Human body heartbeat detection method, device and equipment based on ultra-wideband radar

Abstract

The invention relates to a human body heartbeat detection method, device and equipment based on an ultra-wideband radar, and the method comprises the steps: firstly completing the matching of a heartbeat detection algorithm in a pre-detection stage through employing a pre-detection and algorithm detection two-stage design, and designing an SSA window adaptive switch; a heartbeat detection algorithm of a current detection scene is accurately matched based on the clothes condition of a detected area, and whether self-adaptive SSA dynamic parameter adjustment is started or not is supported, so that the detection process supports a designed SSA window self-adaptive adjustment strategy based on signal energy distribution; accurate heartbeat detection processing in a real-time detection scene is realized by using a self-adaptive matching basic detection algorithm or a high-order detection algorithm, and on the premise of not excessively increasing calculation burden, the signal-to-noise ratio of heartbeat signals is practically improved through the self-adaptive matching algorithm and an SSA window self-adaptive adjustment strategy, and meanwhile, the reliability of human body heartbeat detection is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of non-contact heartbeat detection, and relates to a human heartbeat detection method, device and equipment based on ultra-wideband radar. Background Art

[0002] Non-contact human heartbeat detection based on ultra-wideband radar (UWB) utilizes the strong penetration and anti-multipath interference characteristics of UWB radar. After collecting the human chest echo signal, the CA-CFAR (Cell-Averaging Constant FalseAlarm Rate) algorithm is used to remove static clutter, and the target chest micro-motion signal is extracted through clutter filtering. Then, the improved singular spectrum analysis (SSA) is used to construct the trajectory matrix and reconstruct the singular values ​​to retain the main components of the signal to suppress noise. Then, variational mode decomposition (VMD) is used to separate the breathing and heartbeat signals through constraints to avoid modal aliasing problems. The overall signal processing process usually includes sampling, echo matrix construction, filtering, reconstruction, and frequency domain analysis.

[0003] However, heartbeat signals are usually weak and easily affected by environmental noise and respiratory harmonics, which can easily lead to errors in frequency estimation. This is because the chest displacement caused by the heartbeat is only about millimeter level, the signal amplitude is extremely low, and its noise source is also relatively complex, often including equipment noise, environmental reflection clutter and respiratory harmonics (overlapping with the heartbeat frequency band). In addition, traditional filtering and SSA may not be able to completely and effectively separate high-frequency noise from low-frequency physiological signals. With the urgent need for non-contact health monitoring (such as telemedicine and disaster relief), how to more effectively improve the reliability of human heartbeat detection has become the core issue for the popularization and application of such technologies. The weak signal enhancement technology introduced can promote the widespread application of UWB radar in medical, security and other fields. Summary of the invention

[0004] In view of the problems existing in the above-mentioned traditional technologies, the present invention proposes a human heartbeat detection method based on ultra-wideband radar, a human heartbeat detection device based on ultra-wideband radar, and a human heartbeat detection equipment, which can more effectively improve the reliability of human heartbeat detection.

[0005] In order to achieve the above objectives, the embodiments of the present invention adopt the following technical solutions: On the one hand, a human heartbeat detection method based on ultra-wideband radar is provided, comprising the steps of: The UWB radar is used to pre-detect the measured area of ​​the target. According to the pre-detected radar echo indication, the UWB radar is moved to the calibrated distance point of the measured area and then locked. The UWB radar is a radar on the pan-tilt base, and the measured area is the human chest cavity. According to the pre-detected radar echo energy loss, the clothing thickness range of the measured area is determined by looking up a preset energy thickness table; the energy thickness table includes a mapping between different echo energy losses and clothing thickness ranges; According to the clothing thickness range, a matching heartbeat detection algorithm is called from the detection algorithm library and it is determined whether to turn on the SSA window adaptive switch; the heartbeat detection algorithm includes a basic detection algorithm and a high-order detection algorithm; After receiving the detection echo of the measured area to the ultra-wideband radar and suppressing the static clutter of the detection echo by using the background difference method, the chest micro-motion signal of the measured area is extracted from the detection echo by using the heartbeat detection algorithm; The heartbeat detection algorithm is used to construct the trajectory matrix, reconstruct the singular values ​​and perform variational mode decomposition on the chest micro-motion signal; if the SSA window adaptive switch is turned on, the SSA sliding window is adjusted according to the set SSA window adaptive adjustment strategy, otherwise the SSA sliding window remains unchanged; The frequency domain heartbeat signal after variational mode decomposition is inversely transformed into a time domain heartbeat signal output.

[0006] On the other hand, a human heartbeat detection device based on ultra-wideband radar is also provided, comprising: The pre-detection module is used to pre-detect the detected area of ​​the target by using the ultra-wideband radar, and move the ultra-wideband radar to the calibrated distance point of the detected area according to the pre-detected radar echo instruction and then lock it; the ultra-wideband radar is a radar on the pan-tilt base, and the detected area is the human chest cavity; A loss table lookup module is used to look up a preset energy thickness table to determine the clothing thickness interval of the measured area according to the pre-detected radar echo energy loss; the energy thickness table includes a mapping between different echo energy losses and clothing thickness intervals; An algorithm switch module is used to call a matching heartbeat detection algorithm from a detection algorithm library according to the clothing thickness range and determine whether to turn on the SSA window adaptive switch; the heartbeat detection algorithm includes a basic detection algorithm and a high-order detection algorithm; The micro-motion extraction module is used to receive the detection echo of the measured area to the ultra-wideband radar and use the background difference method to suppress the static clutter of the detection echo, and then extract the chest micro-motion signal of the measured area from the detection echo through the heartbeat detection algorithm; The signal processing module is used to construct the trajectory matrix, reconstruct the singular value and perform variational mode decomposition on the chest micro-motion signal through the heartbeat detection algorithm; if the SSA window adaptive switch is turned on, the SSA sliding window is adjusted according to the set SSA window adaptive adjustment strategy, otherwise the SSA sliding window is maintained unchanged; The detection output module is used to inversely transform the frequency domain heartbeat signal after variational mode decomposition into a time domain heartbeat signal output.

[0007] On the other hand, a human heartbeat detection device is also provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned human heartbeat detection method based on ultra-wideband radar when executing the computer program.

[0008] One of the above technical solutions has the following advantages and beneficial effects: The above-mentioned human heartbeat detection method, device and equipment based on ultra-wideband radar adopts a two-stage design of pre-detection and algorithm detection. First, in the pre-detection stage, the matching of the heartbeat detection algorithm is completed and an SSA window adaptive switch is designed to achieve accurate matching of the heartbeat detection algorithm of the current detection scene based on the clothing condition of the measured area and support whether to turn on adaptive SSA dynamic parameter adjustment, so that the detection process supports the designed SSA window adaptive adjustment strategy based on signal energy distribution, and uses the basic detection algorithm or high-order detection algorithm of adaptive matching to achieve accurate heartbeat detection processing in real-time detection scenarios. Without excessively increasing the computational burden, the adaptive matching algorithm and the SSA window adaptive adjustment strategy are used to effectively improve the signal-to-noise ratio of the heartbeat signal and at the same time improve the reliability of human heartbeat detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the conventional technology, the drawings required for use in the embodiments or the conventional technology descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0010] Figure 1 1 is a flow chart of a human heartbeat detection method based on ultra-wideband radar in one embodiment; Figure 2 A detection logic diagram of a human heartbeat detection method based on ultra-wideband radar in one embodiment; Figure 3 A schematic diagram of a process flow for implementing frequency domain enhancement in an embodiment; Figure 4 is a module block diagram of a human heartbeat detection device based on ultra-wideband radar in one embodiment; Figure 5 Schematic diagram of the module structure of a human heartbeat detection device in one embodiment. DETAILED DESCRIPTION

[0011] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with the accompanying drawings and Examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by those skilled in the art of the present invention. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0012] It should be noted that the reference to "embodiment" in this article means that the specific features, structures or characteristics described in conjunction with the embodiment may be included in at least one embodiment of the present invention. The presentation of this phrase at various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It will be appreciated by those skilled in the art that the embodiments described herein may be combined with other embodiments. The term "and / or" used in the specification of the present invention refers to any combination of one or more of the items listed in association and all possible combinations, and includes these combinations.

[0013] The implementation modes of the present invention will be described in detail below with reference to the accompanying drawings in the embodiments of the present invention.

[0014] In order to further solve the problem of low signal-to-noise ratio of heartbeat signals, the following possible solutions have been found in actual research: Deep learning denoising, which can effectively adapt to complex noise environments by pre-training CNN / RNN models to identify noise patterns and adaptively suppress interference. This method can effectively adapt to complex noise environments in an end-to-end optimization manner, but this method requires a large amount of labeled training data and usually consumes a large amount of computing resources. Multi-sensor data fusion, which can complement and enhance heartbeat signals by combining infrared thermal imaging or piezoelectric sensors, can improve robustness by using multi-dimensional data, but the system complexity may increase and the problem of multi-source data synchronization needs to be solved. Dynamic parameter optimization, which can reduce manual intervention and improve the generalization ability of the algorithm by automatically adjusting the SSA window or VMD mode number according to the signal spectrum characteristics, but requires a specially designed efficient adaptive threshold mechanism. Frequency domain enhancement technology, which can effectively suppress non-target frequency band interference and retain signal details by combining wavelet transform or synchronous compression transform (SST), needs to balance time-frequency resolution and computational efficiency. This application will propose a new human heartbeat detection method based on ultra-wideband radar based on the research solution. The following is an explanation of the overall idea of ​​the method.

[0015] In one embodiment, Figure 1 As shown, a human heartbeat detection method based on ultra-wideband radar is provided, which may include the following steps S10 to S20: S10, using an ultra-wideband radar to pre-detect a measured area of ​​the measured target, and according to the pre-detected radar echo indication, the ultra-wideband radar moves to a calibrated distance point of the measured area and then locks; the ultra-wideband radar is a radar on a pan-tilt base, and the measured area is a human chest cavity.

[0016] It can be understood that the operating frequency band of the UWB radar can be 3.1GHz to 10.6GHz, its bandwidth can reach 7.5GHz, and its distance resolution is about 2cm. It has the characteristics of strong penetration (suitable for non-line-of-sight scenarios), resistance to multipath interference and low power consumption. However, the amplitude of the heartbeat signal is extremely low (chest displacement is about 0.1mm to 0.5mm), and it is easily drowned out by environmental noise. The target to be measured is the human body that currently needs to perform human heartbeat detection. This embodiment uses the human chest area as the measured area for human heartbeat detection to ensure that the radar echo can contain heartbeat information to the greatest extent in the initial stage. The calibration distance point can be the optimal non-contact detection distance point from the human chest that is pre-calibrated based on the detection experiment statistics of the ultra-wideband radar, which is used to further ensure the significance of the heartbeat information contained in the radar echo in the initial stage. In this embodiment, the ultra-wideband radar is installed on the pan-tilt base, and can be operated by the measurement operator by hand-held or placed on a slide rail to perform detection activities. Combined with the position locking of the calibration distance point, the vibration noise introduced to the radar echo by factors such as hand-held vibration or ground vibration can be removed, thereby improving the purity of the received radar echo from the operational level.

[0017] S12, according to the pre-detected radar echo energy loss, look up a preset energy thickness table to determine the clothing thickness range of the measured area; the energy thickness table includes a mapping between different echo energy losses and clothing thickness ranges.

[0018] It is understandable that in actual detection applications, different people often wear different types of clothing materials and different quantities of clothing when undergoing human heartbeat detection. The detection signal of low-frequency (such as 3.1GHz to 10.6GHz) ultra-wideband radar has strong penetration and can penetrate most thin clothing such as cotton and chemical fiber. However, its penetration ability is weakened or even ineffective in scenes such as thick down jackets, multi-layered clothing and metal-coated clothing (such as windproof jackets). For example, experiments show that the signal attenuation of cotton T-shirts for low-frequency ultra-wideband radar is about 1dB to 3dB, while that of thick sweaters can reach 5dB to 10dB.

[0019] However, conventional technologies often consider that when a person is undergoing human heartbeat detection, the person is not wearing clothes, is wearing thin clothes, or that complex signal enhancement and filtering algorithms are used to weaken the influence of clothes on radar echoes, which results in limited technical applications or redundant and complex calculations. In this embodiment, the differences in the types of clothing materials and the number of clothes worn by different people when undergoing human heartbeat detection in actual detection applications are fully studied. According to the situation of the person wearing clothes when undergoing human heartbeat detection (such as less than one piece of clothing and more than two pieces of clothing), the thickness of the clothes covered by the measured area is pre-divided into two levels of clothing thickness intervals, and the radar echo energy loss of the same ultra-wideband radar in the above two different clothing thickness intervals is calibrated through sampling experiments, and the corresponding relationship between the above different clothing thickness intervals and the radar echo energy loss is made into a preset energy thickness table to establish a mapping relationship between different clothing thickness intervals and radar echo energy loss.

[0020] S14, calling a matching heartbeat detection algorithm from a detection algorithm library according to the clothing thickness range and determining whether to turn on the SSA window adaptive switch; the heartbeat detection algorithm includes a basic detection algorithm and a high-order detection algorithm.

[0021] It can be understood that the detection algorithm library may include basic detection algorithm units and high-order detection algorithm units integrated by software. These algorithm units will automatically execute corresponding algorithm processing when they are called by the system. The basic detection algorithm may include the existing improved singular spectrum analysis (SSA, which suppresses noise by retaining the main components of the signal through trajectory matrix construction and singular value reconstruction) and variational mode decomposition (VMD, which separates breathing and heartbeat signals through constraints to avoid modal aliasing problems), while the high-order detection algorithm is an improved version of the basic detection algorithm, in which a new frequency domain enhancement processing is designed to comprehensively improve the heartbeat detection capability of detecting echoes in low signal-to-noise ratio scenarios. For example, when the clothing thickness interval is an interval that causes a relatively large energy loss of radar echoes, the high-order detection algorithm will be matched for subsequent processing, otherwise the basic detection algorithm will be matched for subsequent processing. The adaptive detection algorithm is called according to the clothing thickness interval of the current measured area in the algorithm library matching manner, which can meet the accurate detection requirements while flexibly saving computing resources and computing time, and achieve the effect of algorithm adaptation.

[0022] Among them, the SSA window adaptive switch is an algorithm switch used to determine whether to enable the set SSA window adaptive adjustment strategy in the current detection, so that the detection can adjust and improve the algorithm parameters of the singular spectrum analysis according to the real-time signal characteristics, thereby reducing manual intervention and improving the detection effect. The SSA window adaptive switch can be automatically turned on or off according to the matched heartbeat detection algorithm, or it can be turned on or off manually, regardless of the currently matched heartbeat detection algorithm, to ensure the improvement of the detection effect.

[0023] S16, after receiving the detection echo of the measured area to the ultra-wideband radar and suppressing static clutter on the detection echo by using the background difference method, the chest micro-motion signal of the measured area is extracted from the detection echo by using the heartbeat detection algorithm.

[0024] It can be understood that after matching the heartbeat detection algorithm currently required to be called, the formal heartbeat detection of the measured area is entered. The ultra-wideband radar locks on the calibration distance point of the measured area and transmits the radar detection signal to the measured area. After receiving the corresponding detection echo, the background difference method is used to perform static clutter suppression on the detection echo. Specifically, the static background template can be generated by averaging the detection echo for a long time (for example, 5s), and the real-time echo signal is subtracted from the background template to eliminate the reflection of fixed objects in the measured area and its surrounding positions. The detection echo after static clutter suppression can be expressed as follows : ; in, is the initial received detection echo before static clutter suppression, t Indicates time, T is the sampling interval, N Then, the chest micro-motion signal of the detected area is extracted from the detection echo by the heartbeat detection algorithm, and the specific extraction process of the chest micro-motion signal can be the same as the chest micro-motion signal extraction process in the traditional technology.

[0025] S18, constructing the trajectory matrix, reconstructing the singular values ​​and performing variational mode decomposition on the chest micro-motion signal through the heartbeat detection algorithm; wherein, if the SSA window adaptive switch is turned on, the SSA sliding window is adjusted according to the set SSA window adaptive adjustment strategy, otherwise the SSA sliding window is maintained unchanged.

[0026] It can be understood that the trajectory matrix construction and singular value reconstruction are the existing processing of improved singular spectrum analysis (SSA), variational mode decomposition, namely VMD, and its specific processing flow can be understood in the same way as the existing SSA and VMD processing flow in the art to complete the separation of respiratory and heartbeat signals. At this time, if the SSA window adaptive switch is automatically or manually turned on, the improved singular spectrum analysis process will adjust the SSA sliding window according to the SSA window adaptive adjustment strategy set in this application, so as to accurately and dynamically adapt the SSA analysis processing under different noise levels, thereby significantly improving the signal-to-noise ratio of the heartbeat signal.

[0027] S20, inversely transforming the frequency domain heartbeat signal after variational mode decomposition into a time domain heartbeat signal for output.

[0028] It can be understood that the inverse transform can use the inverse FFT transform to convert the frequency domain signal back to the time domain signal output. The above-mentioned human heartbeat detection method based on ultra-wideband radar adopts a two-stage design of pre-detection and algorithm detection. First, in the pre-detection stage, the matching of the heartbeat detection algorithm is completed and the SSA window adaptive switch is designed to achieve accurate matching of the heartbeat detection algorithm of the current detection scene based on the clothing condition of the measured area and support whether to turn on the adaptive SSA dynamic parameter adjustment, so that the detection process supports the designed SSA window adaptive adjustment strategy based on signal energy distribution, and uses the basic detection algorithm or high-order detection algorithm of adaptive matching to achieve accurate heartbeat detection processing in real-time detection scenarios. Under the premise of not excessively increasing the computational burden, the adaptive matching algorithm and the SSA window adaptive adjustment strategy are used to effectively improve the signal-to-noise ratio of the heartbeat signal and improve the reliability of human heartbeat detection.

[0029] In one embodiment, Figure 2 As shown in FIG. 1 , the process of determining whether to enable the SSA window adaptation switch includes: If the clothing thickness range matches a high-order detection algorithm from the detection algorithm library, the SSA window adaptation switch is turned on.

[0030] It can be understood that in this embodiment, the system can automatically determine whether to turn on the SSA window adaptive switch by confirming whether the currently matched heartbeat detection algorithm is a high-order detection algorithm, that is, when the clothing thickness interval determined by the current table lookup matches a high-order detection algorithm from the detection algorithm library, the system automatically turns on the SSA window adaptive switch, so that during the processing of the high-order detection algorithm, the SSA window adaptive adjustment is turned on for the process of improving the singular spectrum analysis, thereby ensuring that the signal-to-noise ratio of the heartbeat signal in the detection echo is significantly improved in this scenario. Conversely, when the system confirms that the currently matched heartbeat detection algorithm is not a high-order detection algorithm but a basic detection algorithm, the SSA window adaptive switch is automatically turned off, so that the heartbeat detection in the current scenario can be directly completed through the basic detection algorithm, saving the system's computing resources while ensuring that the basic heartbeat signal detection reliability requirements are met.

[0031] In one embodiment, further, Figure 2 As shown, the set SSA window adaptive adjustment strategy includes: When the short-term energy of the chest micro-motion signal in the SSA sliding window is lower than the set first energy threshold and the main frequency ratio is lower than 50%, the SSA sliding window is reduced by one level at a set time interval within the set first window size range; the first window size range is 20ms to 15ms; When the short-term energy of the chest micro-motion signal in the SSA sliding window is higher than the second energy threshold and the main frequency accounts for more than 70%, the SSA sliding window is expanded by one level at every set time interval within the set second window size range; the second window size range is 30ms to 40ms; the first energy threshold is less than the second energy threshold.

[0032] It can be understood that in order to adjust the parameters of the SSA algorithm in real time according to the signal characteristics and significantly reduce the intervention of manual parameter adjustment, a new dynamic parameter optimization design is given in this specification. In this embodiment, the dynamic adaptive selection of the SSA sliding window is first designed, and the signal energy of the chest micro-motion signal in the SSA sliding window is calculated to obtain the short-time energy To determine the noise level, short-time energy The calculation method can be as follows: ;

[0033] in, is the discrete time point of the chest micro-motion signal in the SSA sliding window k The sample value at k is a discrete time index, indicating the k sampling points, k from arrive t , which means covering all sampling points in the current SSA sliding window,L is the size of the SSA sliding window. The energy proportion of the heartbeat frequency band (0.8 Hz to 2.5 Hz, i.e., the main frequency) in the chest micro-motion signal can be analyzed by FFT (Fourier transform), which can be used to effectively identify respiratory harmonic interference.

[0034] Specifically, a dedicated SSA window adaptive adjustment strategy is designed as follows: (1) High noise scenario (i.e., STE < first energy threshold and main frequency ratio < 50%): within the set first window size range, the SSA sliding window is reduced at set time intervals (the set first window size range is 20ms to 15ms) to reduce the integration time of noise energy in complex noise scenarios, effectively avoid the accumulation of noise in a long time window to dominate the trajectory matrix, deeply fit the fast-changing characteristics of the weak heartbeat signal, and use the energy in a short window to distribute more evenly and concentrate the signal energy to better retain its time domain details and improve the signal-to-noise ratio of the heartbeat signal.

[0035] (2) Low-noise scenario (i.e., STE> second energy threshold and main frequency ratio> 70%): within the set second window size range, the SSA sliding window is expanded at set time intervals (the second window size range is 30ms to 40ms). By adaptively expanding the SSA sliding window, more signal energy is distributed in the long window in low-noise scenarios, and more signal details are retained to improve the signal-to-noise ratio of the heartbeat signal. The sliding window size is updated every 1 second or less.

[0036] The set time interval may be 1 second or less than 1 second, and the system may automatically trigger the SSA sliding window adjustment based on sliding statistics (such as the moving average of STE); each gear may be 1ms or 5ms, and may be selected based on the speed of improvement of the signal-to-noise ratio of the echo heartbeat signal in the experimental measurement, or may be set by the tester based on the empirical value when the system is powered on and initialized. The first energy threshold may be set as follows: by pre-collecting signal samples at different noise levels (such as high noise and low noise), and marking the noise level of each signal sample (such as the signal-to-noise ratio SNR), the short-time energy (STE) and the main frequency ratio (the percentage of the main frequency energy in the total energy) of all signal samples are calculated, the STE histogram of high noise and low noise samples is drawn, the critical value for distinguishing the two is determined, and after analyzing the distribution law of the main frequency ratio in the noise environment, the STE value that makes the signal sample classification (high noise) the highest accuracy is selected as the set first energy threshold. The setting of the second energy threshold is similar, except that the STE value that makes the signal sample classification (low noise) the highest accuracy is selected as the set second energy threshold.

[0037] Optionally, the pseudo code corresponding to the above SSA window adaptive adjustment strategy may be as follows:

[0038] By further identifying high-noise signals through the first energy threshold and initiating the corresponding SSA window adaptive adjustment, and further identifying low-noise signals through the second energy threshold, the interference of different noise levels on subsequent signal analysis and processing (such as SSA reconstruction) can be accurately and effectively reduced. At the same time, the use of fixed large windows in noise-dominated scenarios can be avoided to prevent noise energy accumulation from causing signal distortion, greatly improving the signal-to-noise ratio of the heartbeat signal while improving the robustness of the algorithm.

[0039] In one embodiment, the above-mentioned human heartbeat detection method based on ultra-wideband radar further includes the steps of: When the short-time energy of the chest micro-motion signal within a set number of consecutive SSA sliding windows is higher than the first energy threshold and the main frequency proportion is lower than 50%, the first energy threshold is updated to the average signal energy of the chest micro-motion signal within a set number of consecutive SSA sliding windows.

[0040] Specifically, considering that when the ultra-wideband radar continuously detects the same measured area or detects the measured areas of different measured targets, the signal energy may drift as a whole due to external influences (such as the emergence of new interference signals) or radar internal influences, the first energy threshold adaptive dynamic adjustment mechanism is further designed and introduced in this embodiment to adaptively update the first energy threshold according to the real-time signal energy characteristics to automatically match the current detection environment changes, avoiding frequent manual recalibration of the first energy threshold. The set number can be but is not limited to 5 or 8, so as to avoid misjudgment due to too short judgment time while taking into account the energy threshold update efficiency and accuracy. It can be understood that when the main frequency ratio remains unchanged but the short-term energy of the chest micro-motion signal in the continuous set number of SSA sliding windows is significantly lower than the first energy threshold (such as lower than half of the first energy threshold), the same first energy threshold adaptive dynamic adjustment mechanism can also be used to achieve a similar detection environment change adaptation effect.

[0041] In one embodiment, the above-mentioned human heartbeat detection method based on ultra-wideband radar further includes the steps of: When the short-time energy of the chest micro-motion signal within a set number of consecutive SSA sliding windows is lower than the second energy threshold and the main frequency accounts for more than 70%, the second energy threshold is updated to the average signal energy of the chest micro-motion signal within a set number of consecutive SSA sliding windows.

[0042] Specifically, considering that when the ultra-wideband radar continuously detects the same measured area or detects the measured areas of different measured targets, the signal energy may drift as a whole due to external influences or radar internal influences (such as power supply reduction), the second energy threshold adaptive dynamic adjustment mechanism is further designed and introduced in this embodiment to adaptively update the second energy threshold according to the real-time signal energy characteristics to automatically match the current detection environment changes and avoid frequent manual recalibration of the second energy threshold. The set number can be, but is not limited to, 5 or 8, so as to avoid misjudgment due to too short judgment time while taking into account the energy threshold update efficiency and accuracy. It can be understood that when the main frequency ratio remains unchanged but the short-term energy of the chest micro-motion signal in a set number of SSA sliding windows is significantly higher than the second energy threshold (such as higher than 1 times the second energy threshold), the same second energy threshold adaptive dynamic adjustment mechanism can also be used to achieve a similar detection environment change adaptation effect.

[0043] In one embodiment, the set SSA window adaptive adjustment strategy includes: The short-term energy of the chest micro-motion signal in the current SSA sliding window is matched with each standard sample with a noise level annotation in the signal energy sample library to determine the noise level of the chest micro-motion signal in the current SSA sliding window; the noise level includes high noise or low noise; When the noise level of the chest micro-motion signal in the SSA sliding window is high noise, the SSA sliding window is reduced by one level at a set time interval within a set first window size range; the first window size range is 20ms to 15ms; When the noise level of the chest micro-motion signal in the SSA sliding window is low noise, the SSA sliding window is expanded by one level at every set time interval within the set second window size range; the second window size range is 30ms to 40ms.

[0044] It can be understood that another SSA window adaptive adjustment strategy is also designed in this embodiment. By collecting and annotating chest micro-motion signal samples in advance to build a signal energy sample library, a mapping relationship between standard samples of different signal energies and noise levels is established, wherein the main frequency ratio of the standard sample is also <50% and >70% (the two main frequency ratio judgment boundaries are the optimal boundaries that take into account both judgment accuracy and calculation amount through experimental statistics) as the distinction criteria. Therefore, when the short-term energy of the chest micro-motion signal in the current SSA sliding window matches the standard sample of the corresponding signal energy, the system will automatically compare whether the main frequency ratio of the chest micro-motion signal in the current SSA sliding window meets the main frequency ratio constraint of the standard sample. If so, the noise level of the standard sample is used to calibrate the noise level of the chest micro-motion signal in the current SSA sliding window. Otherwise, the chest micro-motion signal in the current SSA sliding window is discarded, and the chest micro-motion signal in the next SSA sliding window is directly judged and processed.

[0045] Through the aforementioned SSA window adaptive adjustment strategy, the interference of different noise levels on subsequent signal analysis and processing can be accurately and effectively reduced, while avoiding the use of fixed large windows in noise-dominated scenarios to prevent noise energy accumulation from causing signal distortion, greatly improving the signal-to-noise ratio of the heartbeat signal while improving the robustness of the algorithm.

[0046] In one embodiment, Figure 3 As shown in the figure, the basic detection algorithm includes improved singular spectrum analysis and variational mode decomposition, and the high-order detection algorithm includes improved singular spectrum analysis, variational mode decomposition and frequency domain enhancement; among them, the process of frequency domain enhancement includes: S101, using Morlet wavelet to perform time-frequency decomposition on the signal after singular value reconstruction to obtain a high-resolution time-frequency diagram; S103, compressing the energy of the high-resolution time-frequency graph along the direction of the instantaneous frequency derivative to obtain a high-resolution time-frequency graph after frequency redistribution; S105, retaining the signals in the frequency band of 0.8 Hz to 2.5 Hz on the high-resolution time-frequency diagram after frequency redistribution and setting the remaining signals to zero, thereby completing the extraction of the heartbeat frequency band; S107, using the designed IIR notch filter to adaptively filter out the intrusive respiratory harmonic frequency from the signal extracted from the heartbeat frequency band; the center frequency of the IIR notch filter is the dynamically detected respiratory harmonic frequency; S109, outputting the time-domain heartbeat signal obtained after the variational mode decomposition by replacing the time-domain heartbeat signal obtained after filtering out the intrusive respiratory harmonic frequency.

[0047] It can be understood that when the signal after singular value reconstruction is subjected to wavelet transform, the Morlet wavelet is used to decompose the signal in time and frequency to obtain a high-resolution time-frequency diagram, and then the frequency is redistributed by compressing the energy along the instantaneous frequency derivative direction to enhance the sparsity of the heartbeat frequency, retaining the 0.8Hz to 2.5Hz frequency band, and setting the rest to zero to complete the frequency band extraction. Compared with the traditional FFT, it has a higher resolution in the time-frequency domain and can more effectively separate the overlapping respiratory harmonics and heartbeat signals.

[0048] Among them, the adaptive comb filter is designed based on the principle that the fundamental frequency of the respiratory signal (such as 0.1Hz to 0.5Hz) and its harmonics (such as 1Hz, 1.5Hz) may still invade the heartbeat frequency band, and the respiratory frequency can be detected dynamically. , set the notch filter to suppress the heartbeat frequency band and The harmonics can be specifically realized by using an IIR notch filter, whose center frequency is set to the detected respiratory harmonic frequency and whose bandwidth is ±0.1Hz to further ensure its filtering performance. Finally, since the time domain heartbeat signal obtained by frequency domain enhancement in the high-order detection algorithm is more accurate and reliable than the time domain heartbeat signal obtained after variational mode decomposition, the time domain heartbeat signal after filtering out the intrusive respiratory harmonic frequency can be output instead of the time domain heartbeat signal obtained after variational mode decomposition.

[0049] In some embodiments, the verification example may be as follows: The quantitative indicators used in the performance verification include signal-to-noise ratio and heart rate error (RMSE). Among them, the signal-to-noise ratio : ; in, For at time point t The heartbeat signal amplitude at n ( t ) is at time point t The heart rate error (RMSE) is calculated by comparing the UWB detection result with the R peak (ECG) measurement value and calculating the root mean square error (BPM). The test scenarios are as follows: static scenario, the target is in a sitting or lying position; dynamic scenario, the target is slightly shaking.

[0050] Data collection: 10 subjects (5 sitting still, 5 lightly exercising) synchronously collected UWB radar signals and ECG reference data in low noise (laboratory) and high noise (Wi-Fi interference environment) scenarios.

[0051] Traditional UWB heartbeat detection method (the previous generation technology related to this application as a baseline): fixed window SSA ( L=30) + bandpass filtering (0.8Hz to 2.5Hz). The human heartbeat detection method (new method) based on ultra-wideband radar in this application adopts dynamic SSA window + frequency domain enhancement + adaptive comb filtering. Among the evaluation indicators, the signal-to-noise ratio (SNR) calculates the signal-to-noise energy ratio of the heartbeat frequency band (0.8Hz to 2.5Hz), and the heart rate error (RMSE) is specifically the root mean square error (BPM) between the UWB detected heart rate and the ECG heart rate.

[0052] Experimental data (as shown in Table 1) and result analysis: Table 1

[0053] According to Table 1, it can be found that the experiment shows that when the SSA window adaptive switch is turned on, the SNR of the human heartbeat detection method based on ultra-wideband radar in low-noise and high-noise scenarios is increased by 35% and 125% respectively compared with the previous generation technology. The frequency domain enhancement technology effectively suppresses respiratory harmonics and environmental noise, and the dynamic window optimization reduces motion interference. The RMSE is also reduced by 66% compared with the previous generation technology in high-noise scenarios; although the human heartbeat detection method based on ultra-wideband radar increases the delay from 5ms / frame to 10ms / frame due to dynamic calculation, it is still within an acceptable range (<50ms). It can be seen that dynamic parameter optimization + frequency domain enhancement technology significantly improves the accuracy of heartbeat detection in complex environments, especially suitable for sports or high-noise scenarios, and provides a direct and reliable practical basis for the application of the above-mentioned technology in medical monitoring and smart homes.

[0054] It should be understood that although the above process Figure 1 and Figure 3 The steps in the flowchart are shown in the order indicated by the arrows, but the steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of the steps, and the steps can be executed in other orders. Figure 1 and Figure 3 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequentially, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0055] In one embodiment, Figure 4As shown, a human heartbeat detection device 100 based on ultra-wideband radar is provided, including a pre-detection module 11, a loss table lookup module 13, an algorithm switch module 15, a micro-motion extraction module 17, a signal processing module 19 and a detection output module 21. Among them, the pre-detection module 11 is used to use the ultra-wideband radar to pre-detect the detected area of ​​the detected target, and according to the pre-detected radar echo indication, the ultra-wideband radar moves to the calibration distance point of the detected area and then locks; the ultra-wideband radar is a radar on the pan-tilt base, and the detected area is the human chest. The loss table lookup module 13 is used to look up the preset energy thickness table according to the pre-detected radar echo energy loss to determine the clothing thickness range of the detected area; the energy thickness table includes a mapping of different echo energy losses and clothing thickness ranges. The algorithm switch module 15 is used to call the matching heartbeat detection algorithm from the detection algorithm library according to the clothing thickness range and determine whether to turn on the SSA window adaptive switch; the heartbeat detection algorithm includes a basic detection algorithm and a high-order detection algorithm. The micro-motion extraction module 17 is used to receive the detection echo of the measured area to the ultra-wideband radar and use the background difference method to suppress static clutter on the detection echo, and then extract the chest micro-motion signal of the measured area from the detection echo through the heartbeat detection algorithm. The signal processing module 19 is used to construct the trajectory matrix, reconstruct the singular value and perform variational mode decomposition on the chest micro-motion signal through the heartbeat detection algorithm; wherein, if the SSA window adaptive switch is turned on, the SSA sliding window is adjusted according to the set SSA window adaptive adjustment strategy, otherwise the SSA sliding window is maintained unchanged. The detection output module 21 is used to inversely transform the frequency domain heartbeat signal after variational mode decomposition into a time domain heartbeat signal output.

[0056] The above-mentioned human heartbeat detection device 100 based on ultra-wideband radar adopts a two-stage design of pre-detection and algorithm detection. First, in the pre-detection stage, the matching of the heartbeat detection algorithm is completed and an SSA window adaptive switch is designed to achieve accurate matching of the heartbeat detection algorithm of the current detection scene based on the clothing condition of the measured area and support whether to turn on the adaptive SSA dynamic parameter adjustment, so that the detection process supports the designed SSA window adaptive adjustment strategy based on signal energy distribution, and uses the basic detection algorithm or high-order detection algorithm of adaptive matching to achieve accurate heartbeat detection processing in real-time detection scenarios. Without excessively increasing the computational burden, the adaptive matching algorithm and the SSA window adaptive adjustment strategy are used to effectively improve the signal-to-noise ratio of the heartbeat signal and at the same time, improve the reliability of human heartbeat detection.

[0057] In one embodiment, the algorithm switch module 15 turns on the SSA window adaptive switch when the clothing thickness interval matches a high-order detection algorithm from the detection algorithm library.

[0058] For specific limitations of a human heartbeat detection device 100 based on ultra-wideband radar, please refer to the corresponding limitations of the various embodiments of the human heartbeat detection method based on ultra-wideband radar above, which will not be repeated here.

[0059] Figure 5 Schematic diagram of a human heartbeat detection device 4 provided in one embodiment of the present application. Figure 5 As shown, the human heartbeat detection device 4 of this embodiment includes: a processor 401, a memory 402, and a computer program 403 stored in the memory 402 and executable on the processor 401. When the processor 401 executes the computer program 403, the steps in the above-mentioned human heartbeat detection method embodiments based on ultra-wideband radar can be implemented. Alternatively, when the processor 401 executes the computer program 403, the functions of each module / unit in the above-mentioned human heartbeat detection device 100 embodiment based on ultra-wideband radar are implemented.

[0060] The human heartbeat detection device 4 can be a computer device such as a desktop computer, a notebook, a palm computer, a cloud server, etc. The human heartbeat detection device 4 can include but is not limited to a processor 401 and a memory 402. Those skilled in the art can understand that Figure 5 The present invention is merely an example of the human heartbeat detection device 4 and does not constitute a limitation on the human heartbeat detection device 4 . The present invention may include more or fewer components than those shown in the figure, or different components.

[0061] The processor 401 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0062] The memory 402 may be an internal storage unit of the human heartbeat detection device 4, for example, a hard disk or memory of the human heartbeat detection device 4. The memory 402 may also be an external storage device of the human heartbeat detection device 4, for example, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the human heartbeat detection device 4. The memory 402 may also include both an internal storage unit and an external storage device of the human heartbeat detection device 4. The memory 402 is used to store computer programs and other programs and data required by the computer device.

[0063] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units.

[0064] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. The computer program may include computer program code, and the computer program code may be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of practice. For example, in some areas, the computer-readable medium may not include electric carrier signals and telecommunication signals.

[0065] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus dynamic random access memory (RambusDRAM, referred to as RDRAM) and interface dynamic random access memory (DRDRAM).

[0066] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0067] The above embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of protection of the invention. It should be pointed out that, for those of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the attached claims.

Claims

1. A human heartbeat detection method based on ultra-wideband radar, characterized in that: Includes steps: The ultra-wideband radar is used to pre-detect the measured area of ​​the measured target, and the ultra-wideband radar is moved to the calibrated distance point of the measured area according to the pre-detected radar echo indication and then locked; the ultra-wideband radar is a radar of the pan-tilt base, and the measured area is the human chest cavity; Determine the clothing thickness range of the measured area by looking up a preset energy thickness table according to the pre-detected radar echo energy loss; The energy thickness table includes a mapping of different echo energy losses and clothing thickness intervals; According to the clothing thickness interval, a matching heartbeat detection algorithm is called from the detection algorithm library and it is determined whether to turn on the SSA window adaptive switch; the heartbeat detection algorithm includes a basic detection algorithm and a high-order detection algorithm; After receiving the detection echo of the measured area to the ultra-wideband radar and performing static clutter suppression on the detection echo by using a background difference method, a chest micro-motion signal of the measured area is extracted from the detection echo by using the heartbeat detection algorithm; The heartbeat detection algorithm is used to construct a trajectory matrix, reconstruct singular values, and perform variational mode decomposition on the chest micro-motion signal; wherein, if the SSA window adaptive switch is turned on, the SSA sliding window is adjusted according to the set SSA window adaptive adjustment strategy, otherwise the SSA sliding window is maintained unchanged; The frequency domain heartbeat signal after variational mode decomposition is inversely transformed into a time domain heartbeat signal output.

2. The human heartbeat detection method based on ultra-wideband radar according to claim 1 is characterized in that: The process of determining whether to enable the SSA window adaptation switch includes: If the clothing thickness interval matches a high-order detection algorithm from the detection algorithm library, the SSA window adaptive switch is turned on.

3. The human heartbeat detection method based on ultra-wideband radar according to claim 2 is characterized in that: The set SSA window adaptive adjustment strategies include: When the short-term energy of the chest micro-motion signal in the SSA sliding window is lower than the set first energy threshold and the main frequency ratio is lower than 50%, the SSA sliding window is reduced by one level at a set time interval within the set first window size range; the first window size range is 20ms to 15ms; When the short-time energy of the chest micro-motion signal in the SSA sliding window is higher than the second energy threshold and the main frequency accounts for more than 70%, the SSA sliding window is expanded by one level at every set time interval within the set second window size range; the second window size range is 30ms to 40ms; the first energy threshold is less than the second energy threshold.

4. The human heartbeat detection method based on ultra-wideband radar according to claim 2 is characterized in that: The set SSA window adaptive adjustment strategies include: Matching the short-term energy of the chest micro-motion signal in the current SSA sliding window with each standard sample with a noise level label in the signal energy sample library to determine the noise level of the chest micro-motion signal in the current SSA sliding window; the noise level includes high noise or low noise; When the noise level of the chest micro-motion signal in the SSA sliding window is high noise, the SSA sliding window is reduced by one level at a set time interval within a set first window size range; the first window size range is 20ms to 15ms; When the noise level of the chest micro-motion signal in the SSA sliding window is low noise, the SSA sliding window is expanded by one level at every set time interval within a set second window size range; the second window size range is 30ms to 40ms.

5. The human heartbeat detection method based on ultra-wideband radar according to claim 3 is characterized in that: The method further comprises the steps of: When the short-time energy of the chest micro-motion signal within a set number of consecutive SSA sliding windows is higher than the first energy threshold and the main frequency accounts for less than 50%, the first energy threshold is updated to the average signal energy of the chest micro-motion signal within a set number of consecutive SSA sliding windows.

6. The human heartbeat detection method based on ultra-wideband radar according to claim 3 is characterized in that: The method further comprises the steps of: When the short-time energy of the chest micro-motion signal within a set number of consecutive SSA sliding windows is lower than the second energy threshold and the main frequency accounts for more than 70%, the second energy threshold is updated to the average signal energy of the chest micro-motion signal within a set number of consecutive SSA sliding windows.

7. The human heartbeat detection method based on ultra-wideband radar according to any one of claims 1 to 6, characterized in that: The basic detection algorithm includes improved singular spectrum analysis and variational mode decomposition, and the high-order detection algorithm includes improved singular spectrum analysis, variational mode decomposition and frequency domain enhancement; wherein the frequency domain enhancement process includes: The Morlet wavelet is used to perform time-frequency decomposition on the signal after singular value reconstruction to obtain a high-resolution time-frequency diagram; Compressing the energy of the high-resolution time-frequency graph along the direction of the instantaneous frequency derivative to obtain a high-resolution time-frequency graph after frequency redistribution; The signals in the frequency band of 0.8 Hz to 2.5 Hz on the high-resolution time-frequency diagram after frequency redistribution are retained and the remaining signals are set to zero to complete the extraction of the heartbeat frequency band; The designed IIR notch filter is used to adaptively filter out the intrusive respiratory harmonic frequency from the signal extracted from the heartbeat frequency band; the center frequency of the IIR notch filter is the dynamically detected respiratory harmonic frequency; The time-domain heartbeat signal obtained after the variational mode decomposition is replaced by the time-domain heartbeat signal after the intrusive respiratory harmonic frequency is filtered out for output.

8. A human heartbeat detection device based on ultra-wideband radar, characterized in that: include: A pre-detection module is used to perform pre-detection on a detected area of ​​a detected target using an ultra-wideband radar, and according to the pre-detected radar echo indication, the ultra-wideband radar is moved to a calibrated distance point of the detected area and then locked; the ultra-wideband radar is a radar of a pan-tilt base, and the detected area is a human chest cavity; A loss table lookup module, used to look up a preset energy thickness table to determine the clothing thickness interval of the measured area according to the pre-detected radar echo energy loss; the energy thickness table includes a mapping between different echo energy losses and clothing thickness intervals; An algorithm switch module, used to call a matching heartbeat detection algorithm from a detection algorithm library according to the clothing thickness interval and determine whether to turn on the SSA window adaptive switch; The heartbeat detection algorithm includes a basic detection algorithm and a high-order detection algorithm; A micro-motion extraction module is used to receive the detection echo of the measured area to the ultra-wideband radar and use the background difference method to suppress static clutter on the detection echo, and then extract the chest micro-motion signal of the measured area from the detection echo by using the heartbeat detection algorithm; A signal processing module, used for performing trajectory matrix construction, singular value reconstruction and variational mode decomposition on the chest micro-motion signal through the heartbeat detection algorithm; wherein, if the SSA window adaptive switch is turned on, the SSA sliding window is adjusted according to the set SSA window adaptive adjustment strategy, otherwise the SSA sliding window is maintained unchanged; The detection output module is used to inversely transform the frequency domain heartbeat signal after variational mode decomposition into a time domain heartbeat signal output.

9. The human heartbeat detection device based on ultra-wideband radar according to claim 8, characterized in that: The algorithm switch module turns on the SSA window adaptive switch when a high-order detection algorithm is matched in the clothing thickness interval from the detection algorithm library.

10. A human heartbeat detection device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the human heartbeat detection method based on ultra-wideband radar as described in any one of claims 1 to 7 are implemented.

Images