Fall detection method and device, electronic equipment and storage medium

Abstract

The invention provides a fall detection method and device, electronic equipment and a storage medium, and relates to the technical field of electromagnetic wave detection, and the method comprises the steps: determining the target mass center height and target motion information corresponding to a target object in each judgment node based on the obtained multi-frame to-be-detected signal, the judgment node is any one of at least two judgment nodes obtained by carrying out equal-frame-number separation on multiple frames of signals to be detected, and each judgment node comprises the signals to be detected of a continuous first preset number of frames; based on the target centroid height and the target motion information, determining state information corresponding to the target object in each judgment node; sorting the plurality of pieces of state information according to the determination moments corresponding to the plurality of pieces of state information, and determining a state information sequence corresponding to the plurality of pieces of state information; and determining a tumble detection result of the target object based on the state information sequence. The fall detection accuracy can be improved.

Description

Technical Field

[0001] The present invention relates to the field of electromagnetic wave detection technology, and in particular to a fall detection method, device, electronic equipment and storage medium. Background Art

[0002] With the advent of the aging era, the safety of empty-nest elderly people has received more and more attention from the society. As the elderly are physically weak, if they cannot get timely help after falling, it is very easy to aggravate related complications, such as fractures, heart disease and brain damage.

[0003] At present, deep learning, machine learning and other algorithms are generally used. The above algorithms have poor generalization ability for the environment and make single-dimensional judgments based on the height or speed of the target, which can easily cause false detections and missed detections, resulting in low fall detection accuracy. Summary of the invention

[0004] The present invention provides a fall detection method, device, electronic device and storage medium, which are used to solve the defect of low fall detection accuracy in the prior art and improve the fall detection accuracy.

[0005] The present invention provides a fall detection method, comprising:

[0006] Based on the acquired multiple frames of the signal to be tested, determine the target centroid height and target motion information corresponding to the target object in each judgment node, wherein the judgment node is any one of at least two judgment nodes obtained by dividing the multiple frames of the signal to be tested by an equal number of frames, and each of the judgment nodes includes a first preset number of consecutive frames of the signal to be tested;

[0007] Based on the target centroid height and the target motion information, determining the state information corresponding to the target object in each of the judgment nodes;

[0008] sorting the plurality of state information according to the respective determination times corresponding to the plurality of state information, and determining a state information sequence corresponding to the plurality of state information;

[0009] Based on the state information sequence, a fall detection result of the target object is determined.

[0010] According to the fall detection method provided by the present invention, the target centroid height and target motion information corresponding to the target object in each judgment node are determined based on the acquired multiple frames of the test signal, including:

[0011] Based on multiple frames of the signal to be tested acquired by the radar, determining at least one point cloud information corresponding to each frame of the signal to be tested, wherein the point cloud information includes a measurement parameter between the target object and the radar;

[0012] Based on each of the point cloud information, determine a first centroid height and a first radial velocity corresponding to the signal to be measured in each frame;

[0013] Traversing each of the judgment nodes, and determining a target centroid height of the judgment node based on the first centroid heights and forgetting factors corresponding to all the signals to be tested in the judgment node, wherein the forgetting factor is used to characterize the reliability of the signal to be tested;

[0014] The target motion information of the judgment node is determined based on the target centroid height and the first radial velocity corresponding to the last frame of the signal to be measured in the judgment node.

[0015] According to the fall detection method provided by the present invention, the measurement parameters include: distance data, azimuth data, pitch data, radial velocity data and signal-to-noise ratio data between the target object and the radar;

[0016] The determining, based on each of the point cloud information, a first centroid height and a first radial velocity corresponding to each frame of the signal to be measured comprises:

[0017] Determine the height data of each point cloud information in the world coordinate system based on the installation height of the radar, the distance data, the azimuth data and the elevation data;

[0018] Determine the second centroid height corresponding to the signal to be tested in each frame based on the height data and signal-to-noise ratio data corresponding to all the point cloud information in the signal to be tested in the same frame;

[0019] Determine the second radial velocity corresponding to the signal to be measured in each frame based on the radial velocity data and the signal-to-noise ratio data of each point cloud information;

[0020] Traversing the signal to be tested in each frame, and determining the first centroid height of the signal to be tested in the current frame based on the second centroid heights corresponding to the signal to be tested in the current frame and the signals to be tested in the previous second preset number of frames;

[0021] The first radial velocity of the signal to be tested in the current frame is determined based on the second radial velocities corresponding to the signal to be tested in the current frame and the signals to be tested in a second preset number of consecutive frames before that.

[0022] According to the fall detection method provided by the present invention, the target motion information of the judgment node is determined based on the target centroid height and the first radial velocity corresponding to the last frame of the signal to be tested in the judgment node, including:

[0023] Determine the target centroid descent speed corresponding to the current judgment node based on the target centroid heights corresponding to the current judgment node and the previous judgment node;

[0024] Determine the node speed corresponding to the current judgment node based on the first radial speed corresponding to the last frame of the signal to be tested in the current judgment node and the target center of mass descending speed;

[0025] Determine the maximum node speed among the node speeds corresponding to the current judgment node and the third preset number of consecutive judgment nodes as the maximum target centroid speed of the current judgment node;

[0026] Determine the target center of mass acceleration corresponding to the current judgment node based on the node speeds corresponding to the current judgment node and the previous judgment node;

[0027] Based on the target center of mass height, the target center of mass descent speed, the target center of mass maximum speed and the target center of mass acceleration of the current judgment node, the target motion information corresponding to the current judgment node is determined.

[0028] According to the fall detection method provided by the present invention, determining the state information corresponding to the target object in each judgment node based on the target centroid height and the target motion information includes:

[0029] For each of the judgment nodes, the target center of mass height corresponding to the judgment node, the target center of mass descent speed in the target motion information, the target center of mass maximum speed and the target center of mass acceleration are matched with their respective corresponding preset thresholds in turn to determine the state information corresponding to the target object.

[0030] According to the fall detection method provided by the present invention, determining the fall detection result of the target object based on the state information sequence includes:

[0031] In a case where a preset state sequence is matched in the state information sequence, it is determined that the fall detection result of the target object is a fall state.

[0032] According to the fall detection method provided by the present invention, the state information corresponding to the target object in the judgment node includes: any one of a first state, a second state, a third state and a fourth state, wherein the first state is used to characterize that the target object is in a non-falling state, the second state is used to characterize that the target object is in a falling state, the third state is used to characterize that the target object is in a state at the moment of falling to the ground, and the fourth state is used to characterize that the target object is in a fallen state of falling to the ground;

[0033] The preset state sequence includes: the first state to the fourth state continuously, and the duration of the fourth state is greater than or equal to the first duration threshold;

[0034] or,

[0035] The first state, the second state and the fourth state are continuous, and the duration of the fourth state is greater than or equal to a second duration threshold.

[0036] The present invention also provides a fall detection device, comprising:

[0037] A determination module, configured to determine, based on the acquired multiple frames of the signal to be tested, a target centroid height and target motion information corresponding to the target object in each judgment node, wherein the judgment node is any one of at least two judgment nodes obtained by dividing the multiple frames of the signal to be tested by an equal number of frames, and each of the judgment nodes includes a first preset number of consecutive frames of the signal to be tested;

[0038] A state judgment module, used to determine the state information corresponding to the target object in each judgment node based on the target centroid height and the target motion information;

[0039] The state detection module is used to sort the multiple state information according to the determination time corresponding to each of the multiple state information, determine the state information sequence corresponding to the multiple state information; and determine the fall detection result of the target object based on the state information sequence.

[0040] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, any of the above-mentioned fall detection methods is implemented.

[0041] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the fall detection method as described in any one of the above is implemented.

[0042] The fall detection method, device, electronic device and storage medium provided by the present invention determine each judgment node at intervals of a first preset number of consecutive frames through the acquired multiple frames of signals to be detected, fully extract the target centroid height and target motion information of each judgment node, improve the generalization ability of the environment from the data dimension, reduce the misjudgment rate, and determine the state information of each judgment node through the target centroid height and target motion information, and sort based on the determination time of each judgment node to determine the state information sequence, match based on the state information sequence, and determine the fall detection result of the target object. The fall detection process is more consistent with the actual fall process, ensuring the rationality of the fall detection. At the same time, the false alarm rate of the fall detection is reduced, thereby improving the accuracy of the fall detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1 This is one of the flow charts of the fall detection method provided by an embodiment of the present invention;

[0045] Figure 2 This is a second flow chart of the fall detection method provided by an embodiment of the present invention;

[0046] Figure 3 is an auxiliary schematic diagram for determining the maximum velocity of the target center of mass provided by an embodiment of the present invention;

[0047] Figure 4 This is a third flow chart of the fall detection method provided by an embodiment of the present invention;

[0048] Figure 5 is a schematic diagram of the structure of a fall detection device provided by an embodiment of the present invention;

[0049] Figure 6 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] In view of the problem of low accuracy of fall detection in the prior art, an embodiment of the present invention provides a fall detection method. Figure 1 is one of the flow charts of the fall detection method provided by an embodiment of the present invention, such as Figure 1 As shown, the method includes:

[0052] Step 110, based on the acquired multi-frame test signal, determine the target center of mass height and target motion information corresponding to the target object in each judgment node, the judgment node is any one of at least two judgment nodes obtained by dividing the multi-frame test signal by an equal number of frames, and each of the judgment nodes includes a first preset number of consecutive frames of the test signal.

[0053] Specifically, the above-mentioned multi-frame test signal is a continuous intermediate frequency signal. The radar continuously transmits electromagnetic waves in frames. After the transmission signal reaches the target object, the target object reflects the echo signal to the radar. After receiving the echo signal, the radar mixes the echo signal with the transmission signal to obtain an analog signal, which is the continuous intermediate frequency signal. After obtaining the multi-frame test signal, the multi-frame test signal is divided into equal frames to obtain at least two judgment nodes, each of which includes the test signal of the first preset number of consecutive frames.

[0054] also, Figure 2 FIG. 2 is a flow chart of a fall detection method provided by an embodiment of the present invention. Figure 2 As shown in the figure, after the radar determines the signal to be measured, the signal to be measured needs to be filtered through a low-pass filter, and the analog / digital converter ADC is used to convert the analog / digital data to obtain the three-dimensional radar cube data. The three-dimensional radar cube data is signal processed, and the parameters after signal processing are data processed. After filtering out the data that does not belong to the target object, the measurement parameters between the target object and the radar can be obtained.

[0055] The above three-dimensional radar cubic data includes fast time domain dimension data, slow time domain dimension data and antenna dimension data. When the three-dimensional radar cubic data is subjected to signal processing, the fast time domain dimension data is first subjected to FFT operation, and then the slow time domain dimension data is compressed. The antenna dimension data and the fast time domain dimension data after operation are subjected to the first CaponBF (Capon Beamforming) operation to obtain RAM (Range-Azimuth Map). Through the CFAR (Constant False Alarm Rate Detector) operation, the target object can be detected from the RAM, and multiple distance cells corresponding to the detected object can be obtained. The energy of the distance cell corresponding to the target object is relatively large, and the data of the distance cell is determined as the distance data between the target object and the radar. At the same time, the signal-to-noise ratio of the distance cell is determined as the signal-to-noise ratio between the target object and the radar. In addition, the azimuth data between the target object and the radar can be obtained, which can be understood as the azimuth angle of the radar where the target object is located. After that, the information corresponding to the distance data and azimuth data in the 3D radar cube data is extracted, and the information is subjected to a second CaponBF (Capon Beamforming) operation to obtain the pitch data between the target object and the radar, that is, the pitch angle when the radar detects the target object. After that, the slow time domain dimension data in the 3D radar cube data is extracted, and the slow time domain dimension data is subjected to FFT operation to determine the radial velocity data of the target object along the radar detection direction.

[0056] It should be noted that beamforming refers to the weighted combination of spatial signals to improve the signal-to-noise ratio of the received signal in the desired direction while suppressing interference from other directions. CaponBF is one of the methods used to constrain the response of the desired signal to be a constant and minimize the total power of the output signal. RAM (Range-Azimuth Map) is the distribution diagram of the echo energy in the two-dimensional plane of range and azimuth after the radar is processed in the range and azimuth dimensions. CFAR is a common radar target detector. The important feature of this detector is to maintain a constant false alarm probability and determine whether the target object exists by dynamically adjusting the threshold level, thereby avoiding overloading the radar target detector due to too many false tracks.

[0057] After determining the measurement parameters corresponding to each frame of the test signal through the above-mentioned multiple frames of the test signal, the continuous multiple frames of the test signal are divided into multiple judgment nodes at intervals of the first preset number of frames k, that is, each continuous k frames of the test signal constitute a judgment node, and the continuous multiple frames of the test signal constitute a continuous plurality of judgment nodes. For each judgment node, the kth frame is taken as the current frame, and the measurement parameters corresponding to the current frame and the previous k-1 frames of the test signal can be fully explored to ensure the accuracy of the subsequent fall detection result.

[0058] Through the measurement parameters corresponding to the above-mentioned signals to be tested in each frame, the target center of mass height and target motion information corresponding to each judgment node can be determined. Since the target object has many reflection points, the center of mass position of the target object can be selected as the point to be detected, and the height of the target object can be determined by determining the center of mass height of the target. By determining the motion information corresponding to the target center of mass as the motion information of the target object. Compared with the prior art that makes judgments based on single-dimensional parameters and is easily disturbed by complex environments, resulting in low detection accuracy, in an embodiment of the present invention, by determining the center of mass height and target motion information, the characteristic parameters corresponding to the target object are fully extracted to perform fall detection and judgment, thereby improving the detection accuracy and the generalization ability of the environment and expanding the scope of application.

[0059] Step 120: Determine the state information corresponding to the target object in each judgment node based on the target centroid height and the target motion information.

[0060] Specifically, after determining the target centroid height and target motion information corresponding to each judgment node, the target centroid height and target motion information can be matched with their respective corresponding thresholds in turn through the threshold judgment method, thereby determining the state information corresponding to the target object in each judgment node. The state information is used to characterize whether the target object is in a fallen state or whether the target object is in a falling state.

[0061] Step 130: Sort the plurality of state information according to the determination time points corresponding to the plurality of state information, and determine a state information sequence corresponding to the plurality of state information.

[0062] Specifically, after determining the state information corresponding to each judgment node, the state information sequence is determined by sorting them based on the order of precedence according to the determination times corresponding to each of the multiple state information, so as to facilitate the subsequent determination of the fall detection result of the detection target object through the state information sequence.

[0063] Step 140: Determine a fall detection result of the target object based on the state information sequence.

[0064] Specifically, after determining the state information sequence, the fall judgment can be performed according to the state information sequence, avoiding the increase of the false alarm rate of detection due to the misjudgment of the state information of a single point when the state information of a single judgment node or a single frame of the signal to be tested is used for judgment. When the state information sequence is used for fall judgment, the state information of each judgment node is used to judge whether the target object has experienced the whole process from the state of not falling to the state of having fallen. This process is reflected by a plurality of continuous state information, which is more in line with the characteristics of falling, so as to improve the accuracy of fall detection and reduce the false alarm rate.

[0065] Optionally, determining the target centroid height and target motion information corresponding to the target object in each judgment node based on the acquired multiple frames of the signal to be tested includes:

[0066] Based on multiple frames of the signal to be tested acquired by the radar, determining at least one point cloud information corresponding to each frame of the signal to be tested, wherein the point cloud information includes a measurement parameter between the target object and the radar;

[0067] Based on each of the point cloud information, determine a first centroid height and a first radial velocity corresponding to the signal to be measured in each frame;

[0068] Traversing each of the judgment nodes, and determining a target centroid height of the judgment node based on the first centroid heights and forgetting factors corresponding to all the signals to be tested in the judgment node, wherein the forgetting factor is used to characterize the reliability of the signal to be tested;

[0069] The target motion information of the judgment node is determined based on the target centroid height and the first radial velocity corresponding to the last frame of the signal to be measured in the judgment node.

[0070] Specifically, after the radar acquires multiple frames of the signal to be tested, and the multiple frames of the signal to be tested are subjected to low-pass filtering, analog / digital conversion, signal processing, and data processing, at least one point cloud information can be obtained. In the case where one frame of the signal to be tested obtains one frame of point cloud information, the signal to be tested and the point cloud information correspond one to one. In the case where one frame of the signal to be tested obtains at least two frames of point cloud information, the signal to be tested and the point cloud information are in a one-to-many relationship, that is, after the above processing is performed on each frame of the signal to be tested, M point cloud information {PointCloud 1 , PointCloud 2 …PointCloud M}, M is an integer greater than or equal to 2. The jth point cloud information PointCloud j The data includes the distance data, azimuth data, elevation data, radial velocity data v between the target object and the radar. j and signal-to-noise ratio data SNR j , j is an integer greater than or equal to 1 and less than or equal to M. After obtaining the above point cloud information, the signal-to-noise ratio data SNR in each point cloud information can be further j In a weighted manner, the first centroid height and the first radial velocity corresponding to the signal to be tested in each frame are obtained. In order to fully exploit the signal to be tested in the current frame and the previous frames, each continuous k frames of the signal to be tested constitute a judgment node. After the continuous multiple frames of the signal to be tested are separated into a plurality of continuous judgment nodes, the target centroid height of each judgment node can be determined by formula (1) according to the first centroid height and forgetting factor corresponding to all the signals to be tested in each judgment node. Formula (1) is:

[0071]

[0072] in, Indicates the target centroid height corresponding to the qth judgment node, H i Indicates the height of the first centroid corresponding to the last frame of the signal to be tested in the qth judgment node, H i-k+1 represents the first centroid height corresponding to the first frame of the signal to be tested in the qth judgment node, k represents the number of frames of the signal to be tested in the ith judgment node, λ represents the forgetting factor of the last frame of the signal to be tested, and 0<λ<1. The larger λ is, the higher the first centroid height H corresponding to the last frame of the signal to be tested in the qth judgment node. i The higher the reliability, 1-λ represents the forgetting factor of the previous frame of the last frame of the signal to be tested. Similarly, (1-λ) k-1It represents the forgetting factor of the first frame of the signal to be tested in the qth judgment node. The farther the other frames of the signal to be tested in the qth judgment node are from the last frame of the signal to be tested, the smaller the forgetting factors corresponding to the other frames of the signal to be tested are, and the lower the rationality of determining the first centroid height according to the other frames of the signal to be tested is, resulting in lower reliability of the first centroid heights corresponding to the other frames of the signal to be tested.

[0073] After determining the target centroid height of each judgment node, the target motion information corresponding to the judgment node may be determined according to the target centroid height corresponding to the judgment node and the first radial velocity corresponding to the last frame of the signal to be tested in the judgment node.

[0074] Optionally, the above-mentioned radar can be a millimeter wave radar, which has the advantages of small size, non-contact, all-weather, not limited by lighting conditions and not infringing on user privacy. It has important application value in the field of medical health, especially in fall detection.

[0075] Optionally, each of the above-mentioned frames of the signal to be tested is a signal that can detect at least one point cloud information. After obtaining multiple frames of continuous initial signals, if the initial signal of the frame does not detect at least one point cloud information, the initial signal of the frame can be discarded and the judgment can be postponed to the next initial signal of the frame. If the initial signal of the frame detects at least one point cloud information, the initial signal of the frame can be determined as the signal to be tested.

[0076] Optionally, the measurement parameters include: distance data, azimuth data, elevation data, radial velocity data and signal-to-noise ratio data between the target object and the radar;

[0077] The determining, based on each of the point cloud information, a first centroid height and a first radial velocity corresponding to each frame of the signal to be measured comprises:

[0078] Determine the height data of each point cloud information in the world coordinate system based on the installation height of the radar, the distance data, the azimuth data and the elevation data;

[0079] Determine the second centroid height corresponding to the signal to be tested in each frame based on the height data and signal-to-noise ratio data corresponding to all the point cloud information in the signal to be tested in the same frame;

[0080] Determine the second radial velocity corresponding to the signal to be measured in each frame based on the radial velocity data and the signal-to-noise ratio data of each point cloud information;

[0081] Traversing the signal to be tested in each frame, and determining the first centroid height of the signal to be tested in the current frame based on the second centroid heights corresponding to the signal to be tested in the current frame and the signals to be tested in the previous second preset number of frames;

[0082] The first radial velocity of the signal to be tested in the current frame is determined based on the second radial velocities corresponding to the signal to be tested in the current frame and the signals to be tested in a second preset number of consecutive frames before that.

[0083] Specifically, after determining the point cloud information corresponding to each frame of the signal to be tested, the point cloud data can be mapped to the world coordinate system according to the installation height of the radar, the distance data, azimuth data and pitch data in each point cloud information to determine the world coordinates of each point cloud data, and the height data x of each point cloud information in the world coordinate system can be determined through the world coordinates. j . In determining the height data x j Then, taking the same frame of the measured signal corresponding to M point cloud information as an example, the signal-to-noise ratio data SNR in each point cloud information is used j With height data x j The weighted method uses formula (2) to determine the second centroid height of the corresponding signal to be measured, and uses the signal-to-noise ratio data SNR in each point cloud information j With radial velocity data v j The weighted method uses formula (3) to determine the second radial velocity corresponding to the signal to be measured. Formula (2) is:

[0084]

[0085] Among them, X i (j) represents the second centroid height corresponding to the i-th frame of the signal to be tested, and M represents the amount of point cloud information in the i-th frame of the signal to be tested.

[0086] Formula (3) is:

[0087]

[0088] in, represents the second radial velocity corresponding to the i-th frame of the signal to be measured.

[0089] After determining the second centroid height and second radial velocity corresponding to each frame of the signal to be tested, in order to reduce the impact of the measurement error within a frame on the fall detection result, for the continuous multiple frames of the signal to be tested, the average value of the second centroid heights corresponding to the signals to be tested of the second preset number of frames are selected as the first centroid height corresponding to the signal to be tested of the current frame, and the average value of the second radial velocities corresponding to the signals to be tested of the second preset number of frames are selected as the first radial velocity corresponding to the signal to be tested of the current frame. That is, when the second preset number is N-1, traverse each frame of the signal to be tested, and use formula (4) to take the average value of the second centroid heights corresponding to the signal to be tested of the current frame and the previous N-1 frames as the first centroid height corresponding to the signal to be tested of the current frame. Use formula (5) to take the average value of the second radial velocities corresponding to the signal to be tested of the current frame and the previous N-1 frames as the first radial velocity corresponding to the signal to be tested of the current frame. Formula (4) is:

[0090]

[0091] Among them, H i (j) represents the first centroid height corresponding to the i-th frame of the signal to be tested, and N is an integer greater than or equal to k.

[0092] Formula (5) is:

[0093]

[0094] in, Indicates the first radial velocity corresponding to the i-th frame of the signal to be measured.

[0095] Optionally, determining the target motion information of the judgment node based on the target centroid height and the first radial velocity corresponding to the last frame of the signal to be tested in the judgment node includes:

[0096] Determine the target centroid descent speed corresponding to the current judgment node based on the target centroid heights corresponding to the current judgment node and the previous judgment node;

[0097] Determine the node speed corresponding to the current judgment node based on the first radial speed corresponding to the last frame of the signal to be tested in the current judgment node and the target center of mass descending speed;

[0098] Determine the maximum node speed among the node speeds corresponding to the current judgment node and the third preset number of consecutive judgment nodes as the maximum target centroid speed of the current judgment node;

[0099] Determine the target center of mass acceleration corresponding to the current judgment node based on the node speeds corresponding to the current judgment node and the previous judgment node;

[0100] Based on the target center of mass height, the target center of mass descent speed, the target center of mass maximum speed and the target center of mass acceleration of the current judgment node, the target motion information corresponding to the current judgment node is determined.

[0101] Specifically, traverse each judgment node and determine the target centroid height Then, using formula (6), according to the target centroid height of the current judgment node and the target centroid height of the previous judgment node Determine the target center of mass descent speed corresponding to the current judgment node When the target object is in a falling state, the center of mass drops. At this time, the target center of mass of the target object drops at a speed of Formula (6) is:

[0102]

[0103] Among them, T frame It indicates the time corresponding to one frame of the signal to be tested, which can be determined by the number of frames of the signal to be tested determined by the radar within a preset time period.

[0104] Determine the target center of mass descent speed After that, based on the fact that when the target object falls, in addition to the target center of mass falling speed perpendicular to the ground, it also includes the target center of mass radial speed along the radar direction. In order to avoid the problem that the radar cannot accurately measure the speed when the target object falls along the radar tangential direction, resulting in low detection accuracy, in the embodiment of the present invention, since the reliability of the last frame of the signal to be tested in the judgment node is the highest, the first radial speed corresponding to the last frame of the signal to be tested in the current judgment node is used as the target center of mass radial speed along the radar direction in the current judgment node. Formula (7) is used to weight the target center of mass falling speed and the first radial speed corresponding to the last frame of the signal to be tested in the current judgment node to determine the node speed of the current judgment node. Formula (7) is:

[0105]

[0106] Among them, V q (i) represents the node speed of the qth judgment node, that is, the node speed of the current judgment node, Indicates the radial velocity of the target mass center along the radar direction in the current judgment node, that is, the first radial velocity corresponding to the last frame of the signal to be tested in the current judgment node. w represents the velocity weight, which is generally set between 0.7 and 0.9. Based on the application of the embodiment of the present invention to the fall detection of the target object, the target mass center descent velocity perpendicular to the ground is set. With greater weight.

[0107] In addition, using formula (8), according to the node speed V of the current judgment node q (i) and the node velocity V of the previous judgment node q-1 (i) Determine the target center of mass acceleration a corresponding to the current judgment node q , formula (8) is:

[0108]

[0109] Figure 3 is an auxiliary schematic diagram for determining the maximum velocity of the target center of mass provided by an embodiment of the present invention, such as Figure 3 As shown in the figure, the target object falls in the process of falling, and the target center of mass falls at a speed perpendicular to the ground. The absolute value of will reach the maximum, causing the node speed to reach the maximum value. Figure 3 When the target center of mass height of the same judgment node is at point A, It is not the lowest. After several judgment nodes, the target centroid height Reach the lowest point B, that is, the target center of mass descending speed and the target centroid height There is a misalignment phenomenon in the time of reaching the extreme value, which causes the maximum value of the node speed to arrive before the lowest target centroid height. Therefore, when determining the node speed V of each judgment node q (i) To ensure the accuracy of the determination of the maximum speed of the target center of mass in the motion information, based on the node speeds corresponding to the current judgment node and the third consecutive preset number of judgment nodes, the maximum node speed is determined as the maximum speed of the target center of mass. That is, taking the third preset number as m as an example, m≤k, so as to improve the accuracy of subsequent determination of state information based on the target center of mass height and target motion information.

[0110] Optionally, the determining, based on the target centroid height and the target motion information, the state information corresponding to the target object in each of the judgment nodes includes:

[0111] For each of the judgment nodes, the target center of mass height corresponding to the judgment node, the target center of mass descent speed in the target motion information, the target center of mass maximum speed and the target center of mass acceleration are matched with their respective corresponding preset thresholds in turn to determine the state information corresponding to the target object.

[0112] Specifically, Figure 4 FIG. 3 is a flow chart of a fall detection method provided by an embodiment of the present invention. Figure 4 As shown, determining the state information corresponding to the target object includes the following steps:

[0113] S1. For each judgment node, compare the target centroid height with the first height threshold H 1_Thr For comparison, execute S1.1 and S1.2 below;

[0114] S1.1. When the target centroid height is greater than or equal to the first height threshold H 1_Thr In the case of, determining that the state flag corresponding to the target object is the first state;

[0115] S1.2, when the target centroid height is less than the first height threshold H 1_Thr In the case of 2_Thr For comparison, execute S1.21 and S1.22 below;

[0116] S1.21, when the target centroid height is less than the second height threshold H 2_Thr In the case of , determining that the state flag corresponding to the target is in the fourth state;

[0117] S1.22, when the target centroid height is greater than or equal to the second height threshold H 2_Thr In the case of , determine whether the target center of mass descent speed is less than 0, and execute the following S1.221 and S1.222;

[0118] S1.221. When the target mass center descending speed is greater than or equal to 0, determine that the state flag corresponding to the target is the first state;

[0119] S1.222, when the target mass center descending speed is less than 0, the target mass center maximum speed and the first speed threshold V 1_Thr For comparison, execute S1.2221 and S1.2222 below;

[0120] S1.2221, when the maximum speed of the target center of mass is less than or equal to the first speed threshold V 1_Thr In the case of, determining that the state flag bit corresponding to the target is the first state;

[0121] S1.2222, when the maximum speed of the target center of mass is greater than the first speed threshold V 1_Thr In the case of 2_Thr For comparison, execute the following S1.22221 and S1.22222;

[0122] S1.22221, when the maximum speed of the target center of mass is greater than the second speed threshold V 2_Thr In the case of , determining that the state flag bit corresponding to the target is in the third state;

[0123] S1.22222, when the maximum speed of the target center of mass is less than or equal to the second speed threshold V 2_ThrIn the case of Thr For comparison, execute the following S1.222221 and S1.222222;

[0124] S1.222221, when the absolute value of the target center of mass acceleration is greater than the acceleration threshold a Thr In the case of , determining that the state flag bit corresponding to the target is in the third state;

[0125] S1.222222, when the absolute value of the target center of mass acceleration is less than or equal to the acceleration threshold a Thr In the case of, it is determined that the state identification bit corresponding to the target is the second state.

[0126] It should be noted that the first height threshold H 1_Thr Greater than the second height threshold H 2_Thr , the first speed threshold V 1_Thr Less than the second speed threshold V 2_Thr .

[0127] The above-mentioned first state can indicate that the target object is in a normal non-falling state such as walking, standing and sitting, and the non-falling state indicates that the target object has no safety risk in the judgment node to which it belongs. The above-mentioned second state indicates that the target object is in a falling state, and there is a certain safety risk. The above-mentioned third state indicates that the target object is in a state at the moment of falling to the ground. When falling to the ground, since the speed suddenly becomes 0, the acceleration changes greatly in the third state. In the third state, the corresponding risk factor of the target object is higher. The above-mentioned fourth state indicates that the target user is in a fallen state. At this time, the target center of mass height is the lowest.

[0128] Optionally, the above status information can be represented by a status identification bit, for example, the status identification bit of the first state can be 1, the status identification bit of the second state can be 2, the status identification bit of the third state can be 3, and the status identification bit of the fourth state can be 4; or, the status identification bit of the first state can be a, the status identification bit of the second state can be b, the status identification bit of the third state can be c, the status identification bit of the fourth state can be d, and so on. The embodiment of the present invention is not limited to this.

[0129] Optionally, determining a fall detection result of the target object based on the state information sequence includes:

[0130] In a case where a preset state sequence is matched in the state information sequence, it is determined that the fall detection result of the target object is a fall state.

[0131] Specifically, after determining the state information sequence, multiple consecutive state information in the state information sequence are matched with the preset state sequence. If the preset state sequence is matched in the state information sequence, it indicates that the target object has experienced the entire process from the never-fallen state to the fallen state. The fall detection result of the target object can be determined as the fall state, thereby improving the accuracy of fall detection and reducing the false alarm rate.

[0132] Optionally, the state information corresponding to the target object in the judgment node includes: any one of a first state, a second state, a third state and a fourth state, wherein the first state is used to characterize that the target object is in a non-falling state, the second state is used to characterize that the target object is in a falling state, the third state is used to characterize that the target object is in a state at the moment of falling to the ground, and the fourth state is used to characterize that the target object is in a fallen state of falling to the ground;

[0133] The preset state sequence includes: the first state to the fourth state continuously, and the duration of the fourth state is greater than or equal to the first duration threshold;

[0134] or,

[0135] The first state, the second state and the fourth state are continuous, and the duration of the fourth state is greater than or equal to a second duration threshold.

[0136] Specifically, the above-mentioned preset state sequence includes a continuous first state, a second state, a third state and a fourth state, and the target object is in the fourth state for a long time, and the duration is greater than or equal to the first duration threshold, indicating that the target object has experienced a complete fall process, and has been in a fallen state for a long time after falling, or, if the state information sequence includes a continuous first state, a second state and a fourth state, and the target object is in the fourth state for a long time, and the duration is greater than or equal to the second duration threshold, indicating that the target object has experienced a complete fall process, and has been in a fallen state for a long time after falling. At this time, it can be determined that the fall detection result of the target object is a fall state, and an alarm mechanism can be triggered to remind family members or medical personnel to implement targeted measures until the first state is detected, indicating that the target object has been out of danger after targeted measures are implemented by family members or medical personnel. In the period between the continuous fourth state and the detection of the first state, the alarm mechanism is always triggered, and after the first state is detected, the alarm mechanism is no longer triggered.

[0137] It should be noted that, since the third state lasts for a very short time and is difficult to be detected, the second time threshold is set to be greater than the first time threshold to ensure the reliability of the fall detection result.

[0138] The fall detection method provided by the embodiment of the present invention determines each judgment node at intervals of a first preset number of frames through the acquired multiple frames of signals to be detected, fully extracts the target centroid height and target motion information of each judgment node, improves the generalization ability of the environment from the data dimension, reduces the misjudgment rate, and determines the state information of each judgment node through the target centroid height and target motion information, and sorts the determination time of each judgment node to determine the state information sequence, matches based on the state information sequence, and determines the fall detection result of the target object. The fall detection process is more consistent with the actual fall process, ensuring the rationality of the fall detection. At the same time, the false alarm rate of the fall detection is reduced, thereby improving the accuracy of the fall detection.

[0139] The fall detection device provided by the present invention is described below. The fall detection device described below and the fall detection method described above can refer to each other.

[0140] The embodiment of the present invention further provides a fall detection device, Figure 5 is a schematic diagram of the structure of a fall detection device provided by an embodiment of the present invention. Figure 5 As shown, the fall detection device 500 includes: a determination module 510, a state judgment module 520 and a state monitoring module, wherein:

[0141] A determination module 510 is used to determine the target centroid height and target motion information corresponding to the target object in each judgment node based on the acquired multiple frames of the signal to be tested, wherein the judgment node is any one of at least two judgment nodes obtained by dividing the multiple frames of the signal to be tested by an equal number of frames, and each of the judgment nodes includes a first preset number of consecutive frames of the signal to be tested;

[0142] A state judgment module 520, configured to determine the state information corresponding to the target object in each judgment node based on the target centroid height and the target motion information;

[0143] The state detection module 530 is used to sort the multiple state information according to the determination time corresponding to each of the multiple state information, determine the state information sequence corresponding to the multiple state information; and determine the fall detection result of the target object based on the state information sequence.

[0144] The fall detection device provided by the embodiment of the present invention determines each judgment node at intervals of a first preset number of frames through the acquired multiple frames of signals to be detected, fully extracts the target centroid height and target motion information of each judgment node, improves the generalization ability of the environment from the data dimension, reduces the misjudgment rate, and determines the state information of each judgment node through the target centroid height and target motion information, and sorts the determination time of each judgment node to determine the state information sequence, matches based on the state information sequence, and determines the fall detection result of the target object. The fall detection process is more consistent with the actual fall process, ensuring the rationality of the fall detection. At the same time, the false alarm rate of the fall detection is reduced, thereby improving the accuracy of the fall detection.

[0145] Optionally, the determination module 510 is specifically configured to:

[0146] Based on multiple frames of the signal to be tested acquired by the radar, determining at least one point cloud information corresponding to each frame of the signal to be tested, wherein the point cloud information includes a measurement parameter between the target object and the radar;

[0147] Based on each of the point cloud information, determine a first centroid height and a first radial velocity corresponding to the signal to be measured in each frame;

[0148] Traversing each of the judgment nodes, and determining a target centroid height of the judgment node based on the first centroid heights and forgetting factors corresponding to all the signals to be tested in the judgment node, wherein the forgetting factor is used to characterize the reliability of the signal to be tested;

[0149] The target motion information of the judgment node is determined based on the target centroid height and the first radial velocity corresponding to the last frame of the signal to be measured in the judgment node.

[0150] Optionally, the measurement parameters include: distance data, azimuth data, pitch data, radial velocity data and signal-to-noise ratio data between the target object and the radar.

[0151] Optionally, the determination module 510 is specifically configured to:

[0152] Determine the height data of each point cloud information in the world coordinate system based on the installation height of the radar, the distance data, the azimuth data and the elevation data;

[0153] Determine the second centroid height corresponding to the signal to be tested in each frame based on the height data and signal-to-noise ratio data corresponding to all the point cloud information in the signal to be tested in the same frame;

[0154] Determine the second radial velocity corresponding to the signal to be measured in each frame based on the radial velocity data and the signal-to-noise ratio data of each point cloud information;

[0155] Traversing the signal to be tested in each frame, and determining the first centroid height of the signal to be tested in the current frame based on the second centroid heights corresponding to the signal to be tested in the current frame and the signals to be tested in the previous second preset number of frames;

[0156] The first radial velocity of the signal to be tested in the current frame is determined based on the second radial velocities corresponding to the signal to be tested in the current frame and the signals to be tested in a second preset number of consecutive frames before that.

[0157] Optionally, the determination module 510 is specifically configured to:

[0158] Determine the target centroid descent speed corresponding to the current judgment node based on the target centroid heights corresponding to the current judgment node and the previous judgment node;

[0159] Determine the node speed corresponding to the current judgment node based on the first radial speed corresponding to the last frame of the signal to be tested in the current judgment node and the target center of mass descending speed;

[0160] Determine the maximum node speed among the node speeds corresponding to the current judgment node and the third preset number of consecutive judgment nodes as the maximum target centroid speed of the current judgment node;

[0161] Determine the target center of mass acceleration corresponding to the current judgment node based on the node speeds corresponding to the current judgment node and the previous judgment node;

[0162] Based on the target center of mass height, the target center of mass descent speed, the target center of mass maximum speed and the target center of mass acceleration of the current judgment node, the target motion information corresponding to the current judgment node is determined.

[0163] Optionally, the state determination module 520 is specifically configured to:

[0164] For each of the judgment nodes, the target center of mass height corresponding to the judgment node, the target center of mass descent speed in the target motion information, the target center of mass maximum speed and the target center of mass acceleration are matched with their respective corresponding preset thresholds in turn to determine the state information corresponding to the target object.

[0165] Optionally, the state detection module 530 is specifically configured to:

[0166] In a case where a preset state sequence is matched in the state information sequence, it is determined that the fall detection result of the target object is a fall state.

[0167] Optionally, the state information corresponding to the target object in the judgment node includes: any one of a first state, a second state, a third state and a fourth state, wherein the first state is used to characterize that the target object is in a non-falling state, the second state is used to characterize that the target object is in a falling state, the third state is used to characterize that the target object is in a state at the moment of falling to the ground, and the fourth state is used to characterize that the target object is in a fallen state of falling to the ground;

[0168] The preset state sequence includes: the first state to the fourth state continuously, and the duration of the fourth state is greater than or equal to the first duration threshold;

[0169] or,

[0170] The first state, the second state and the fourth state are continuous, and the duration of the fourth state is greater than or equal to a second duration threshold.

[0171] Figure 6 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention, such as Figure 6 As shown, the electronic device may include: a processor 610, a communication interface 620, a memory 630 and a communication bus 640, wherein the processor 610, the communication interface 620 and the memory 630 communicate with each other through the communication bus 640. The processor 610 may call the logic instructions in the memory 630 to execute the fall detection method, which includes:

[0172] Based on the acquired multiple frames of the signal to be tested, determine the target centroid height and target motion information corresponding to the target object in each judgment node, wherein the judgment node is any one of at least two judgment nodes obtained by dividing the multiple frames of the signal to be tested by an equal number of frames, and each of the judgment nodes includes a first preset number of consecutive frames of the signal to be tested;

[0173] Based on the target centroid height and the target motion information, determining the state information corresponding to the target object in each of the judgment nodes;

[0174] sorting the plurality of state information according to the respective determination times corresponding to the plurality of state information, and determining a state information sequence corresponding to the plurality of state information;

[0175] Based on the state information sequence, a fall detection result of the target object is determined.

[0176] In addition, the logic instructions in the above-mentioned memory 630 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0177] On the other hand, the present invention further provides a computer program product, the computer program product includes a computer program, the computer program can be stored on a computer-readable storage medium, when the computer program is executed by a processor, the computer can execute the fall detection method provided by the above methods, the method includes:

[0178] Based on the acquired multiple frames of the signal to be tested, determine the target centroid height and target motion information corresponding to the target object in each judgment node, wherein the judgment node is any one of at least two judgment nodes obtained by dividing the multiple frames of the signal to be tested by an equal number of frames, and each of the judgment nodes includes a first preset number of consecutive frames of the signal to be tested;

[0179] Based on the target centroid height and the target motion information, determining the state information corresponding to the target object in each of the judgment nodes;

[0180] sorting the plurality of state information according to the respective determination times corresponding to the plurality of state information, and determining a state information sequence corresponding to the plurality of state information;

[0181] Based on the state information sequence, a fall detection result of the target object is determined.

[0182] In another aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the fall detection method provided by the above methods is implemented, and the method includes:

[0183] Based on the acquired multiple frames of the signal to be tested, determine the target centroid height and target motion information corresponding to the target object in each judgment node, wherein the judgment node is any one of at least two judgment nodes obtained by dividing the multiple frames of the signal to be tested by an equal number of frames, and each of the judgment nodes includes a first preset number of consecutive frames of the signal to be tested;

[0184] Based on the target centroid height and the target motion information, determining the state information corresponding to the target object in each of the judgment nodes;

[0185] sorting the plurality of state information according to the respective determination times corresponding to the plurality of state information, and determining a state information sequence corresponding to the plurality of state information;

[0186] Based on the state information sequence, a fall detection result of the target object is determined.

[0187] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0188] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0189] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A fall detection method, characterized in that: include: Based on the acquired multiple frames of the signal to be tested, determine the target centroid height and target motion information corresponding to the target object in each judgment node, wherein the judgment node is any one of at least two judgment nodes obtained by dividing the multiple frames of the signal to be tested by an equal number of frames, and each of the judgment nodes includes a first preset number of consecutive frames of the signal to be tested; Based on the target centroid height and the target motion information, determining the state information corresponding to the target object in each of the judgment nodes; sorting the plurality of state information according to the respective determination times corresponding to the plurality of state information, and determining a state information sequence corresponding to the plurality of state information; Based on the state information sequence, a fall detection result of the target object is determined.

2. The fall detection method according to claim 1, characterized in that: The method of determining the target centroid height and target motion information corresponding to the target object in each judgment node based on the acquired multiple frames of the test signal includes: Based on multiple frames of the signal to be tested acquired by the radar, determining at least one point cloud information corresponding to each frame of the signal to be tested, wherein the point cloud information includes a measurement parameter between the target object and the radar; Based on each of the point cloud information, determine a first centroid height and a first radial velocity corresponding to the signal to be measured in each frame; Traversing each of the judgment nodes, and determining a target centroid height of the judgment node based on the first centroid heights and forgetting factors corresponding to all the signals to be tested in the judgment node, wherein the forgetting factor is used to characterize the reliability of the signal to be tested; The target motion information of the judgment node is determined based on the target centroid height and the first radial velocity corresponding to the last frame of the signal to be measured in the judgment node.

3. The fall detection method according to claim 2, characterized in that: The measurement parameters include: distance data, azimuth data, elevation data, radial velocity data and signal-to-noise ratio data between the target object and the radar; The determining, based on each of the point cloud information, a first centroid height and a first radial velocity corresponding to each frame of the signal to be measured comprises: Determine the height data of each point cloud information in the world coordinate system based on the installation height of the radar, the distance data, the azimuth data and the elevation data; Determine the second centroid height corresponding to the signal to be tested in each frame based on the height data and signal-to-noise ratio data corresponding to all the point cloud information in the signal to be tested in the same frame; Determine the second radial velocity corresponding to the signal to be measured in each frame based on the radial velocity data and the signal-to-noise ratio data of each point cloud information; Traversing the signal to be tested in each frame, and determining the first centroid height of the signal to be tested in the current frame based on the second centroid heights corresponding to the signal to be tested in the current frame and the signals to be tested in the previous second preset number of frames; The first radial velocity of the signal to be tested in the current frame is determined based on the second radial velocities corresponding to the signal to be tested in the current frame and the signals to be tested in a second preset number of consecutive frames before the signal to be tested.

4. The fall detection method according to claim 3, characterized in that: The determining the target motion information of the judgment node based on the target centroid height and the first radial velocity corresponding to the last frame of the signal to be tested in the judgment node includes: Determine the target centroid descent speed corresponding to the current judgment node based on the target centroid heights corresponding to the current judgment node and the previous judgment node; Determine the node speed corresponding to the current judgment node based on the first radial speed corresponding to the last frame of the signal to be tested in the current judgment node and the target center of mass descending speed; Determine the maximum node speed among the node speeds corresponding to the current judgment node and the third preset number of consecutive judgment nodes as the maximum target centroid speed of the current judgment node; Determine the target center of mass acceleration corresponding to the current judgment node based on the node speeds corresponding to the current judgment node and the previous judgment node; Based on the target center of mass height, the target center of mass descent speed, the target center of mass maximum speed and the target center of mass acceleration of the current judgment node, the target motion information corresponding to the current judgment node is determined.

5. The fall detection method according to any one of claims 1 to 4, characterized in that: The determining, based on the target centroid height and the target motion information, the state information corresponding to the target object in each of the judgment nodes includes: For each of the judgment nodes, the target center of mass height corresponding to the judgment node, the target center of mass descent speed in the target motion information, the target center of mass maximum speed and the target center of mass acceleration are matched with their respective corresponding preset thresholds in turn to determine the state information corresponding to the target object.

6. The fall detection method according to claim 5, characterized in that: The determining, based on the state information sequence, a fall detection result of the target object includes: In a case where a preset state sequence is matched in the state information sequence, it is determined that the fall detection result of the target object is a fall state.

7. The fall detection method according to claim 6, characterized in that: The state information corresponding to the target object in the judgment node includes: any one of a first state, a second state, a third state and a fourth state, wherein the first state is used to represent that the target object is in a non-falling state, the second state is used to represent that the target object is in a falling state, the third state is used to represent that the target object is in a state at the moment of falling to the ground, and the fourth state is used to represent that the target object is in a fallen state of falling to the ground; The preset state sequence includes: the first state to the fourth state continuously, and the duration of the fourth state is greater than or equal to the first duration threshold; or, The first state, the second state and the fourth state are continuous, and the duration of the fourth state is greater than or equal to a second duration threshold.

8. A fall detection device, characterized in that: include: A determination module, configured to determine, based on the acquired multiple frames of the signal to be tested, a target centroid height and target motion information corresponding to the target object in each judgment node, wherein the judgment node is any one of at least two judgment nodes obtained by dividing the multiple frames of the signal to be tested by an equal number of frames, and each of the judgment nodes includes a first preset number of consecutive frames of the signal to be tested; A state judgment module, used to determine the state information corresponding to the target object in each judgment node based on the target centroid height and the target motion information; A state detection module, used for sorting the plurality of state information according to the respective determination times corresponding to the plurality of state information, and determining a state information sequence corresponding to the plurality of state information; Based on the state information sequence, a fall detection result of the target object is determined.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the fall detection method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the fall detection method according to any one of claims 1 to 7 is implemented.

Images