interval losses and histograms in an embodiment of the present invention; Fig. 3 shows an apparatus for extracting heartbeat data based on wireless radar signals in an embodiment of the present invention; It is a figure which shows the electronic device in the Example of this invention.

The foregoing and other features of the present invention will become apparent with reference to the accompanying drawings and the following description. While the specification and drawings disclose specific embodiments of the invention, it is to be understood that they illustrate only some of the embodiments that may employ the principles of the invention and that the invention is not so described. It is not intended to be limited to any particular embodiment, i.e., the invention includes all modifications, variations and alternatives falling within the scope of the appended claims.

In embodiments of the present invention, the radar may be, but is not limited to, millimeter wave (mm Wave) radar. The radar transmits electromagnetic waves through a transmitting antenna and receives corresponding reflected waves (radar echo information) after being reflected by different objects. By analyzing the radar echo information, we can effectively extract the information such as the position between the object and the radar, the radial movement speed, etc., which can meet the needs of many application scenarios.

In embodiments of the present invention, objects to be detected may be people of different ages, such as the elderly, children, elderly and/or caregivers, children and/or guardians. It can be. The present invention is not limited to these examples, and the object to be detected may also be an animal having biological characteristics. A description will be given below using the human body as an example.

FIG. 1 is a diagram showing an electrocardiographic signal acquired by electrocardiogram (ECG), which has at least two pieces of R-wave information (wave peaks) 101, as shown in FIG. The inter-beat interval (IBI) and heart rate can be calculated from the electrocardiographic signal.

In an embodiment of the present invention, the heartbeat data includes a detected heartbeat interval (which may be expressed using IBI) and a heart rate, wherein the heart rate can represent an index of resting heart rate; Heartbeat intervals can be used to analyze heart rate variability. Note that related art can be referred to for specific contents of these concepts.

<Example of the first aspect>
Embodiments of the present invention provide a method for extracting heartbeat data based on wireless radar signals.

FIG. 2 is a diagram showing a method for extracting heartbeat data based on wireless radar signals in an embodiment of the present invention, as shown in FIG. 2, the method includes the following operations (steps).

201: Acquire radio radar signals detected by radar;
202: performing band-pass filtering in a first frequency band on the obtained wireless radar data, wherein the first frequency band is a predetermined frequency band related to the heartbeat to be detected;
203: Calculate the frequency spectrum energy at each time accumulated by the time window of the data after bandpass filtering;
204: Perform feature extraction on the data after bandpass filtering to obtain bandpass filtering feature data, and/or perform feature extraction on the frequency spectrum energy at each time in the second frequency band to obtain frequency obtaining spectral energy signature data;
205: Input the bandpass filtering feature data in the time domain and/or the frequency spectrum energy feature data in the frequency domain into a neural network-based detection model, and obtain waveform data after detection using the detection model; and
206: Obtaining the heartbeat interval and heart rate of the detection object according to the waveform data.

Note that FIG. 2 described above is merely for exemplifying the embodiment of the present invention, and the present invention is not limited to this. For example, the execution order between operations (steps) can be adjusted appropriately, or some operations can be increased or decreased. Moreover, those skilled in the art are not limited to the description of FIG. 2 above, and can make appropriate modifications based on the above contents.

In some embodiments, a single frequency modulated continuous wave radar system is employed to sense the external space via radio signals transmitted by the radar transmitter. The object to be detected is within the radiation range of the radar device and can, for example, sit or lie stationary at a certain distance from the radar device, for example less than 60 cm. Although one radar device may monitor a certain detection target, the present invention is not limited to this. For example, a radar device with enhanced functionality can simultaneously monitor multiple detection targets.

In some embodiments, a radar receiver can receive a radar echo (a wireless radar signal), and a data acquisition card acquires the wireless radar signal at a first sampling rate S1 to obtain wireless radar data. be able to. The first sampling rate S1 may be relatively high, for example 2500Hz.

FIG. 3 is a diagram showing an example of radio radar data in an embodiment of the present invention. As shown in FIG. 3, the radio radar data includes, for example, an in-phase signal radar_I(n) and a quadrature signal radar_Q(n). .

In some embodiments, the first frequency band is a predetermined frequency band for heartbeats to be detected, e.g., the first frequency band is [8, 100] Hz, or [-100, -8]. [8, 100] Hz. A bandpass filter can provide a basis for accurate extraction of heartbeat data by canceling some of the effects of low frequency respiratory signals and high frequency noise signals.

For example, by performing band-pass filtering on radar_I(n) and radar_Q(n), data modulated mainly by heartbeat waves after band-pass filtering, that is, radar_I_bp(n) and radar_Q_bp(n) ).

In some embodiments, the DC frequency offset may also be eliminated from the data after bandpass filtering.

For example:
radar_I_bp(n)=radar_I_bp(n)-mean(radar_I_bp(n)); and
radar_Q_bp(n)=radar_Q_bp(n)-mean(radar_Q_bp(n))
can be used to eliminate the DC offset.

FIG. 4 is a diagram showing an example of data after band-pass filtering according to the embodiment of the present invention. As shown in FIG. 4, the data after band-pass filtering is, for example, the in-phase signal radar_I_bp( n) and the quadrature signal radar_Q_bp(n).

In some embodiments, the data after band-pass filtering with a sliding time window is transformed from a time domain signal to a frequency domain signal to obtain the frequency spectrum energy at each time. For example, the short-time Fourier transform (STFT), wavelet transform, etc. can transform the data after band-pass filtering from a time domain signal to a frequency domain signal.

For example, transforming the processed IQ complex signal from the time domain to the frequency domain, the complex signal is
S(n)=radar_I_bp(n)+j*radar_Q_bp(n)
is. A time window may be represented by a sequence, where n represents each time instant.

For example, a complex signal S(n) is divided into n-stage (segment) k-point sequences S(k), and each sequence S(k, i) is subjected to a frequency domain transform (for example, FFT). Acquire the frequency spectrum z(t(i), f(m), i) of the sequence, and perform the merger on the frequency spectrum of each sequence to obtain the frequency spectrum z(t(n) of the complex signal S(n) , f(m)).

Among them, S(k, i) represents the i-th sequence of sequence length k, where i=0, 1, 2, . , and the number of points may be a fixed value, such as 512. z(t(i), f(m), i) is the ith sequence (i.e., time is t(i), t(i)=i*f _s , f _s is the sampling frequency) ) at each frequency point f(m) (m=0, 1, 2, . . . , k−1).

FIG. 5 is a diagram showing an example of frequency spectrum energy at multiple times in the embodiment of the present invention. Among them, the segmented time window is e.g. 204.8ms (512-point FFT), the sliding step length is 0.4ms, e.g. It can guarantee that the sampling rate matches the sampling rate of the IQ signal. For example, within the range [-30,-8][8,30] Hz, we can obtain the frequency spectrum energy at each time (as indicated by the gray area in FIG. 5).

FIG. 6 is a diagram showing an example of an electrocardiographic signal obtained by ECG in the embodiment of the present invention. As shown in FIGS. 5 and 6, the frequency spectrum energy distribution in FIG. 5 substantially matches the electrocardiogram signal distribution in FIG. This provides an initial verification that embodiments of the present invention can improve the accuracy of heartbeat data.

In the following, we first describe how to obtain the frequency spectrum energy feature data.

In some embodiments, the second frequency band can be a sub-frequency band (sub-frequency band) within the first frequency band. For example, the second frequency band is [8,30]Hz, or [-30,-8][8,30]Hz. Aggregation and analysis of the frequency spectrum within the second frequency band can further remove the effects of noise signals and further improve the accuracy of the heartbeat data.

であり、
また、例えば、周波数帯域[-30，-8]Hzの周波数スペクトルエネルギーに対して和を求めた後のデータ、即ち、

であり；
・前記第二周波数帯域内の一部周波数帯域の各時刻の周波数スペクトルエネルギーの最大値を求めた後のデータであり、
例えば、周波数帯域[8，30]Hzの周波数スペクトルエネルギーの最大値を求めた後のデータ、即ち、

であり、
また、例えば、周波数帯域[-30，-8]Hzの周波数スペクトルエネルギーの最大値を求めた後のデータ、即ち、

であり；
・前記第二周波数帯域内の全部周波数帯域の各時刻の周波数スペクトルエネルギーに対して和を求めた後のデータであり、
例えば、周波数帯域[8，30]Hz及び[-30，-8]Hzの周波数スペクトルエネルギーに対して和を求めた後のデータ、即ち、

であり；
・前記第二周波数帯域内の全部周波数帯域の各時刻の周波数スペクトルエネルギーの最大値を求めた後のデータであり、
例えば、周波数帯域[8，30]Hz及び[-30，-8]Hzの周波数スペクトルエネルギーの最大値を求めた後のデータ、即ち、

である。 In some embodiments, feature extraction is performed on the frequency spectrum energy at each time point within the second frequency band to obtain frequency spectrum energy feature data, wherein the frequency spectrum energy feature data is as follows: including one or more of, i.e.
- The data after obtaining the sum of the frequency spectrum energy at each time of the partial frequency band within the second frequency band,
For example, the summed data for the frequency spectrum energy in the frequency band [8, 30] Hz, that is,

and
Also, for example, the data after summing the frequency spectrum energy in the frequency band [-30, -8] Hz, that is,

is;
- The data after obtaining the maximum value of the frequency spectrum energy at each time of the partial frequency band within the second frequency band,
For example, the data after obtaining the maximum value of the frequency spectrum energy in the frequency band [8, 30] Hz, that is,

and
Also, for example, data after obtaining the maximum value of the frequency spectrum energy in the frequency band [-30, -8] Hz, that is,

is;
- the data after calculating the sum of the frequency spectrum energy at each time of all the frequency bands in the second frequency band,
For example, the data after summing for the frequency spectrum energy of the frequency bands [8, 30] Hz and [-30, -8] Hz, i.e.

is;
- Data after obtaining the maximum value of the frequency spectrum energy at each time of all frequency bands in the second frequency band,
For example, the data after obtaining the maximum value of the frequency spectrum energy in the frequency bands [8, 30] Hz and [-30, -8] Hz, that is,

is.

Although the frequency spectrum energy feature data has been exemplified above, the present invention is not limited to this, and other feature data after performing feature extraction based on the frequency spectrum energy may be used. Some or all of these frequency spectrum energy feature data can be input into a neural network-based detection model, and the detection model can be used to obtain post-detection waveform data.

The following describes how to obtain the bandpass filtering feature data.

であり、
そのうち、S(n)=radar_I_bp(n)+j*radar_Q_bp(n)である。 In some embodiments, bandpass filtered feature data can also be obtained by performing feature extraction on the data after bandpass filtering. The bandpass filtering feature data may include one or more of the following:
- is data after performing band-pass filtering on the quadrature signal of the radar data,
For example, radar_Q(n);
- Data after performing band-pass filtering on the in-phase signal of the radar data,
For example, radar_I_bp(n);
- data obtained after obtaining an amplitude value by performing band-pass filtering on the radar data,
for example,

and
Among them, S(n)=radar_I_bp(n)+j*radar_Q_bp(n).

FIG. 7 is a diagram showing an example of bandpass filtering feature data in an embodiment of the present invention, data radar_I_bp after bandpass filtering and DC offset removal, bandpass filtering and DC offset removal; The data radar_Q_bp after the amplitude values have been calculated and the data iq_am after the amplitude values have been obtained are shown.

Although the bandpass filtering feature data has been exemplified above, the present invention is not limited to this, and other feature data after feature extraction has been performed based on data after bandpass filtering may be used. . By inputting some or all of these bandpass filtering feature data into a neural network-based detection model and using the detection model to detect the frequency spectrum energy feature data and the bandpass filtering feature data, post-detection waveform data can be obtained.

FIG. 8 is a diagram illustrating another method of extracting heartbeat data based on wireless radar signals in an embodiment of the present invention, in which both bandpass filtering feature data and frequency spectrum energy feature data are input to the detection model. It shows the case of using as feature data. As shown in FIG. 8, the method includes the following steps (operations).

801: Collect radio radar signals detected by radar;
802: performing band-pass filtering of a first frequency band on the obtained wireless radar data, wherein the first frequency band is a predetermined frequency band related to the heartbeat to be detected;
803: Calculate the frequency spectrum energy at each time accumulated by the time window of the data after bandpass filtering;
804: Acquire frequency spectrum energy feature data by performing feature extraction on the frequency spectrum energy at each time in the second frequency band.

As shown in FIG. 8, the method further includes the following steps.

805: Obtain bandpass filtering feature data by performing feature extraction on data after bandpass filtering;
806: input the frequency spectrum energy feature data and the bandpass filtering feature data into a neural network-based detection model, and obtain post-detection waveform data using the detection model; and
807: Obtaining the heartbeat interval and heart rate of the detection target based on the waveform data.

Note that FIG. 8 described above is merely for exemplifying the embodiment of the present invention, and the present invention is not limited thereto. For example, the execution order between operations can be adjusted appropriately, or some operations can be increased or decreased. Moreover, those skilled in the art are not limited to the description of FIG. 8 above, and can make appropriate modifications based on the above contents.

In some embodiments, prior to inputting feature data (frequency spectrum energy feature data and/or bandpass filtering feature data) into a neural network-based detection model, a second sampling rate S2 is further applied to the feature data. wherein the second sampling rate S2 is lower than the first sampling rate S1.

For example, the first sampling rate S1 is 2500Hz and the second sampling rate S2 is 200Hz. A relatively high first sampling rate enables the acquisition of wireless radar signals that can accurately reflect heartbeat characteristics, thus obtaining accurate characteristic data, and a relatively low second sampling rate permits the acquisition of accurate characteristic data. It can be free of distortion and reduce the complexity of the deep learning model.

In some embodiments, normalization may also be performed on the resampled feature data.

である。 For example, radar feature information is normalized using a min-max normalization method. Taking radar_I_bp as an example,

is.

In some embodiments, feature data within one time period (time period) may be input to the detection model, wherein the one time period is a predetermined time period that can be ensured to contain at least two pieces of R-wave information ( For example, 2s). This can further guarantee the accuracy of the detection data and the response time is fast.

In some embodiments, the waveform data of the detection model output is triangular wave data, which allows simple processing of the waveform data.

FIG. 9 is a diagram showing an example of comparison between an electrocardiogram signal obtained by ECG and triangular wave data in the embodiment of the present invention. As shown in FIG. 9, the loss between the triangular wave data in the embodiment of the present invention and the electrocardiographic signal acquired by ECG (loss, as indicated by the broken line in FIG. 9) is within a certain range. Heartbeat data accuracy in embodiments of the present invention is high because it is controlled internally.

In some embodiments, adding noise to triangular wave data output from the detection model; and detecting the maximum value of triangular wave data after adding noise to obtain the heartbeat interval and heart rate to be detected. Also good.

For example, to avoid detecting points with the same maxima,
ypred=ypred+rand(length(ypred))*0.0001
You can add minute noise to triangular wave data like

In some embodiments, the heartbeat interval and heart rate to be detected can be obtained based on the triangular wave data after adding noise. Specifically, the maximal value of the triangular wave data is associated with the R-wave information of the electrocardiographic signal obtained from the electrocardiogram, and the heartbeat interval of the detection target is obtained based on the time interval between adjacent maximal values. and the heart rate of the detection target can be obtained based on the obtained heartbeat interval of the detection target. As a result, the maximum value of the triangular wave after noise addition is detected, and the position of the point of the maximum value corresponds to the position of the R wave in the electrocardiogram, thereby further improving detection accuracy.

であり、現在の時刻の心拍数は現在の時刻のIBIにより取得でき、即ち、

である。 For example, the IBI at the current time can be obtained by the value of adjacent R-wave intervals (RR interval) within a window of average 5s. Suppose the data sampling rate is s ₀ , the window time length is 5s, there are m R waves, and the corresponding positions of the R waves are {l ₀ ,l ₁ ,...,l _m-1 } ,

, and the heart rate at the current time can be obtained by the IBI at the current time, i.e.

is.

FIG. 10 is a diagram showing an example of heart rate and heartbeat interval loss and histogram in an embodiment of the present invention. As shown in FIG. 10, the scheme of an embodiment of the present invention allows accurate extraction of heartbeat data from wireless radar signals.

Having exemplified how to extract heartbeat data, a brief description of the detection model and training follows.

In some embodiments, the neural network-based detection model may be a deep learning neural network model, such as Resnet, VGG, etc., may be a 1D fully convolutional network, and the loss function may be can be KLDivLoss, etc. In the training process, the training may be performed using the electrocardiographic signal detected by the ECG as the true value.

Embodiments of the present invention may obtain a set/set of optimal parameters by a supervised training (learning) method, and then apply the parameters to the detection model. The embodiments of the present invention do not limit the structure of the detection model, which can be referred to related art. Also, the specific training method is not limited. For example, SGD (Stochastic Gradient Descent) optimization, Adam (Adaptive
Moment Estimation) optimization etc. may be used.

Although only each step or process according to the present invention has been described above, the present invention is not limited to these. The heartbeat data extraction method may further include other steps or processes, and the specific contents of these steps or processes can be referred to the prior art.

Although each of the above-described embodiments is for illustratively describing the embodiments of the present invention, the present invention is not limited to these, and appropriate modifications and the like are made based on each of the above-described embodiments. can be For example, each of the above-described embodiments may be used alone, or a plurality of the above-described embodiments may be used in combination.

As can be seen from the above embodiments, the bandpass filtering feature data and/or the frequency spectrum energy feature data are obtained based on the wireless radar signal, and the bandpass filtering feature data in the time domain and/or the frequency spectrum energy feature data in the frequency domain are obtained. The data may be input into a neural network-based detection model, the detection model may be used to obtain post-detection waveform data; and the heartbeat interval and heart rate of a detection target may be obtained based on the waveform data. It can accurately acquire heartbeat data without contact and realize resting heart rate and heart rate variability monitoring, so heartbeat monitoring is not only convenient at any time, but also provides a comfortable user experience. and high accuracy.

<Example of second aspect>
An embodiment of the present invention provides an apparatus for extracting heartbeat data based on wireless radar signals, wherein the same content as the embodiment of the first aspect is omitted herein.

FIG. 11 is a schematic diagram of an apparatus for extracting heartbeat data based on wireless radar signals in an embodiment of the present invention, as shown in FIG. 11, an apparatus 1100 for extracting heartbeat data based on wireless radar signals includes Including such things as:

Acquisition unit 1101; Acquire radio radar signals detected by radar;
filtering unit 1102: band-pass filtering a first frequency band for the obtained wireless radar signal, the first frequency band being a predetermined frequency band for the detected heartbeat;
Calculation unit 1103: calculating the frequency spectrum energy at each time accumulated by the time window of the data after bandpass filtering;
Feature extraction unit 1104: perform feature extraction on the bandpass filtered data to obtain bandpass filtered feature data, and/or perform feature extraction on the frequency spectrum energy at each time point within the second frequency band. to obtain frequency spectrum energy feature data;
detection unit 1105: inputting the bandpass filtering feature data in the time domain and/or the frequency spectrum energy feature data in the frequency domain into a neural network-based detection model to obtain waveform data after detection using the detection model; and Obtaining unit 1106: obtaining the heartbeat interval and heart rate of the detection object according to the waveform data.

In some embodiments, the computing unit 1103 converts the band-pass filtered data from a time domain signal to a frequency domain signal with a sliding time window to obtain the frequency spectrum energy at each time.

In some embodiments, computation unit 1103 uses a short-time Fourier transform or wavelet transform to transform the bandpass filtered data from a time domain signal to a frequency domain signal.

In some embodiments, the frequency spectrum energy feature data includes one or more of the following:
Data after calculating the sum of the frequency spectrum energy at each time of the partial frequency band within the second frequency band;
Data after obtaining the maximum value of frequency spectrum energy at each time in a partial frequency band within the second frequency band;
data after summing the frequency spectrum energy at each time of all frequency bands within said second frequency band; and the maximum value of frequency spectrum energy at each time of all frequency bands within said second frequency band This is the data after the calculation.

In some embodiments, the bandpass filtering feature data includes one or more of the following:
Data after performing bandpass filtering on quadrature signals of the radar data;
Data after performing band-pass filtering on an in-phase signal of the radar data; and Data after obtaining an amplitude value by performing band-pass filtering on the radar data.

In some embodiments, collection unit 1101 samples at a first sampling rate, and the apparatus further includes:
resampling unit 1107: resampling the feature data at a second sampling rate before inputting the feature data into the neural network-based detection model, wherein the second sampling rate is equal to the first sampling rate; and normalization unit 1108: performs normalization on the resampled feature data.

In some embodiments, the detection unit 1105 inputs feature data in one time period into the detection model, of which the one time period is a predetermined time period that can be ensured to contain at least two R-wave information of a heartbeat. It's time.

In some embodiments, the waveform data of the detection model output is triangular wave data.

In some embodiments, the acquisition unit 1106 further adds noise to the triangular wave data of the detection model output; and heart rate.

In some embodiments, the acquiring unit 1106 specifically associates the maxima of the triangular wave data with the R-wave information of the electrocardiographic signal acquired by the electrocardiogram, and calculates the time interval between adjacent maxima. and obtaining the heartbeat interval of the detection target based on the obtained heartbeat interval of the detection target, and obtaining the heartbeat rate of the detection target based on the obtained heartbeat interval of the detection target.

Although each component or module according to the present invention has been described above, the present invention is not limited to these. The device 1100 for extracting heartbeat data based on wireless radar signals may further include other components or modules, and the specific content of these components or modules can be referred to related art.

For convenience, FIG. 11 only shows the connection relationship or signal direction between each component or module, but it should be understood by those skilled in the art that various related techniques such as bus connections can be adopted. That is. Each component or module described above may be implemented by hardware such as a processor, a memory, etc., but embodiments of the present invention are not limited thereto.

Although each of the above-described embodiments is for illustratively describing the embodiments of the present invention, the present invention is not limited to these, and appropriate modifications and the like are made based on each of the above-described embodiments. can be For example, each of the above-described embodiments may be used alone, or a plurality of the above-described embodiments may be used in combination.

As can be seen from the above embodiments, the bandpass filtering feature data and/or the frequency spectrum energy feature data are obtained based on the wireless radar signal, and the bandpass filtering feature data in the time domain and/or the frequency spectrum energy feature data in the frequency domain are obtained. The data may be input into a neural network-based detection model, the detection model may be used to obtain post-detection waveform data; and the heartbeat interval and heart rate of a detection target may be obtained based on the waveform data. It can accurately acquire heartbeat data without contact and realize resting heart rate and heart rate variability monitoring, so heartbeat monitoring is not only convenient at any time, but also provides a comfortable user experience. and high accuracy.

<Embodiment of the third aspect>
An embodiment of the present invention provides an electronic device, which includes an apparatus 1100 for extracting heartbeat data based on wireless radar signals according to embodiments of the second aspect, the contents of which are incorporated herein. Examples of the electronic device include computers, servers, workstations, laptops, smart phones, etc., but embodiments of the present invention are not limited thereto.

FIG. 12 is a diagram showing an electronic device according to an embodiment of the invention. As shown in FIG. 12, electronic device 1200 may include a processor (eg, central processor CPU) 1210 and memory 1220 , with memory 1220 connected to central processor 1210 . Among them, the memory 1220 can store various data, and can also store a program 1221 for information processing, and can execute the program 1221 under the control of the processor 1210 .

In some embodiments, the functionality of apparatus 1100 for extracting heartbeat data based on wireless radar signals may be implemented integrated in processor 1210 . Therein, the processor 1210 may be configured to perform the method of extracting heartbeat data based on wireless radar signals as described in the embodiments of the first aspect.

In some embodiments, the apparatus 1100 for extracting heartbeat data based on wireless radar signals may be arranged independently from the processor 1210, e.g., the apparatus for extracting heartbeat data based on wireless radar signals. 1100 may be configured as a chip connected to a processor 1210 and under the control of the processor 1210 may realize the functionality of the apparatus 1100 for extracting heartbeat data based on wireless radar signals.

For example, the processor 1210 may be configured to: acquire wireless radar signals sensing objects detected by radar; wherein the first frequency band is a predetermined frequency band related to the heartbeat to be detected; calculating the frequency spectrum energy of the data after bandpass filtering at each time accumulated by the time window; Performing feature extraction on data after bandpass filtering to obtain bandpass filtering feature data, and/or performing feature extraction on frequency spectral energy at each time in the second frequency band to obtain frequency spectral energy obtaining feature data; inputting the bandpass filtering feature data in the time domain and/or the frequency spectrum energy feature data in the frequency domain into a neural network-based detection model, and using the detection model to obtain waveform data after detection; and obtaining the heartbeat interval and heart rate of the detection target based on the waveform data.

In addition, as shown in FIG. 12, the electronic device 1200 may further include an input/output (I/O) device 1230, a display 1240, etc., of which the functions of these components are similar to those of the prior art, so here The detailed explanation is omitted here. Note that the electronic device 1200 does not need to include all the components shown in FIG. 12, and the electronic device 1200 may further include components not shown in FIG. 12, which can be referred to related art.

An embodiment of the present invention further provides a computer readable program, wherein, when executing the program in an electronic device, the program instructs a computer to perform the wireless communication according to the embodiment of the first aspect in the electronic device. Execute a method for extracting heartbeat data based on radar signals.

An embodiment of the present invention further provides a storage medium storing a computer readable program, wherein the computer readable program causes a computer to generate a heart rate signal based on the wireless radar signal according to the embodiment of the first aspect in an electronic device. Execute a method for extracting beat data.

In addition, the apparatus and methods described above may be implemented by software or hardware, or by a combination of hardware and software. The invention further relates to a computer readable program as follows, which, when executed by a logic component, causes the logic component to implement the device or component described above, or implement the various methods or steps described above. The logic component may be, for example, an FPGA (Field Programmable Gate Array), a microprocessor, a processor used in a computer, or the like. The present invention further relates to a storage medium storing the above program, such as a hard disk, magnetic disk, optical hard disk, DVD, flash memory, and the like.

Furthermore, one or more combinations of the functional blocks described in the figures and/or one or more combinations of the functional blocks can be combined into a general purpose processor, digital signal processing unit, etc. to perform the functions described herein. device (DSP), application specific integrated circuit (ASIC), field programmable gate array (FPGA) or other programmable logic component, discrete gate or transistor logic component, discrete hardware assembly or any other suitable combination It may be realized. Also, one or more combinations of the functional blocks and/or one or more combinations of the functional blocks described in the drawings may also be combined with computing devices, e.g., DSP and microprocessor combinations, multiple microprocessors. It may be configured as a processor, one or more microprocessors communicatively coupled with a DSP, or any other combination of components.

In addition, the following additional remarks are disclosed regarding the above-described embodiments and the like.

(Appendix 1)
A method for extracting heartbeat data based on wireless radar signals, comprising:
Collecting radio radar signals from detection targets detected by the radar;
band-pass filtering a first frequency band for the acquired wireless radar data, wherein the first frequency band is a predetermined frequency band for the heartbeat to be detected;
calculating the frequency spectrum energy at each instant accumulated by the time window of the data after bandpass filtering;
Performing feature extraction on data after bandpass filtering to obtain bandpass filtering feature data, and/or performing feature extraction on frequency spectral energy at each time in the second frequency band to obtain frequency spectral energy obtain feature data;
inputting the bandpass filtering feature data in the time domain and/or the frequency spectrum energy feature data in the frequency domain into a neural network-based detection model to obtain post-detection waveform data using the detection model; and the waveform. A method comprising obtaining a heartbeat interval and heart rate of the detection target based on data.

(Appendix 2)
The method of Appendix 1, wherein
Computing the frequency spectrum energy at each time accumulated by the time window of the data after bandpass filtering is
transforming the bandpass filtered data from a time domain signal to a frequency domain signal with a sliding time window to obtain the frequency spectrum energy at each time instant.

(Appendix 3)
The method of Appendix 2,
transforming the bandpass filtered data from a time domain signal to a frequency domain signal using a short-time Fourier transform (STFT) or wavelet transform.

(Appendix 4)
The method according to any one of Appendices 1 to 3,
The frequency spectrum energy feature data is
Data after calculating the sum of the frequency spectrum energy at each time of the partial frequency band within the second frequency band;
Data after obtaining the maximum value of frequency spectrum energy at each time in a partial frequency band within the second frequency band;
data after summing the frequency spectrum energy at each time of all frequency bands within said second frequency band; and the maximum value of frequency spectrum energy at each time of all frequency bands within said second frequency band post-asking data, including one or more of:

(Appendix 5)
The method according to any one of Appendices 1 to 4, further comprising:
A method comprising canceling a DC frequency offset for data after bandpass filtering.

(Appendix 6)
The method according to any one of Appendices 1 to 5,
The bandpass filtering feature data is
Data after performing bandpass filtering on quadrature signals of the radar data;
data after performing band-pass filtering on the in-phase signal of said radar data; and data after performing band-pass filtering on said radar data to obtain an amplitude value. including, method.

(Appendix 7)
The method according to any one of Appendices 1 to 6,
The method, wherein the wireless radar signal is sampled based on a first sampling rate.

(Appendix 8)
The method of Appendix 7, wherein
Before inputting feature data into the neural network-based detection model, the method further comprises:
resampling the feature data based on a second sampling rate, wherein the second sampling rate is lower than the first sampling rate; and normalizing the resampled feature data. ,Method.

(Appendix 9)
The method according to any one of Appendices 1 to 8,
The method of inputting feature data within one time period into the detection model, wherein the one time period is a predetermined time period (eg, 2s) that can be ensured to contain at least two R-wave information of a heartbeat.

(Appendix 10)
The method according to any one of Appendixes 1 to 9,
The method, wherein the waveform data output by the detection model is triangular wave data.

(Appendix 11)
The method of Appendix 10, further comprising:
adding noise to triangular wave data output from the detection model; and detecting a maximum value of triangular wave data after adding noise to obtain the heartbeat interval and heart rate of the detection target.

(Appendix 12)
11. The method of Supplementary Note 11,
Detecting the maximum value of triangular wave data after adding noise and obtaining the heartbeat interval and heart rate of the detection target include:
Correlating the maximum value with R-wave information of an electrocardiographic signal obtained from an electrocardiogram;
obtaining a heartbeat interval of the detectable target based on a time interval between adjacent maxima; and obtaining a heartbeat rate of the detectable target based on the obtained heartbeat interval of the detectable target.

(Appendix 13)
An electronic device comprising a memory and a processor,
A computer program is stored in the memory,
13. An electronic device, wherein the processor executes the computer program to perform the method of extracting heartbeat data based on wireless radar signals according to any one of Appendixes 1-12.

(Appendix 14)
A storage medium storing a computer readable program is provided;
A storage medium, wherein the computer-readable program causes a computer to perform, in an electronic device, the method of extracting heartbeat data based on wireless radar signals according to any one of Appendixes 1-12.

Although preferred embodiments of the present invention have been described above, the present invention is not limited to such embodiments, and all modifications to the present invention fall within the technical scope of the present invention as long as they do not depart from the spirit of the present invention.

Claims (10)

An apparatus for extracting heartbeat data based on wireless radar signals, comprising:
an acquisition unit for acquiring radio radar signals that have sensed objects to be detected by the radar;
a filtering unit for band-pass filtering a collected wireless radar signal in a first frequency band, said first frequency band being a predetermined frequency band for said heartbeat to be detected;
a computing unit for computing the frequency spectrum energy at each instant accumulated by the time window of the data after bandpass filtering;
performing feature extraction on the data after the bandpass filtering to obtain bandpass filtering feature data, and/or performing feature extraction on the frequency spectrum energy at each time point in the second frequency band to obtain the frequency spectrum a feature extraction unit for obtaining energy feature data;
a detection unit that inputs the bandpass filtering feature data in the time domain and/or the frequency spectrum energy feature data in the frequency domain into a neural network-based detection model, and obtains detected waveform data using the detection model; and An apparatus comprising an acquisition unit for obtaining a heartbeat interval and heart rate of the detection target based on the waveform data. A device according to claim 1, wherein
The apparatus, wherein the calculation unit converts the band-pass filtered data from a time domain signal to a frequency domain signal with a sliding time window to obtain the frequency spectrum energy at each time. A device according to claim 2, wherein
The apparatus of claim 1, wherein the computation unit transforms the bandpass filtered data from a time domain signal to a frequency domain signal using a short-time Fourier transform or a wavelet transform. A device according to claim 1, wherein
The frequency spectrum energy feature data is
Data after obtaining the sum of the frequency spectrum energy at each time of the partial frequency band within the second frequency band;
Data after obtaining the maximum value of frequency spectrum energy at each time in a partial frequency band within the second frequency band;
Data after calculating the sum of frequency spectrum energy at each time of all frequency bands within said second frequency band; and obtaining the maximum value of frequency spectrum energy at each time of all frequency bands within said second frequency band A device containing one or more of the following data. A device according to claim 1, wherein
The bandpass filtering feature data is
Data after performing bandpass filtering on quadrature signals of the radar data;
data after performing bandpass filtering on the in-phase signal of said radar data; and data after performing bandpass filtering on said radar data to obtain an amplitude. ,Device. A device according to claim 1, wherein
The collection unit samples at a first sampling rate, the device comprising:
a resampling unit for resampling the feature data at a second sampling rate before inputting the feature data into the neural network-based detection model, wherein the second sampling rate is higher than the first sampling rate; a resampling unit; and a normalization unit that performs normalization on the resampled feature data. A device according to claim 1, wherein
The apparatus, wherein the detection unit inputs feature data in one time period into the detection model, wherein the one time period is a predetermined time period that can be ensured to contain at least two R-wave information of a heartbeat. A device according to claim 1, wherein
The waveform data output from the detection model is triangular wave data,
The acquisition unit further comprises:
adding noise to the triangular wave data output from the detection model; and detecting a maximum value of the triangular wave data after adding noise to obtain the heartbeat interval and heart rate of the detection target. A device according to claim 8, wherein
The acquisition unit associates the maximum value with R-wave information of an electrocardiographic signal acquired from an electrocardiogram, acquires the heartbeat interval of the detection target based on the time interval between adjacent maximum values, and obtaining the heart rate of the detection target based on the acquired heartbeat interval of the detection target. A method for extracting heartbeat data based on wireless radar signals, comprising:
Collecting radio radar signals from detection targets detected by the radar;
performing bandpass filtering on the collected wireless radar data in a first frequency band, the first frequency band being a predetermined frequency band associated with the heartbeat to be detected;
calculating the frequency spectrum energy at each instant accumulated by the time window of the data after bandpass filtering;
performing feature extraction on the data after the bandpass filtering to obtain bandpass filtering feature data, and/or performing feature extraction on the frequency spectrum energy at each time point in the second frequency band to obtain the frequency spectrum obtaining energy signature data;
inputting the bandpass filtering feature data in the time domain and/or the frequency spectrum energy feature data in the frequency domain into a neural network-based detection model to obtain post-detection waveform data using the detection model; and the waveform. A method comprising obtaining a heartbeat interval and heart rate of the detection target based on data.

Images