Multi-person respiration detection and rapid person identification system based on millimeter wave radar

Abstract

The invention relates to the technical field of personnel identification, and discloses a multi-person breathing detection and rapid personnel identification system based on a millimeter wave radar, and the multi-person breathing detection based on the millimeter wave radar comprises the following steps: for a to-be-detected human body breathing target existing in a radar vision field scene; obtaining a coordinate value of the target through two-dimensional Fourier transform; original data before static clutter filtering is utilized, and two-dimensional Fourier transform is carried out on the original data again to obtain a new RA spectrum; a breathing micro-displacement waveform is obtained through a matched filter; the velocity waveform of chest breathing micro-motion is obtained through the first-order derivative of the displacement; biomedicine-assisted breathing feature efficient extraction: nine features are extracted from a single-cycle breathing waveform. According to the multi-person respiration detection and rapid personnel identification system based on the millimeter wave radar, the limitation of a traditional method on short-time respiration feature extraction is broken through, the effectiveness and timeliness of feature extraction are improved, and a solid foundation is provided for subsequent implementation of the rapid and accurate personnel identification system.

Description

Technical Field

[0001] The present invention relates to the technical field of personnel identification, and particularly to a multi-person respiration detection and rapid personnel identification system based on millimeter-wave radar. Background Art

[0002] With the development of technology and the improvement of human living standards, personnel identification technology (Person Identification, PID) has an increasingly wide range of demands and application prospects in scenarios such as security systems, access control systems, medical monitoring, and human-computer interaction. The core idea of PID technology is to identify different personnel and authenticate identity information through the unique and stable specific characteristics of individuals. According to the types of characteristics relied on, it can be mainly divided into the following three categories: personnel identification based on biological characteristics, personnel identification based on dynamic characteristics, and personnel identification based on vital signs. Personnel identification based on vital signs mainly uses respiration and heartbeat characteristics to achieve personnel identification. Respiration and heartbeat, as human vital signs, have the characteristics of spontaneous generation, small micro-movement amplitude, and difficulty in forgery. Such information contains less privacy information and can be extracted and recognized even in a natural static state. Therefore, compared with biological characteristics and dynamic characteristics, personnel identification based on vital signs has the characteristics of high security and wide applicable scenarios, effectively making up for the deficiencies of personnel identification based on biological characteristics and dynamic characteristics. Respiration characteristics, as an important and stable vital sign, can effectively reflect the respiration habits between different individuals and show significant differences among different individuals and different physiological and health states, which provides a strong biomedical theoretical basis for using respiration as vital sign information for personnel identification. Therefore, respiration signs are also increasingly used by researchers to achieve PID.

[0003] For the intelligent classification and recognition of respiration information, according to different implementation means, it can be roughly divided into two categories: The first category is respiration sign extraction + classifier, that is, first, the respiration signs are mined and extracted, and then these signs are used as the input of the classifier to achieve the classification and recognition of respiration information; the second category is the pure "black-box" method, that is, using the powerful learning ability of highly complex classification networks such as Convolutional Neural Network (CNN), directly taking the detected signal waveform as the input to achieve the classification and recognition of respiration. The advantages and disadvantages of these two existing methods are summarized and presented.

[0004] The first type of method screens out discriminative respiratory features through manual design or modeling, and then realizes respiratory recognition based on a simple classification and recognition network. According to the different natures of the extracted features, this type of method can be further divided into three methods: time-domain feature extraction, frequency-domain / time-frequency domain feature extraction, and other feature extraction. The time-domain feature extraction method uses the statistical time-domain parameters of the respiratory waveform (such as respiratory time, respiratory cycle, respiratory amplitude, etc.) or morphological features (such as waveform slope, waveform shape, waveform area, etc.) as the classification basis. This type of method is simple to calculate and the features are highly interpretable. However, there is still room for further improvement in terms of the number of extractable features, information sources, and real-time performance (strong dependence on time). The frequency-domain / time-frequency domain feature extraction method mainly extracts the frequency-domain energy distribution (such as main frequency, harmonic energy ratio, etc.) or time-frequency joint features (such as wavelet coefficient entropy, marginal spectrum energy, etc.). Since this type of method requires data of a specific duration, it also has a dependence on time and relies on the calculation accuracy of time-frequency analysis methods. The other feature extraction method mainly quantifies the dynamic complexity of the respiratory signal through the complexity measurement parameters of the signal (such as sample entropy, multi-scale entropy, etc.). Such features are sensitive to weak non-linear changes, but the physical or physiological interpretability of this method is very poor, and the accuracy of feature calculation often depends on empirical tuning.

[0005] The second type of method directly inputs the signals obtained in respiratory detection into a high-complexity classification network (such as CNN) for classification and recognition. In recent years, with the continuous iterative development of high-complexity neural network technology, many researchers have also conducted a large number of studies on the second type of method, designing various network structures and models to achieve personnel recognition based on respiratory information. However, as is well known, an important problem with this type of "black box" method lies in its high dependence on training samples and computing resources, which results in poor performance in terms of implementation cost and efficiency.

[0006] In summary, although the current human respiratory sign classification and recognition system based on millimeter-wave bio-radar has some applications in the market, there are obviously multiple problems that urgently need to be broken through and solved. These problems and deficiencies can be summarized into the following two aspects: 1) Insufficient effectiveness of respiratory sign extraction; 2) Insufficient timeliness of respiratory sign recognition. Specifically, first of all, most of the existing methods interpret or utilize respiratory sign information from the perspective of signal processing or pure "black box", and do not closely link the extraction and analysis process of respiratory signs with biomedical mechanisms, resulting in ineffective and inaccurate feature extraction in the existing technology. Secondly, the existing sign recognition methods, whether based on feature extraction or the "black box" type CNN, usually rely on data of a long time and several respiratory cycles. This technical defect makes the existing methods unable to meet the high-timeliness requirements of daily personnel recognition applications, and it is also one of the important problems that plague the field of millimeter-wave bio-radar technology. Summary of the invention

[0007] In view of the deficiencies in the prior art, the present invention provides a multi-person breathing detection and rapid personnel identification system based on millimeter-wave radar, which solves the problems in the above-mentioned background technology.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions: a multi-person breathing detection and rapid personnel identification system based on millimeter wave radar, including: multi-person breathing detection based on millimeter wave radar and efficient extraction of breathing features assisted by biomedicine; Multi-person breathing detection based on millimeter-wave radar: For the human breathing target to be detected in the radar field of view, assume that the radial distance between the mth target and the center of the array is Rm, and the horizontal incident angle of the target echo is θm; Fourier transform is performed in the fast time dimension and channel dimension respectively; according to and , reconstruct the radial distance Rm and horizontal incident angle θm of each target; The coordinate values ​​of the target are obtained by two-dimensional Fourier transform; after the position coordinates of each target are obtained, the data of each respiratory target is separated from the original data, and the respiratory waveform detection operation is performed based on the separated data; The original data before static clutter filtering is used to re-transform it into a two-dimensional Fourier transform to obtain a new RA spectrum, and then the target position is obtained. , extract new RA spectra in Location data; Through the matched filter, it is eliminated before the neighborhood integration, and then the neighborhood integration is performed to obtain the respiratory micro-motion displacement waveform ; According to the relationship between displacement and velocity, the velocity waveform of the chest breathing micro-motion is obtained through the first-order derivative of displacement; Efficient extraction of biomedically assisted respiratory features: 9 features extracted from a single-cycle respiratory waveform, of which the 9 features can be divided into 4 categories: basic displacement waveform, transition phase, instantaneous velocity, and lung emptying phase.

[0009] Preferably, after obtaining the coordinate value of the target through two-dimensional Fourier transform, the target peak point position in the RA spectrum is found through a constant false alarm rate detector and peak detection.

[0010] Preferably, before performing the two-dimensional Fourier transform, a static clutter filtering technique is used to remove the static background in the scene.

[0011] Preferably, before performing phase extraction, DC compensation is performed on the DC term, and DC compensation is achieved by a linear least squares estimation method.

[0012] Preferably, the basic displacement waveform category, which includes the characteristic breathing frequency, expiratory area, and breathing ratio; The breathing frequency represents the number of breaths per unit time; in the waveform of a single breathing cycle, the breathing frequency is expressed as the reciprocal of the period.

[0013] Preferably, the transition stage category, which includes the area and duration of the characteristic inhalation-to-exhalation transition stage; These two characteristics are extracted based on the four-segment segmentation method of a single-cycle breath, focusing on the inhalation-to-exhalation transition stage, that is, the stage from the start of inhalation deceleration to the end of early exhalation.

[0014] Preferably, the instantaneous velocity category is based on the characteristics of the instantaneous velocity of chest movement, including the number of characteristic expiratory velocity peaks, the moment of maximum velocity during inhalation, and the maximum velocity during inhalation.

[0015] Preferably, the lung emptying stage category; this category includes the displacement of the characteristic early exhalation stage based on the lung emptying characteristics.

[0016] The present invention provides a multi-person breathing detection and rapid personnel identification system based on millimeter-wave radar. It has the following beneficial effects: Aiming at the two problems of insufficient effectiveness and timeliness of feature extraction in the classification and recognition of respiratory signs in existing research, a high-efficiency respiratory feature extraction scheme assisted by biomedical-related principles is proposed. This scheme relies on the respiratory micro-motion waveform within a single breathing cycle (including one inhalation and one exhalation), deeply excavates the effective respiratory features in a short time, and through the expansion of the time-domain respiratory feature extraction dimension and the correlation analysis of the respiratory micro-motion waveform features and respiratory biomedical features within a single cycle, breaks through the limitations of traditional methods in short-time respiratory feature extraction, improves the effectiveness and timeliness of feature extraction, and provides a solid foundation for the subsequent realization of a fast and accurate personnel identification system.

[0017] In addition, combining the millimeter-wave radar respiratory micro-motion waveform detection in a multi-person scenario with the high-efficiency respiratory feature extraction assisted by biomedicine, this paper proposes a multi-person breathing detection and rapid personnel identification system based on millimeter-wave radar. This system is based on an FMCW radar, detects the position information and respiratory signals of multiple human targets within the field of view, and through 9 effective features extracted from the single-cycle respiratory signal and a simple SVM classification network, realizes the efficient and accurate identification of target personnel within the time of a single breathing cycle (about 3 seconds). This system well meets the dual requirements of accuracy and real-time performance in actual PID applications, fully proves its wide applicability and high identification efficiency, and has the potential for popularization and application in actual scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is the overall algorithm flowchart of the present invention; Figure 2 It is a schematic diagram of the breathing target data separation process of the present invention; Figure 3 The overall flow chart of the respiratory waveform detection based on phase extraction of the present invention; Figure 4 It is a single-cycle respiratory displacement waveform diagram of the present invention; Figure 5 It is the respiratory speed waveform diagram of the present invention; Figure 6 9 characteristic graphs extracted from the displacement and velocity waveforms of the breathing of the present invention; Figure 7 For the actual test scenario of the present invention, three targets were randomly seated in three seats; Figure 8 2T4RMIMOFMCW radar used in the experiment of the present invention; Figure 9 For the collection scenario of the training data of the present invention, three targets (A, B, and C) sit alone in front of the radar for data collection; Figure 10 The results of the millimeter wave radar multi-person breathing detection and single waveform extraction of the present invention; Figure 11 This is the confusion matrix of the classification results of the present invention, and the overall accuracy is shown in the yellow area. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] The present invention provides a technology: a multi-person breathing detection and rapid personnel identification system based on millimeter-wave radar, including: multi-person breathing detection based on millimeter-wave radar and efficient extraction of breathing features assisted by biomedicine.

[0021] Multi-person breathing detection based on millimeter wave radar For multiple human breathing targets to be detected in the radar field of view, assuming that the radial distance between the mth target and the array center is Rm, and the horizontal incident angle of the target echo is θm. In this scenario, the intermediate frequency (IF) signal of the FMCW radar in the slow time dimension can be approximated as:

[0022] (3-3) in, represents the speed of light, Denote the thoracic displacement of the m-th target along the slow time and the incident angle , is the wavelength of the carrier frequency, which is used as the starting point for subsequent inversion of the respiratory signal.

[0023] Perform Fourier transforms on the formula respectively in the fast time dimension and the channel dimension to obtain:

[0024] (3-4) wherein, and respectively represent the phase terms after Fourier transforms in the fast time dimension and the channel dimension in the formula , and it is a Dirac function. The SsS in the formula is the reconstructed range-azimuth spectrum (RA spectrum). According to and , the radial distance Rm and the horizontal incident angle θm of each target can be reconstructed as:

[0025] (3-5)

[0026] (3-6) However, in the actual implementation process, since the measurement scene does not only contain human targets but also some other static backgrounds. Therefore, before performing the two-dimensional Fourier transform, it is necessary to first remove the static background in the scene through static clutter filtering technology. After obtaining the coordinate values of the targets through the two-dimensional Fourier transform, it is also necessary to find the position of the target peak point in the RA spectrum through a Constant False Alarm Rate (CFAR) detector and peak detection, so as to finally achieve the detection of the targets. The overall algorithm flow is as shown in the attached Figure 1 of the specification.

[0027] After obtaining the position coordinates of each target, it is also necessary to separate the data of each respiratory target from the original data and perform the operation of respiratory waveform detection based on the separated data. Since the time-domain averaging method is used in static clutter filtering, the phase of the original signal in the slow time dimension is changed, and the respiratory waveform detection utilizes the phase change in the slow time dimension. Therefore, in order to ensure the accuracy of subsequent respiratory waveform detection, it is necessary to use the original data before static clutter filtering, that is, the formula , perform a new two-dimensional Fourier transform on it to obtain a new RA spectrum, and then according to the target positions obtained as shown in the formula and the formula , extract the new RA spectrum at ​ Data of the position. Extraction The data of the position can be achieved by integrating within the neighborhood of each extreme point of the target position. The process of separating the respiration target data is as shown in the appendix of the specification Figure 2 as follows

[0028] It can be seen that in Equation in addition to the phase term related to the micro displacement , there are also an amplitude term and a phase term related to Rm. When Rm has been calculated through Equation , we can thus design a matching filter to eliminate it before performing the neighborhood integration. The expression of the matching filter is

[0029] (3-7) After passing through the filter shown in Equation , the neighborhood integration is then carried out

[0030] (3-8) By extracting the phase of this equation, the respiration micro displacement waveform is obtained

[0031] (3-9) wherein represents the phase extraction operation. According to the relationship between displacement and velocity, the velocity waveform of thoracic respiration micro motion can be conveniently obtained through the first derivative of the displacement

[0032] (3-17) In the measured data, the equal sign in Equation does not hold exactly. This is because in an actual radar system, including the direct wave between Tx and Rx and other factors such as devices, a direct current (DC) term is usually generated. Therefore, before performing the phase extraction, it is necessary to perform DC compensation on this DC term to eliminate its negative impact on the phase extraction accuracy, and the DC compensation can be achieved through the method of linear least squares estimation. The algorithm flow of the respiration waveform detection based on phase extraction is as Figure 3 shown

[0033] Efficient extraction of respiration characteristics assisted by biomedicine According to the relevant definitions of biomedicine, we have extracted a total of as shown in the appendix of the specification Figures 4 - 6Nine features shown in , where six features are from the displacement waveform of Equation (3-9) and three features are from the velocity waveform of Equation (3-10). It should be noted that all features are from a single respiratory cycle. All these nine features can be divided into four categories: basic displacement waveform category, transition phase category, instantaneous velocity category, and lung emptying phase category. Category 1: Basic displacement waveform category. This category includes Feature #1: Respiratory rate, #2: Expiratory area, and #3: Respiratory ratio. The respiratory rate, also known as the breathing rate, represents the number of breaths per unit time. In the waveform of a single respiratory cycle, the respiratory rate is expressed as the reciprocal of the period, as follows:

[0034] where Ti and Te represent the durations of inspiration and expiration, respectively. The respiratory rate is affected by respiratory mechanics, genetic factors, and the interaction between the respiratory and cardiovascular systems, so there are significant differences among different individuals.

[0035] The expiratory area is defined as the area between the displacement curve and the time axis during the expiratory phase, as Figure 4 shown. This area represents the cumulative thoracic displacement and can be expressed as:

[0036] where tb and tc represent the start and end times of the expiratory phase, respectively. The expiratory area can reflect the tidal volume (TV) of an individual to a certain extent and is mainly affected by gender, age, respiratory stimulation, and elastic load.

[0037] The respiratory ratio, also known as the inspiration-to-expiration ratio (IE ratio), reflects the proportion of time occupied by inspiration and expiration within a single cycle. The variation of this feature is affected by gender, exercise, and anthropometric factors, and its expression is:

[0038] Category 2: Transition phase category. This category includes Features #4 and #5: Area and duration of the inspiration-to-expiration transition phase. These two features are extracted based on a four-segment segmentation method for single-cycle breathing, focusing on the inspiration-to-expiration transition phase, that is, the phase from the start of inspiratory deceleration to the end of early expiration, as Figure 4 shown. The expressions for these two features are:

[0039]

[0040] Among them, t1 and t2 represent the start and end times of the inhalation-to-exhalation transition phase respectively. These two characteristics, area and duration, reflect the total displacement of the chest wall during the transition phase of breathing and the breathing efficiency of the human body during this transition phase. Chest displacement is an important indicator for evaluating respiratory muscle activation and lung function, and is related to the state of respiratory muscle activity and vital signs. In addition, individuals with different exercise habits also have differences in respiratory muscle strength and lung function, thus affecting the breathing efficiency during this transition phase.

[0041] Category 3: Instantaneous velocity type. The three characteristics of this category are proposed for the first time in this paper and are characteristics based on the instantaneous velocity of chest movement, including Feature #6: Number of peak exhalation velocities, #7: Moment of maximum velocity during inhalation, and #8: Maximum velocity during inhalation. These characteristics are as Figure 5 shown and can be expressed respectively as:

[0042]

[0043]

[0044] Among them, pi represents the i-th peak point in the exhalation phase, and ta represents the start time of the inhalation phase. Chest velocity reflects the change in air flow during breathing. Air flow is affected by the compliance and resistance of the respiratory system and the anatomical structure of the airway, and has individual differences. Exhalation is a passive process driven by elastic recoil, accompanied by a large pressure loss, resulting in air flow turbulence and generating multiple velocity peaks. These peaks reflect the differences in the elasticity and stability of the individual's chest and lungs. In contrast, the inhalation process is relatively smooth because it is driven by the active contraction of the respiratory muscle group, so the maximum velocity during the inhalation phase is extracted. This makes and have certain value in evaluating the performance of the diaphragm and external intercostal muscles, and help to distinguish the strength and endurance of respiratory muscles.

[0045] Category 4: Lung emptying phase type. This category is based on lung emptying characteristics and is also a newly proposed characteristic in this study, including Feature #9: Displacement in the early exhalation phase. This feature helps to evaluate the short-term changes in chest wall movement at the beginning of exhalation and is expressed as:

[0046] Among them, td is the end time of the initial exhalation phase. This feature is jointly affected by the coordinated relaxation of the diaphragm and intercostal muscles and helps to distinguish the respiratory muscle function and chest wall mechanical characteristics among different individuals.

[0047] In addition, studies have shown that lung function and cardiorespiratory fitness are related to various factors including age, height, weight, smoking status, blood pressure, and heart rate, and are also affected by factors such as genetics, body composition, trunk size, exercise habits, and physical health status. These factors show significant heterogeneity among different individuals and stability within a short period for a single individual, which provides further support for the rationality of the above feature extraction.

[0048] Experimental verification Experimental verification scenario setting. As shown in the instruction manual Figure 7 : Actual test scenario, three targets are randomly seated on three seats. As shown in the instruction manual Figure 8 : The 2T4RMIMOFMCW radar used in the experiment. As shown in the instruction manual Figure 9 : Acquisition scenario of training data, three targets (A, B, and C) are separately seated in front of the radar for data acquisition.

[0049] The experimental verification scenario setting is as shown in the instruction manual Figures 7 - 9 As shown. All data is collected through a 2T4R FMCW millimeter-wave radar chip. There are 3 seats placed at different positions in front of the radar sensor, and 3 participants can randomly choose to sit down. The 3 participants are individuals of the same gender, with similar body types, ages, and health conditions. Such individuals have similar physiological states, and thus similar breathing characteristics, making them relatively difficult to identify. The entire data acquisition process includes two steps: 1) The 3 participants (labeled as A, B, and C respectively) are separately seated in front of the radar sensor to obtain training data; 2) The 3 participants sit on the three seats simultaneously and randomly change their positions to obtain test data. The entire algorithm is implemented through MATLAB code. First, the one-dimensional thoracic displacement waveform of the target is obtained through the phase extraction algorithm and single-cycle separation is performed, and then the one-dimensional thoracic velocity waveform is obtained by taking the derivative; finally, based on the single-cycle thoracic displacement and velocity waveforms, the features as shown in the instruction manual Figures 4 - 6 As shown are extracted, and an SVM model with a linear kernel function is used to complete the classification. The results of multi-person breathing detection are as shown in the instruction manual Figure 10 As shown. The radar parameters are as shown in the following table.

[0050] Nine features extracted from the displacement and velocity waveforms of breathing System parameters Value RF bandwidth 24.5 GHz ~ 26.5 GHz Chirp sweep time 420 μs <![CDATA[ADC * Sampling count]]> 128 Chirp transmit interval 2.5 ms Antenna array layout 2T4R Tx interval 23.08 mm Rx interval 5.77 mm Finally, 1251 and 971 single-cycle training data and test data are obtained through experimental measurement, forming the training set and the test set. The data distribution is as follows: There are 502, 442, and 307 samples in the training set from participant A, participant B, and participant C respectively; there are 351, 358, and 262 samples in the test set from participant A, participant B, and participant C respectively. The recognition and classification results are as shown in the instruction manualFigure 11 As shown, it can be seen that the proposed features achieve an accuracy rate of 98.56% even in the identification of personnel with similar physiological states. In addition, for the entire process of breathing detection and identification, only a single-cycle breathing waveform is required, and high-accuracy breathing identification can be completed in about 3 seconds.

[0051] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the element.

[0052] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-person breathing detection and rapid personnel identification system based on millimeter-wave radar, characterized in that: Including: Efficient extraction of respiratory characteristics for multi-person respiration detection based on millimeter-wave radar and biomedical assistance; Multi-person respiration detection based on millimeter-wave radar: For the human respiration targets to be measured in the radar field of view scenario, assume that the radial distance of the m-th target from the array center is Rm, and the horizontal incident angle of the target echo is θm; Implemented by performing Fourier transforms in the fast time dimension and the channel dimension respectively; According to and , the radial distance Rm and the horizontal incident angle θm of each target are reconstructed; Obtain the coordinate values of the targets through two-dimensional Fourier transform; After obtaining the position coordinates of each target, separate the data of each respiration target from the original data, and perform the operation of respiration waveform detection based on the separated data; Using the original data before static clutter filtering, perform a two-dimensional Fourier transform on it again to obtain a new RA spectrum, and then according to the obtained target position , extract the data of the new RA spectrum at the position of ; It is eliminated by a matched filter before neighborhood integration, and then neighborhood integration is performed to obtain a respiratory micro-motion displacement waveform ; According to the relationship between displacement and velocity, obtain the velocity waveform of thoracic respiratory micro-motion through the first derivative of displacement; Efficient extraction of respiratory characteristics with biomedical assistance: 9 characteristics extracted within a single-cycle respiration waveform, and these 9 characteristics can be divided into 4 categories: basic displacement waveform category, transition stage category, instantaneous velocity category, and lung emptying stage category.

2. The multi-person respiration detection and rapid personnel identification system based on millimeter-wave radar according to claim 1, wherein: After obtaining the coordinate values of the targets through two-dimensional Fourier transform, find the position of the target peak point in the RA spectrum through a constant false alarm rate detector and peak detection.

3. The multi-person respiration detection and rapid personnel identification system based on millimeter-wave radar according to claim 1, characterized in that: Before performing two-dimensional Fourier transform, remove the static background in the scene through static clutter filtering technology.

4. The multi-person respiration detection and rapid personnel identification system based on millimeter-wave radar according to claim 1, characterized in that: Before performing phase extraction, perform DC compensation on this DC term, and implement DC compensation through the method of linear least squares estimation.

5. The multi-person respiration detection and rapid personnel identification system based on millimeter-wave radar according to claim 1, characterized in that: Basic displacement waveform category, this category includes characteristic respiratory frequency, exhalation area, and respiration ratio; Respiratory frequency, representing the number of breaths per unit time; In the waveform of a single respiratory cycle, the respiratory frequency is expressed as the reciprocal of the period.

6. The multi-person respiration detection and rapid personnel identification system based on millimeter-wave radar according to claim 1, characterized in that: Transition stage category, this category includes the area and duration of the characteristic inhalation-to-exhalation transition stage; These two characteristics are extracted based on the four-segment segmentation method of a single-cycle respiration, focusing on the inhalation-to-exhalation transition stage, that is, the stage from the start of inhalation deceleration to the end of early exhalation.

7. The multi-person respiration detection and rapid personnel identification system based on millimeter-wave radar according to claim 1, characterized in that: The instantaneous velocity category is a characteristic based on the instantaneous velocity of chest movement, Including the number of characteristic exhalation velocity peak times, the moment of maximum velocity during inhalation, and the maximum velocity during inhalation.

8. The multi-person respiration detection and rapid personnel identification system based on millimeter-wave radar according to claim 1, wherein: Lung emptying stage category; This category includes the displacement in the characteristic early exhalation stage based on lung emptying characteristics.

Images