Non-contact exercise physiological sensing method and exercise physiological sensing radar

Abstract

A non-contact motion physiological sensing method and a motion physiological sensing radar. The non-contact motion physiological sensing method is executed by a processor in a signal processing device, comprising: obtaining a digital signal; obtaining a phase map and a vibration frequency map according to the digital signal, the phase map presenting an energy distribution varying with a distance relative to the motion physiological sensing radar and a phase variation, and the vibration frequency map presenting an energy distribution varying with a distance relative to the motion physiological sensing radar and a vibration frequency variation; selecting at least one candidate position with an energy intensity exceeding an energy threshold from the vibration frequency map; selecting a target position from the candidate positions; obtaining one or more target phase data according to the target position; and inputting the one or more target phase data into a machine learning model to obtain a physiological parameter prediction result. The non-contact motion physiological sensing method and the motion physiological sensing radar can accurately sense the physiological parameter in the motion state of the subject in a non-contact manner.

Description

TECHNICAL FIELD

[0001] The present application relates to radar technology, and in particular, to a non-contact motion physiological sensing method and a motion physiological sensing radar. BACKGROUND

[0002] There are many wearable or direct contact physiological parameter measurement devices that can monitor physiological parameters (such as heart rate) in daily life activities. However, long-term wearing of wearable or contact devices can make the subject feel uncomfortable. Although there are non-contact measurement methods, when the subject is in a motion state, the shaking of the body can easily interfere with the measurement, affecting the measurement accuracy.

[0003] Therefore, there is a need to provide a non-contact motion physiological sensing method and a motion physiological sensing radar to solve the above problems. SUMMARY

[0004] Therefore, according to some embodiments, a non-contact motion physiological sensing method is performed by a processor in a signal processing device, comprising: obtaining a digital signal; obtaining a phase map and a vibration frequency map according to the digital signal, wherein the phase map presents an energy distribution varying with a phase change and a distance change relative to a motion physiological sensing radar, and the vibration frequency map presents an energy distribution varying with a vibration frequency change and a distance change relative to the motion physiological sensing radar; selecting at least one candidate position with an energy intensity exceeding an energy threshold from the vibration frequency map; selecting a target position from the at least one candidate position, the target position being one of the at least one candidate position with a vibration frequency meeting a physiological parameter range; obtaining one or more target phase data in a distance range in the phase map according to the target position; and inputting the one or more target phase data into a machine learning model to obtain a physiological parameter prediction result.

[0005] According to some embodiments, a motion physiological sensing radar includes a transmitting unit, a receiving unit, and a signal processing module. The transmitting unit transmits an incident radar signal. The receiving unit receives a reflected radar signal. The signal processing module obtains a digital signal corresponding to the reflected radar signal, obtains a phase map and a vibration frequency map from the digital signal, selects at least one candidate position from the vibration frequency map, selects a target position from the at least one candidate position, obtains one or more target phase data within a distance range in the phase map according to the target position, and inputs the one or more target phase data into a machine learning model to obtain a physiological parameter prediction result. The phase map presents an energy distribution varying with a distance relative to the motion physiological sensing radar and a phase variation. The vibration frequency map presents an energy distribution varying with a distance relative to the motion physiological sensing radar and a vibration frequency variation. The target position is a position having a vibration frequency conforming to a physiological parameter range and having a maximum energy intensity among the at least one candidate position.

[0006] According to some embodiments, the target position is a position having a vibration frequency conforming to a physiological parameter range and having a maximum energy intensity among the at least one candidate position.

[0007] According to some embodiments, an energy threshold value is calculated for each distance strip block in the phase map, and each phase energy value in each distance strip block is compared with the energy threshold value of the corresponding distance strip block to select at least one candidate position exceeding the energy threshold value.

[0008] According to some embodiments, the energy threshold value is determined according to an energy average value or an energy maximum value of the corresponding distance strip block.

[0009] According to some embodiments, the energy threshold value is a sum of a times the energy average value and b times the energy maximum value, where a+b=1, a and b are positive numbers.

[0010] According to some embodiments, the energy threshold value is a times the energy average value, where a is a positive number.

[0011] According to some embodiments, the vibration frequency map is obtained by performing a fast Fourier transform on the phase distribution of each distance in the phase map.

[0012] According to some embodiments, a fast Fourier transform is performed on the digital signal to obtain a distance profile map, where the distance profile map presents an energy distribution varying with a distance relative to the motion physiological sensing radar and a time variation.

[0013] According to some embodiments, a phase unwrapping is performed on the distance profile map to obtain the phase map.

[0014] According to some embodiments, the target phase data are a plurality of distance bins in the phase map, and one of the phase maps comprises the target position.

[0015] According to some embodiments, N target positions are selected from the at least one candidate position, and N is greater than 1, wherein the N target positions are the candidate positions having vibration frequencies within the physiological parameter range and having top N energy intensities; according to each target position, one or more target phase data in a corresponding distance range in the phase map are obtained; and the one or more target phase data of each target position are input into the machine learning model to obtain a physiological parameter prediction result corresponding to each target position.

[0016] In summary, according to some embodiments, the non-contact motion physiological sensing method and the motion physiological sensing radar can accurately sense physiological parameters in the motion state of the subject in a non-contact manner. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 A use state diagram of the motion physiological sensing radar according to some embodiments.

[0018] Figure 2 A diagram illustrating a radar signal.

[0019] Figure 3 A block diagram of the frequency-modulated continuous wave radar according to some embodiments.

[0020] Figure 4 A diagram illustrating an incident radar signal and a reflected radar signal.

[0021] Figure 5 A signal processing diagram according to some embodiments.

[0022] Figure 6 A flowchart of the non-contact motion physiological sensing method according to some embodiments.

[0023] Figure 7 A block diagram of the signal processing device according to some embodiments.

[0024] Figure 8A A gray-scale version of a raw color map of the distance topography map according to some embodiments.

[0025] Figure 8B A black-and-white version of a raw color map of the distance topography map according to some embodiments.

[0026] Figure 9A A gray-scale version of a raw color map of the phase map according to some embodiments.

[0027] Figure 9B A black-and-white version of a raw color map of a phase map according to some embodiments.

[0028] Figure 10A A gray-scale version of a raw color map of a vibration frequency map according to some embodiments.

[0029] Figure 10B A black-and-white version of a raw color map of a vibration frequency map according to some embodiments.

[0030] Figure 11 A vibration frequency distribution of distance bins according to some embodiments.

[0031] Figure 12 Loss-over-epoch plots of a machine learning model on a training set and a validation set according to some embodiments.

[0032] Figure 13 A physiological parameter prediction result plot according to some embodiments.

[0033] Figure 14 A Bland-Altman plot according to some embodiments.

[0034] Main component symbol explanation:

[0035] 10 kinematic physiological sensing radar

[0036] 10' frequency modulated continuous wave radar

[0037] 11 transmitting unit

[0038] 12 receiving unit

[0039] 13 demodulating unit

[0040] 14 analog-to-digital converter

[0041] 15 processing unit

[0042] 16 signal processing module

[0043] 60 signal processing device

[0044] 61 processor

[0045] 62 storage device

[0046] 63 program

[0047] 64 machine learning model

[0048] 90 target

[0049] FH incident radar signal

[0050] FN reflected radar signal

[0051] A1, A2 two-dimensional matrix

[0052] B pulse bandwidth

[0053] Ct, Cr chirped pulse

[0054] C1, C2, C3, Cn chirped pulse

[0055] D1, D2, Dn sequence

[0056] P1, P2, Pn frequency domain signal

[0057] S slope

[0058] SC chirped pulse

[0059] SD digital signal

[0060] SI intermediate frequency signal

[0061] SP frequency domain signal

[0062] S700-S714 steps

[0063] Tc duration time

[0064] τ delay time

[0065] V th energy threshold DETAILED DESCRIPTION

[0066] With respect to the term "connected" as used herein, this means that two or more elements are either in direct physical or electrical contact with one another or that two or more elements are not in direct contact with one another, but that they are still in contact in another way such as by way of a thermal or electrical conductor.

[0067] Referring to Figure 1 , Figure 1 is a schematic diagram of a use case of a motion physiologic sensing radar 10 according to some embodiments. The motion physiologic sensing radar 10 emits a radar signal (hereinafter referred to as "incident radar signal FH"). The incident radar signal FH is emitted towards a target 90 and is modulated by the motion of the target 90 (e.g., a subject) and reflected back to the motion physiologic sensing radar 10, hereinafter referred to as "reflected radar signal FN". The reflected radar signal FN can then be analyzed to detect one or more information of the target 90. The information can be, for example, velocity, distance, orientation, physiologic information (e.g., heart beat, respiration), etc.

[0068] In some embodiments, the motion physiology sensing radar 10 can be a frequency modulated continuous wave (FMCW) radar, a continuous wave (CW) radar, or an ultra-wideband (UWB) radar. The following will be described by way of example of a frequency modulated continuous wave radar.

[0069] With reference to Figure 2 , Figure 2 For the purpose of illustrating the radar signal, the upper half presents the amplitude of the incident radar signal FH versus time, and the lower half presents the frequency of the incident radar signal FH versus time. The incident radar signal FH comprises a plurality of chirp pulses SC. For the sake of clarity of the drawing, Figure 2 Only one chirp pulse SC is presented. Here, the chirp pulse SC is a linear frequency modulation pulse signal, which refers to a sinusoidal wave whose frequency increases linearly with time. In some embodiments, the frequency of the chirp pulse SC is increased in a non-linear manner. For the sake of convenience of description, the following will be described in a linear manner. As shown in Figure 2 The chirp pulse SC increases linearly from a start frequency (e.g., 77 GHz) to an end frequency (e.g., 81 GHz) according to a slope S within a time duration Tc (e.g., 40 microseconds). The start frequency and the end frequency can be selected from the millimeter wave band (i.e., 30 GHz to 300 GHz). The difference between the start frequency and the end frequency is the pulse bandwidth B.

[0070] With reference to Figure 3 and Figure 4 . Figure 3 For the purpose of illustrating the radar signal, the upper half presents the amplitude of the incident radar signal FH versus time, and the lower half presents the frequency of the incident radar signal FH versus time. The incident radar signal FH comprises a plurality of chirp pulses SC. For the sake of clarity of the drawing, Figure 2 Only one chirp pulse SC is presented. Here, the chirp pulse SC is a linear frequency modulation pulse signal, which refers to a sinusoidal wave whose frequency increases linearly with time. In some embodiments, the frequency of the chirp pulse SC is increased in a non-linear manner. For the sake of convenience of description, the following will be described in a linear manner. As shown in Figure 2 The chirp pulse SC increases linearly from a start frequency (e.g., 77 GHz) to an end frequency (e.g., 81 GHz) according to a slope S within a time duration Tc (e.g., 40 microseconds). The start frequency and the end frequency can be selected from the millimeter wave band (i.e., 30 GHz to 300 GHz). The difference between the start frequency and the end frequency is the pulse bandwidth B.

[0070] With reference to Figure 3 and Figure 4 . Figure 3 For the purpose of illustrating the radar signal, the upper half presents the amplitude of the incident radar signal FH versus time, and the lower half presents the frequency of the incident radar signal FH versus time. The incident radar signal FH comprises a plurality of chirp pulses SC. For the sake of clarity of the drawing, Figure 4To illustrate the incident radar signal FH and the reflected radar signal FN, a schematic diagram is shown. The frequency-modulated continuous-wave radar 10' comprises a transmitting unit 11, a receiving unit 12, a demodulating unit 13, an analog-to-digital converter 14, and a processing unit 15. The transmitting unit 11 is configured to transmit the incident radar signal FH, including a transmitting antenna and a signal synthesizer. The signal synthesizer is configured to generate the incident radar signal FH including a chirp pulse Ct and transmit via the transmitting antenna. The receiving unit 12 includes a receiving antenna configured to receive the reflected radar signal FN including at least one chirp pulse Cr. The chirp pulse Cr can be considered as a delayed version of the chirp pulse Ct. The demodulating unit 13, the analog-to-digital converter 14, and the processing unit 15 are configured to process the received reflected radar signal FN, which can be collectively referred to as a signal processing module 16. The demodulating unit 13 is connected to the transmitting unit 11 and the receiving unit 12, including a mixer and a low-pass filter. The mixer couples the chirp pulse Ct of the incident radar signal FH and the chirp pulse Cr corresponding to the reflected radar signal FN, which can generate two coupled signals of the sum and the difference of the frequencies of the two chirp pulses Ct, Cr. The low-pass filter low-pass filters the coupled signals to remove high-frequency components to obtain the coupled signal of the difference of the frequencies of the two chirp pulses Ct, Cr, which is hereinafter referred to as an intermediate frequency signal SI. The analog-to-digital converter 14 converts the intermediate frequency signal SI into a digital signal SD (e.g. as shown in FIG. 2). The processing unit 15 performs digital signal processing on the digital signal SD. The processing unit 15 can be, for example, a central processing unit (CPU), a graphics processing unit (GPU), or other programmable general-purpose or special-purpose microprocessors, digital signal processors (DSPs), programmable controllers, application specific integrated circuits (ASICs), programmable logic devices (PLDs), or other similar devices, chips, integrated circuits, and combinations thereof. Figure 5 The frequency f0 of the intermediate frequency signal SI can be represented as Equation 1, where S is the slope and τ is the delay time between transmitting the incident radar signal FH and receiving the reflected radar signal FN. Thus, τ can be represented as Equation 2, where d is the distance between the radar transmitting antenna and the target 90, and c is the speed of light. Substituting Equation 2 into Equation 1 can obtain Equation 3. From Equation 3, it can be known that the frequency f0 of the intermediate frequency signal SI implicitly contains the distance information (i.e. the distance between the frequency-modulated continuous-wave radar 10' and the target 90).

[0071] Referring to Figure 4 The frequency f0 of the intermediate frequency signal SI can be represented as Equation 1, where S is the slope and τ is the delay time between transmitting the incident radar signal FH and receiving the reflected radar signal FN. Thus, τ can be represented as Equation 2, where d is the distance between the radar transmitting antenna and the target 90, and c is the speed of light. Substituting Equation 2 into Equation 1 can obtain Equation 3. From Equation 3, it can be known that the frequency f0 of the intermediate frequency signal SI implicitly contains the distance information (i.e. the distance between the frequency-modulated continuous-wave radar 10' and the target 90).

[0072] f0 = S - τ... Equation 1

[0073] τ = 2d / c... Equation 2

[0074] f0 = 2Sd / c... Equation 3

[0075] Referring to Figure 5 , Figure 5 is a schematic diagram of signal processing according to some embodiments. Here, the chirp pulses SC are sequentially numbered as C1, C2, C3,..., Cn, n being a positive integer. The analog-to-digital converter 14 converts the received intermediate frequency signals SI corresponding to each chirp pulse C1~Cn into digital signals SD (represented as sequences D1, D2,..., Dn, n being a positive integer, respectively), each chirp pulse Cx (x = 1~n) having a corresponding sequence Dx (x = 1~n). The sequences Dx (x = 1~n) of the digital signals SD can be represented as one-dimensional arrays (row matrices) respectively. Sequentially arranging these row arrays Dx (x = 1~n) vertically can form a two-dimensional matrix A1. It is understood that the digital signals SD can also be arranged as column arrays, and sequentially arranging these column arrays horizontally can also form another two-dimensional matrix. The values of the two-dimensional matrix A1 represent signal intensity (amplitude). The index value x of the column of the two-dimensional matrix A1 corresponds to the order of the chirp pulse SC. The index value of the row of the two-dimensional matrix A1 has the meaning of time, i.e., the row array of the two-dimensional matrix A1 is a time-domain signal (a set of digital data related to time).

[0076] The processing unit 15 performs a Fast Fourier Transform (FFT) (hereinafter referred to as "range Fourier transform") on each row matrix of the two-dimensional matrix A1 (i.e., the two-dimensional matrix A1 formed by the digital signals SD) to obtain a frequency-domain signal SP (represented as P1, P2,..., Pn, n being a positive integer, respectively), i.e., a two-dimensional matrix A2. Therefore, the row matrix of the two-dimensional matrix A2 corresponds to the frequency spectrum distribution in response to a chirp pulse Cx. As mentioned above, the frequency of the intermediate frequency signal SI implies distance information. That is, the index value of the row of the two-dimensional matrix A2 has the meaning of distance. The values of the two-dimensional matrix A2 represent the intensity of each frequency on the frequency spectrum, which can represent the intensity of the radar signal reflected by targets 90 at different distances from the frequency modulated continuous wave radar 10'. As shown in FIG. 2, the filled boxes in the two-dimensional matrix A2 are at peak values (i.e., the values exceed a threshold value), indicating that there are targets 90 at the corresponding distances at these frequencies. The distance between the frequency modulated continuous wave radar 10' and the target 90 can be calculated from the frequency at the peak value. Further, the large-scale motion information (such as average speed) can be calculated from the distance changes of a particular target 90 calculated at different time points. Figure 5

[0077] ​The above is described by way of example with the transmitting unit 11 having one transmitting antenna and the receiving unit 12 having one receiving antenna. However, in some embodiments, the transmitting unit 11 has multiple transmitting antennas to transmit multiple incident radar signals FH. Similarly, in some embodiments, the receiving unit 12 has multiple receiving antennas.

[0078] By way of reference Figure 6 And Figure 7 . Figure 6 A flowchart of a non-contact motion physiological sensing method according to some embodiments. Figure 7 A block diagram of a signal processing device 60 according to some embodiments. The signal processing device 60 comprises a processor 61 and a storage device 62. The storage device 62 is a computer readable storage medium storing a program 63 for execution by the processor 61 to perform a non-contact motion physiological sensing method. In some embodiments, the signal processing device 60 is the aforementioned frequency modulated continuous wave radar 10', and the processor 61 is the aforementioned processing unit 15. In some embodiments, the signal processing device 60 is an edge device or a cloud server, i.e. after the frequency modulated continuous wave radar 10' obtains the digital signal SD, the digital signal SD is transmitted to the edge device or the cloud server for digital signal processing.

[0079] As mentioned above, the analog-to-digital converter 14 can convert the received intermediate frequency signals SI corresponding to each chirp pulse Cx into a digital signal SD, so that the processor 61 can obtain a digital signal SD corresponding to the reflected radar signal FN (step S700). Then, the aforementioned range Fourier transform (step S701) can be performed to obtain a range profile map (step S702). As shown in Figure 8A , Figure 8A A gray scale version of a raw color map of a range profile map according to some embodiments. As shown in Figure 8B , Figure 8B A black and white version of a raw color map of a range profile map according to some embodiments. The range profile map presents the energy distribution as a function of distance (horizontal axis) and time (vertical axis) relative to the frequency modulated continuous wave radar 10', where the energy difference is represented by the color depth.

[0080] According to the digital signal SD, in addition to obtaining the range profile map, a phase map and a vibration frequency map can also be further obtained. In step S703, the range profile map is subjected to phase unwrapping to obtain a phase map (step S704). As shown in Figure 9A , Figure 9A A gray scale version of a raw color map of a phase map according to some embodiments. As shown in Figure 9B ,Figure 9B This is a black-and-white schematic diagram of the original color image of the phase map according to some embodiments. The phase map presents the energy distribution as a function of distance (horizontal axis) and phase (vertical axis) relative to the frequency-modulated continuous wave radar 10', with energy differences represented by color depth. Next, in step S705, a fast Fourier transform is performed on the phase distribution (i.e., range bins) at each distance of the phase map to obtain the vibration frequency map (step S706). Figure 10A As shown, Figure 10A This is a grayscale illustration of the original color image of a vibration frequency map based on some embodiments. For example... Figure 10B As shown, Figure 10B This is a black-and-white schematic diagram of the original color image of a vibration frequency map according to some embodiments. The vibration frequency map shows the energy distribution as a function of distance (horizontal axis) and vibration frequency (vertical axis) relative to the frequency-modulated continuous wave radar 10', with energy differences represented by shades of color.

[0081] After obtaining the vibration frequency map, in step S707, frequencies with energy intensities exceeding an energy threshold V are selected from the vibration frequency map. th At least one candidate position (step S708). For example... Figure 11 As shown, Figure 11 This is a schematic diagram of the vibration frequency distribution of a distance strip block according to some embodiments. Figure 11 The middle presents an energy threshold V that exceeds the threshold. th The peak of the wave, therefore, the distance bar is selected as a candidate location. In other words, step S707 is to match each distance bar in the phase map with the energy threshold V. th Comparison, if it exceeds the energy threshold V th If the distance is specified, the corresponding distance bar will be selected as the candidate position.

[0082] In some embodiments, the energy threshold V th The threshold is a floating threshold. For each distance bar in the phase map, an individual energy threshold V is calculated. th Energy threshold V th It is determined based on the average or maximum energy of the corresponding distance strip. For example, the energy threshold is the sum of a times the average energy and b times the maximum energy, where a + b = 1, and a and b are positive numbers. Another example is that the energy threshold is a times the average energy, where a is a positive number.

[0083] The candidate positions obtained in the foregoing step S708 can be multiple, and thus it is necessary to further determine which one should be selected to exclude the interference signal. In step S709, one or more are selected from the candidate positions to obtain one or more target positions (step S710). The target position is one of the candidate positions having a vibration frequency that meets a physiological parameter range. The physiological parameter range can be, for example, a breathing frequency range (such as 10-20 times per minute), a heartbeat frequency range (such as 60-100 times per minute), and the like.

[0084] Specifically, in some embodiments, it is detected that there is one target 90 in the field. Each candidate position having a vibration frequency that meets a physiological parameter range is found, and the one having the largest energy size of the oscillation frequency range is selected. The selected candidate position (distance) is the position of the target 90 (i.e., the target position).

[0085] In some embodiments, it is detected that there are multiple targets 90 in the field. N target positions to be detected are selected from the candidate positions and N is greater than 1, wherein the N target positions to be detected are the ones having a vibration frequency that meets a physiological parameter range and having the top N largest energy intensities among the candidate positions. The target positions to be detected are the positions of the targets 90 (i.e., the target positions).

[0086] After the position of one or more targets is determined, the corresponding one or more target phase data (or target phase data to be detected) can be obtained (step S711). Considering that the detection of an object in motion can be biased due to possible misjudgment. In step S711, according to each target position, a target phase data in the corresponding distance range in the phase map is obtained (step S712). In some embodiments, according to each target position, a target phase data in the corresponding distance range in the phase map is obtained, wherein the target phase data includes the distance strip block of the target position. In other embodiments, according to each target position, multiple target phase data in the corresponding distance range in the phase map are obtained, wherein the target phase data includes the distance strip block of the target position and one or more distance strip blocks adjacent to the target position. For example, taking the distance strip block of the target position as the center, two distance strip blocks are taken on both sides, and the target phase data includes five distance strip blocks.

[0087] In step S713, the target phase data of each target position to be detected is input into the machine learning model 64 to obtain a physiological parameter prediction result (step S714). For example, the breathing frequency or the heartbeat frequency is predicted. In some embodiments, the target phase data is normalized before being input into the machine learning model 64.

[0088] In an embodiment, the machine learning model 64 adopts a MobileNetV3 model. The training samples are collected from the usage data of three kinds of exercise equipment (treadmill, elliptical machine, and treadmill). Each kind of exercise equipment collects 30 radar data, a total of 90 radar data. Each piece of radar data includes data of four exercise intensities (rest, slow, medium, and fast), and each exercise intensity lasts for two minutes. The frequency-modulated continuous wave radar 10 is installed at a height of 1.6-2 meters, and the distance from the subject is 0.7-0.9 meters. During the collection process, the subject wears a heart rate monitor to synchronously obtain real-time heart rate as a labeled sample. Referring to Figure 12 , Figure 12 The loss-epoch variation diagram of the machine learning model 64 according to some embodiments on the training set and the validation set can be seen to have good convergence of the model. The total training data is 7047, the learning rate is 10 -5 , the training loss is 397.11, and the test loss is 201.52. Referring to Figure 13 , Figure 13 The physiological parameter prediction result diagram according to some embodiments is shown. The accuracy is 91.23%, the root mean square error is 14.2 (times / min, bpm), and the standard error is 8.15 (times / min, bpm). Referring to Figure 14 , Figure 14 The Bland-Altman diagram according to some embodiments is shown to compare the consistency of the predicted heart rate with the actual heart rate monitor measured value. The sample number is 723. The maximum difference is 27.84, the minimum difference is -25.69, the average difference is -2.3, the standard deviation is 11.98, and the 95% confidence interval of the regression line is [-27.78, 21.19]. It can be seen that the predicted heart rate is close to the actual heart rate monitor measured value.

[0089] The above-mentioned non-contact exercise physiological sensing method is in a sliding window manner to obtain the digital signal SD and perform processing. In some embodiments, the window size is 10 seconds, and the time step is one second.

[0090] In summary, the non-contact exercise physiological sensing method and the exercise physiological sensing radar 10 according to some embodiments can accurately sense the physiological parameters of the subject in the exercise state in a non-contact manner.

Claims

1. A non-contact motion physiological sensing method performed by a processor in a signal processing device, the non-contact motion physiological sensing method comprising: obtaining a digital signal; obtaining a phase map and a vibration frequency map from the digital signal, wherein the phase map presents an energy distribution varying with a phase and a distance relative to a motion physiological sensing radar, and the vibration frequency map presents an energy distribution varying with a vibration frequency and a distance relative to the motion physiological sensing radar; selecting at least one candidate position having an energy intensity exceeding an energy threshold from the vibration frequency map; selecting a target position from the at least one candidate position, the target position being one of the at least one candidate position having a vibration frequency meeting a physiological parameter range; obtaining one or more target phase data in a distance range of the target position from the phase map; and inputting the one or more target phase data into a machine learning model to obtain a physiological parameter prediction result; wherein the step of selecting the at least one candidate position from the vibration frequency map comprises: calculating the energy threshold for each distance strip of the phase map respectively; and comparing an energy value of each phase on each distance strip with the energy threshold corresponding to the distance strip to select the at least one candidate position exceeding the energy threshold from the vibration frequency map. 2.The non-contact motion physiological sensing method of claim 1, wherein the digital signal corresponds to a reflected radar signal of the motion physiological sensing radar; and wherein the target position is one of the at least one candidate position having the vibration frequency meeting the physiological parameter range and having a maximum energy intensity. 3.The non-contact motion physiological sensing method of claim 1, wherein the energy threshold is determined according to an energy average value or an energy maximum value of the corresponding distance strip. 4.The non-contact motion physiological sensing method of claim 3, wherein the energy threshold is a sum of a times the energy average value and b times the energy maximum value, where a+b=1, a and b are positive numbers. 5.The non-contact motion physiological sensing method of claim 3, wherein the energy threshold is a times the energy average value, where a is a positive number. 6.The non-contact motion physiological sensing method of claim 1, wherein the vibration frequency map is obtained by performing a fast Fourier transform on a phase distribution of each distance of the phase map. 7.The non-contact motion physiological sensing method of claim 1, wherein the step of obtaining the phase map comprises: performing a fast Fourier transform on the digital signal to obtain a distance profile map, wherein the distance profile map presents an energy distribution varying with a time and a distance relative to the motion physiological sensing radar. 8.The non-contact motion physiological sensing method of claim 7, wherein the step of obtaining the phase map further comprises: performing a phase unwrapping on the distance profile map to obtain the phase map. ​ 9. The non-contact physiological sensing method of claim 1, wherein the one or more target phase data are a plurality of distance bins in the phase map, one of the plurality of distance bins is the distance bin of the target position.

10. The non-contact physiological sensing method of claim 1, further comprising: selecting N target positions from the at least one candidate position, where N is greater than 1, wherein the N target positions are the N top energy intensity of the at least one candidate position having the vibration frequency within the physiological parameter range; for each of the target positions, obtaining one or more target phase data within a corresponding distance range in the phase map; and inputting the one or more target phase data of each of the target positions into the machine learning model to obtain a physiological parameter prediction result corresponding to each of the target positions.

11. A physiological sensing radar, comprising: a transmitting unit configured to transmit an incident radar signal; a receiving unit configured to receive a reflected radar signal; and a signal processing module configured to obtain a digital signal corresponding to the reflected radar signal, obtain a phase map and a vibration frequency map from the digital signal, select at least one candidate position from the vibration frequency map, select a target position from the at least one candidate position, obtain one or more target phase data within a distance range in the phase map from the target position, and input the one or more target phase data into a machine learning model to obtain a physiological parameter prediction result; wherein the phase map presents an energy distribution varying with respect to distance and phase, the vibration frequency map presents an energy distribution varying with respect to distance and vibration frequency, the target position is the one of the at least one candidate position having a vibration frequency within a physiological parameter range and having a maximum energy intensity; wherein the signal processing module calculates an energy threshold for each distance bin in the phase map, and compares an energy value of each phase in each distance bin with the energy threshold corresponding to the distance bin to select the at least one candidate position exceeding the energy threshold from the vibration frequency map.

12. The physiological sensing radar of claim 11, wherein the energy threshold is determined according to an energy average value or an energy maximum value of the corresponding distance bin.

13. The physiological sensing radar of claim 12, wherein the energy threshold is a sum of a times the energy average value and b times the energy maximum value, where a+b=1, a and b are positive numbers.

14. The physiological sensing radar of claim 12, wherein the energy threshold is a times the energy average value, where a is a positive number. ​ 15. The kinesiological sensing radar of claim 11, wherein the vibration frequency map is obtained by performing a fast Fourier transform on a phase profile at each range of the phase map.

16. The kinesiological sensing radar of claim 11, wherein the signal processing module performs a fast Fourier transform on the digital signal to obtain a range profile map, wherein the range profile map presents an energy distribution varying with respect to a range and time from the kinesiological sensing radar.

17. The kinesiological sensing radar of claim 16, wherein the signal processing module performs a phase unwrapping on the range profile map to obtain the phase map.

18. The kinesiological sensing radar of claim 11, wherein the one or more target phase data are a plurality of range bins in the phase map, one of the plurality of range bins is the range bin of the target location.

19. The kinesiological sensing radar of claim 11, wherein: the signal processing module selects N target locations to be tested from the at least one candidate location, and N is greater than 1, wherein the N target locations to be tested are the N range bins with the vibration frequency meeting the physiological parameter range and with the top N energy intensities from the at least one candidate location; the signal processing module obtains one or more target phase data in a corresponding range of the phase map for each of the target locations to be tested; and the signal processing module inputs the one or more target phase data of each of the target locations to be tested into the machine learning model to obtain a physiological parameter prediction result corresponding to each of the target locations to be tested. ​

Images