Radar-based fall detection method, apparatus, device, and medium

Abstract

The present application relates to a radar-based fall detection method, device, equipment and medium, comprising: obtaining the intermediate frequency signal of the human target activity based on the radar sensing device, and extracting the 4D point cloud of the human target through the intermediate frequency signal; calculating the first height of the human target according to the 4D point cloud of the human target, and judging whether the height of the human target decreases according to the first height of the human target; when the height decreases, the dynamic and static state of the human target is judged; calculating the second height of the human target according to the dynamic and static state of the human target; judging whether the human target lies on the ground according to the second height of the human target, and if the human target lies on the ground for a set time, it is determined that a life-threatening fall event has occurred. The present application proposes two-stage fall detection to determine whether a life-threatening fall occurs, which can greatly reduce the false positive rate of fall detection.

Description

Technical Field

[0001] This invention relates to a radar-based fall detection method, device, equipment, and medium, and pertains to the field of IoT intelligent non-contact sensing technology. Background Technology

[0002] In modern society, falls are one of the leading causes of accidental injury among the elderly. Statistics show that one-third of people over 65 experience a fall each year. Accurately detecting fall behavior in the elderly is becoming increasingly important, and numerous researchers have proposed various fall detection methods.

[0003] Current fall detection methods include wearable device-based methods, vision-based methods, and wireless signal-based methods. Wearable device-based methods require the elderly to wear the devices at all times, and these devices only function properly when worn correctly and powered. For elderly people living at home, the frequent charging of these devices is very cumbersome. Vision-based methods, which use cameras, are generally quite accurate, but these methods face privacy invasion issues, especially with cameras installed in bathrooms. Wireless signal-based fall detection methods have attracted widespread attention in recent years due to their privacy-preserving characteristics. These methods continuously monitor human behavior by analyzing wireless signals reflected from the human body, such as Wi-Fi and radar signals, to detect falls. Current fall detection methods are all based on a core assumption: a fall is accompanied by a sudden body movement, causing a short and strong signal fluctuation in the wireless signal. Therefore, researchers have proposed time-frequency analysis-based methods and deep learning-based methods to identify fall behavior by detecting this signal fluctuation.

[0004] However, existing methods primarily focus on detecting the act of falling itself, which doesn't necessarily mean the person is in danger. If someone falls and gets up on their own, it means they are uninjured and can seek help, making an alarm unnecessary. The truly dangerous situation arises when someone falls and, due to loss of consciousness or serious injury, is unable to stand and remains lying on the ground. Therefore, these life-threatening falls are what truly need to be detected and alerted. Summary of the Invention

[0005] The present invention aims to at least solve one of the technical problems existing in the prior art. Therefore, in view of the above-mentioned problems, the object of the present invention is to provide a radar-based fall detection method, apparatus, device, and medium capable of accurately detecting life-threatening falls.

[0006] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows:

[0007] In a first aspect, the present invention provides a radar-based fall detection method, comprising:

[0008] The intermediate frequency signal of human target activity is obtained based on radar sensing equipment, and the 4D point cloud of human target is extracted from the intermediate frequency signal;

[0009] The first height of the human body is calculated based on the 4D point cloud of the human body target, and the human body is judged to have experienced a height drop event based on the first height of the human body target.

[0010] When a descent occurs, the dynamic and static status of the human target is assessed.

[0011] The second height of the human target is calculated based on the dynamic and static states of the human body;

[0012] The second height of the human body is used to determine whether the human body is lying on the ground. If the human body lies on the ground for a set period of time, it is determined that a life-threatening fall has occurred.

[0013] Furthermore, 4D point clouds of the human target are extracted using intermediate frequency signals, including:

[0014] The intermediate frequency signal is calculated using a fast Fourier transform to obtain the distance between the human target and the radar sensing device.

[0015] The intermediate frequency signal after the fast Fourier transform is subjected to a second fast Fourier transform along the slow time dimension to obtain the velocity of the human target and the range-velocity spectrum of the human target relative to the radar sensing device.

[0016] Target points for human body reflection are selected based on distance-velocity spectrum.

[0017] Phase compensation is performed on the human target to calculate the angle of the human target;

[0018] By combining the distance, velocity, and angle information of each human body's reflected target point relative to the radar sensing device, a 4D point cloud of the human target is obtained. The 4D point cloud includes 3D spatial information and velocity.

[0019] Furthermore, based on the 4D point cloud computing of the human target, the first height of the human target includes:

[0020] Perform coordinate system transformation on the 4D point cloud, transforming it from the radar coordinate system to the world coordinate system;

[0021] Clustering methods are used to denoise the 4D point cloud, resulting in a clean 4D point cloud in the world coordinate system.

[0022] The median height of the three highest points in a clean 4D point cloud is taken as the first height of the human body.

[0023] Furthermore, when a descent event occurs, the motion and static status of the human target are assessed, including:

[0024] If a valid 4D point cloud of a human target can be extracted, the human target is considered to be in motion; otherwise, the human target is considered to be stationary. A valid 4D point cloud is defined as a point cloud containing more than a set number of points.

[0025] Furthermore, the second height of the human target is calculated based on the human's dynamic and static states, including:

[0026] If the human target is in motion, the second height of the human target is obtained by calculating the median height of the three highest points in the 4D point cloud.

[0027] If the human target is currently stationary, the micro-motion positioning method is used to locate the human body by finding micro-motions in the perception space, thus obtaining the second height of the human target.

[0028] Furthermore, the micro-motion positioning method is used to locate the human body by detecting micro-motions in the sensory space, thus obtaining the second height of the human target, including:

[0029] The distance calculation method for micro-moving targets based on dynamic energy includes: obtaining the distance-time spectrum of the human target relative to the radar sensing device within a long observation window; performing a fast Fourier transform on the data of each distance domain within the time window; the energy of non-zero frequencies in the spectrum obtained by accumulating the fast Fourier transforms is the dynamic energy; and the distance domain with the strongest dynamic energy is the target distance.

[0030] The method for calculating the angle of a micro-movement target based on dynamic energy includes: obtaining the two-dimensional angle spectrum of the human target at each time point; using all the two-dimensional angle spectra within a time window, calculating the spectrum of each angle using fast Fourier transform, accumulating the energy of non-zero frequencies as the dynamic energy of that angle, and the angle with the strongest dynamic energy is the angle where the target is located.

[0031] Based on the distance and angle information of the human target, the micro-movement target of the human target is located, and the second height of the human target is obtained by coordinate transformation.

[0032] Furthermore, based on the second height of the human target, it is determined whether the human body is lying on the ground. If the human body remains lying on the ground for a set period of time, a life-threatening fall event is determined to have occurred, including:

[0033] The judgment is made based on the second height of the human target. If the second height of the human target is lower than the set value, it is considered that the person is lying on the ground. If the human target lies on the ground for more than a preset time without standing up, it is considered that a dangerous fall has occurred, and an alarm is issued in time. Otherwise, the detected decrease in the second height of the human target is considered not a dangerous fall.

[0034] Secondly, the present invention also provides a radar-based fall detection device, comprising:

[0035] The point cloud acquisition unit is configured to obtain the intermediate frequency signal of human target activity based on radar sensing equipment, and extract the 4D point cloud of human target through the intermediate frequency signal;

[0036] The first height calculation unit is configured to calculate the first height of the human target based on the 4D point cloud of the human target, and to determine whether the human body has experienced a height drop event based on the first height of the human target.

[0037] The motion and static state judgment unit judges the motion and static state of the human target when a descent event occurs.

[0038] The second height calculation unit is configured to calculate the second height of the human target based on the dynamic and static state of the human body;

[0039] The fall time determination unit is configured to determine whether a person is lying on the ground based on the second height of the human target. If the person lies on the ground for a set time, a life-threatening fall event is determined to have occurred.

[0040] Thirdly, the present invention also provides an electronic device comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for performing any of the methods.

[0041] Fourthly, the present invention also provides a computer-readable storage medium for storing one or more programs, said one or more programs including instructions that, when executed by a computing device, cause the computing device to perform any of the methods.

[0042] Because the present invention adopts the above technical solution, it has the following characteristics:

[0043] 1. This invention enables long-term stable monitoring of life-threatening falls without requiring any sensors on the human body. By proposing a two-stage fall detection method, in the first stage, a fall is distinguished from similar behaviors in daily life by detecting the decrease in height of the human body. In the second stage, the behavior detected in the first stage is determined to be a life-threatening fall by judging whether the person is lying on the ground. This can greatly reduce the false alarm rate of fall detection and can operate stably in various environments.

[0044] 2. This invention requires no training and therefore no data collection, and can be quickly deployed in various environments.

[0045] In summary, this invention can robustly detect life-threatening falls in real-world scenarios and can be widely applied to non-contact human fall detection. Attached Figure Description

[0046] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts. In the drawings:

[0047] Figure 1 This is a flowchart of a fall detection method based on millimeter-wave radar according to an embodiment of the present invention.

[0048] Figure 2 This is a structural diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0049] It should be understood that the terminology used herein is for the purpose of describing particular exemplary embodiments only and is not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “described” as used herein may also include the plural forms. The terms “comprising,” “including,” “containing,” and “having” are inclusive and therefore indicate the presence of the stated features, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, elements, components, and / or combinations thereof. The method steps, processes, and operations described herein are not construed as requiring them to be performed in a particular order described or illustrated unless the order of performance is explicitly indicated. It should also be understood that additional or alternative steps may be used.

[0050] For ease of description, spatial relative terms may be used in this text to describe the relationship of one element or feature relative to another element or feature as shown in the figure. These relative terms include, for example, "inside," "outside," "middle," "outer," "below," "above," etc. Such spatial relative terms are intended to include different orientations of the device in use or operation, other than those depicted in the figure. This invention defines a life-threatening fall as a fall in which the person is unable to stand up independently.

[0051] Because existing methods fail to detect life-threatening situations following a fall, this invention provides a radar-based fall detection method, apparatus, device, and medium, comprising: obtaining an intermediate frequency signal of human target activity based on a radar sensing device; extracting a 4D point cloud of the human target from the intermediate frequency signal; calculating a first height of the human target based on the 4D point cloud; determining whether a height drop event has occurred based on the first height; judging the dynamic and static state of the human target when a height drop event occurs; calculating a second height of the human target based on the dynamic and static state; determining whether the human target is lying on the ground based on the second height; and determining whether a life-threatening fall event has occurred if the human target remains lying on the ground for a set period of time. Therefore, this invention can accurately detect life-threatening falls. This invention proposes that only falls where the person cannot stand up after falling are truly dangerous falls, and accordingly designs a two-stage fall detection: in the first stage, detecting a height drop distinguishes a fall from similar falls in daily life; in the second stage, determining whether the behavior detected in the first stage is a life-threatening fall is determined by whether the person is lying on the ground.

[0052] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the invention and to fully convey the scope of the invention to those skilled in the art.

[0053] Before detection, this invention requires the placement of a radar sensing device in an activity space, such as a room. This radar sensing device can be fixedly placed on the ceiling, wall, or corner of the room, or mounted on the side, depending on actual needs; no limitation is made here. The radar sensing device should be a monostatic radar, possessing both a transmitting antenna and a receiving antenna. The radar sensing device needs to have multiple antennas, i.e., the capability of single-transmit multiple-receive (SIMO) or multiple-transmit multiple-receive (MIMO). The radar signal frequency band includes, but is not limited to, the millimeter-wave band. When the radar sensing device collects the sensing signal, the target must be within the effective coverage area of ​​the signal. It should be noted that the radar signal frequency band described in this embodiment is the millimeter-wave band, but this is not an limitation.

[0054] Example 1: As Figure 1 As shown, the fall detection method based on millimeter-wave radar provided in this embodiment includes:

[0055] S1. Data acquisition and preprocessing to obtain the intermediate frequency signal of human target activity.

[0056] Specifically, the above data acquisition and preprocessing obtains the mid-frequency signal of human target activity, including:

[0057] Millimeter-wave radar sensing equipment sends FMCW (Frequency Modulated Continuous Wave) signals into the activity space, and the FMCW signals are reflected back to the radar sensing equipment by the target human body.

[0058] The received signal and the transmitted signal from the radar sensing device are mixed and then passed through a low-pass filter to obtain the intermediate frequency signal.

[0059] The intermediate frequency (IF) signals from all receiving antennas are sampled to obtain the IF signals of human target activity. It should be noted that in this embodiment, a basic unit for calculating the distance, velocity, and angle of a human target is a frame, which contains several IF signals. For example, one frame contains 128 IF signals. This is just one example, and the method is not limited to this.

[0060] S2. Extracting 4D point cloud of human target

[0061] Specifically, the above-mentioned extraction of the 4D point cloud of the human target includes:

[0062] S21. Obtain the distance between the human target and the radar sensing device.

[0063] In this embodiment, since the distance between the human target and the radar sensing device is proportional to the frequency of the intermediate frequency signal, the frequency of each intermediate frequency signal is calculated using Fast Fourier Transform (FFT) to obtain the distance between the human target and the radar sensing device.

[0064] S22. Obtain the distance-velocity spectrum of the human target relative to the radar sensing device.

[0065] In this embodiment, the radial velocity of the human target relative to the radar sensing device is proportional to the frequency of the 1D FFT signal (the intermediate frequency signal after using Fast Fourier Transform) along the slow time dimension within a frame. Therefore, a second FFT is performed on the signal after the 1D FFT along the slow time dimension to obtain the velocity of the human target, and then the range-velocity spectrum of the human target relative to the radar sensing device can be obtained. The magnitude of each point value on the range-velocity spectrum represents the probability of the presence of a human reflective target at the corresponding velocity and distance.

[0066] S23. Target points for human body reflection are selected based on the distance-velocity spectrum.

[0067] In this embodiment, an adaptive threshold algorithm is used to filter target points reflected from the human body in the range-velocity spectrum. Since this invention focuses on dynamic human bodies, points with zero velocity in the range-velocity spectrum will be judged as non-target points.

[0068] Furthermore, the adaptive threshold algorithm in this embodiment can employ a constant false alarm rate (CFAR) algorithm.

[0069] S24. Perform phase compensation on human targets.

[0070] In this embodiment, the angle is estimated using the phase difference caused by the human target on different antennas. Due to the use of Time Division Multiplexing Multiple Transmit Multiple Receive (TDM-MIMO) technology, the movement of the human target will cause additional phase differences on the intermediate frequency signals corresponding to different transmit antennas. This phase difference can be expressed as:

[0071]

[0072] To ensure accurate angle calculations, this phase error needs to be eliminated. Because T itv This indicates the transmission gap between different transmitting antennas, which is determined during radar parameter configuration. And f... doppler The calculation was also completed during the second velocity FFT, so phase compensation can be performed on each human reflection target point according to the above formula. This error can be eliminated based on the target velocity calculated in the velocity estimation.

[0073] S25. Calculate the angle of the human target.

[0074] In this embodiment, two-dimensional digital beamforming (DBF) is used to calculate the target's angle. For each human reflection target point, a two-dimensional angle spectrum can be obtained, where one dimension is the elevation angle and the other dimension is the azimuth angle. The elevation angle and azimuth angle corresponding to the peak values ​​in this spectrum represent the angle at which the target point is located.

[0075] S26. Obtain the 4D point cloud of the human target.

[0076] In this embodiment, the 4D point cloud of the moving human target is obtained by combining the distance, speed and angle information of each human target point relative to the radar sensing device. In this embodiment, the 4D point cloud refers to the 3D spatial information acceleration.

[0077] S3. Calculate the height of the human body based on the 4D point data of the human target.

[0078] Specifically, the height of the human body calculated using 4D point cloud computing based on the human target includes:

[0079] S31, Coordinate system transformation.

[0080] Since the calculated distances and angles of points in the point cloud are relative to the radar sensing device, calculating the height of the human body requires knowing its distance relative to the ground. Therefore, a coordinate system transformation of the point cloud is needed, from the radar coordinate system to the world coordinate system. Based on the known installation location and tilt angle information of the radar sensing device, a rotation matrix R and a translation vector T are constructed to transform the radar coordinate system to the world coordinate system. Then, for the i-th target point, the transformed coordinates in the world coordinate system (x, y, y) are... i ,y i ,z i ) and radar coordinate system (x i ′,y i ′,z i There is a relationship:

[0081]

[0082] S32, Point Cloud Denoising.

[0083] In this embodiment, the point cloud contains a lot of noise due to the multipath effect. To ensure high computational accuracy, noise removal is necessary. Compared to the point cloud of a human body, this noisy point cloud is very sparse. Therefore, this embodiment uses a clustering algorithm (DBSCAN) to remove noisy point clouds, but this is not a limitation.

[0084] S33. Calculate the height of the human target.

[0085] In this embodiment, after coordinate system transformation and denoising, a clean point cloud in the world coordinate system is obtained. At this point, the height of the human body can be estimated by taking the median height of the three highest points in the point cloud. Because of the change in the reflection point (head or shoulder), this height estimate is only a rough estimate and is not strictly equal to the height of the human body, with an error of about 10cm. This error is negligible compared to the height change when the human body falls.

[0086] S4. Determine whether a height drop event has occurred based on the body's height information.

[0087] In this embodiment, by accumulating the height changes of the point cloud over a set time period, such as the past 5 seconds, if the maximum height of the human body decreases by more than a preset distance, such as 80 centimeters, within 5 seconds, it is considered that the height of the human body has decreased significantly, and a potentially fatal event may have occurred, requiring a second-stage judgment. If the decrease does not exceed 80 centimeters, then Anhui S2 is triggered to continue extracting the target point cloud and continuously monitor the height of the human body.

[0088] S5. When a descent event occurs, determine the motion or static state of the human target.

[0089] In this embodiment, after a person experiences a drop in height, it's unclear whether they've simply fallen onto a bed, picked up something from the ground, or actually fallen. Therefore, further perception and judgment of the person's state are needed. However, the person may be stationary, such as lying on a bed or on the ground, or they may be moving, such as picking something up and standing up, or struggling to get up after falling. 4D point clouds can only be extracted when the target is in obvious motion; when stationary, 4D point clouds cannot be used for human perception.

[0090] In this embodiment, the distinction between static and dynamic states is as follows: when a human body is not moving, it is considered "static"; otherwise, it is considered "dynamic." Because different dynamic and static states require different algorithms for perception, a dynamic / static determination should be performed first. The method for determining dynamic / static states in this embodiment is: if a valid 4D point cloud of a human target can be extracted, the human target is considered to be moving; otherwise, the human target is considered static. In this embodiment, a valid 4D point cloud is defined as one with more than 10 points in the point cloud.

[0091] S6. Calculate the height of the human body at this time by selecting 4D point cloud or micro-motion human body positioning technology based on the dynamic state of the human body after the descent.

[0092] Specifically, this embodiment proposes a micro-motion human body positioning technology, which uses micro-motions such as breathing and heartbeat to locate a stationary human body. The calculation of the person's height based on the dynamic state of the body after descent, choosing between 4D point cloud or micro-motion human body positioning technology, includes:

[0093] S61. If the human target is in motion at this time, then use the methods of S2 and S3 to calculate the height of the human body.

[0094] S62. If the human target is currently stationary, the micro-motion localization method is used to locate the human body by searching for micro-motions in the perception space. Localization of micro-motions can be divided into two steps: distance calculation and angle calculation. Because micro-motions have relatively low speeds compared to larger limb movements, traditional frame-based localization methods are no longer applicable. To accurately locate low-frequency micro-motions in complex indoor static environments, a longer time window is needed to obtain better frequency resolution, distinguishing low-frequency micro-motions from the zero-frequency static environment. Therefore, data from different frames is combined to obtain data for a long observation time window, specifically including:

[0095] S621. A method for calculating the distance of a micro-moving target based on dynamic energy, including:

[0096] First, obtain the distance-time spectrum of the human target relative to the radar sensing device within a long observation window, for example, greater than 6 seconds;

[0097] Then, an FFT is performed on the data in each range domain within the time window, and the energy of the non-zero frequencies in the spectrum obtained by the FFT is accumulated, which is the dynamic energy. Because the longer time window ensures that the energy of micro-motion can be distinguished from the energy of static objects, the range domain with the strongest dynamic energy is the range domain where the target is located, that is, the distance of the target.

[0098] S622. A method for calculating the angle of a micro-moving target based on dynamic energy, including:

[0099] First, obtain the two-dimensional angle spectrum of the human target at each time point;

[0100] Then, using all the two-dimensional angle spectra within a time window, the spectrum of each angle is calculated using FFT, and the energy of non-zero frequencies is accumulated as the dynamic energy of that angle. The angle with the strongest dynamic energy is the angle where the target is located.

[0101] S623. Based on the distance and angle information of the human target, accurately locate the micro-movement target of the human target.

[0102] S624. Human height is obtained after coordinate transformation.

[0103] In this embodiment, similar to the point cloud method, in order to calculate the height of the target, it is also necessary to transform the calculated target coordinates in the radar coordinate system to the world coordinate system.

[0104] S7. Determine whether the person is lying on the ground based on the latest calculated height of the human target after the descent; if the person remains lying on the ground for a period of time, it is determined that a life-threatening fall has occurred, and an alarm should be promptly issued to relatives and community medical staff.

[0105] In this embodiment, based on the height of the human target obtained in S6, if the height of the human body is lower than a set value, such as 50cm, it is considered that the person is lying on the ground. If the person lies on the ground for more than a preset time, such as 10 seconds, without standing up, it is considered that a dangerous fall has occurred. At this time, an alarm should be issued to the relatives of the person who fell and medical personnel in the community. The alarm can be issued to relevant personnel through mobile APP, computer software, or telephone. Otherwise, the height drop event detected in S4 is considered not a fall, no alarm is required, and the process returns to S2 to continuously extract human body point cloud data and monitor changes in human body height.

[0106] Example 2: Following the fall detection method based on millimeter-wave radar provided in Example 1, this example provides a fall detection device based on millimeter-wave radar. The device provided in this example can implement the fall detection method based on millimeter-wave radar in Example 1. This device can be implemented through software, hardware, or a combination of both. For ease of description, this example is described by dividing the functionality into various units. Of course, in implementation, the functions of each unit can be implemented in one or more software and / or hardware components. For example, the device may include integrated or separate functional modules or units to execute the corresponding steps in the methods of Example 1. Since the device in this example is basically similar to the method example, the description process of this example is relatively simple. Relevant details can be found in the description of the parts in Example 1. The example of the fall detection device based on millimeter-wave radar provided by this invention is merely illustrative.

[0107] The radar-based fall detection device provided in this embodiment includes:

[0108] The point cloud acquisition unit is configured to obtain the intermediate frequency signal of human target activity based on radar sensing equipment, and extract the 4D point cloud of human target through the intermediate frequency signal;

[0109] The first height calculation unit is configured to calculate the first height of the human target based on the 4D point cloud of the human target, and to determine whether the human body has experienced a height drop event based on the first height of the human target.

[0110] The motion and static state judgment unit judges the motion and static state of the human target when a descent event occurs.

[0111] The second height calculation unit is configured to calculate the second height of the human target based on the dynamic and static state of the human body;

[0112] The fall time determination unit is configured to determine whether a person is lying on the ground based on the second height of the human target. If the person lies on the ground for a set time, a life-threatening fall event is determined to have occurred.

[0113] Example 3: This example provides an electronic device corresponding to the fall detection method based on millimeter-wave radar provided in Example 1. The electronic device can be an electronic device for the client, such as a mobile phone, laptop, tablet computer, desktop computer, etc., to execute the method of Example 1.

[0114] like Figure 2 As shown, the electronic device includes a processor, memory, communication interface, and bus. The processor, memory, and communication interface are connected via the bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Component (EISA) bus, etc. The memory stores a computer program that can run on the processor. When the processor runs the computer program, it executes the method of Embodiment 1. The implementation principle and technical effects are similar to those of Embodiment 1, and will not be repeated here. Those skilled in the art will understand that... Figure 2 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computing device on which the present application is applied. The specific computing device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0115] In a preferred embodiment, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and 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 a 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.) to execute all or part of the steps of the methods described 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), and optical discs.

[0116] In a preferred embodiment, the processor can be any type of general-purpose processor such as a central processing unit (CPU) or a digital signal processor (DSP), and is not limited thereto.

[0117] Example 4: This example provides a computer program product. The computer program product may include a computer program stored on a computer-readable storage medium. The computer program includes program instructions. When the program instructions are executed by the computer, the computer can execute the method provided in Example 1 above. Its implementation principle and technical effects are similar to those in Example 1, and will not be repeated here.

[0118] In a preferred embodiment, the computer-readable storage medium may be a tangible device for holding and storing instructions used by an instruction execution device, such as, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any combination thereof. The computer-readable storage medium stores computer program instructions that cause a computer to perform the method provided in Embodiment 1 above.

[0119] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In the description of this specification, the terms "a preferred embodiment," "furthermore," "specifically," "in this embodiment," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the embodiments in this specification. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described can be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0120] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (devices), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0121] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0122] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0123] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A radar-based fall detection method, characterized in that, include: Based on the intermediate frequency (IF) signal of human target activity obtained from radar sensing equipment, the 4D point cloud of the human target is extracted from the IF signal, including: The intermediate frequency signal is calculated using a fast Fourier transform to obtain the distance between the human target and the radar sensing device. The intermediate frequency signal after the fast Fourier transform is subjected to a second fast Fourier transform along the slow time dimension to obtain the velocity of the human target and the range-velocity spectrum of the human target relative to the radar sensing device. Target points for human body reflection are selected based on distance-velocity spectrum. Phase compensation is performed on the human target to calculate the angle of the human target; By combining the distance, velocity, and angle information of each human body's reflected target point relative to the radar sensing device, a 4D point cloud of the human target is obtained. The 4D point cloud includes 3D spatial information and velocity. The first height of the human body is calculated based on the 4D point cloud of the human body target, and the human body is judged to have experienced a height drop event based on the first height of the human body target. When a descent event occurs, the dynamic and static states of the human target are assessed, including: If a valid 4D point cloud of a human target can be extracted, the human target is considered to be moving; otherwise, the human target is considered to be stationary. The definition of a valid 4D point cloud is that the number of points in the point cloud exceeds a set number. The second height of a human target is calculated based on the dynamic and static states of the human body, including: If the human target is in motion, the second height of the human target is obtained by calculating the median height of the three highest points in the 4D point cloud. If the human target is currently stationary, a micro-motion localization method is used to locate the human body by detecting micro-motions in the perception space, thus obtaining the second height of the human target. This includes: calculating the distance based on a dynamic energy-based micro-motion target distance calculation method, which involves: obtaining the distance-time spectrum of the human target relative to the radar sensing device within a long observation window; performing a Fast Fourier Transform on the data in each range domain within the observation window, accumulating the non-zero frequency energy in the spectrum obtained by the Fast Fourier Transform as the dynamic energy, and determining the target distance in the range domain with the strongest dynamic energy; calculating the angle based on a dynamic energy-based micro-motion target angle calculation method, which involves: obtaining the two-dimensional angle spectrum of the human target at each time point; using all two-dimensional angle spectra within an observation window, calculating the spectrum of each angle using a Fast Fourier Transform, accumulating the non-zero frequency energy as the dynamic energy of that angle, and determining the angle with the strongest dynamic energy as the target angle; and, based on the distance and angle information of the human target, completing the localization of the micro-motion human target and performing coordinate transformation to obtain the second height of the human target. The second height of the human body is used to determine whether the human body is lying on the ground. If the human body lies on the ground for a set period of time, it is determined that a life-threatening fall has occurred.

2. The radar-based fall detection method according to claim 1, characterized in that, Based on the 4D point cloud computing of the human target, the first height of the human target includes: Perform coordinate system transformation on the 4D point cloud, transforming it from the radar coordinate system to the world coordinate system; Clustering methods are used to denoise the 4D point cloud, resulting in a clean 4D point cloud in the world coordinate system. The median height of the three highest points in a clean 4D point cloud is taken as the first height of the human body.

3. The radar-based fall detection method according to claim 1, characterized in that, The system determines whether a person is lying on the ground based on the second height of the target. If the person remains lying on the ground for a set period of time, a life-threatening fall event is considered to have occurred, including: The judgment is made based on the second height of the human target. If the second height of the human target is lower than the set value, it is considered that the person is lying on the ground. If the human target lies on the ground for more than a preset time without standing up, it is considered that a dangerous fall has occurred, and an alarm is issued in time. Otherwise, the detected decrease in the second height of the human target is considered not a dangerous fall.

4. A detection device for implementing the radar-based fall detection method according to any one of claims 1 to 3, characterized in that, include: The point cloud acquisition unit is configured to obtain the intermediate frequency signal of human target activity based on radar sensing equipment, and extract the 4D point cloud of human target through the intermediate frequency signal; The first height calculation unit is configured to calculate the first height of the human target based on the 4D point cloud of the human target, and to determine whether the human body has experienced a height drop event based on the first height of the human target. The motion and static state judgment unit judges the motion and static state of the human target when a descent event occurs. The second height calculation unit is configured to calculate the second height of the human target based on the dynamic and static state of the human body; The fall time determination unit is configured to determine whether a person is lying on the ground based on the second height of the human target. If the person lies on the ground for a set time, a life-threatening fall event is determined to have occurred.

5. An electronic device, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for performing any of the methods described in claims 1 to 3.

6. A computer-readable storage medium for storing one or more programs, characterized in that, The one or more programs include instructions that, when executed by a computing device, cause the computing device to perform any of the methods described in claims 1 to 3.

Images