Biological detection method and related apparatus

Abstract

The application discloses a biological body detection method and related device, the biological body detection method comprises the following steps: acquiring a point cloud frame sequence of a management area by using a radar; extracting a first point cloud cluster sequence of each regional unit in the management area from the point cloud frame sequence; for the first point cloud cluster sequence of at least part of the regional units, clustering the first point cloud cluster sequence to obtain a second point cloud cluster sequence of a plurality of potential biological bodies, the second point cloud cluster sequence of the potential biological bodies comprising a plurality of first point sequences of the potential biological bodies; for each potential biological body, obtaining a first micro-motion feature of the potential biological body based on a first phase sequence and / or an amplitude sequence corresponding to each first point sequence of the potential biological body; and clustering each potential biological body based on a similarity measurement between the first micro-motion features of each potential biological body to obtain a plurality of final biological bodies. The above scheme can avoid the situation that the same biological body across regional units is detected as multiple biological bodies.

Description

Technical Field

[0001] This application relates to the field of target detection technology, and in particular to a biological detection method and related apparatus. Background Technology

[0002] Radar-based organism detection refers to using radar to collect point cloud frames of a managed area to detect the presence of organisms in the managed area.

[0003] Radar-based organism detection methods in related technologies can be broadly categorized into two types. The first type utilizes radar to acquire point cloud frames of the managed area. Based on the position of each point in the point cloud frame, clusters are formed to obtain point cloud clusters representing several organisms. In this method, multiple organisms that are geographically adjacent may be detected as a single organism. Geographical proximity includes situations where organisms are close together (e.g., two organisms sitting side-by-side on a double chair) or overlapping (e.g., two organisms walking close together). The second type utilizes radar to acquire point cloud frames of the managed area. The managed area is divided into several regional units, and a first point cloud cluster for each regional unit is extracted from the point cloud frame. For each regional unit, based on the position of each point in the first point cloud cluster, clusters are formed to obtain a second point cloud cluster representing several organisms. This method is a regional unit-level clustering, which avoids the situation where organisms geographically adjacent across regional units are detected as a single organism.

[0004] Through long-term research, the inventors of this application have discovered that while clustering at the regional unit level can prevent organisms across regional units from being detected as a single organism, it also introduces the possibility of the same organism being detected as multiple organisms across regional units, resulting in insufficient detection accuracy. Summary of the Invention

[0005] This application provides a biological detection method and related apparatus, which can solve the problem in the related art where the same biological entity across regional units is detected as multiple biological entities, resulting in insufficient detection accuracy.

[0006] This application provides a method for detecting organisms, comprising: acquiring a point cloud frame sequence of a managed area using radar; extracting a first point cloud cluster sequence of each regional unit in the managed area from the point cloud frame sequence; clustering the first point cloud cluster sequences of at least some regional units to obtain a second point cloud cluster sequence of several potential organisms, wherein the second point cloud cluster sequence of a potential organism includes a number of first point sequences of the potential organism; for each potential organism, obtaining a first micro-motion feature of the potential organism based on the first phase sequence and / or amplitude sequence corresponding to each first point sequence of the potential organism; and clustering each potential organism based on a similarity measure between the first micro-motion features of each potential organism to obtain a number of final organisms.

[0007] This application provides a biological detection device, comprising: an acquisition module, an extraction module, a first clustering module, an acquisition module, and a second clustering module. The acquisition module is used to acquire a point cloud frame sequence of a managed area using radar; the extraction module is used to extract a first point cloud cluster sequence of each regional unit in the managed area from the point cloud frame sequence; the first clustering module is used to cluster the first point cloud cluster sequences of at least some regional units to obtain a second point cloud cluster sequence of several potential organisms, wherein the second point cloud cluster sequence of a potential organism includes several first point sequences of the potential organism; the acquisition module is used to acquire a first micro-motion feature of each potential organism based on a first phase sequence and / or amplitude sequence corresponding to each first point sequence of the potential organism; the second clustering module is used to cluster each potential organism based on a similarity measure between the first micro-motion features of each potential organism to obtain several final organisms.

[0008] This application provides an electronic device, including a memory and a processor, wherein the processor is used to execute program instructions stored in the memory to implement the above-described method.

[0009] This application provides a computer-readable storage medium having program instructions stored thereon, which, when executed by a processor, implement the above-described method.

[0010] This application provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0011] The above scheme separates the first point cloud cluster sequence of each regional unit in the management area from the point cloud frame sequence; it then performs intra-regional clustering on the first point cloud cluster sequence to obtain the second point cloud cluster sequence of each potential organism within the regional unit; based on the first micro-motion features of each potential organism in each regional unit, it performs cross-regional clustering on the second point cloud cluster sequence of each potential organism to obtain several final organisms. On the one hand, because it is regional unit-level clustering, it avoids the situation where multiple organisms with adjacent locations are detected as a single organism. On the other hand, since cross-regional clustering is performed after intra-regional clustering, the multiple potential organisms (partial parts of the final organism) obtained from clustering in different regional units are merged during the cross-regional clustering stage, thus avoiding the situation where the same organism is detected as multiple organisms. Furthermore, compared to using point cloud frames as the organism detection unit, using point cloud frame sequences as the organism detection unit improves detection accuracy because the point cloud frame sequence covers more organism information.

[0012] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this application. Attached Figure Description

[0013] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with this application and, together with the specification, serve to explain the technical solutions of this application.

[0014] Figure 1 This is a schematic flowchart of an embodiment of the biological detection method provided in this application; Figure 2 This is a flowchart illustrating a specific example of interference detection in this application; Figure 3 This is a schematic diagram of the structure of an embodiment of the biological detection device of this application; Figure 4 This is a schematic diagram of the structure of an embodiment of the electronic device of this application; Figure 5 This is a schematic diagram of the structure of an embodiment of the computer-readable storage medium of this application. Detailed Implementation

[0015] The embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0016] In the following description, specific details such as particular system architectures, interfaces, and technologies are presented for illustrative purposes rather than for limiting purposes, in order to provide a thorough understanding of this application.

[0017] In this document, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship. Furthermore, "many" in this document means two or more. The term "at least one" in this document means any combination of at least two of any one or more of a plurality of objects. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C. Finally, the term "several" in this document means any integer greater than 0, such as 1, 2, 3, 4, 5, ...

[0018] Radar-based biological detection methods in related technologies can be divided into two categories: The first method involves deploying radar within the managed area to collect point cloud frames. Based on the position of each point in the point cloud frame, these points are clustered to obtain point cloud clusters representing several organisms. In this method, multiple organisms that are geographically adjacent may be detected as a single organism. Geographical proximity includes situations where organisms are close to each other and side-by-side (e.g., two organisms sitting side-by-side on a double chair) or overlapping (e.g., two organisms walking close together).

[0019] The second method utilizes radar to acquire point cloud frames of the managed area. The managed area is divided into several regional units, and the first point cloud cluster of each regional unit is extracted from the point cloud frames. For each regional unit, based on the position of each point in the first point cloud cluster, the points within the regional unit are clustered to obtain the second point cloud clusters of several organisms. This method is a regional unit-level clustering, which can avoid the situation where organisms that are geographically adjacent across regional units are detected as a single organism.

[0020] Through long-term research, the inventors of this application have discovered that while clustering at the regional unit level can prevent organisms that cross regional units from being detected as a single organism, it also introduces the additional problem of the same organism being detected in different regional units, thus being detected as multiple organisms.

[0021] Based on this, this application proposes a novel radar-based method for detecting biological organisms. Examples of this method are described below.

[0022] Figure 1 This is a schematic flowchart of an embodiment of the biological detection method provided in this application. Figure 1 As shown, in this embodiment, biological detection may include the following steps: S110: The point cloud frame sequence of the managed area is obtained by radar acquisition.

[0023] The execution subject of this embodiment is a biological detection device, which can be any electronic device with biological detection capabilities.

[0024] The management area can be any indoor or outdoor location that requires biological detection.

[0025] The point cloud frame / point cloud cluster / point / phase / amplitude sequence mentioned in this application refers to a time series composed of point cloud frames / point cloud clusters / points / phase / amplitude at multiple consecutive time points. For example, the point cloud frame sequence includes a time series composed of point cloud frames at three consecutive time points t1, t2, and t3.

[0026] S120: Extract the first point cloud cluster sequence of each regional unit in the management area from the point cloud frame sequence.

[0027] The region unit can be a three-dimensional region unit or a two-dimensional region unit. In some embodiments, the management region can be discretized into a two-dimensional planar mesh, and each region unit in the two-dimensional planar mesh can be used as a two-dimensional region unit. In some embodiments, the management region can be voxelized into a three-dimensional cubic mesh, and each region unit in the three-dimensional cubic mesh can be used as a three-dimensional region unit.

[0028] The first point cloud cluster sequence refers to a sequence composed of first point cloud clusters at multiple consecutive time points. The first point cloud cluster at a given time point is composed of points falling into the region cell in the point cloud frame at that time point.

[0029] S130: For the first point cloud cluster sequence of at least some regional units, the first point cloud cluster sequence is clustered to obtain a second point cloud cluster sequence of several potential organisms.

[0030] The second point cloud cluster sequence of a potential organism includes several first point sequences of the potential organism.

[0031] Understandably, although the processing involves sequences, the duration of multiple consecutive moments is relatively short. Generally, the position of an organism changes little over a short period, and its regional unit remains unchanged. Therefore, the first point cloud cluster sequence of a regional unit encompasses several potential organisms, and performing regional unit-level clustering on the first point cloud cluster sequence yields the second point cloud cluster sequence of several potential organisms.

[0032] The first point sequence refers to a sequence consisting of the first points at multiple consecutive moments. The first point is the point in the first point cloud cluster at the corresponding moment that belongs to a potential organism.

[0033] S140: For each potential organism, based on the first phase sequence and / or amplitude sequence corresponding to each first point sequence of the potential organism, obtain the first micro-motion feature of the potential organism.

[0034] Understandably, the first micro-motion feature characterizes the minute motion information of a potential organism at multiple consecutive moments. This minute motion information can include information such as shaking, breathing, heartbeat, displacement, velocity, geometry (body shape, posture), and behavior.

[0035] The first phase sequence is composed of the phase corresponding to the first point at each time step in the first point sequence, and the amplitude sequence is composed of the amplitude corresponding to the first point at each time step in the first point sequence. The phase and amplitude corresponding to the first point can be extracted from the echo signal corresponding to the first point.

[0036] It is understandable that the principle of radar acquiring point cloud frames is as follows: the radar emits electromagnetic waves into the managed area, receives the echo signals reflected by organisms, and converts the echo signals into individual points, which together form a point cloud frame. Therefore, in this embodiment, the phase and amplitude corresponding to the first point can be extracted from the echo signal before it is converted into the first point.

[0037] S150: Based on the similarity measure between the first micro-motion features of each potential organism, cluster each potential organism to obtain several final organisms.

[0038] The above scheme separates the first point cloud cluster sequence of each regional unit in the management area from the point cloud frame sequence; it then performs intra-regional clustering on the first point cloud cluster sequence to obtain the second point cloud cluster sequence of each potential organism within the regional unit; based on the first micro-motion features of each potential organism in each regional unit, it performs cross-regional clustering on the second point cloud cluster sequence of each potential organism to obtain several final organisms. On the one hand, because it is regional unit-level clustering, it avoids the situation where multiple organisms with adjacent locations are detected as a single organism. On the other hand, since cross-regional clustering is performed after intra-regional clustering, the multiple potential organisms (partial parts of the final organism) obtained from clustering in different regional units are merged during the cross-regional clustering stage, thus avoiding the situation where the same organism is detected as multiple organisms. Furthermore, compared to using point cloud frames as the organism detection unit, using point cloud frame sequences as the organism detection unit improves detection accuracy because the point cloud frame sequence covers more organism information.

[0039] The following is a further expansion of S110: The management area can be equipped with one radar or multiple radars with overlapping fields of view. Multiple radars can be two, three, or more. When multiple radars are deployed, each radar acquires its own point cloud frame sequence, and these sub-point cloud frame sequences are then fused to obtain the final point cloud frame sequence.

[0040] Understandably, a single radar, limited by its inherent physical field of view or subject to obstruction, cannot fully cover the managed area, resulting in blind spots. On one hand, the fields of view of multiple radars can compensate for each other, eliminating blind spots. For example, if one radar cannot capture points in a sub-region of the managed area due to a blind spot, but another radar's field of view covers that sub-region, it can capture those points. The resulting fused point cloud frame sequence contains the points from that sub-region, thus eliminating it from the blind spot. On the other hand, the point cloud frame sequence obtained by fusing sub-point cloud frame sequences from multiple radars has higher reliability. Furthermore, deploying multiple radars can prevent the inability to detect organisms due to the failure of a single radar.

[0041] In some embodiments, if multiple radars are set up in the management area, some radars can be designated as primary radars and others as secondary radars. The acquisition confidence level of the primary radars is higher than that of the secondary radars.

[0042] In some embodiments, the radar set up in the management area includes a first radar and a second radar. Based on this, S110 includes: acquiring a first sub-point cloud frame sequence and a second sub-point cloud frame sequence of the management area using the first radar and the second radar respectively; performing alignment processing on the first sub-point cloud frame sequence and the second sub-point cloud frame sequence; and fusing the aligned first sub-point cloud frame sequence and the second sub-point cloud frame sequence to obtain a point cloud frame sequence.

[0043] The first sub-point cloud frame sequence refers to a sequence composed of multiple consecutive first sub-point cloud frames at the first time step. The second sub-point cloud frame sequence refers to a sequence composed of multiple consecutive second sub-point cloud frames at the second time step.

[0044] It is understandable that the acquisition times of the first and second radars are difficult to be completely synchronized; that is, the first sub-point cloud frame sequence corresponds to multiple consecutive first moments, and the second sub-point cloud frame sequence corresponds to multiple consecutive second moments. Alignment processing can align the first and second sub-point cloud frame sequences to multiple consecutive common moments. Alignment processing can be achieved, but is not limited to, interpolation calculations.

[0045] Among them, the point cloud frames at each time point in the point cloud frame sequence are the fusion results of the first sub-point cloud frame at the corresponding time point in the first sub-point cloud frame sequence after alignment processing and the second sub-point cloud frame at the corresponding time point in the second sub-point cloud frame sequence.

[0046] The following is an example illustrating S110: Specific example 0: Q1. The first sub-point cloud frame sequence and the second sub-point cloud frame sequence are obtained by using the first radar and the second radar respectively.

[0047] Q2. The first sub-point cloud frame sequence is interpolated from multiple consecutive first time points to multiple consecutive common time points to obtain the aligned first sub-point cloud frame sequence. The second sub-point cloud frame sequence is interpolated from multiple consecutive second time points to multiple consecutive common time points to obtain the aligned second sub-point cloud frame sequence. Using one of the common time points... For example: Q21. Determine the relationship between multiple consecutive first moments and... Two consecutive first moments , .

[0048] Q22, Yes The first sub-point cloud frame, The first sub-point cloud frame interpolation calculation yielded The first sub-point cloud frame. Where: right First sub-point in the first sub-point cloud frame , First sub-point in the first sub-point cloud frame Perform interpolation calculations to obtain First sub-point in the first sub-point cloud frame : ; in, , , It's a distance value. , , It is an angle value.

[0049] The interpolation calculation for the second sub-point cloud frame is similar and will not be elaborated here.

[0050] Q23, will The first sub-point cloud frame, The second sub-point cloud frame is transformed to the global coordinate system.

[0051] The first sub-point in the first sub-point cloud frame (abbreviated as) () is the first radar measurement point, which is converted into two-dimensional coordinates in the global coordinate system xy as follows. : .

[0052] The second sub-point in the second sub-point cloud frame (abbreviated as) () is the measurement point of the second radar, and the two-dimensional coordinates of the second radar relative to the first radar are: Furthermore, the azimuth zero point of the second radar is parallel to the second radar. This is converted to two-dimensional coordinates in the global coordinate system xy as follows: : .

[0053] Q3. Merge the first sub-point cloud frames at each time step in the aligned first sub-point cloud frame sequence and the corresponding second sub-point cloud frames in the second sub-point cloud frame sequence to obtain point cloud frames at each time step, forming a point cloud frame sequence. Then, take one of the consecutive common time steps... For example: The first sub-point cloud frame is ,in , representing the nth first sub-point. The second sub-point cloud frame is .

[0054] Will and Merge, and obtain Point cloud frames .

[0055] The following is a further expansion of S140: In some embodiments, in S140, the first micro-motion feature of the potential organism can be obtained based on the first phase sequence / amplitude sequence corresponding to a first point sequence of the potential organism.

[0056] In some embodiments, in S140, the first micro-motion feature of the potential organism can be obtained based on the synthesized phase sequence / synthetic amplitude sequence of the first phase sequence / amplitude sequence corresponding to a portion of the first point sequence of the potential organism.

[0057] In some embodiments, in S140, the first micro-motion feature of the potential organism can be obtained based on the synthesized phase sequence / synthetic amplitude sequence of the first phase sequence / amplitude sequence corresponding to all first point sequences of the potential organism.

[0058] The process of obtaining the first micro-motion feature from the synthesized phase sequence or the first phase sequence may include: performing a Fourier transform on the corresponding phase sequence to obtain a micro-Doppler spectrum; extracting the respiratory frequency band and / or the heartbeat frequency band from the micro-Doppler spectrum; statistically analyzing the respiratory frequency band to obtain the respiratory micro-motion feature, and / or statistically analyzing the heartbeat frequency band to obtain the heartbeat micro-motion feature. The respiratory micro-motion feature and the heartbeat micro-motion feature represent respiratory information and heartbeat information, respectively.

[0059] In some embodiments, respiratory micromotion features are extracted using a synthetic phase sequence of all first-point sequences. Based on this, S140 includes: S141: The first phase sequences corresponding to the potential organism are fused to obtain a synthetic phase sequence, wherein the synthetic phase corresponding to each time moment in the synthetic phase sequence is obtained by fusing the phases of the potential organism at the same time moment in each first phase sequence.

[0060] The fusion method for phases at the same moment can be, but is not limited to, weighted averaging. The weights for each phase in the weighted average can be the same or different. If different, the weights for each phase can be determined based on, but is not limited to, the signal-to-noise ratio of the corresponding echo signal / first point.

[0061] S142: Perform Fourier transform based on the synthesized phase sequence to obtain the micro-Doppler spectrum.

[0062] In some embodiments, S142 includes S1421. S1421: Perform a Fourier transform on the synthesized phase sequence to obtain the micro-Doppler spectrum.

[0063] In some embodiments, S142 includes S1422. S1422: The synthesized phase sequence is resampled according to at least two coprime sampling intervals to obtain at least two synthesized phase sub-sequences; Fourier transforms are performed on each synthesized phase sub-sequence to obtain each sub-microDoppler spectrum; the sub-microDoppler spectra are fused to obtain the microDoppler spectrum.

[0064] The method for fusing the microDoppler spectra of each sub-sub can be, but is not limited to, element-wise multiplication.

[0065] Understandably, resampling at multiple coprime sampling intervals to obtain the micro-Doppler spectrum is equivalent to extending the sampling time, significantly improving the resolution of the micro-Doppler spectrum without increasing the actual sampling time. Consequently, the respiratory frequency band extracted from the micro-Doppler spectrum is more accurate, and the respiratory micro-motion characteristics obtained from the statistical analysis of the respiratory frequency band are also more accurate.

[0066] S143: Extracting respiratory frequency bands from microDoppler spectra.

[0067] S144: Statistical analysis of respiratory frequency bands yields respiratory micro-motion characteristics, which include at least one of respiratory frequency characteristics, respiratory pattern characteristics, respiratory morphology characteristics, respiratory amplitude, and pattern characteristics.

[0068] Among them, respiratory rate characteristics represent respiratory frequency, respiratory pattern characteristics represent respiratory pattern, respiratory morphology characteristics represent respiratory morphology, and respiratory amplitude and pattern characteristics represent respiratory amplitude and respiratory pattern. Respiratory amplitude can be reflected by the signal amplitude of the respiratory frequency band, and respiratory pattern can be reflected by the phase change pattern of the respiratory frequency band.

[0069] In some embodiments, respiratory amplitude and pattern characteristics can be obtained based on the micro-Doppler spectrum obtained in S1421, and respiratory frequency characteristics, respiratory pattern characteristics, and respiratory morphology characteristics can be obtained based on the micro-Doppler spectrum obtained in S1422.

[0070] In some embodiments, prior to S142, the process may further include: performing low-pass filtering on each synthetic phase in the synthetic phase sequence. It is understood that low-pass filtering can detrend the synthetic phase sequence. Detrending removes low-frequency components caused by overall organismal movement or baseline drift, highlighting the periodic phase changes caused by respiration.

[0071] The following are examples illustrating S141-S144: Specific example 1: The point cloud frame sequence consists of M consecutive point cloud frames. A second point cloud cluster sequence of K potential organisms is obtained from the point cloud frame sequence. ,in The second point cloud cluster sequence represents the k-th potential organism.

[0072] A1, Regarding From the i-th first point sequence, obtain its corresponding first phase sequence as follows: ,in This represents the first phase at the m-th time.

[0073] A2, Yes We take a weighted average of the first phase sequences corresponding to each first point sequence to obtain... Corresponding synthetic phase sequence : .in, It is the synthesized phase corresponding to the m-th time step, which is composed of the phases of the m-th time steps in each first-point sequence. The weighted average is obtained as follows: It is the weight of the phase at time m, and the signal-to-noise ratio at the corresponding first point. Positive correlation, for example .

[0074] A3, Yes Each synthesized phase is low-pass filtered to obtain : .in, It is a low-pass filter with a low cutoff frequency.

[0075] A4, Yes Perform a short-time Fourier transform (STFT) to obtain the micro-Doppler spectrum. .in, It is a window function (such as the Hanning window).

[0076] A5. Extracting respiratory frequency bands from microDoppler spectrum .

[0077] A6. Yes Statistical analysis was performed to obtain respiratory rate characteristics. (Center frequency / peak frequency), respiratory pattern characteristics (Spectral entropy characteristics), respiratory morphology characteristics , respiratory amplitude and pattern characteristics .in, It refers to the respiratory frequency band features extracted from the respiratory frequency band. , These are the mean and variance of each characteristic value in the respiratory frequency band characteristics.

[0078] Specific example 2: Unlike specific example 1, after A3, it also includes sampling intervals based on two coprime samples. and (Right now )right Resampling yields two synthetic phase subsequences. , .

[0079] In A4, Fourier transforms are performed on the two synthesized phase subsequences to obtain the two sub-microDoppler spectra. , The microDoppler spectrum is obtained by multiplying the two sub-microDoppler spectra element by element. .

[0080] It is understandable that resampling at two coprime sampling intervals to obtain the micro-Doppler spectrum is equivalent to changing the sampling time from... The expansion has been made to This significantly improves the resolution of the micro-Doppler spectrum without increasing the actual sampling time. Consequently, the respiratory frequency band extracted from the micro-Doppler spectrum is more accurate, and the respiratory micro-motion characteristics obtained from the statistical analysis of the respiratory frequency band are also more accurate.

[0081] In some embodiments, the radar includes a first radar and a second radar, and the amplitude sequence includes a first phase sequence corresponding to the first radar and a second phase sequence corresponding to the second radar. Based on this, S140 includes: S241: The phase sequences of each first radar corresponding to the potential organism are fused to obtain the first synthetic phase sequence, and the phase sequences of each second radar corresponding to the potential organism are fused to obtain the second synthetic phase sequence.

[0082] Specifically, fusion of the phase sequences of each first radar refers to fusing the phases of each first radar phase sequence at the same moment into a first composite phase. Similarly, fusion of the phase sequences of each second radar refers to fusing the phases of each second radar phase sequence at the same moment into a second composite phase.

[0083] S242: Subtract each first synthetic phase in the first synthetic phase sequence from the corresponding second synthetic phase in the second synthetic phase sequence to obtain the corresponding synthetic phase difference, and obtain the behavioral micro-motion characteristics based on each synthetic phase difference.

[0084] Among them, behavioral micro-movement characteristics characterize the behavioral information of potential organisms.

[0085] In some embodiments, the individual synthesized phase differences can be directly used as behavioral micro-motion features.

[0086] In some embodiments, each synthesized phase difference can be subtracted from the pre-calibrated value of the inherent path difference to update each synthesized phase difference, and the updated synthesized phase differences can be used as behavioral micro-motion features.

[0087] In some embodiments, the radar includes a first radar and a second radar, and the amplitude sequence includes a first amplitude sequence corresponding to the first radar and a second amplitude sequence corresponding to the second radar. Based on this, S140 includes: S243: Fuse each of the first amplitude sequences corresponding to the potential organism to obtain a first synthetic amplitude sequence, and fuse each of the second amplitude sequences corresponding to the potential organism to obtain a second synthetic amplitude sequence.

[0088] Specifically, fusing the amplitude sequences of each first radar means merging the amplitudes at the same moment in each first radar amplitude sequence into a first composite amplitude. Similarly, fusing the amplitude sequences of each second radar means merging the amplitudes at the same moment in each second radar amplitude sequence into a second composite amplitude.

[0089] S244: Divide each first synthetic amplitude in the first synthetic amplitude sequence by the corresponding second synthetic amplitude in the second synthetic amplitude sequence to obtain the corresponding synthetic amplitude ratio, which is used as a geometric micro-motion feature.

[0090] Among them, geometric micro-motion features characterize the geometric information of potential organisms.

[0091] The following are examples illustrating S241-S242: Specific example 3: B1. The first radar phase sequences corresponding to the k-th potential organism are fused to obtain the first synthetic phase sequence. The first synthetic phase at the m-th time in the first synthetic phase sequence is: The first radar amplitude sequences corresponding to the k-th potential organism are fused to obtain the first synthetic amplitude sequence. The first synthetic amplitude at time m in the first synthetic amplitude sequence is... .

[0092] B2. Fuse the second radar phase sequences corresponding to the k-th potential organism to obtain the second synthetic phase sequence. The second synthetic phase at the m-th time in the second synthetic phase sequence is: The second radar amplitude sequences corresponding to the k-th potential organism are fused to obtain the second composite amplitude sequence. The second composite amplitude at time m in the second composite amplitude sequence is... .

[0093] B3. Subtract each first synthesized phase from its corresponding second synthesized phase and pre-calibration value to obtain the synthesized phase difference. The second synthesized phase difference at the m-th time is: ,in This is the pre-calibration value. Divide each first synthesized amplitude by the corresponding second synthesized amplitude to obtain the synthesized amplitude ratio. The second synthesized amplitude ratio at time m is... .

[0094] The following is a further expansion of S150: In some embodiments, the first micro-motion feature includes respiratory micro-motion features, behavioral micro-motion features, and geometric micro-motion features. The respiratory and behavioral micro-motion features of the potential organism are obtained based on the first phase sequence corresponding to the potential organism, and the geometric micro-motion features are obtained based on the amplitude sequence corresponding to the potential organism. Based on this, before S150, the process includes: performing dimensionality reduction and nonlinear transformation on the first micro-motion feature to obtain the final first micro-motion feature. The dimensionality reduction and nonlinear transformation can be implemented, but are not limited to, through a fully connected layer.

[0095] For example, the initial first micro-motion feature is By using a fully connected layer to perform dimensionality reduction and nonlinear transformation on the initial first micro-motion feature, a low-dimensional, compact, and highly discriminative final first micro-motion feature is obtained. This final first micro-motion characteristic allows for a better measurement of the likelihood that different potential organisms will become the same final organism.

[0096] In some embodiments, S150 includes: S151: Determine the association measure between potential organisms based at least on the similarity measure between the first micromotion features of each potential organism.

[0097] In some embodiments, the first micro-motion feature is That is, it does not include respiratory amplitude and pattern characteristics. Based on this, before S151, the first micro-motion characteristic can also be combined with the respiratory amplitude and pattern characteristics. splicing, to update the first micro-motion feature .

[0098] In some embodiments, similarity metrics can be directly used as association metrics.

[0099] In some embodiments, the association metric between potential organisms is obtained by weighting the similarity metric and the spatial distance metric between potential organisms. The spatial distance metric is calculated based on the location of the potential organism, which is determined based on the location of each point in the second point cloud sequence of the potential organism.

[0100] One approach is to take a weighted average of the positions of each point in the second-point cloud cluster sequence to obtain the location of the potential organism. Alternatively, the center of each point in the second-point cloud cluster sequence can be used as the location of the potential organism.

[0101] S152: Based on the correlation measure between each potential organism, cluster each potential organism to obtain several final organisms.

[0102] In some embodiments, the clustering of potential organisms is based on a measure of the pairwise associations between each potential organism.

[0103] In some embodiments, the clustering of potential organisms is based on the distribution of association metrics between potential organisms and other potential organisms. The association metric distribution characterizes information such as which potential organisms are associated with, which are not associated with, and the degree of association with each potential organism. Based on this, S152 includes S1521-S1522. S1521: For each potential organism, based on the distribution of association metrics between the potential organism and other potential organisms, obtain the embedding representation of the potential organism in a low-dimensional feature space. S1522: Based on the embedding representation of each potential organism in the low-dimensional feature space, cluster the potential organisms to obtain several final organisms.

[0104] Clustering can be implemented, but is not limited to, using the k-means algorithm.

[0105] The embedding representation in the low-dimensional feature space is obtained by dimensionality reduction of the association metric distribution. The association metric distribution can be calculated, but is not limited to, using the Laplacian matrix. In the case of Laplacian matrix calculation, S1521 may include: constructing a node graph by treating each potential organism as a node, wherein nodes corresponding to two potential organisms in the node graph with an association metric greater than the association metric threshold are connected by an edge, and the edge weight is the association metric of the two connected nodes; constructing an edge weight matrix of the node graph using the edge weights corresponding to each edge; calculating the degree of the same node by counting the edge weights corresponding to each edge, and constructing a degree matrix using the degrees of each node; obtaining the Laplacian matrix based on the degree matrix and the edge weight matrix; and obtaining the embedding representation of each node in the low-dimensional feature space based on the Laplacian matrix.

[0106] In some embodiments, the degree of a node can be obtained by summing the weights of all edges corresponding to the same node. In other embodiments, the degree of a node can be obtained by averaging the weights of all edges corresponding to the same node.

[0107] The following are examples illustrating S151-S152: Specific example 4: C1. Obtain spatial distance measurements between the first micro-motion characteristics of each potential organism: ; in, It is the position of the kth potential organism. It is a scale parameter that controls spatial distance sensitivity.

[0108] C2. Obtain a similarity measure between the first micro-motion features of each potential organism: ; Among them, X k The first micromotor feature representing the k-th potential organism. It is a scale parameter that controls the sensitivity to similarity.

[0109] C3. Weight the similarity and spatial distance metrics between potential organisms to obtain the association metrics between them: ; in, It is a balance parameter.

[0110] C4. Construct a weighted graph Each potential organism is treated as a node, and edges are established between the nodes corresponding to two potential organisms whose association metric is greater than the association metric threshold, resulting in a node graph. It is a set of nodes. It is a set of edges. The edge weight matrix of the node graph is constructed using the edge weights corresponding to each edge. .

[0111] C5. Sum the edge weights corresponding to the same node to obtain the degree of the same node, and construct a degree matrix using the degrees of all nodes. The degree matrix is ​​a diagonal matrix. Its diagonal elements Represents a node The degree.

[0112] C6. Based on the degree matrix and edge weight matrix, the normalized graph Laplacian matrix is ​​obtained. ,in It is an identity matrix.

[0113] C7. Obtain the embedding representation of each node in the low-dimensional feature space based on the graph Laplacian matrix. Specifically: C71. Perform eigenvalue decomposition on the graph Laplacian matrix to obtain the eigenvectors corresponding to the first m smallest eigenvalues. And these eigenvectors are arranged into a feature matrix by columns. : .

[0114] C72, the characteristic matrix Each row is considered as the embedding representation of the corresponding node in the low-dimensional feature space. .

[0115] C73. Using the k-means algorithm to represent the embeddings of each node. Clustering of each node yields Cluster of nodes Each node cluster It represents a final organism.

[0116] The following is a further expansion of S130: In some embodiments, in S130, intra-region clustering can be performed for all region units.

[0117] It is understandable that, through further research by the inventors of this application, it has been discovered that some regional units may not contain potential organisms. Therefore, uniformly performing region-level clustering on all regional units would waste computational resources. Therefore, to avoid wasting computational resources, some embodiments may perform region-level clustering only on a subset of regional units in S130. Based on this, before S130, the following is included: S310: For each regional unit, the biomass intensity of the regional unit is obtained based on the corresponding first point cloud cluster sequence. The biomass intensity characterizes the possibility of potential organisms existing in the regional unit.

[0118] S320: Based on the organism strength of each regional unit, identify the regional units in which potential organisms exist.

[0119] In some embodiments, the organism strength of each region unit can be used to form an original organism strength map; regions in the original organism strength map whose organism strength is greater than the global organism strength threshold are identified as regions containing potential organisms. This approach can be considered as binary segmentation based on the global organism strength threshold.

[0120] In some embodiments, the organism intensities of each region unit can be used to form an original organism intensity map; each organism intensity in the original organism intensity map is taken as the current organism intensity; the local statistical intensity value is obtained by statistically analyzing the organism intensities in the neighborhood of the current organism intensity; a local organism intensity threshold is determined based on the local statistical intensity value; in response to the current organism intensity being greater than the local statistical intensity value, the region unit corresponding to the current organism intensity is determined to be a unit with potential organisms. In some embodiments, the local statistical intensity value can be multiplied by a preset ratio to obtain the local organism intensity threshold. This method can be regarded as binary segmentation based on the local organism intensity threshold, which can be implemented, but is not limited to, using the constant false alarm rate (CA-CFAR) algorithm. In the case of the CA-CFAR algorithm, the preset ratio is based on a preset false alarm probability. Calculated.

[0121] S330: The regional unit where potential organisms exist is taken as at least a partial regional unit.

[0122] In some embodiments, S310 includes S311-S312. S311: Obtain the confidence level of each second point sequence in the first point cloud cluster sequence, wherein the confidence level of the second point sequence is positively correlated with at least one of the radar acquisition confidence level and the signal-to-noise ratio of the second point sequence. S312: Statistically analyze the confidence levels of each second point sequence in the first point cloud cluster sequence to obtain the biomass intensity of the regional unit.

[0123] If the radar includes a first radar and a second radar, then some second point sequences in the first point cloud cluster sequence may have been acquired by the first radar, corresponding to the acquisition confidence level of the first radar, while some second point sequences may have been acquired by the second radar, corresponding to the acquisition confidence level of the second radar. If there is only a single radar, then the acquisition confidence levels of the radars corresponding to each second point sequence are the same.

[0124] In some embodiments, S312 includes: summing the confidence scores of each second point sequence to obtain a total confidence score; and dividing the total confidence score by the number of second point sequences in the first point cloud sequence to obtain the biomass strength of the region unit.

[0125] In some embodiments, S310 includes: obtaining the intensity of each sub-organism of the region unit based on the first point cloud cluster at each time point in the corresponding first point cloud cluster sequence; and taking a weighted average of the intensities of each sub-organism to obtain the organism intensity corresponding to the region unit. The method of obtaining the sub-organism intensity is similar to that of S311-S312, and will not be described in detail here.

[0126] The following are examples illustrating S310-S330: Specific example 5: The regional unit in the management area is , where p and q represent the x-coordinate range and y-coordinate range covered by the regional unit, respectively.

[0127] D1. Obtain the organism strength of the region unit: ; in, , These are the second point sequence corresponding to the first radar in the point cloud frame sequence and its confidence level. , These are the second point sequence corresponding to the second radar in the point cloud frame sequence and its confidence level. The confidence level of the second point sequence is determined by its signal-to-noise ratio and the acquisition confidence level of the corresponding first radar / second radar. It is an indicator function, when the second point sequence belongs to The corresponding first point cloud cluster sequence (located in) The value is 1 if the result is within the range of (the first point in the cloud cluster sequence), and 0 otherwise. In general, the denominator of this formula represents the number of sequences in the second point sequence within the first point cloud cluster sequence, i.e., the number of sequences falling within the range of (the first point in the cloud cluster sequence). The number of second-point sequences is represented by the numerator, which is the sum of the confidence scores of each second-point sequence in the first-point cloud cluster sequence, i.e., the total confidence score.

[0128] D2. Compile the organism strength of each region unit into an original organism strength map; then use the strength of each organism in the original organism strength map as the current organism strength. The local statistical intensity value is obtained by averaging the intensities of all organisms within the neighborhood of the current organism's intensity. ; Compare local statistical intensity values ​​with preset proportions Multiplying them yields the local organism strength threshold. To determine the false alarm probability based on the preset false alarm probability Calculated; in response to the current organism strength being greater than the local statistical strength value ( The region unit corresponding to the current organism strength is determined as the region unit where potential organisms exist.

[0129] D3. Regions containing potential organisms are selected as the regions to be clustered.

[0130] The following is a further expansion of S130: In some embodiments, based on the position of each second point sequence in the first point cloud cluster sequence, the second point sequences in the first point cloud cluster sequence can be clustered to obtain the second point cloud cluster sequences of several potential organisms.

[0131] In some embodiments, based on the second micro-motion features of each second point sequence in the first point cloud cluster sequence, each second point sequence in the first point cloud cluster sequence can be clustered to obtain the second point cloud cluster sequences of several potential organisms.

[0132] It is understandable that there may be stationary organisms (without significant motion information) in the managed area (referred to as stationary organisms). Location-based clustering cannot yield the second point cluster sequence of stationary organisms within the regional unit. However, although stationary organisms lack significant motion information, they possess second micro-motion information caused by respiration, heartbeat, etc. Therefore, clustering based on the second micro-motion characteristics that characterize this second micro-motion information can yield the second point cluster sequence of stationary organisms.

[0133] The micro-motion information represented by the second micro-motion feature and the first micro-motion feature may be the same or different.

[0134] In some embodiments, S130 includes: S131: Based on the second phase sequence corresponding to each second point sequence in the first point cloud cluster sequence, obtain the second micro-motion feature of each second point sequence. The second micro-motion feature represents the velocity information.

[0135] Each point in the second point sequence (also called the second point) corresponds one-to-one with the phase of each point in the second phase sequence.

[0136] The second point sequence refers to a sequence of points at multiple consecutive moments.

[0137] S132: Based on the second micro-motion features of each second point sequence, cluster each second point sequence to obtain the second point cloud cluster sequence of several potential organisms.

[0138] In some embodiments, each second point sequence can be clustered based solely on the similarity between the second micro-motion features of each second point sequence to obtain several potential organism second point cloud cluster sequences.

[0139] In some embodiments, the second point sequences can be clustered based on the similarity between the second micro-motion features of each second point sequence and the spatial distance between each second point sequence to obtain a number of potential organism second point cloud cluster sequences.

[0140] Here, the spatial distance between each second-point sequence refers to the distance between the positions of each second-point sequence, and the position of the second-point sequence is determined based on the position of each point in the second-point sequence. For example, the position of the second-point sequence can be obtained by weighted averaging the positions of each point in the second-point sequence. This method can be implemented using, but is not limited to, the density-based spatial clustering (DBSCAN) algorithm.

[0141] In some embodiments, S132 includes S1321. S1321: For each second point sequence, analyze the phase change at each moment in the second phase sequence corresponding to the second point sequence to obtain the radial displacement at each moment, and the radial displacement at each moment constitutes the second micro-motion feature.

[0142] In some embodiments, in the case of a single radar, S1321 includes: for each second point sequence, subtracting the phase corresponding to the point at the current time from the phase corresponding to the point at the previous time in the second point sequence to obtain the phase difference at the current time in the second point sequence; and calculating the radial displacement at the corresponding time based on the phase difference at each time in the second point sequence.

[0143] In some embodiments, the radar includes a first radar and a second radar, and the second phase sequence includes a third radar phase sequence corresponding to the first radar and a fourth radar phase sequence corresponding to the second radar. The first phase at each moment in the third radar phase sequence, the second phase at each moment in the fourth radar phase sequence, and the points at each moment in the second point sequence correspond one-to-one. Based on this, S1321 includes: for each second point sequence, subtracting the second phase corresponding to the point from the first phase at each moment in the second point sequence to obtain the phase difference at each moment in the second point sequence; for each moment in the second point sequence, using the difference between the phase difference at that moment and the phase difference at the previous moment as the phase difference change at that moment; and calculating the radial displacement at that moment based on the phase difference change at that moment and the phase difference change at a reference moment, wherein the radial displacement at that moment is positively correlated with the phase difference change at that moment.

[0144] It is understandable that, due to the difference in the propagation path of electromagnetic waves, the phase difference between the first phase and the second phase is very sensitive to minute radial displacements; that is, minute radial displacements will cause changes in the phase difference between the first phase and the second phase.

[0145] The following are examples illustrating S131-S132: Specific example 6: E1, regarding the second point sequence The point, from the echo signal corresponding to the first radar Get the first phase corresponding to the point From the echo signal corresponding to the second radar Get the second phase corresponding to the point .

[0146] E2, will minus get phase difference .

[0147] E3, will phase difference and The difference between the phase differences is used as Phase difference change .

[0148] E4, according to Calculation of phase difference change radial displacement . Representative reference time The change in phase difference, where the reference time can be any one of multiple consecutive moments. It is the carrier wavelength of the signals transmitted by the first and second radars. The radial displacements at M consecutive moments constitute the second micro-motion characteristic of the second point sequence. .

[0149] E5. For the U second-point sequences in the first point cloud cluster sequence. Based on the similarity and spatial distance between the second micro-motion features of each second point sequence, the second point sequences are clustered to obtain several potential organism second point cloud cluster sequences. Specifically: E51: Take each of the second point sequences as the current second point sequence. Get the number of other second point sequences in the neighborhood of the current second point sequence position. ,in For indicator functions, Euclidean distance. The neighborhood radius is the response to the number of other second-point sequences. Greater than the sequence number threshold (Right now ), and determine the current second point sequence as the core second point sequence.

[0150] E52: For each core second point sequence, obtain other second point sequences whose second micro-motion feature similarity to the core second point sequence is greater than the similarity threshold, and combine the core second point sequence and other second point sequences whose similarity is greater than the similarity threshold to form a second point cloud cluster sequence.

[0151] It is understandable that if the managed area contains non-living objects with micro-motion information, such as a working electric fan or swaying furniture, or if there are other types of living objects besides the specific type to be detected, such as pets, these non-living objects with micro-motion information or other types of living objects may be detected as the final living object as interference. Therefore, in some embodiments, interference detection can be performed on the final living object to improve the accuracy of the final living object detection.

[0152] Based on this, in some embodiments, S160-S190 are included after S150. S160: For each final organism, extract the third point cloud cluster sequence of the final organism from the point cloud frame sequence. S170: Obtain interference reference indicators for the final organism based on the third point cloud cluster sequence. The interference reference indicators include at least one of velocity information, dwell time, outer contour size, and motion pattern. S180: Determine whether the final organism is an interference based on the degree of difference between the interference reference indicators of the final organism and the interference reference indicators of non-interference objects. In response to being identified as an interference, execute S190. S190: Mark the corresponding location in the management area as the region where the interference is located, and exclude the processing of the region where the interference is located during the organism detection process of subsequent point cloud frame sequences.

[0153] In some embodiments, the velocity information of the final organism can be represented by the velocity micro-motion features of the final organism, which are obtained by fusing the second micro-motion features of each third point sequence in the third point cloud cluster sequence. The acquisition method of the second micro-motion features of the third point sequence is similar to that of the second micro-motion features of the second point sequence, and will not be described in detail here.

[0154] In some embodiments, the final organism is identified as an interfering substance when the difference between the interference reference index of the final organism and the interference reference index of the non-interfering substance is greater than a difference threshold.

[0155] In some embodiments, when the degree of difference between the interference reference index of the final organism and the interference reference index of the non-interference object in multiple point cloud frame sequences is greater than the degree of difference threshold, the final organism is determined to be an interference object.

[0156] The above scheme, after detecting interference, marks the corresponding location in the management area as the area where the interference is located. During the subsequent biological detection process of the point cloud frame sequence, the processing of the area where the interference is located is excluded, thereby avoiding repeated detection of the same interference and saving computational overhead.

[0157] In some embodiments, steps S1000-S1020 are included before step S160. S1000: Determine the comprehensive features of the final organism based on the third point cloud cluster sequence of the final organism. S1010: Obtain the comprehensive similarity between the comprehensive features of the final organism and the comprehensive features of each historical interfering substance in the historical interfering substance feature library. S1020: Determine whether the final organism is an interfering substance based on each comprehensive similarity. If it is an interfering substance, proceed to step S190; otherwise, continue to determine whether it is an interfering substance through steps S160-S180.

[0158] In some embodiments, the final comprehensive similarity can be obtained by weighted averaging of the individual comprehensive similarities. If the final comprehensive similarity is greater than the comprehensive similarity threshold, the final organism is determined to be an interfering object.

[0159] In some embodiments, after S190, the method further includes: storing the comprehensive characteristics of the interfering object into a historical interfering object characteristic database.

[0160] In some embodiments, if the final organism is not a disruptor as detected by S160-S180, it is determined again by S1000-S1020 whether the final organism is a disruptor.

[0161] The following is an example illustrating interference detection: Specific example 7: Please refer to the following: Figure 2 , Figure 2 This is a flowchart illustrating a specific example of interference detection in this application, as shown below. Figure 2 As shown, the steps for interference detection include: F1. Extract the third point cloud cluster sequence of the final organism from the point cloud frame sequence.

[0162] F2. Obtaining the velocity micro-motion characteristics of the final organism based on the third-point cloud cluster sequence of the final organism. Calculate the velocity micro-motion characteristics of the final organism Velocity micro-motion characteristics of non-interfering objects The degree of difference between them, as a measure of the degree of difference in movement: ;in It is a constant.

[0163] F3. Based on the number of point cloud frame sequences that continuously detect the final organism (the number of consecutive third point cloud cluster sequences), determine the dwell time of the final organism and calculate the dwell time of the final organism. Dwell time with non-interference objects The degree of difference between them is used as a measure of the degree of difference in length of stay: ;in It is a constant.

[0164] F4. Based on the velocity micro-motion characteristics of the final organism To obtain the degree of difference in movement patterns between the final organism's movement pattern and the movement patterns of the non-disturbing object: .

[0165] F5. Obtain the outer contour of the final organism based on the third point cloud cluster sequence, and calculate the outer contour size of the final organism. Outer contour dimensions of non-interfering objects The degree of difference between them yields the degree of size difference: ;in, It is a constant.

[0166] F6. The interference confidence level is obtained by weighting the degree of difference in motion, the degree of difference in dwell time, the degree of difference in motion pattern, and the degree of difference in size. ,in It is a weight vector.

[0167] F7. Obtain the final organism's interference confidence score from N consecutive point cloud frame sequences, all of which must be greater than the interference confidence threshold. (Right now If the final organism is determined to be a disruptor, proceed to F8-F9; otherwise, proceed to F10-F12.

[0168] F8. Mark the location of the interfering object in the management area as the area where the interfering object is located, and exclude the processing of the area where the interfering object is located during the biological detection process of subsequent point cloud frame sequences.

[0169] F9. Store the comprehensive characteristics of the interfering objects into the historical interfering object characteristic database. .

[0170] F10. Determine the comprehensive characteristics of the final organism based on the third point cloud cluster sequence of the final organism.

[0171] F11: Obtain the overall similarity between the overall characteristics of the final organism and the overall characteristics of each historical interfering object in the historical interfering object feature library.

[0172] F12: The final overall similarity score is obtained by weighted averaging of all individual similarity scores. ; in, It is the total number of features of interfering objects in the historical interfering object feature database. It is the comprehensive characteristic of the final organism. It is a combination of characteristics of historical interference. It is a constant.

[0173] If the final overall similarity is greater than the overall similarity threshold, the final organism is determined to be a distractor and proceeds to F8-F9; otherwise, the final organism is determined not to be a distractor.

[0174] Figure 3 This is a schematic diagram of the structure of an embodiment of the biological detection device of this application. Figure 3 As shown, the biological detection device includes: a collection module, an extraction module, a first clustering module, an acquisition module, and a second clustering module. Wherein: The acquisition module is used to acquire point cloud frame sequences of the managed area using radar.

[0175] The extraction module is used to extract the first point cloud cluster sequence of each regional unit in the management area from the point cloud frame sequence.

[0176] The first clustering module is used to cluster the first point cloud cluster sequence for at least some regional units to obtain a second point cloud cluster sequence of several potential organisms, wherein the second point cloud cluster sequence of the potential organisms includes a number of first point sequences of the potential organisms.

[0177] The acquisition module is used to acquire the first micro-motion features of each potential organism based on the first phase sequence and / or amplitude sequence corresponding to each first point sequence of the potential organism.

[0178] The second clustering module is used to cluster each potential organism based on the similarity measure between the first micro-motion features of each potential organism, and obtain several final organisms.

[0179] For further detailed descriptions of the biological detection device, please refer to the preceding embodiments, which will not be repeated here.

[0180] Figure 4 This is a schematic diagram of the structure of an embodiment of the electronic device of this application. Figure 4 As shown, the electronic device 50 includes a memory 51 and a processor 52. The processor 52 is used to execute program instructions stored in the memory 51 to implement the steps in any of the above method embodiments. In a specific implementation scenario, the electronic device 50 may include, but is not limited to, a microcomputer or a server. In addition, the electronic device 50 may also include a laptop computer, a tablet computer, or other carrier device, which is not limited here.

[0181] Specifically, processor 52 controls itself and memory 51 to implement the steps in any of the above method embodiments. Processor 52 may also be referred to as a CPU (Central Processing Unit). Processor 52 may be an integrated circuit chip with signal processing capabilities. Processor 52 may also be a general-purpose processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component. A general-purpose processor may be a microprocessor or any conventional processor. Furthermore, processor 52 may be implemented using integrated circuit chips.

[0182] Please see Figure 5 , Figure 5 This is a schematic diagram of a computer-readable storage medium according to an embodiment of the present application. The computer-readable storage medium 60 stores program instructions 601 thereon, which, when executed by a processor, implement the steps in any of the above method embodiments.

[0183] This application also provides a computer program product comprising a computer program that, when executed by a processor, can implement the steps of the methods described in any of the foregoing embodiments. Specifically, the computer program product can be a software or program product containing a computer program, capable of running on a computing device or stored on any available medium.

[0184] In some embodiments, the functions or modules of the apparatus provided in this disclosure can be used to perform the methods described in the above method embodiments. The specific implementation can be referred to the description of the above method embodiments, and for the sake of brevity, it will not be repeated here.

[0185] The description of the various embodiments above tends to emphasize the differences between the various embodiments. The similarities or similarities between them can be referred to, and for the sake of brevity, they will not be repeated here.

[0186] In the several embodiments provided in this application, it should be understood that the disclosed methods and apparatus can be implemented in other ways. For example, the apparatus implementations described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection of devices or units may be electrical, mechanical, or other forms.

[0187] Furthermore, the functional units in the various embodiments of this application 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 integrated unit can be implemented in hardware or as a software functional unit. If the integrated unit is implemented as 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 technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0188] If the technical solution of this application involves personal information, the product using this technical solution has clearly informed the user of the personal information processing rules and obtained the user's voluntary consent before processing the personal information. If the technical solution of this application involves sensitive personal information, the product using this technical solution has obtained the user's separate consent before processing the sensitive personal information, and also meets the requirement of "express consent". For example, at personal information collection devices such as cameras, clear and prominent signs are set up to inform users that they have entered the scope of personal information collection and that personal information will be collected. If an individual voluntarily enters the collection scope, it is deemed that they have agreed to the collection of their personal information; or on the personal information processing device, with clear signs / information informing users of the personal information processing rules, authorization is obtained from the user through pop-up information or by asking the user to upload their personal information; wherein, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the types of personal information processed.

Claims

1. A method for detecting organisms, characterized in that, include: Point cloud frame sequences of the managed area were obtained using radar data acquisition. Extract the first point cloud cluster sequence of each regional unit in the management area from the point cloud frame sequence; For at least some of the said regional units, the first point cloud cluster sequence is clustered to obtain a number of potential organisms' second point cloud cluster sequences, wherein the second point cloud cluster sequence of the potential organisms includes a number of the first point sequences of the potential organisms; For each of the potential organisms, the first micro-motion feature of the potential organism is obtained based on the first phase sequence and / or amplitude sequence corresponding to each first point sequence of the potential organism; Based on the similarity measure between the first micro-motion features of each potential organism, the potential organisms are clustered to obtain several final organisms.

2. The method according to claim 1, characterized in that, The step of obtaining the first micromotion feature of the potential organism based on the first phase sequence and / or amplitude sequence corresponding to each first point sequence of the potential organism includes: The first phase sequences corresponding to the potential organisms are fused to obtain a synthetic phase sequence; A Fourier transform is performed based on the synthesized phase sequence to obtain the micro-Doppler spectrum; Extract the respiratory frequency band from the microDoppler spectrum; Statistical analysis of the respiratory frequency band yields respiratory micro-motion characteristics, which include at least one of respiratory frequency characteristics, respiratory pattern characteristics, respiratory morphology characteristics, respiratory amplitude, and pattern characteristics.

3. The method according to claim 2, characterized in that, Before performing a Fourier transform based on the synthesized phase sequence to obtain the micro-Doppler spectrum, the method further includes: Each synthesized phase in the synthesized phase sequence is low-pass filtered; And / or, the step of performing a Fourier transform based on the synthesized phase sequence to obtain the micro-Doppler spectrum includes: The synthesized phase sequence is resampled at at least two coprime sampling intervals to obtain at least two synthesized phase sub-sequences. Fourier transforms are performed on each of the synthesized phase sub-sequences to obtain the corresponding sub-microDoppler spectra; The microDoppler spectrum is obtained by fusing the individual microDoppler spectra.

4. The method according to claim 1, characterized in that, The radar includes a first radar and a second radar. The first phase sequence includes a first radar phase sequence corresponding to the first radar and a second radar phase sequence corresponding to the second radar. The amplitude sequence includes a first amplitude sequence corresponding to the first radar and a second amplitude sequence corresponding to the second radar. The step of obtaining the first micromotion feature of the potential organism based on the first phase sequence and / or amplitude sequence corresponding to each first point sequence of the potential organism includes: The first radar phase sequences corresponding to the potential organism are fused to obtain a first synthetic phase sequence, and the second radar phase sequences corresponding to the potential organism are fused to obtain a second synthetic phase sequence. Subtract each first synthetic phase in the first synthetic phase sequence from the corresponding second synthetic phase in the second synthetic phase sequence to obtain the corresponding synthetic phase difference, and obtain the behavioral micro-motion feature based on each synthetic phase difference; And / or, The first amplitude sequences corresponding to the potential organism are fused to obtain a first synthetic amplitude sequence, and the second amplitude sequences corresponding to the potential organism are fused to obtain a second synthetic amplitude sequence. Divide each first synthetic amplitude in the first synthetic amplitude sequence by the corresponding second synthetic amplitude in the second synthetic amplitude sequence to obtain the corresponding synthetic amplitude ratio, which serves as the geometric micro-motion feature.

5. The method according to claim 1, characterized in that, The first micro-motion feature includes respiratory micro-motion feature, behavioral micro-motion feature, and geometric micro-motion feature. The respiratory micro-motion feature and behavioral micro-motion feature of the potential organism are obtained based on the first phase sequence corresponding to the potential organism, and the geometric micro-motion feature is obtained based on the amplitude sequence corresponding to the potential organism. And / or, before clustering the potential organisms based on a similarity measure among their first micromotor features to obtain a plurality of final organisms, the method further includes: The first micro-motion feature is subjected to dimensionality reduction and nonlinear transformation to obtain the final first micro-motion feature.

6. The method according to claim 1, characterized in that, The similarity measure based on the first micro-motion features of each potential organism is used to cluster the potential organisms to obtain several final organisms, including: At least based on a similarity measure between the first micro-motion features of each of the potential organisms, a correlation measure between the potential organisms is determined; Based on the correlation metrics between the potential organisms, the potential organisms are clustered to obtain the several final organisms.

7. The method according to claim 6, characterized in that, The association metric between each potential organism is obtained by weighting the similarity metric and the spatial distance metric between each potential organism. The spatial distance metric is calculated based on the position of the potential organism, which is determined based on the position of each point in the second point cloud cluster sequence of the potential organism. And / or, the clustering of the potential organisms based on the association metrics between them to obtain the plurality of final organisms includes: For each potential organism, based on the distribution of the association metric between the potential organism and other potential organisms, the embedding representation of the potential organism in the low-dimensional feature space is obtained; Based on the embedding representation of each potential organism in the low-dimensional feature space, the potential organisms are clustered to obtain the several final organisms.

8. The method according to claim 6, characterized in that, The step of obtaining the embedding representation of the potential organism in the low-dimensional feature space based on the distribution of association metrics between the potential organism and other potential organisms includes: Each of the potential organisms is used as a node to construct a node graph, wherein the nodes corresponding to two potential organisms whose association metric is greater than the association metric threshold are connected by an edge, and the edge weight is the association metric corresponding to the two connected nodes. The edge weight matrix of the node graph is constructed using the edge weights corresponding to each edge. The edge weights corresponding to the same node are statistically analyzed to obtain the degree of the same node, and a degree matrix is ​​constructed using the degrees of the nodes. Based on the degree matrix and the edge weight matrix, the Laplace matrix is ​​obtained; Based on the Laplacian matrix, the embedding representation of each node in the low-dimensional feature space is obtained.

9. The method according to claim 1, characterized in that, Before clustering the first point cloud cluster sequence for at least a portion of the said regional units to obtain a second point cloud cluster sequence of several potential organisms, the method further includes: For each of the aforementioned regional units, the organism intensity of the regional unit is obtained based on the corresponding first point cloud cluster sequence, and the organism intensity characterizes the probability that the potential organism exists in the regional unit. Based on the organism strength of each of the said regional units, the regional units in which the potential organisms exist are determined from each of the said regional units; The region unit where the potential organism exists is considered as at least a portion of the region unit.

10. The method according to claim 9, characterized in that, The step of obtaining the biological intensity of the regional unit based on the corresponding first point cloud cluster sequence includes: The confidence level of each second point sequence in the first point cloud cluster sequence is obtained, and the confidence level of the second point sequence is positively correlated with at least one of the radar acquisition confidence level and the signal-to-noise ratio of the second point sequence. The biomass intensity of the region unit is obtained by statistically analyzing the confidence levels of each second point sequence in the first point cloud cluster sequence.

11. The method according to claim 10, characterized in that, The confidence score analysis of each second point sequence in the first point cloud cluster sequence to obtain the biomass intensity of the region unit includes: The total confidence score is obtained by summing the confidence scores of each second point sequence. The biomass strength of the region unit is obtained by dividing the total confidence score by the number of sequences in the second point sequence of the first point cloud cluster sequence.

12. The method according to claim 1, characterized in that, The second point cluster sequence obtained by clustering the first point cluster sequence into several potential organisms includes: Based on the second phase sequence corresponding to each second point sequence in the first point cloud cluster sequence, the second micro-motion feature of each second point sequence is obtained, and the second micro-motion feature represents the velocity information. Based on the second micro-motion features of each second point sequence, the second point sequences are clustered to obtain the second point cloud cluster sequence of the several potential organisms.

13. The method according to claim 12, characterized in that, The points at each moment in the second point sequence correspond one-to-one with the phases at each moment in the second phase sequence; The step of obtaining the second micro-motion features of each second point sequence based on the second phase sequence corresponding to each second point sequence in the first point cloud cluster sequence includes: For each second point sequence, the phase change of adjacent moments in the second phase sequence corresponding to the second point sequence is analyzed to obtain the radial displacement at each moment, and the radial displacement at each moment constitutes the second micro-motion feature.

14. The method according to claim 13, characterized in that, The radar includes a first radar and a second radar. The second phase sequence includes a third radar phase sequence corresponding to the first radar and a fourth radar phase sequence corresponding to the second radar. The first phase at each moment in the third radar phase sequence, the second phase at each moment in the fourth radar phase sequence, and the points at each moment in the second point sequence correspond one-to-one. For each second point sequence, the analysis of the phase change between adjacent moments in the second phase sequence corresponding to the second point sequence to obtain the radial displacement at each moment includes: For each second point sequence, the first phase corresponding to each point at each time moment in the second point sequence is subtracted from the second phase corresponding to that point to obtain the phase difference at each time moment in the second point sequence. For each moment in the second point sequence, the difference between the phase difference at that moment and the phase difference at the previous moment is taken as the change in phase difference at that moment. The radial displacement at the specified moment is calculated based on the phase difference change at the specified moment and the phase difference change at the reference moment, wherein the radial displacement at the specified moment is positively correlated with the phase difference change at the specified moment.

15. The method according to claim 1, characterized in that, After clustering the potential organisms based on the similarity measure among their first micro-motion features to obtain several final organisms, the process further includes: For each of the final organisms, the third point cloud cluster sequence of the final organism is extracted from the point cloud frame sequence; Based on the third point cloud cluster sequence of the final organism, the interference reference index of the final organism is obtained. The interference reference index includes at least one of velocity information, dwell time, outer contour size, and motion pattern. Based on the degree of difference between the interference reference index of the final organism and the interference reference index of the non-interfering substance, it is determined whether the final organism is an interfering substance; In response to the interference, the corresponding location in the management area is marked as the area where the interference is located, and the processing of the area where the interference is located is excluded during the biological detection process of subsequent point cloud frame sequences.

16. The method according to claim 1, characterized in that, The radar includes a first radar and a second radar, and the acquisition of point cloud frame sequences of the managed area using the radar includes: The first sub-point cloud frame sequence and the second sub-point cloud frame sequence of the management area are respectively acquired using the first radar and the second radar; Alignment processing is performed on the first sub-point cloud frame sequence and the second sub-point cloud frame sequence; The first sub-point cloud frame sequence and the second sub-point cloud frame sequence after alignment are fused to obtain the point cloud frame sequence.

17. An electronic device, characterized in that, It includes a memory and a processor, the processor being configured to execute program instructions stored in the memory to implement the method of any one of claims 1-16.

18. A computer-readable storage medium, characterized in that, It stores program instructions that, when executed by a processor, implement the method of any one of claims 1-16.

19. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1-16.

Images