A human vital sign information robust extraction method and related device

Abstract

This invention discloses a robust method and related equipment for extracting human vital signs information. The method includes: acquiring raw radio frequency data collected by radar; preprocessing the raw radio frequency data to obtain a target data matrix; identifying multipath channels based on the target data matrix; performing parallel phase demodulation on each identified multipath channel to obtain a phase sequence for each multipath channel; performing coherence analysis on the signal components of all phase sequences; adaptively weighting and fusing the phase sequences based on the results of the coherence analysis to obtain a fused vital sign signal; and performing vital sign analysis based on the fused vital sign signal to obtain the target vital sign information. This invention effectively solves the key technical problem of weak vital sign signals being submerged in complex environments through the collaborative processing of multipath signals, enabling high-precision and highly robust non-contact vital sign monitoring, and can be widely applied in the field of radar vital sign monitoring technology.

Description

Technical Field

[0001] This invention relates to the field of radar vital sign monitoring technology, and in particular to a robust method and related equipment for extracting human vital sign information. Background Technology

[0002] With the increasing global trend of population aging and the growing demand for high-quality healthy living, continuous and non-contact vital sign monitoring technology has become a core research direction in the fields of smart healthcare and home-based elderly care. Traditional vital sign monitoring, such as electrocardiogram (ECG) and pulse oximetry measurement, heavily relies on contact sensors and electrode patches. While these technologies can provide accurate physiological parameters, their wired connection or close-fitting wearing methods bring many inconveniences to users, such as restricting daily activities and causing skin discomfort. They are particularly unsuitable for scenarios requiring long-term, uninterrupted monitoring (such as sleep monitoring) or special populations (such as burn patients and newborns). Therefore, developing non-contact vital sign monitoring technology to achieve non-contact, accurate, and all-weather monitoring of key indicators such as respiration and heartbeat has significant social value and broad application prospects.

[0003] Among numerous non-contact sensing technologies, ultra-wideband (IR-UWB) radar technology has attracted considerable attention due to its unique advantages. IR-UWB radar achieves centimeter-level or even millimeter-level high range resolution by emitting extremely narrow electromagnetic pulses on the nanosecond scale. It possesses strong penetrating power (capable of penetrating clothing, bedding, and other obstructions), and consumes very little power, posing no harm to the human body. These characteristics make it an ideal technology choice for achieving high-precision vital sign monitoring without infringing on user privacy.

[0004] Currently, research on vital sign extraction technology based on IR-UWB radar has made some progress. Existing technical solutions typically model the human body as a single reflecting point and assume that the radar signal propagates along a single line-of-sight path. Under this ideal model, vital sign information can be extracted by analyzing the phase or time delay changes modulated by the minute displacements of the human chest cavity caused by breathing and heartbeat in the radar echo signal. However, these solutions have a fundamental limitation: they heavily rely on an ideal electromagnetic wave propagation environment and ignore the multipath effect that is prevalent and unavoidable in radar's primary application scenarios, such as indoor environments (e.g., bedrooms, hospital wards). In real indoor environments, the pulse signal emitted by the radar will be reflected multiple times by surfaces such as walls, ceilings, and furniture, forming multiple propagation paths to the receiver. These multipath components interfere with the direct echo signal from the human body, causing severe amplitude and phase distortions in the received signal and a sharp drop in the signal-to-noise ratio. Especially for the extremely weak heartbeat signal, multipath interference often completely overwhelms it, causing the performance of algorithms based on the single-path assumption to degrade severely or even fail completely. Some existing technologies attempt to circumvent this problem by selecting only the most energetic echo path for analysis. However, this essentially discards useful information from other paths as noise, resulting in a waste of information resources. Furthermore, when the quality of the main path signal is poor due to changes in human posture or occlusion, the robustness of the entire system cannot be guaranteed. Summary of the Invention

[0005] The main objective of this invention is to provide a robust method, apparatus, electronic device, storage medium, and program product for extracting human vital signs information, aiming to solve at least one problem in the prior art.

[0006] To achieve the above objectives, one aspect of this invention proposes a robust method for extracting human vital signs information, the method comprising: The raw radio frequency data collected by the radar is acquired, and the raw radio frequency data is preprocessed to obtain the target data matrix; Multipath channel identification is performed based on the target data matrix, and parallel phase demodulation is performed on each identified multipath channel to obtain the phase sequence of each multipath channel; Coherence analysis of signal components is performed on all phase sequences. Based on the results of the coherence analysis, the phase sequences are adaptively weighted and fused to obtain fused vital sign signals. Vital signs analysis is performed based on fused vital sign signals to obtain target vital sign information.

[0007] In some implementations, the raw radio frequency data is preprocessed to obtain the target data matrix, including the following steps: Based on the raw radio frequency echo data continuously acquired by the radar system in the slow time dimension, a two-dimensional fast-time-slow-time data matrix is ​​constructed. Based on a two-dimensional fast-time-slow-time data matrix, static clutter filtering is performed by coherently accumulating the slow-time dimension to obtain a filtered data matrix. Based on the filtered data matrix, a bandpass filter is applied in the fast time dimension, and then the effective signal energy in the band is retained according to the spectral characteristics of the radar transmitted signal to complete the out-of-band noise suppression operation, thus obtaining a target data matrix containing dynamic target information.

[0008] In some implementations, multipath channel identification based on the target data matrix includes the following steps: The energy of the target data matrix is ​​accumulated along the slow time dimension to obtain a one-dimensional fast time energy distribution profile; Multiple multipath channels are obtained by identifying local energy maxima points on the fast-time energy distribution profile.

[0009] In some implementations, parallel phase demodulation is performed on each identified multipath channel to obtain the phase sequence of each multipath channel, including the following steps: The fast time local information of each multipath channel is converted to the frequency domain by Fourier transform, and then the phase information at the radar carrier frequency is demodulated. Based on the phase information of each multipath channel, the phase sequence of the corresponding multipath channel is continuously integrated in the slow time dimension.

[0010] In some implementations, the fast-time local information of each multipath channel is converted to the frequency domain using Fourier transform, and then the phase information at the radar's carrier frequency is demodulated. This includes the following steps: The signal within a fast time window near the static delay corresponding to the multipath channel is taken as the signal segment within the window. A fast time Fourier transform is performed on the signal segment within the window to obtain the spectrum. The complex value of the spectrum is obtained by extracting the spectral components at the carrier frequency based on the spectrum. Phase information is obtained by converting the phase angle corresponding to the complex value of the spectrum.

[0011] In some implementations, the results of coherence analysis include the coherence of each phase sequence with a pre-selected reference path. Based on the results of the coherence analysis, the phase sequences are adaptively weighted and fused to obtain a fused vital sign signal, including the following steps: Weights are assigned to phase sequences based on coherence, and the weights are positively correlated with coherence. The phase sequences are weighted and fused based on the weights corresponding to each phase sequence to obtain the fused vital sign signal.

[0012] In some implementations, the target vital signs information includes respiratory rate and heart rate. Vital signs analysis based on fused vital sign signals is used to obtain the target vital signs information, including the following steps: Based on the fusion of vital sign signals, respiratory rate is extracted through linear filtering and spectral analysis; Based on the fusion of vital signs signals, the heart rate is extracted through variational mode decomposition.

[0013] In some implementations, the heart rate is extracted through variational mode decomposition based on fused vital sign signals, including the following steps: Taking the set of discrete sub-signals with sparse properties as the decomposition target, the decomposition problem of fused vital sign signals is transformed into a constrained variational problem; The constrained variational problem is transformed into a saddle point finding problem by using a quadratic penalty term and Lagrange multipliers. The saddle point finding problem is solved by the alternating direction multiplier method, and multiple eigenmode functions are obtained. A modality selection strategy based on prior knowledge filters out heartbeat modes from intrinsic modal functions; The heartbeat signal is obtained by linearly superimposing all heartbeat modes; The heartbeat frequency is extracted by performing spectral analysis on the heartbeat signal.

[0014] To achieve the above objectives, another aspect of the present invention provides a robust extraction device for human vital signs information, the device comprising: The data acquisition module is used to acquire the raw radio frequency data collected by the radar, perform data preprocessing on the raw radio frequency data, and obtain the target data matrix. The phase demodulation module is used to identify multipath channels based on the target data matrix, and to perform parallel phase demodulation on each identified multipath channel to obtain the phase sequence of each multipath channel. The signal fusion module is used to perform coherence analysis on the signal components of all phase sequences, and to perform adaptive weighted fusion on the phase sequences based on the results of the coherence analysis to obtain fused vital sign signals. The vital sign extraction module is used to perform vital sign analysis based on fused vital sign signals to obtain target vital sign information.

[0015] To achieve the above objectives, another aspect of the present invention provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the aforementioned method.

[0016] To achieve the above objectives, another aspect of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method.

[0017] To achieve the above objectives, another aspect of the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the aforementioned method.

[0018] The embodiments of the present invention include at least the following beneficial effects: The present invention provides a robust extraction method, apparatus, electronic device, storage medium, and program product for human vital signs information. This scheme acquires raw radio frequency data collected by radar, performs data preprocessing on the raw radio frequency data to obtain a target data matrix; performs multipath channel identification based on the target data matrix, performs parallel phase demodulation on each identified multipath channel to obtain a phase sequence of each multipath channel; performs coherence analysis on the signal components of all phase sequences, and performs adaptive weighted fusion on the phase sequences based on the results of the coherence analysis to obtain a fused vital sign signal; and performs vital sign analysis based on the fused vital sign signal to obtain the target vital sign information. This invention overcomes the limitations of traditional single-path models by identifying and utilizing multiple effective multipath channels, transforming multipath components traditionally considered interference into useful signal sources, thus achieving effective capture and enhancement of vital sign information. Through coherence analysis and adaptive weighted fusion of signals from each path, it can optimize and integrate path components with high signal-to-noise ratio and good signal quality, dynamically suppressing interference from unreliable paths. Furthermore, this invention abandons the simplistic strategy of "selecting the main path and discarding the rest," fully exploring and utilizing the scattered information resources in radar echoes, avoiding information waste, and achieving superior monitoring performance under the same hardware conditions. Specifically, the core contribution of this invention lies in shifting the processing approach from "avoiding multipath" to "utilizing multipath." Through the collaborative processing of multipath signals, it effectively solves the key technical problem of weak vital sign signals being submerged in complex environments, enabling high-precision and highly robust non-contact vital sign monitoring. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of an implementation environment for the robust extraction method of human vital signs information provided in this embodiment of the invention; Figure 2 This is a flowchart illustrating a robust method for extracting human vital signs information provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of the overall data processing flow of the robust extraction method for human vital signs information provided in this embodiment of the invention; Figure 4 This is a schematic diagram of the comparative experiment provided in an embodiment of the present invention; Figure 5 This is a schematic diagram illustrating an example of multipath channel identification and processing intervals provided in an embodiment of the present invention; Figure 6 This is a schematic diagram illustrating a comparison of respiratory and heart rate extraction results provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of the robust extraction device for human vital signs information provided in an embodiment of the present invention; Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of this invention; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this invention as detailed in the appended claims.

[0021] It is understood that the terms “first,” “second,” etc., used in this invention may be used herein to describe various concepts, but unless specifically stated otherwise, these concepts are not limited by these terms. These terms are used only to distinguish one concept from another. For example, first information may also be referred to as second information without departing from the scope of embodiments of the invention, and similarly, second information may also be referred to as first information. Depending on the context, the words “if,” “when,” or “in response to determination” as used herein may be interpreted as “when…” or “when…” or “in response to determination.”

[0022] The terms “at least one,” “multiple,” “each,” “any,” etc., used in this invention, “at least one” includes one, two, or more than two; “multiple” includes two or more than two; “each” refers to each of the corresponding multiple; and “any” refers to any one of the multiple.

[0023] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein is for the purpose of describing embodiments of the invention only and is not intended to limit the invention.

[0024] Among related technologies, vital sign extraction techniques based on IR-UWB radar have a fundamental limitation: they heavily rely on ideal electromagnetic wave propagation environments, neglecting the multipath effect that is prevalent and unavoidable in radar's primary application scenarios (such as bedrooms and hospital wards). In real indoor environments, the pulse signal emitted by the radar undergoes multiple reflections from surfaces such as walls, ceilings, and furniture, forming multiple propagation paths to the receiver. These multipath components interfere with the direct echo signal from the human body, causing severe amplitude and phase distortions in the received signal and a sharp drop in the signal-to-noise ratio. Especially for the extremely weak heartbeat signal, multipath interference often completely overwhelms it, severely degrading or even rendering algorithms based on the single-path assumption ineffective.

[0025] In view of this, this invention provides a robust method and related equipment for extracting human vital signs information. This method acquires raw radio frequency data collected by radar, preprocesses the raw radio frequency data to obtain a target data matrix, identifies multipath channels based on the target data matrix, performs parallel phase demodulation on each identified multipath channel to obtain a phase sequence for each multipath channel, performs coherence analysis on all phase sequences, adaptively weights and fuses the phase sequences based on the results of the coherence analysis to obtain a fused vital signs signal, and performs vital signs analysis based on the fused vital signs signal to obtain the target vital signs information. This invention overcomes the limitations of traditional single-path models by identifying and utilizing multiple effective multipath channels, transforming multipath components traditionally considered interference into useful signal sources, thus achieving effective capture and enhancement of vital sign information. Through coherence analysis and adaptive weighted fusion of signals from each path, it can optimize and integrate path components with high signal-to-noise ratio and good signal quality, dynamically suppressing interference from unreliable paths. Furthermore, this invention abandons the simplistic strategy of "selecting the main path and discarding the rest," fully exploring and utilizing the scattered information resources in radar echoes, avoiding information waste, and achieving superior monitoring performance under the same hardware conditions. Specifically, the core contribution of this invention lies in shifting the processing approach from "avoiding multipath" to "utilizing multipath." Through the collaborative processing of multipath signals, it effectively solves the key technical problem of weak vital sign signals being submerged in complex environments, enabling high-precision and highly robust non-contact vital sign monitoring.

[0026] It is understood that the robust extraction method for human vital signs information provided by this invention can be applied to any computer device with data processing and computing capabilities, and this computer device can be various terminals or servers. When the computer device in the embodiment is a server, the server is an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. Optionally, the terminal can be a smartphone, tablet, laptop, or desktop computer, but it is not limited to these.

[0027] like Figure 1 The diagram shown is a schematic representation of an implementation environment provided by an embodiment of the present invention. (Refer to...) Figure 1 The implementation environment includes at least one terminal 102 and a server 101. The terminal 102 and the server 101 can be connected via a network, either wirelessly or via a wired connection, to complete data transmission and exchange.

[0028] Server 101 can be a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms.

[0029] Additionally, server 101 can also be a node server in a blockchain network. Blockchain is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms.

[0030] Terminal 102 can be a smartphone, tablet computer, laptop computer, desktop computer, smart speaker, smartwatch, etc., but is not limited to these. Terminal 102 and server 101 can be directly or indirectly connected via wired or wireless communication, and this embodiment of the invention does not impose any limitations.

[0031] For example, based on Figure 1 The implementation environment shown in this embodiment of the invention provides a robust method for extracting human vital signs information. The following description uses the application of this robust method for extracting human vital signs information in server 101 as an example. It can be understood that this robust method for extracting human vital signs information can also be applied in terminal 102.

[0032] Reference Figure 2 , Figure 2 This is an optional flowchart of a robust extraction method for human vital signs information provided in an embodiment of the present invention. The executing entity of this robust extraction method for human vital signs information can be any of the aforementioned computer devices (including servers or terminals). Figure 2 The method may include, but is not limited to, steps S100 to S400.

[0033] Step S100: Acquire the raw radio frequency data collected by the radar, perform data preprocessing on the raw radio frequency data, and obtain the target data matrix; It should be noted that in some embodiments, data preprocessing of the raw radio frequency data to obtain the target data matrix may include the following steps: constructing a two-dimensional fast-time-slow-time data matrix based on the raw radio frequency echo data continuously acquired by the radar system in the slow-time dimension; performing static clutter filtering on the slow-time dimension based on the two-dimensional fast-time-slow-time data matrix to obtain a filtered data matrix; and applying bandpass filtering on the fast-time dimension based on the filtered data matrix, thereby retaining the effective signal energy within the band and performing out-of-band noise suppression based on the spectral characteristics of the radar transmitted signal to obtain a target data matrix containing dynamic target information.

[0034] For example, in some specific implementations, data preprocessing can be implemented as follows: preliminary noise and clutter suppression is performed on the raw data acquired by the radar to lay the foundation for subsequent accurate signal extraction. The raw radio frequency echo data continuously acquired by the IR-UWB radar system in the slow time dimension constitutes a two-dimensional fast-time-slow-time data matrix. This echo data typically contains strong reflection clutter generated by static objects in the environment and out-of-band noise. To eliminate their influence, this step first estimates and filters out static clutter components by performing coherent accumulation processing on the slow time dimension; subsequently, bandpass filtering is applied in the fast time dimension to retain effective signal energy within the band and suppress out-of-band noise based on the spectral characteristics of the radar transmitted signal. After processing, a data matrix that has been preliminarily purified and mainly contains dynamic target information is obtained.

[0035] Step S200: Based on the target data matrix, multipath channel identification is performed, and parallel phase demodulation is performed on each identified multipath channel to obtain the phase sequence of each multipath channel; It should be noted that, in some embodiments, multipath channel identification based on the target data matrix may include the following steps: accumulating the energy of the target data matrix along the slow time dimension to obtain a one-dimensional fast time energy distribution profile; and identifying multiple multipath channels based on the local energy maxima points on the fast time energy distribution profile.

[0036] For example, in some specific embodiments, multipath channel identification can be achieved by accumulating the energy of the preprocessed data matrix along the slow-time dimension to obtain a one-dimensional fast-time energy distribution profile. The local energy maxima on this profile characterize the channel response of the signal arriving at the receiver via different time-delay paths, thus identifying the various major multipath channels. This invention identifies multiple energy-significant, target-related multipath channels by analyzing the topological structure of this energy profile.

[0037] It should be noted that in some embodiments, parallel phase demodulation is performed on each identified multipath channel to obtain the phase sequence of each multipath channel, which may include the following steps: converting the fast-time local information of each multipath channel to the frequency domain through Fourier transform, and then demodulating the phase information at the radar carrier frequency; based on the phase information of each multipath channel, continuously integrating in the slow-time dimension to obtain the phase sequence of the corresponding multipath channel.

[0038] For example, in some specific implementations, parallel phase demodulation can be achieved as follows: for each identified multipath channel, its echo signal is modulated by the phase of the minute displacement of the human chest cavity caused by breathing and heartbeat. This invention performs coherent processing on the fast-time local data corresponding to each multipath channel, transforms it to the frequency domain using Fourier transform, and demodulates the phase information at the radar carrier frequency. This process is performed continuously in the slow-time dimension, thereby demodulating an independent original phase sequence reflecting the minute movements of vital signs for each multipath channel.

[0039] It should be noted that in some embodiments, the fast-time local information of each multipath channel is converted to the frequency domain by Fourier transform, and then the phase information at the carrier frequency of the radar is demodulated. This may include the following steps: taking the signal within the fast-time window near the static delay corresponding to the multipath channel as the signal segment within the window, performing a fast-time Fourier transform on the signal segment within the window to obtain the spectrum; extracting the spectral complex value based on the spectral components at the carrier frequency; and converting the phase angle corresponding to the spectral complex value to obtain the phase information.

[0040] For example, in some specific implementations, the basic principle of phase demodulation is as follows: In this embodiment of the invention, the extraction of micro-motion information caused by vital signs from radar echoes hinges on the precise demodulation of the signal phase. The basic physical principle is that when electromagnetic waves emitted by radar are reflected by the surface of the human chest cavity, the periodic micro-displacements of the chest cavity due to breathing and heartbeat cause minute disturbances in the propagation path length of the reflected wave, resulting in a corresponding periodic modulation of the reflected wave's phase. By demodulating this phase change, the motion information of vital signs can be retrieved. 1) Signal model: For an IR-UWB radar system, the transmitted single pulse signal can be expressed as: Its equivalent center frequency (carrier frequency) is In a multipath propagation environment, the receiver in the first... The slow time sampling time (i.e. the nth) The received radio frequency echo signal (each pulse repetition cycle) It can be modeled as all Linear superposition of echoes from multiple propagation paths: (1) In the formula, For fast time variables, and The first The path at time The signal amplitude and propagation delay, This represents additive noise.

[0041] Propagation delay of any path It consists of two parts: static and dynamic. The static part... The dynamic part is determined by the fixed geometric length of the path. This is due to the micro-displacement of the human body target. This is caused by [the following]. The relationship is as follows: (2) in, The speed of light is given, and the factor of 2 indicates the round-trip propagation of the electromagnetic wave. The minute displacement of the target... It is a breathing exercise and heartbeat exercise The linear superposition, i.e. .

[0042] 2) Phase extraction principle based on fast-time Fourier transform: This invention employs a phase demodulation technique in the frequency domain. The theoretical basis of this technique is the time-shift theorem of Fourier transform, which states that the delay of a signal in the time domain is equivalent to its spectrum being multiplied in the frequency domain by a linear phase factor related to both the frequency and the delay.

[0043] In the above, the static delay of each main path can be determined through multipath channel identification. . By any path Using the static delay as a reference, we analyze the signal within a fast time window in its vicinity. Within this window, the signal segment can be approximated as being determined solely by the dynamic delay of that path. Modulation. For the signal segment within this window. Its spectrum can be obtained by performing a fast-time Fourier transform. : (3) in, To transmit pulse The spectrum.

[0044] Since the energy of radar signals is mainly concentrated in the carrier frequency The frequency is in the vicinity, therefore the spectral components at this frequency point are most sensitive to phase changes. (Examine the frequency...) Complex values ​​of the spectrum at: (4) The phase angle of the complex number This refers to the phase modulation amount caused by vital signs, which is what we are looking for: (5) Dynamic latency Substituting into the above equation and using the carrier wavelength By making a substitution, a linear mapping relationship can be established between phase modulation and the target's micro-displacement: (6) This formula is the core theoretical basis for phase demodulation in this invention. It shows that by accurately measuring the phase at the carrier frequency in the frequency domain, the movement of human vital signs can be recovered without distortion. proportional phase sequence .

[0045] In obtaining the original, wrapped After the phase sequence within the interval is obtained, its continuity needs to be restored through a phase dewinding algorithm, and finally a signal that can completely reflect the waveform of breathing and heartbeat is obtained.

[0046] In summary, the phase demodulation method based on fast-time Fourier transform employed in this invention utilizes the duality between time delay and phase in the frequency domain to directly extract phase information at the carrier frequency where signal energy is most concentrated. This method does not rely on precise modeling of the time-domain pulse waveform but directly acts on the phase modulation generated by the physical propagation process of the signal. Therefore, it possesses high accuracy and robustness against noise, providing high-quality raw data input for subsequent multipath signal fusion and analysis.

[0047] Step S300: Perform coherence analysis on the signal components of all phase sequences, and perform adaptive weighted fusion on the phase sequences based on the results of the coherence analysis to obtain fused vital sign signals. It should be noted that the results of coherence analysis include the coherence of each phase sequence with the pre-selected reference path. In some embodiments, the phase sequences are adaptively weighted and fused based on the results of coherence analysis to obtain a fused vital sign signal. This may include the following steps: assigning weights to the phase sequences based on coherence, where the weights are positively correlated with coherence; and performing weighted fusion of the phase sequences based on the weights corresponding to each phase sequence to obtain a fused vital sign signal.

[0048] For example, in some specific embodiments, component coherence analysis can be achieved as follows: multipath phase sequences originating from the same vital sign target theoretically possess high coherence. This step quantitatively assesses the similarity in signal morphology between the phase sequences by calculating the cross-correlation function between them, thereby distinguishing the signal component dominated by vital signs from the incoherent component dominated by noise or distortion.

[0049] Adaptive weighted fusion can achieve the following: Based on the results of coherence analysis, this invention constructs an adaptive weighted fusion mechanism. For the phase sequences demodulated from each path, corresponding fusion weights are assigned according to their coherence with the pre-selected reference path. Components with high coherence are considered high-quality signals and receive higher weights; components with low coherence are suppressed. Through this weighted superposition process, the periodic phase modulation common to all paths and caused by vital signs is coherently enhanced, while independent, random noise components in each path are effectively suppressed due to incoherent superposition, ultimately outputting a fused vital sign signal with significantly improved signal-to-noise ratio.

[0050] Step S400: Perform vital sign analysis based on the fused vital sign signals to obtain target vital sign information; It should be noted that the target vital signs information includes respiratory rate and heart rate. In some embodiments, step S400 may include the following steps: extracting the respiratory rate based on the fused vital signs signal through linear filtering and spectral analysis; and extracting the heart rate based on the fused vital signs signal through variational mode decomposition.

[0051] For example, in some specific implementations, vital sign analysis can be achieved as follows: This stage utilizes the fused high-quality signal to extract respiratory and heart rate separately; a) Respiratory frequency extraction: Respiratory movements, as the main low-frequency, high-amplitude components of vital signs, can be directly extracted through linear filtering and spectral analysis. The fused signal is passed through a bandpass filter matching the respiratory frequency range, and then the spectrum of the filtered signal is estimated. The frequency corresponding to the main peak of the spectrum is the respiratory frequency.

[0052] b) Heartbeat Frequency Extraction: Heartbeat signals are weak and easily interfered with by strong respiratory signals and their harmonics. To achieve effective separation of these two signals, this invention introduces Variational Mode Decomposition (VMD) technology. VMD is an adaptive, non-recursive signal decomposition technique that decomposes the fused signal into a set of intrinsic mode functions (IMFs) with specific center frequencies and compact bandwidths. Based on the prior frequency characteristics of the heartbeat signal, this invention establishes a mode selection criterion to identify and select the mode components carrying heartbeat information from the multiple decomposed IMFs. The selected heartbeat mode components are linearly superimposed to reconstruct a high-purity heartbeat signal. Spectral analysis of this reconstructed signal reveals that the frequency corresponding to its main peak is the final heartbeat frequency.

[0053] It should be noted that in some embodiments, the extraction of heartbeat frequency based on fused vital sign signals through variational mode decomposition may include the following steps: taking the set of discrete sub-signals with sparse properties as the decomposition target, the decomposition problem of fused vital sign signals is transformed into a constrained variational problem; the constrained variational problem is transformed into a saddle point finding problem through a quadratic penalty term and Lagrange multipliers; the saddle point finding problem is solved by the alternating direction multiplier method to obtain multiple intrinsic mode functions; a mode selection strategy based on prior knowledge is used to select heartbeat modes from the intrinsic mode functions; all heartbeat modes are linearly superimposed to obtain the heartbeat signal; and the heartbeat frequency is extracted by spectral analysis of the heartbeat signal.

[0054] For example, in some specific implementations, the heart rate extraction algorithm based on variational mode decomposition can be implemented as follows: The main technical challenge faced by this invention in extracting heartbeat information from the fused signal is the masking effect of strong respiratory signals and their harmonics on weak heartbeat signals. Traditional linear bandpass filters struggle to effectively separate respiratory harmonics when they are close to the fundamental frequency of the heartbeat. To address this issue, this invention introduces Variational Mode Decomposition (VMD), an advanced, non-recursive adaptive signal decomposition method.

[0055] 1) Basic principles of VMD: The core idea of ​​VMD is to take a real-valued input signal... It is decomposed into a series of discrete sub-signals (modes) with specific sparse properties. The set of modes. Unlike Empirical Mode Decomposition (EMD), which relies on recursive "sifting," VMD transforms the signal decomposition problem into a constrained variational problem, the goal of which is to find a set of modes. and its corresponding center frequency , so that: 1. Reconstructability: The sum of all modes can accurately reconstruct the original signal, i.e. .

[0056] 2. Compactness: Each mode In the frequency domain, they are all tightly clustered at their respective center frequencies. around.

[0057] VMD achieves this by minimizing the sum of the estimated bandwidths of all modes. The bandwidth of a mode is measured by the H¹ norm of its analytic signal. Specifically, for each mode... First, its analytic signal is obtained through Hilbert transform, and then its spectrum is transformed by a complex exponent. Multiply the signals and shift them to the vicinity of the baseband. Finally, calculate the L² norm (i.e., energy) of the demodulated signal gradient as a measure of its bandwidth.

[0058] Therefore, the constrained variational problem of VMD can be formulated as follows: (7) in, It is the Dirac function. Represents convolution. It is about time The partial derivatives of .

[0059] 2) Solving using the augmented Lagrange method: To solve the constrained optimization problem described above, VMD introduces a quadratic penalty term and Lagrange multipliers. This transforms the problem into an unconstrained augmented Lagrange expression. The problem of finding saddle points: (8) in, This is a quadratic penalty factor used to balance data fidelity. This problem can be efficiently solved using the Alternating Direction Method of Multipliers (ADMM). ADMM decomposes a complex optimization problem into a series of subproblems that can be iteratively solved in alternating directions until convergence. The iterative process mainly includes: a) Update modality : Fix other modes and center frequency, for each Perform an update. In the frequency domain, this step is equivalent to performing Wiener filtering on the current signal residual, with the filter's center frequency being... .

[0060] b) Update the center frequency : Fixed mode, for each Updated. (Updated version) For its corresponding mode The centroid frequency of the power spectrum.

[0061] c) Update the Lagrange multipliers Perform dual ascending updates based on the current reconstruction error.

[0062] 3) Application in heartbeat extraction: In this embodiment of the invention, VMD is applied to the fused vital sign signals. Separate the heartbeat signal.

[0063] a) Decomposition: As input to the VMD algorithm, the algorithm decomposes the signal into... Each intrinsic mode function (IMF) is a unique feature of a system. and their respective center frequencies Because respiratory signals are low-frequency and high-amplitude, they are typically decomposed into an IMF (Intra-Frequency Mixture) with a lower center frequency. Heartbeat signals and their harmonics, on the other hand, are distributed within an IMF with a higher center frequency.

[0064] b) Modality Optimization: After decomposition, a strategy is needed to automatically identify which IMFs represent the heartbeat. This invention employs a modality optimization strategy based on prior knowledge. This strategy is based on the following criteria: Frequency range: The resting heart rate frequency range of a typical adult is defined (e.g., 1.0–1.8 Hz). Therefore, only those center frequencies are considered. IMFs falling within the preset heart rate range.

[0065] Energy Proportion: Among the candidate IMFs that meet the frequency range criteria, their energy distribution is further analyzed. A true heartbeat mode should have most of its energy concentrated within the heartbeat frequency band. The proportion of energy in the heartbeat frequency band to the total energy of each candidate IMF can be calculated, or the significance of its peak value within the frequency band can be used to screen out the most likely heartbeat modes.

[0066] c) Reconstruction and Analysis: Reconstruct and analyze all selected heartbeat modes Linear superposition is performed to reconstruct a pure and complete heartbeat signal. Because VMD can effectively separate respiratory harmonics into other modes, the reconstructed heartbeat signal has a high signal-to-noise ratio and low harmonic interference. Finally, this reconstructed heartbeat signal... By performing spectral analysis, the frequency corresponding to the main peak in the spectrum is the precise heartbeat frequency.

[0067] In summary, this invention utilizes VMD as the core algorithm for heartbeat extraction. Its principle lies in transforming the complex signal separation problem into a variational optimization problem with a solid theoretical foundation. By adaptively decomposing the signal into a set of bandwidth-limited modes and combining this with a mode selection strategy based on prior knowledge, it can effectively overcome the interference of strong breathing harmonics, achieving precise separation and extraction of weak heartbeat signals—a feat difficult to achieve with traditional filtering methods.

[0068] To explain in detail the principle of the technical solution of the present invention, the overall process of the present invention will be described below with reference to some specific embodiments. It is easy to understand that the following is an explanation of the technical principle of the present invention and should not be regarded as a limitation of the present invention.

[0069] First, it should be noted that the lack of a systematic method for effectively identifying, processing, and utilizing multipath information in existing technologies is one of the core technological bottlenecks currently hindering the large-scale practical application of IR-UWB radar life sign monitoring technology.

[0070] In view of this, and addressing the shortcomings of IR-UWB radar in real indoor environments, such as low accuracy and poor robustness in vital sign monitoring due to multipath effects, this invention provides a systematic multipath signal processing method. This method actively identifies and extracts vital sign information from multiple effective propagation paths, performs correlation analysis and adaptive weighted fusion on this information, and combines advanced signal decomposition techniques to accurately and stably separate and extract respiratory and heartbeat signals from complex multipath echoes. This significantly improves the accuracy and reliability of non-contact vital sign monitoring systems in real-world application scenarios.

[0071] In some specific embodiments, this invention proposes a method for high-precision and robust vital sign monitoring based on ultra-wideband (IR-UWB) radar in multipath environments. The core of this method lies in systematically identifying, extracting, analyzing, and fusing multipath signal components to enhance weak vital sign information, and combining this with advanced signal processing algorithms to achieve accurate calculation of respiratory and heart rate. The overall technical flow of this invention is as follows: Figure 3 As shown, the process mainly includes three key stages: data preprocessing, extraction of vital sign micro-motion information based on multipath processing, and vital sign analysis. 1) Data preprocessing: This stage aims to perform preliminary noise and clutter suppression on the raw radar data, laying the foundation for subsequent accurate signal extraction. The raw radio frequency echo data continuously acquired by the IR-UWB radar system in the slow-time dimension constitutes a two-dimensional fast-time-slow-time data matrix. This echo data typically contains strong reflection clutter generated by static objects in the environment, as well as out-of-band noise. To eliminate their influence, this step first estimates and filters out static clutter components by performing coherent accumulation processing on the slow-time dimension; subsequently, bandpass filtering is applied in the fast-time dimension to retain effective signal energy within the band and suppress out-of-band noise based on the spectral characteristics of the radar transmitted signal. The resulting data matrix is ​​a preliminarily purified data matrix primarily containing dynamic target information. 2) Extraction of vital sign micro-motion information based on multipath processing: This stage is the core step of the invention and the most significant difference between the multipath processing scheme proposed in this invention and traditional schemes. Its purpose is to robustly extract the micro-motion information modulated by vital signs from the complex mixed echoes caused by multipath propagation. The specific steps are as follows: a) Multipath Channel Identification: By accumulating the energy of the preprocessed data matrix along the slow-time dimension, a one-dimensional fast-time energy distribution profile is obtained. The local energy maxima on this profile characterize the channel response of the signal arriving at the receiver via different time-delay paths, thus identifying the main multipath channels. This invention identifies multiple energy-significant, target-related multipath channels by analyzing the topology of this energy profile.

[0072] b) Parallel Phase Demodulation: For each identified multipath channel, its echo signal is modulated by the phase of the minute displacement of the human chest cavity caused by breathing and heartbeat. This invention performs coherent processing on the fast-time local data corresponding to each multipath channel, transforms it to the frequency domain through Fourier transform, and demodulates the phase information at the radar carrier frequency. This process is performed continuously in the slow-time dimension, thereby demodulating an independent original phase sequence reflecting the minute movements of vital signs for each multipath channel.

[0073] c) Component coherence analysis: Multipath phase sequences originating from the same vital sign target theoretically possess high coherence. This step quantitatively assesses their similarity in signal morphology by calculating the cross-correlation function between each phase sequence, thereby distinguishing the signal component dominated by vital signs from the incoherent component dominated by noise or distortion.

[0074] d) Adaptive Weighted Fusion: Based on the results of coherence analysis, this invention constructs an adaptive weighted fusion mechanism. For the phase sequences demodulated from each path, corresponding fusion weights are assigned according to their coherence with the pre-selected reference path. Components with high coherence are considered high-quality signals and receive higher weights; components with low coherence are suppressed. Through this weighted superposition process, the periodic phase modulation common to all paths and caused by vital signs is coherently enhanced, while independent, random noise components in each path are effectively suppressed due to incoherent superposition, ultimately outputting a fused vital sign signal with significantly improved signal-to-noise ratio.

[0075] 3) Vital signs analysis: This stage utilizes the fused high-quality signal to extract respiratory and heart rate separately; a) Respiratory frequency extraction: Respiratory movements, as the main low-frequency, high-amplitude components of vital signs, can be directly extracted through linear filtering and spectral analysis. The fused signal is passed through a bandpass filter matching the respiratory frequency range, and then the spectrum of the filtered signal is estimated. The frequency corresponding to the main peak of the spectrum is the respiratory frequency.

[0076] b) Heartbeat Frequency Extraction: Heartbeat signals are weak and easily interfered with by strong respiratory signals and their harmonics. To achieve effective separation of these two signals, this invention introduces Variational Mode Decomposition (VMD) technology. VMD is an adaptive, non-recursive signal decomposition technique that decomposes the fused signal into a set of intrinsic mode functions (IMFs) with specific center frequencies and compact bandwidths. Based on the prior frequency characteristics of the heartbeat signal, this invention establishes a mode selection criterion to identify and select the mode components carrying heartbeat information from the multiple decomposed IMFs. The selected heartbeat mode components are linearly superimposed to reconstruct a high-purity heartbeat signal. Spectral analysis of this reconstructed signal reveals that the frequency corresponding to its main peak is the final heartbeat frequency.

[0077] In some specific application scenarios, to verify the effectiveness of the technical solution described in this invention, comparative experiments were conducted in a typical indoor multipath environment. The experimental environment was a standard indoor office, furnished with desks, cabinets, and other conventional furniture, constituting a typical multipath propagation scenario. A test subject sat approximately 1.2 meters in front of the IR-UWB radar, with the radar antenna pointed at their chest. During the experiment, radar echo data and heart rate data from a contact heart rate monitor were simultaneously collected. To evaluate the technical effect of this invention, the following two processing procedures were applied to the same set of collected data, as detailed in the following steps: Figure 4 As shown, and obtained as follows Figure 5 , Figure 6 The simulation results are shown below: The present invention (multipath signal processing method) follows the aforementioned complete technical process, namely, multipath channel identification, parallel phase demodulation of each path, adaptive weighted fusion based on coherence analysis, and calculation of vital sign parameters based on the fused signal. The comparative scheme (single-path signal processing method) represents a common processing strategy in the prior art. This scheme only identifies and selects the strongest single energy region in the average fast time-energy profile, and performs subsequent vital sign analysis entirely based on the phase information extracted from this region.

[0078] Figure 5 The diagram illustrates the multipath channel identification and processing interval, where the gray area represents a fast-time energy distribution profile, clearly showing the distribution of direct waves and multiple multipath echoes along the range dimension. Based on the scheme of this invention, five multipath channels can be identified, and five original phase sequences can be extracted. To avoid interference sequences causing damage (such as multipath channel 3 in this experiment) and to further improve the quality of the fused phase sequences, this scheme uses an adaptive weighting method to obtain the final phase sequences; while traditional phase sequence extraction schemes can only select a single maximum energy window and obtain one phase sequence. Figure 4 It can also be seen that the fused phase sequence is significantly more stable than the single phase sequence and can avoid interference caused by invalid echoes or low-quality echoes.

[0079] Figure 6 The normalized spectra of respiratory and heartbeat signals are shown. For the respiratory signal spectrum, the two spectral curves highly overlap in the low-frequency range (0.1 Hz ~ 0.3 Hz), both exhibiting a clear and sharp main peak at approximately 0.19 Hz. Based on this main peak frequency, the estimated respiratory frequency obtained by both the proposed and comparative methods is 11.4 BPM. The signal-to-noise ratio and shape of the two spectral peaks show good consistency. This result indicates that for high-energy respiratory signals, both processing methods can achieve accurate frequency extraction.

[0080] For the comparison of the VMD reconstructed heartbeat signal spectrum, the spectrum of the comparative scheme (red dashed line) shows that its energy distribution is relatively dispersed. Within the heartbeat frequency band, there are multiple spectral peaks with similar amplitudes, the most significant peak being located at approximately 1.18 Hz, corresponding to a heart rate of 70.6 BPM. The noise floor is high throughout the band, resulting in an unsatisfactory signal-to-noise ratio for the main peak, and its frequency estimate deviates significantly from the reference value. The spectrum of the present invention (blue solid line) exhibits significantly different characteristics. At approximately 1.325 Hz, there is a highly concentrated, very sharp spectral peak with an amplitude far exceeding other components in the spectrum. The heart rate corresponding to this frequency is 79.5 BPM, which is highly consistent with the reference value of 78.8 BPM. At the same time, the overall noise floor of this spectrum is effectively suppressed, resulting in a very high signal-to-noise ratio for the main heartbeat peak.

[0081] Comprehensive experimental analysis demonstrates the significant advantages of the proposed technical solution. In typical indoor multipath environments, through a systematic multipath signal processing framework, this invention effectively transforms multipath interference into usable information resources. Through identification, extraction, coherence analysis, and adaptive fusion, it fundamentally improves the signal-to-noise ratio and robustness of the original vital sign signals. Furthermore, by combining advanced spectral analysis methods centered on variational mode decomposition (VMD), it achieves precise separation of strong respiratory harmonics and weak heartbeat signals. Experimental results clearly confirm that, compared to traditional single-path methods, this invention can obtain heartbeat signals with higher spectral purity and more accurate frequency positioning, thus achieving high-precision non-contact vital sign monitoring in complex real-world scenarios.

[0082] In summary, this invention constructs a systematic processing framework from coherent utilization of multipath channels to deep separation of signal modes. This framework treats multipath channels as multiple independent observations of the same vital sign event, and achieves effective information gain through coherence analysis and fusion. Furthermore, by utilizing the nonlinear coupling characteristics of heartbeat and respiratory harmonics in the frequency domain, variational mode decomposition (VMD) is used instead of traditional linear filters to achieve accurate separation of the two in the intrinsic mode space of the signal.

[0083] Existing technologies for UWB radar vital sign monitoring in real indoor scenarios often fail to effectively address the strong interference caused by multipath effects, resulting in low accuracy and poor robustness of monitoring results in complex environments. Compared with existing technologies, this invention has the following significant advantages: 1) Higher accuracy and robustness: Through multipath fusion, this invention fundamentally improves the signal-to-noise ratio of the original signal, effectively resisting the effects of single-path signal fading or distortion, and can obtain accurate and reliable vital sign monitoring results even in complex real indoor environments.

[0084] 2) Enhanced signal separation capability: The introduction of VMD technology can adaptively separate strong respiratory harmonics from weak heartbeat signals, solving the common spectral aliasing problem in vital sign information extraction and ensuring the accuracy of heartbeat frequency extraction.

[0085] 3) Better information utilization efficiency: This invention makes full use of all identifiable multipath echoes, achieving maximum utilization of the information collected by the radar. Compared with the traditional method of discarding multipath information, it has higher signal gain and information integrity.

[0086] like Figure 7 As shown, this embodiment of the invention also provides a robust human vital sign information extraction device 900, which can implement the above-described method. This device may include: The data acquisition module 910 is used to acquire the raw radio frequency data collected by the radar, perform data preprocessing on the raw radio frequency data, and obtain the target data matrix. The phase demodulation module 920 is used to identify multipath channels based on the target data matrix, and to perform parallel phase demodulation on each identified multipath channel to obtain the phase sequence of each multipath channel. The signal fusion module 930 is used to perform coherence analysis on the signal components of all phase sequences, and to perform adaptive weighted fusion on the phase sequences based on the results of the coherence analysis to obtain fused vital sign signals. The vital sign extraction module 940 is used to perform vital sign analysis based on fused vital sign signals to obtain target vital sign information.

[0087] It is understood that the content of the above method embodiments is applicable to the present device embodiments. The specific functions implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0088] This invention also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described above. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0089] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0090] like Figure 8 As shown, Figure 8 The hardware structure of an electronic device 1000 according to another embodiment is illustrated. The electronic device 1000 includes: The processor 1001 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (aSIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present invention. The memory 1002 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RaM). The memory 1002 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1002 and is called and executed by the processor 1001. Input / output interface 1003 is used to implement information input and output; The communication interface 1004 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 1005 transmits information between various components of the device (e.g., processor 1001, memory 1002, input / output interface 1003, and communication interface 1004); The processor 1001, memory 1002, input / output interface 1003 and communication interface 1004 are connected to each other within the device via bus 1005.

[0091] The electronic device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0092] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0093] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0094] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0095] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0096] The robust extraction method, apparatus, electronic device, storage medium, and program product for human vital signs information provided in this invention embodiment acquire raw radio frequency data collected by radar, preprocess the raw radio frequency data to obtain a target data matrix; perform multipath channel identification based on the target data matrix, perform parallel phase demodulation on each identified multipath channel to obtain a phase sequence for each multipath channel; perform coherence analysis on the signal components of all phase sequences, and perform adaptive weighted fusion of the phase sequences based on the results of the coherence analysis to obtain a fused vital sign signal; and perform vital sign analysis based on the fused vital sign signal to obtain target vital sign information. This invention overcomes the limitations of traditional single-path models by identifying and utilizing multiple effective multipath channels, transforming multipath components traditionally considered interference into useful signal sources, thus achieving effective capture and enhancement of vital sign information. Through coherence analysis and adaptive weighted fusion of signals from each path, it can optimize and integrate path components with high signal-to-noise ratio and good signal quality, dynamically suppressing interference from unreliable paths. Furthermore, this invention abandons the simplistic strategy of "selecting the main path and discarding the rest," fully exploring and utilizing the scattered information resources in radar echoes, avoiding information waste, and achieving superior monitoring performance under the same hardware conditions. Specifically, the core contribution of this invention lies in shifting the processing approach from "avoiding multipath" to "utilizing multipath." Through the collaborative processing of multipath signals, it effectively solves the key technical problem of weak vital sign signals being submerged in complex environments, enabling high-precision and highly robust non-contact vital sign monitoring.

[0097] The preferred embodiments of the present invention have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and spirit of the present invention should be within the scope of the claims of the present invention.

Claims

1. A robust method for extracting human vital signs information, characterized in that, The method includes the following steps: The raw radio frequency data collected by the radar is acquired, and the raw radio frequency data is preprocessed to obtain the target data matrix; Multipath channel identification is performed based on the target data matrix, and parallel phase demodulation is performed on each identified multipath channel to obtain the phase sequence of each multipath channel; Coherence analysis of signal components is performed on all the phase sequences, and adaptive weighted fusion is performed on the phase sequences based on the results of the coherence analysis to obtain fused vital sign signals; Based on the fused vital sign signals, vital sign analysis is performed to obtain target vital sign information.

2. The method according to claim 1, characterized in that, The process of preprocessing the raw radio frequency data to obtain the target data matrix includes the following steps: Based on the raw radio frequency echo data continuously acquired by the radar system in the slow time dimension, a two-dimensional fast-time-slow-time data matrix is ​​constructed. Based on the two-dimensional fast-time-slow-time data matrix, static clutter filtering is completed by coherently accumulating the slow-time dimension to obtain the filtered data matrix. Based on the filtered data matrix, a bandpass filter is applied in the fast time dimension, and then the effective signal energy in the band is retained according to the spectral characteristics of the radar transmitted signal to complete the out-of-band noise suppression operation, thereby obtaining the target data matrix containing dynamic target information.

3. The method according to claim 1, characterized in that, The multipath channel identification based on the target data matrix includes the following steps: The energy of the target data matrix is ​​accumulated along the slow time dimension to obtain a one-dimensional fast time energy distribution profile; Multiple multipath channels are identified based on the local energy maxima points on the fast-time energy distribution profile.

4. The method according to claim 1, characterized in that, The step of performing parallel phase demodulation on each identified multipath channel to obtain the phase sequence of each multipath channel includes the following steps: The fast-time local information of each multipath channel is converted to the frequency domain by Fourier transform, and then the phase information at the radar's carrier frequency is demodulated. Based on the phase information of each multipath channel, the phase sequence corresponding to the multipath channel is continuously integrated in the slow time dimension.

5. The method according to claim 4, characterized in that, The step of converting the fast-time local information of each multipath channel to the frequency domain through Fourier transform, and then demodulating the phase information at the radar's carrier frequency, includes the following steps: The signal within a fast time window near the static delay corresponding to the multipath channel is taken as the signal segment within the window, and a fast time Fourier transform is performed on the signal segment within the window to obtain the spectrum; Based on the spectrum, the spectral components at the carrier frequency are extracted to obtain the complex spectral value; The phase information is obtained by converting the phase angle corresponding to the complex value of the spectrum.

6. The method according to claim 1, characterized in that, The results of the coherence analysis include the coherence of each phase sequence with a pre-selected reference path. The adaptive weighted fusion of the phase sequences based on the results of the coherence analysis to obtain a fused vital sign signal includes the following steps: Weights are assigned to the phase sequence based on the coherence, and the weights are positively correlated with the coherence; The phase sequences are weighted and fused based on the weights corresponding to each phase sequence to obtain a fused vital sign signal.

7. The method according to claim 1, characterized in that, The target vital sign information includes respiratory rate and heart rate. The step of analyzing vital signs based on the fused vital sign signals to obtain the target vital sign information includes the following steps: Based on the fused vital sign signals, the respiratory rate is extracted through linear filtering and spectral analysis. Based on the fused vital signs signal, the heart rate is extracted through variational mode decomposition.

8. The method according to claim 7, characterized in that, The step of extracting the heart rate based on the fused vital sign signals through variational mode decomposition includes the following steps: Using the set of discrete sub-signals with sparse properties as the decomposition target, the decomposition problem of the fused vital sign signal is transformed into a constrained variational problem; The constrained variational problem is transformed into a saddle-point finding problem by using a quadratic penalty term and Lagrange multipliers. The saddle point finding problem is solved by the alternating direction multiplier method, yielding multiple eigenmode functions; A modality selection strategy based on prior knowledge filters out heartbeat modes from the intrinsic modal functions; The heartbeat signal is obtained by linearly superimposing all the described heartbeat modes. The heartbeat frequency is obtained by performing spectral analysis on the heartbeat signal.

9. A robust extraction device for human vital signs information, characterized in that, The device includes: The data acquisition module is used to acquire the raw radio frequency data collected by the radar, and to preprocess the raw radio frequency data to obtain the target data matrix. The phase demodulation module is used to identify multipath channels based on the target data matrix, and to perform parallel phase demodulation on each identified multipath channel to obtain the phase sequence of each multipath channel. The signal fusion module is used to perform coherence analysis on the signal components of all the phase sequences, and to perform adaptive weighted fusion on the phase sequences based on the results of the coherence analysis to obtain fused vital sign signals. The vital sign extraction module is used to perform vital sign analysis based on the fused vital sign signals to obtain target vital sign information.

10. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method according to any one of claims 1 to 8.

Images