Identity and sign information matching method based on radar and SARFID

Abstract

The invention provides an identity and sign information matching method based on radar and SARFID. The method comprises the following steps: acquiring a radar signal and an SARFID signal; pre-processing the SARFID signal: positioning a tag by using tag data readings collected from different moving positions on a reader antenna to obtain a tag position representing a human target position; the data fusion between the radar and the SARFID comprises the following steps of: 1, acquiring radar signals of a plurality of human targets at different positions by using the radar to obtain a multi-target point cloud cluster based on the radar; 2, determining the position of each radar point cloud of the target point cloud cluster based on the radar; 3, for a human target, calculating the distance between the radar point cloud and the label position; 4, obtaining a radar center point of a human target by using a DBSCAN algorithm; and obtaining the position of the human target in which the radar and the RFID of the human target are fused.

Description

Technical Field

[0001] The present invention relates to the technical field of millimeter wave radar and RFID, and in particular to an identity and vital sign information matching method based on radar and SARFID. Background Art

[0002] Radio frequency identification (RFID) and millimeter wave radar are important components of interawareness integration. Due to their portability, comfort and all-weather capabilities, RFID and millimeter wave radar are widely used in smart home and smart elderly care to detect physiological information (breathing and heartbeat) and locate the target.

[0003] Millimeter wave radar technology can be used to detect important information such as breathing, heartbeat, gestures, and the identities of multiple people. Reference [1] uses millimeter wave radar to monitor breathing and heartbeat, and proposes a two-dimensional synthetic aperture radar imaging recognition system based on mobile radar for multi-target vital sign monitoring and target position detection, which can be used in complex outdoor environments. In reference [2], a single-input multi-output continuous wave radar system and adaptive digital beamforming technology are used to simultaneously detect the breathing rate of multiple human targets at unknown locations. Generally speaking, in millimeter wave radar signal processing, the detected echo data is usually mathematically modeled to obtain the location and physiological information of the target. If you want to further confirm the identity of the physiological information, you must use a neural network and a data set. Because this is a multi-classification problem. However, most references [3] show that the performance (accuracy) of the neural network training decreases as the number of people increases or when facing unknown identities.

[0004] RFID technology is also used for physiological information detection. Reference [4] proposed an algorithm for sensing human heart rate, which achieves accurate estimation of heart rate variability by constructing a tag array attached to the chest and a heart rate signal mapping model. However, when using RFID to detect physiological information (respiration and heartbeat), it is necessary to attach the tag to a fixed position on the clothing (chest) because the skin greatly affects the sensing accuracy of the RFID tag. In other words, when estimating the accuracy of vital signs, in addition to considering the impact of respiration and heartbeat signals on the channel, it is also necessary to consider the type and placement of the tag, which increases uncertainty. Because RFID tags are directional and carry unique identity information, RFID technology is easy to classify multiple targets and operate at longer distances, so RFID is more suitable for multi-target positioning. Reference [5] proposed a holographic synthetic aperture (SAR) RFID positioning system with aperture position error compensation without knowing the exact trajectory of the synthetic aperture. This makes it possible to detect the target location and provide unique identity information, but compared with millimeter-wave radar, the accuracy of using RFID to detect physiological information is affected by the performance and placement of the RFID tag.

[0005] In summary, in response to the above problems, the present invention proposes an identity and vital sign information matching method based on radar and SARFID (Synthetic Aperture Radio Frequency Identification), called M-VIMS. It combines millimeter wave radar and RFID, uses location information as a link, and achieves matching of the target's physiological information and identity information, and the identity recognition accuracy remains constant at 100%.

[0006] Related Literature

[0007] [1]Yan J, Zhang G, Hong H, et al. Phase-Based Human Target 2-DIdentification With a Mobile FMCW Radar Platform [J]. IEEE Transactions onMicrowave Theory and Techniques, 2019, 67(12): 5348–5359.

[0008] [2] Xiong J, Hong H, Zhang H, et al. Multitarget Respiration Detection With Adaptive Digital Beamforming Technique Based on SIMO Radar[J]. IEEE Transactions on Microwave Theory and Techniques, 2020, 68(11): 4814–4824.

[0009] [3] Zhao H, Ma Y, Wang X. MN-UIV: Multimodal Neural Network Enabling User Identity Verification Based on Millimeter Wave Radar[J]. IEEE Internet of Things Journal, 2024, 11(15): 26304–26313.

[0010] [4] Wang C, Xie L, Wang W, et al. Rf-ecg: Heart rate variability assessment based on cots rfid tag array[J]. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, 2018, 2(2): 1–26.

[0011] [5] Zhao R, Wang D, Zhang Q, et al. PEC: Synthetic Aperture RFID Localization with Aperture Position Error Compensation[C]. In 2019 16th Annual IEEE International Conference on Sensing, Communication, and Networking (SECON), 2019: 1–9.

[0012] [6] Ren W, Hou K Y, Wang G, et al. Multi-person respiration detection method using FMCW millimeter wave radar[J]. Journal of Signal Processing, 2021, 37(09): 1581-1588.

[0014] [7]

[181] Motroni A,Nepa P,Tripicchio P,et al.A Multi-Antenna SAR-basedmethod for UHF RFID Tag Localization via UGV[C].In 2018IEEE InternationalConference on RFID Technology&Application(RFID-TA),2018:1–6. Summary of the invention

[0015] The purpose of the present invention is to provide a method for matching identity and vital signs information based on radar and SARFID, combining radar and RFID, using location information as a link, to achieve matching of physiological information and identity information of human targets, and maintaining a constant accuracy rate of 100% for identity recognition. The technical solution is as follows:

[0016] A method for matching identity and vital signs information based on radar and SARFID comprises the following steps:

[0017] Collect radar signals and SARFID signals;

[0018] Preprocessing of SARFID signals: Using tag data readings collected from different positions of the movement on the reader antenna to locate the tag, obtain the tag position representing the human target position;

[0019] The data fusion between radar and SARFID includes the following steps:

[0020] Step 1: Use radar to collect radar signals of multiple human targets at different locations to obtain radar-based multi-target point cloud clusters;

[0021] Step 2: Determine the position of each radar point cloud of the radar-based target point cloud cluster;

[0022] Step 3: For human targets, calculate the distance between the radar point cloud and the label position;

[0023] Step 4: Use the DBSCAN algorithm to obtain the radar center point of the human target as follows:

[0024] For the k-th human target, a distance threshold is set, and the distance between each point cloud in the multi-target point cloud cluster measured by the radar and the label position is calculated, and the radar point cloud is screened according to the distance threshold; by setting the density reachable radius threshold and the point cloud quantity threshold, the screened radar point cloud is clustered to obtain the final radar point cloud measured for the k-th target; based on the obtained final radar point cloud, the radar center point of the k-th human target is calculated;

[0025] Step 5: According to the radar center point of the k-th human target and the tag position of the k-th human target obtained by SARFID, the human target position of the k-th human target fused by radar and RFID is obtained.

[0026] Furthermore, the method for collecting radar signals and SARFID signals is as follows: the starting points of the radar and reader antennas are the same, the reader moves in a straight line to continuously collect tag data readings to obtain SARFID signals; the radar collects radar signals at different positions; for two consecutive tag data readings, the distance moved by the reader antenna is less than one-quarter of the wavelength distance.

[0027] Further, the radar signal is preprocessed to determine the location of the human target obtained by the radar signal and obtain the vital sign information, the method is as follows:

[0028] Step 1: Use time division multiplexing multiple input multiple output (TDM-MIMO) technology to form multi-target intermediate frequency signals, and perform three-dimensional discrete time Fourier transform on the received intermediate frequency signals;

[0029] Step 2: Remove clutter and determine the location of the human target obtained by the radar signal;

[0030] Step 3: Extract the phase waveform of the human target physiological signal;

[0031] Step 4: Decompose the phase waveform of the human target physiological signal through variational mode decomposition (VMD) to obtain the respiratory waveform and heartbeat waveform; according to the respiratory frequency and heartbeat frequency, set the appropriate frequency range respectively, select the modal components of the respiratory signal and heartbeat signal, and select the maximum value from the corresponding modal components as the frequency of breathing and heartbeat.

[0032] Furthermore, clutter is removed by static clutter elimination method and two-dimensional constant false alarm detection (2D-CFAR) method.

[0033] Furthermore, for the breathing signal and the heartbeat signal, the modal components between 0.1-0.5 and 0.9-2 Hz are selected respectively.

[0034] Furthermore, the method of step 2 is: for a point cloud cluster consisting of K targets, the total number of radar point clouds is i, and the point cloud cluster of the kth target is expressed as

[0035] A k ={(x1,y1),(x2,y2),…,(x i ,y i )}

[0036] Assume that the position of the radar relative to the origin when the radar collects data for the sth time is

[0037] B (s) =(x,y)

[0038] Through the coordinate system transformation, relative to the origin, the point cloud position is

[0039] Furthermore, the method of step 3 is:

[0040] Assume that the tag position of the kth human target measured by SARFID technology is

[0041] C=(x k ,y k )

[0042] The distance between the radar point cloud of the kth human target measured by radar and the label position is

[0043] d k =||AC||2.

[0044] Furthermore, in step 4, the distance threshold is half of the human body width; the threshold of the density reachable radius is the human body width, and the threshold of the number of point clouds is in the range of 25-50.

[0045] Furthermore, in step 4, the radar center point of the target is obtained by the following formula, where l is the total number of final radar point clouds:

[0046]

[0047] The beneficial effects of the present invention are as follows: The radar and SARFID-based identity and vital signs information matching method proposed by the present invention, called M-VIMS, uses location information as a link to match personal identity information with breathing and heartbeat information. It is applicable to multi-target scenarios, so that the identity recognition accuracy is fixed at 100%. This method avoids the trouble of establishing a data set and building a neural network, and also avoids the influence of tag location and performance. This method screens human targets (removes environmental noise, such as wall reflections), reduces positioning deviations in multi-target scenarios, and reduces physiological signal detection errors. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 Overall flow chart

[0049] Figure 2 Radar signal processing flow chart

[0050] Figure 3 SARFID schematic diagram

[0051] Figure 4 Experimental diagram

[0052] Figure 5 Comparison of identity recognition accuracy experimental results

[0053] Figure 6 Diagram of screening human targets

[0054] Figure 7 Comparison of experimental results of positioning error

[0055] Figure 8 Respiration and heart rate measurement results DETAILED DESCRIPTION

[0056] The present invention will now be further described with reference to the accompanying drawings and embodiments.

[0057] The present invention uses location information as a link to match personal identity information with breathing and heartbeat information. First, the multi-target radar echo signal is processed using multiple input multiple output (MIMO) technology and digital beamforming (DBF) technology, and the location information is obtained using static clutter elimination and two-dimensional constant false alarm detection (2D-CFAR). According to the threshold screening of 2D-CFAR, point cloud clusters of multiple targets are formed. After obtaining the location information corresponding to multiple targets, the variational mode decomposition (VMD) technology is used to obtain physiological signal information (breathing and heartbeat). Secondly, SARFID technology is used to obtain the tag position, that is, the location of the target. Since the RFID tag carries its own identity (ID) information, the ID and the location of the target are bound. Then, in order to merge the location detected by the millimeter wave radar and the location detected by the RFID, the distance between the two locations is calculated, and the radar point cloud is screened. The density-based noise application spatial clustering (DBSCAN) method is used to distinguish the remaining point clouds corresponding to each target for clustering, and the average value is calculated to obtain the final position. Finally, with the location information as the connection, the identity information and the physiological signal information are matched one by one.

[0058] The identity and vital sign information matching method based on radar and SARFID of the present invention mainly consists of three parts. Figure 1 As shown in the figure, (1) preprocessing of radar signal. (2) preprocessing of SARFID signal. (3) data fusion between radar and SARFID.

[0059] (1) The process of radar signal preprocessing is as follows Figure 2 As shown, the following steps are included:

[0060] Step 1: In order to improve the angular resolution and multi-target positioning capability of the radar, the present invention adopts time division multiplexing multiple input multiple target output (TDM-MIMO) technology to form the intermediate frequency signal of multiple targets. The received intermediate frequency signal is subjected to three-dimensional discrete time Fourier transform.

[0061] Step 2: After Fourier transform, the static clutter removal method and the two-dimensional constant false alarm detection (2D-CFAR) method are used to remove clutter and determine the location of the human target. Multiple target point clouds, i.e. the location of the target, are formed by determining the detection threshold of 2D-CFAR.

[0062] Step 3: After determining the target position, separate the phase waveforms of multiple target physiological signals through digital beam forming (DBF), phase unwrapping and variance judgment technology. The prior art already has this technology, for example, documents 1 and 2 both involve this technology. The embodiment of the present invention adopts the solution provided by document [6].

[0063] Step 4: Decompose the phase waveform by variational mode decomposition (VMD) to obtain the breathing and heartbeat waveforms. By setting appropriate parameters, the modal components between 0.1-0.5 and 0.9-2 Hz are used as breathing signals and heartbeat signals respectively, and the maximum value is selected as the frequency of breathing and heartbeat.

[0064] (2) The preprocessing process of SARFID signal is as follows Figure 3 As shown, the following steps are included:

[0065] Step 1: In the line-of-sight path, when the reader antenna moves along a straight line, the distance between the reader antenna and the tag changes due to the movement of the reader antenna. Due to phase ambiguity, for two consecutive tag data readings, the distance moved by the reader antenna should be less than one-quarter of the wavelength distance.

[0066] Step 2: Use the tag data readings collected from different moving positions on the reader antenna to locate the tag and obtain the target position. The embodiment of the present invention adopts the solution provided by the document [7].

[0067] (3) Data fusion between radar and SARFID, such as Figure 4 As shown in the figure, the experimental scene consists of millimeter wave radar and RFID. It includes the following steps:

[0068] Step 1: Through the above technical steps, a point cloud cluster of multiple targets based on millimeter-wave radar is obtained. Assuming a point cloud cluster consisting of K targets and a point cloud cluster with i point clouds, the point cloud cluster of the kth target can be expressed as

[0069] A k ={(x1,y1),(x2,y2),…,(x i ,y i )}

[0070] Step 2: The starting point of the radar and reader antennas is the same, and the RFID continuously collects data along a straight line at a constant speed. The millimeter-wave radar collects data three times at three locations. During the collection process, repeated target point clouds will appear. Therefore, the radar position relative to the origin of the sth collection data is

[0071] B (s) =(x,y)

[0072] Through the coordinate system transformation, relative to the origin, the positions of all point clouds are

[0073]

[0074] Step 3: Assume that the kth target position measured by SARFID technology is

[0075] C=(x k ,y k )

[0076] The distance between the target point cloud cluster and the label position measured by the radar is

[0077] d k =‖AC‖2

[0078] Step 4: Use the DBSCAN (Density-Based Spatial Clustering of Applications with Noise, DBSCAN) algorithm to obtain the radar center point of each target. The method is as follows: for the kth target, set a distance threshold, where the distance threshold is half the width of the human body, and calculate the distance between the target point cloud cluster measured by the radar and the label position, and filter out the radar point cloud according to the distance threshold; by setting the density reachable radius threshold and the point cloud number threshold, cluster the filtered radar point cloud to obtain the final radar point cloud. The threshold of the density reachable radius is the width of the human body, and the threshold of the number of point clouds is in the range of 25-50. Then, based on the final radar point cloud obtained, the radar center point of this target is obtained by the following formula, where l is the total number of final radar point clouds:

[0079]

[0080] Step 5: Average the final radar multiple target point cloud center points and the positions obtained by SARFID to obtain the final target position that is a fusion of radar and RFID.

[0081] Experimental Results

[0082] In order to verify the feasibility of the algorithm, an experimental scenario is composed of millimeter wave radar and SARFID, such as Figure 4As shown. To verify the accuracy of the algorithm, a millimeter wave radar and a contact sensing device (SCG) were used, one for measurement and the other for reference. The starting point of the millimeter wave radar and the reader antenna is the same. The RFID tag is placed in the middle below the human heart and lungs. The millimeter wave radar collects target position and physiological signals at positions (0, 0), (0.75, 0) and (1.75, 0). The reader antenna collects three tag data at a speed of 0.05m / s and an equal interval of 0.02m.

[0083] Experimental results

[0084] (1) The accuracy of identity recognition is constant at 100%. Figure 5 As shown in the figure, the present invention compares several commonly used deep learning architectures and verifies the performance of the algorithm. First, the frequency of physiological signals (breathing and heartbeat) is obtained by preprocessing the millimeter wave radar signal. A data set consists of data from three targets. Subsequently, the Conv1d, BiLSTM and Encoder networks are trained respectively, with a cross loss function and 5-fold cross validation. The experimental results are shown in Figure 5 As shown. Figure 5 It can be seen that the average recognition accuracy of Conv1d, BiLSTM and Encoder networks for identity information is 84%, 64% and 98% respectively. In contrast, since the tag location information and tag ID information are bound in RFID technology, the identity recognition rate remains unchanged at 100%. Compared with the M-VIMS algorithm proposed in the present invention, the identity information recognition accuracy of different architectures trained by neural networks is low and unsatisfactory. These three factors can be explained as follows. First, as long as the recognition of identity information is achieved by establishing a data set and using a neural network, its recognition accuracy will not maintain 100% accuracy. Second, if the identity information of the person being tested changes, such as in a meeting scenario where the identity information of the test person changes every day, the practicality of the neural network and the data set will be significantly reduced. Third, as the number of people increases, the accuracy of personnel identity information recognition decreases. In contrast, RFID technology itself has an identification number, so it does not face problems such as reduced recognition rate, construction of data sets and neural networks.

[0085] (2) Screening of human targets. Figure 6 As shown in (a), before determining the positions of the three targets, noise (such as wall reflections) needs to be filtered out. Figure 6 (b) shows the positioning results from millimeter wave radar, RFID and actual target position. The algorithm proposed in this invention filters out non-human targets.

[0086] (3) Reduce positioning deviation. Since the algorithm in the present invention confirms the identity information by confirming the position, the key to the algorithm lies in whether the position is consistent, that is, whether the position is consistent, and the position measured by millimeter wave radar and SARFID is the same. The actual positions of the three targets are (0.25, 1.25), (1.25, 1.25) and (2.25, 1.25), respectively. The positions of the three targets detected by radar are (0.239, 1.298), (1.21, 1.37) and (2.1, 1.29), respectively. The positions of the three targets detected using SARFID are (0.29, 1.2), (1.2, 1.28) and (2.28, 1.21), respectively. Compared with the actual positions of the three targets, Figure 7 As shown, the deviations between the positions detected by the radar and the actual positions are 0.049m, 0.126m and 0.155m, respectively, with an average deviation of 0.11m. The deviations between the positions detected by the SARFID technology and the actual positions are 0.064m, 0.058m and 0.05m, respectively, with an average deviation of 0.05733m. The center point bias after radar point cloud clustering is greater than that of the SARFID technology because the width of the human body is about 0.5m. The tags in the SARFID technology are unique and directional, so their deviations are relatively small compared to radar technology. After averaging the positions obtained by the radar and SARFID, three positions are obtained, with coordinates of (0.2645, 1.249), (1.205, 1.325) and (2.19, 1.25), respectively. The deviations from the actual position are 0.015m, 0.087m and 0.06m, respectively, with an average deviation of 0.054m. Compared with radar and SARFID technology, the position deviation of the algorithm proposed in the present invention is significantly reduced.

[0087] (4) Reduce physiological information errors. After determining the identity and location information, the next step is to evaluate the accuracy of the extracted physiological signals to achieve a one-to-one correspondence between identity, location, and physiological data. When the target is stationary, the point cloud coverage of each target is about 50 cm, indicating that the point cloud is not only generated by breathing and heartbeat, but also by other limb movements. The width of the human body is about 50 cm, with the heart and lungs located in the center. The diameter of the human chest cavity (which houses the heart and lungs) is usually about 25.3 cm. The tag is also placed in the center of the body. Using the directionality of the tag, the position range measured by RFID can limit the detection area of ​​the radar and filter out the noise of limb movement. In previous experiments, due to position deviation, only the phase with the highest energy value within the position deviation range of each frame in the range angle map was used for phase unwrapping to determine the breathing and heartbeat rates. This method calculates the distance between each point cloud and the RFID tag measurement position and deletes any point cloud that exceeds this distance. The contact sensor measures the actual breathing and heartbeat rates, while the M-VIMS algorithm estimates these rates using the number of heartbeats per minute (bpm) as an evaluation metric. The error is calculated as the difference between the actual value and the estimated value. Figure 8 The results show that the average respiration and heartbeat errors produced by the traditional algorithm were 0.86bpm and 1.17bpm, respectively. The traditional algorithm here refers to the algorithm that only uses radar detection. In contrast, after applying the M-VIMS algorithm for matching, the average errors of respiration and heartbeat were reduced to 0.71bpm and 0.97bpm, respectively, which are 0.15bpm and 0.2bpm lower than the radar-only method. These findings show that the M-VIMS algorithm provides accurate physiological signal detection for multiple targets.

[0088] according to Figure 5 , Figure 7 and Figure 8 From the results, it is clear that identity, location, and physiological information can be effectively matched. For example, the breathing and heartbeat frequencies of target 1 at position (0.2645, 1.249) are 0.275Hz and 1.1Hz respectively. The static information matching of the other two targets is also similar.

Claims

1. A method for matching identity and vital signs information based on radar and SARFID, comprising the following steps: Collect radar signals and SARFID signals; Preprocessing of SARFID signals: Using tag data readings collected from different positions of the movement on the reader antenna to locate the tag, obtain the tag position representing the human target position; The data fusion between radar and SARFID includes the following steps: Step 1: Use radar to collect radar signals of multiple human targets at different locations to obtain radar-based multi-target point cloud clusters; Step 2: Determine the position of each radar point cloud of the radar-based target point cloud cluster; Step 3: For human targets, calculate the distance between the radar point cloud and the label position; Step 4: Use the DBSCAN algorithm to obtain the radar center point of the human target as follows: For the kth human target, a distance threshold is set, and the radar point cloud is screened according to the distance threshold by calculating the distance between each point cloud in the multi-target point cloud cluster measured by the radar and the label position; By setting the density reachable radius threshold and the point cloud quantity threshold, the screened radar point clouds are clustered to obtain the final radar point cloud measured for the k-th target; based on the obtained final radar point cloud, the radar center point of the k-th human target is calculated; Step 5: According to the radar center point of the k-th human target and the tag position of the k-th human target obtained by SARFID, the human target position of the k-th human target fused by radar and RFID is obtained.

2. The identity and vital sign information matching method according to claim 1, characterized in that: The method for collecting radar signals and SARFID signals is as follows: the starting points of the radar and reader antennas are the same, the reader moves in a straight line to continuously collect tag data readings and obtain SARFID signals; the radar collects radar signals at different positions; for two consecutive tag data readings, the distance moved by the reader antenna is less than one-quarter of the wavelength distance.

3. The identity and vital sign information matching method according to claim 1, characterized in that: The radar signal is preprocessed to determine the location of the human target obtained by the radar signal and obtain the vital signs information. The method is as follows: Step 1: Use time division multiplexing multiple input multiple output (TDM-MIMO) technology to form multi-target intermediate frequency signals, and perform three-dimensional discrete time Fourier transform on the received intermediate frequency signals; Step 2: Remove clutter and determine the location of the human target obtained by the radar signal; Step 3: Extract the phase waveform of the human target physiological signal; Step 4: Decompose the phase waveform of the human target physiological signal through variational mode decomposition (VMD) to obtain the respiratory waveform and heartbeat waveform; according to the respiratory frequency and heartbeat frequency, set the appropriate frequency range respectively, select the modal components of the respiratory signal and heartbeat signal, and select the maximum value from the corresponding modal components as the frequency of breathing and heartbeat.

4. The identity and vital sign information matching method according to claim 1, characterized in that: Clutter is removed by static clutter elimination method and two-dimensional constant false alarm rate detection (2D-CFAR) method.

5. The identity and vital sign information matching method according to claim 3, characterized in that: For the breathing signal and heartbeat signal, the modal components between 0.1-0.5 and 0.9-2 Hz are selected respectively.

6. The identity and vital sign information matching method according to claim 1, characterized in that: The method of step 2 is: for a point cloud cluster consisting of K targets, the total number of radar point clouds is i, and the point cloud cluster of the kth target is expressed as A k ={(x1,y1),(x2,y2),…,(x i ,y i )} Assume that the position of the radar relative to the origin when the radar collects data for the sth time is B (s) =(x,y) Through the coordinate system transformation, relative to the origin, the point cloud position is 7. The identity and vital sign information matching method according to claim 6, characterized in that: The method for step 3 is: Assume that the tag position of the kth human target measured by SARFID technology is C=(x k ,y k ) The distance between the radar point cloud of the kth human target measured by radar and the label position is 8. The identity and vital sign information matching method according to claim 1, characterized in that: In step 4, the distance threshold is half of the human body width; the threshold of the density reachable radius is the human body width, and the threshold of the number of point clouds is in the range of 25-50.

9. The identity and vital sign information matching method according to claim 7, characterized in that: In step 4, the radar center point of the target is obtained by the following formula, where l is the total number of final radar point clouds:

Images