Human fall detection method, device and readable storage medium in bathroom environment

Abstract

The application provides a human body fall detection method, device and readable storage medium in a bathroom environment, the method comprising: extracting a Doppler spectrum width feature from a range-Doppler spectrum corresponding to a bathroom radar echo signal; wherein the Doppler spectrum width feature is used to represent the width of the detected target speed distribution; determining whether the shower device in the current bathroom is in a working state based on the Doppler spectrum width feature; if the current shower device is in a working state, generating a 4D point cloud image representing human body action information based on the range-Doppler spectrum; and determining the detection result of the human body fall event according to the 4D point cloud image. Through the implementation of the application scheme, the flow state of the shower in the bathroom is judged by extracting the Doppler spectrum width feature corresponding to the radar echo signal, the human body target and the water flow target are distinguished, and the human body action is recognized and processed based on the 4D point cloud imaging technology, so that high-precision detection of human body fall can be realized.

Description

Technical Field

[0001] This application relates to the field of radar data processing technology, and in particular to a method, device and readable storage medium for detecting human falls in a bathroom environment. Background Technology

[0002] Falls are a common accident in the daily lives of the elderly due to the decline in physical function. If falls can be detected in time and treated effectively, the health risks and pressure on medical resources can be reduced.

[0003] In the bathroom shower environment, the risk of falls for the elderly is significantly increased due to slippery floors, steamy conditions, limited space, and few supports. When a fall occurs, the privacy and soundproofing of the bathroom often make it difficult for outsiders to detect the fall and provide timely assistance. Therefore, developing a fall detection system capable of real-time monitoring in this high-risk environment has significant practical and social value.

[0004] In related technologies, common fall detection devices can be mainly divided into two types: contact-based (e.g., accelerometers, gyroscopes, etc.) and non-contact-based (e.g., visual sensors, audio sensors, infrared sensors, Wi-Fi devices, etc.). While contact-based sensors offer high sensitivity and real-time performance, they suffer from drawbacks such as needing to be worn constantly and poor comfort during extended periods. Most non-contact sensors, while avoiding direct user contact, present issues such as privacy breaches and susceptibility to environmental influences. Non-contact radar sensors offer advantages such as all-day operation and privacy protection, and are unaffected by external factors like light, temperature, and sound. However, radar signals acquired in a bathroom environment are easily affected by environmental factors such as shower water flow, leading to lower accuracy in fall detection in bathroom settings. Summary of the Invention

[0005] This application provides a method, device, and readable storage medium for detecting human falls in a bathroom environment, which can at least solve the problem of low accuracy in detecting human falls based on radar signals in a bathroom environment.

[0006] The first aspect of this application provides a method for detecting human falls in a bathroom environment, including: Doppler spectral width features are extracted from the range-Doppler spectrum corresponding to the bathroom radar echo signal; the Doppler spectral width features are used to characterize the breadth of the velocity distribution of the detected target. Based on the Doppler spectral width characteristics, determine whether the shower equipment in the bathroom is currently in operation; If the shower equipment is currently in operation, a 4D point cloud image representing human motion information is generated based on the distance-Doppler spectrum. The detection results of human fall events are determined based on 4D point cloud images.

[0007] The second aspect of this application provides an electronic device, including a memory and a processor, wherein the processor is used to execute a computer program stored in the memory, and when the processor executes the computer program, it implements the steps of the human fall detection method in a bathroom environment provided in the first aspect of this application.

[0008] The third aspect of this application provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, it implements the steps of the human fall detection method in a bathroom environment provided in the first aspect of this application.

[0009] As can be seen from the above, according to the method, equipment, and readable storage medium for detecting human falls in a bathroom environment provided in this application, Doppler spectral width features are extracted from the range-Doppler spectrum corresponding to the bathroom radar echo signal. The Doppler spectral width features are used to characterize the breadth of the velocity distribution of the detected target. Based on the Doppler spectral width features, it is determined whether the shower equipment in the bathroom is currently in operation. If the shower equipment is currently in operation, a 4D point cloud image representing human motion information is generated based on the range-Doppler spectrum. Based on the 4D point cloud image, the detection result of the human fall event is determined. Through the implementation of this application, the on / off state of the shower head in the bathroom environment is identified on the range-Doppler spectrum of the radar echo signal. When it is determined that the shower head is in operation, i.e., when there is continuous water flow from the shower head in the bathroom, 4D human point cloud imaging technology is used to identify the specific posture adjustment of the human body. This data processing can effectively suppress the interference of the shower head water flow on human motion recognition, greatly improving the accuracy of human fall detection. Attached Figure Description

[0010] Figure 1 A schematic diagram of the basic process for a method for detecting human falls in a bathroom environment, provided in the first embodiment of this application; Figure 2 Distance to an unoccupied bathroom with the shower turned off - Doppler spectrum; Figure 3 Distance to an unoccupied bathroom with the shower in operation - Doppler spectrum; Figure 4 Distance to a bathroom where someone is present and the shower is off - Doppler spectrum; Figure 5 Distance to a bathroom where someone is present and the shower is in operation - Doppler spectrum; Figure 6 The distance-Doppler spectrum to be processed in the first embodiment of this application; Figure 7 This is the spectral diagram corresponding to the statistical matrix in the first embodiment of this application; Figure 8 This is a marked distance-Doppler spectrum in the first embodiment of this application; Figure 9 A detailed flowchart illustrating a method for detecting human falls in a bathroom environment, provided in the second embodiment of this application; Figure 10 A schematic diagram of the structure of an electronic device provided in the third embodiment of this application. Detailed Implementation

[0011] To make the inventive objectives, features, and advantages of this application more apparent and understandable, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0012] In the description of the embodiments of this application, it should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings. They are only for the convenience of describing the embodiments of this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting the present invention.

[0013] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0014] In the embodiments of this application, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. For those skilled in the art, the specific meaning of the above terms in the embodiments of this application can be understood according to the specific circumstances.

[0015] To address the issue of low accuracy in radar-based fall detection in related technologies, the first embodiment of this application provides a fall detection method in a bathroom environment, such as... Figure 1 This is a basic flowchart illustrating the human fall detection method in a bathroom environment provided in this embodiment. The human fall detection method in a bathroom environment includes the following steps: Step 101: Extract the Doppler spectral width feature from the range-Doppler spectrum corresponding to the bathroom radar echo signal.

[0016] Specifically, the Doppler spectral width feature is used to characterize the breadth of the velocity distribution of the detected target. In this embodiment, in a bathroom shower environment, millimeter-wave radar can be used to detect human falls and distinguish between shower states (on / off). When the shower is on, the continuous movement of water causes the Doppler frequency distribution to broaden, resulting in an increased Doppler spectral width feature value. When the shower is off but the user makes a rapid movement, a brief period of high-speed movement may occur, which may also produce a large Doppler spectral width feature. Therefore, when detecting human falls, the interference of the shower water flow on the detection must be eliminated first.

[0017] Step 102: Based on the Doppler spectral width characteristics, determine whether the shower equipment in the bathroom is currently in operation.

[0018] Specifically, if the shower equipment in the bathroom is on, 4D point cloud imaging and other technologies can be used for further fall detection. If the shower equipment is off, conventional fall detection methods (such as posture angle change analysis) can be used directly to detect falls, ensuring accuracy while reducing processing complexity. Of course, in other implementations, 4D point cloud imaging and other technologies can still be used for fall detection even when the shower equipment is off; this is not a limitation.

[0019] Step 103: If the shower equipment is currently in operation, generate a 4D point cloud image representing human motion information based on the distance-Doppler spectrum.

[0020] Specifically, by using millimeter-wave radar information to create 4D point cloud images of human movements, information such as the target's distance, azimuth, pitch, and velocity can be obtained. By constructing dynamic point cloud images, the movement of the human body can be continuously tracked from a spatiotemporal perspective, accurately capturing the movement characteristics of the human body and providing high-precision data support for subsequent analysis of human behavior patterns.

[0021] Step 104: Determine the detection results of human fall events based on the 4D point cloud image.

[0022] Specifically, in the process of analyzing human behavior patterns based on 4D point cloud images, it can be first determined whether the human body is in a low posture state; if the human body is in a low posture state, point cloud image features can be further extracted to perform human fall recognition.

[0023] In one embodiment of this invention, the aforementioned radar echo signal is a discrete echo signal associated with a slow-time stack, a fast-time stack, and an antenna stack. Accordingly, before the step of extracting the Doppler spectral width feature from the range-Doppler spectrum corresponding to the bathroom radar echo signal, the method further includes: performing an FFT transform on the bathroom radar echo signal in the fast-time dimension to generate an initial range-dimensional spectrum; performing static clutter suppression processing on the range-dimensional spectrum to obtain the target range-dimensional spectrum; performing an FFT transform on the target range-dimensional spectrum in the slow-time stack to obtain range-Doppler spectrum data; and performing incoherent accumulation processing on the range-Doppler spectrum data to output the range-Doppler spectrum.

[0024] Specifically, radar continuously transmits electromagnetic wave signals into space. These signals are scattered or reflected by objects and received by the radar receiver. After passing through a signal amplifier, mixer, and ADC sampling, a discrete echo signal containing information in the range, velocity, and angle dimensions is obtained. The echo signal received by the radar module can be represented as... ,in, Let be the slow time dimension, indicating the th A linear frequency modulated continuous wave signal; For fast time dimension, it means the first... One sampling point; Let be the antenna dimension, and let represent the th . The received signals from each channel are analyzed. Fast Fourier Transforms are performed along the fast and slow times of the echo signals, followed by static clutter suppression and incoherent accumulation, to obtain the range-Doppler spectrum containing target information. .

[0025] It should be noted that the distance-Doppler spectrum in the bathroom exhibits different spectral width characteristics under different conditions, such as... Figure 2-5 As shown, through comparative observation, it can be seen that: (1) the numerical difference between the shower water flow interference and the human target in the range-Doppler spectrum is small, which is due to the similarity of their radar cross sections (RCS); (2) the shower water flow interference has a wider distribution range in the spectrum and is distributed in the positive Doppler frequency range, while the human target is more concentrated near the zero Doppler frequency in the spectrum. Therefore, the shower status can be detected based on the range-Doppler spectrum to determine whether the shower equipment in the bathroom is in working condition.

[0026] In one embodiment of this invention, the Doppler spectral width feature can be extracted from the distance-Doppler spectrum; the Doppler spectral width feature can be expressed as:

[0027]

[0028]

[0029] in, Indicates the characteristics of Doppler spectral width. Indicates the distance-Doppler spectrum. Indicates the index of the Doppler cell. Indicates the index of the distance cell. Represents the average coefficient. Indicates the number of Doppler elements. Indicates the number of distance units.

[0030] Furthermore, in one embodiment of this example, the step of determining whether the shower equipment in the current bathroom is in working state based on the Doppler spectral width feature includes: performing sliding window detection on the distance-Doppler spectrum based on the Doppler spectral width feature; determining that the shower equipment in the current bathroom is in working state when the number of target data frames that meet the first preset condition in the time window is greater than a preset frame number threshold; wherein, the target data is distance-Doppler spectrum data, and the first preset condition is: the Doppler spectral width feature corresponding to the target data is greater than a preset feature threshold.

[0031] Specifically, when When this occurs, it indicates that there is a large Doppler bandwidth at this time, where, A preset feature threshold is used. For the Doppler spectrum width feature, the feature value can sometimes be relatively large when the shower head is off. This may be due to continuous Doppler frequency components caused by a person quickly squatting or swinging their arms. Therefore, this embodiment detects the shower status using a sliding window. The frame data contains satisfy When this is the case, it can be determined that the shower equipment is in working condition. Indicates the window length. This indicates the trigger threshold.

[0032] In one embodiment of this example, the step of generating a 4D point cloud image representing human motion information based on the range-Doppler spectrum includes: detecting the interference distribution information corresponding to the shower head water flow in the range-Doppler spectrum, removing the interference distribution information from the range-Doppler spectrum to obtain the target range-Doppler signal; and processing the target range-Doppler signal based on a digital beamforming algorithm to obtain a 4D point cloud image representing human motion information.

[0033] Specifically, due to the interference of shower water flow, before performing point cloud imaging, it is necessary to first determine the information corresponding to the water flow target and the human body target, and separate the water flow target and the human body target to suppress the interference of shower water flow on recognition and improve the accuracy of human fall recognition.

[0034] Further, in one embodiment of this example, the step of detecting the interference distribution information corresponding to the shower head water flow in the range-Doppler spectrum includes: creating a statistical matrix corresponding to the range-Doppler spectrum; wherein the statistical matrix is ​​the same size as the range-Doppler spectrum; based on the statistical matrix, detecting interference targets for each frame of data in the range-Doppler spectrum, determining all target locations where interference targets appear in the range-Doppler spectrum and the frequency of interference targets appearing at each target location; marking the range-Doppler spectrum by combining the target locations and frequencies to obtain a marked range-Doppler spectrum; and determining the interference distribution information corresponding to the shower head water flow based on the marking information in the marked range-Doppler spectrum.

[0035] For example, when performing 4D point cloud imaging, the following steps can be followed: (1) Create a statistical matrix with the same size as the distance-Doppler spectrum and initial values ​​of zero for its elements. And set the statistical duration to Among them, the distance-Doppler spectrum is as follows: Figure 6 As shown, the statistical matrix like Figure 7 As shown; (2) Target detection is performed using a two-dimensional OS-CFAR detector on each frame's range-Doppler spectrum. The detected interfering target can be represented as: ,in Doppler index representing the interfering target. Indicates distance index, The number of interfering targets detected. And based on each... The detection results within the specified time period were used to re-count the frequency of interfering targets at each location in the range-Doppler spectrum, and the data was recorded. ; (3) In A breadth-first search algorithm is then used to find and label different connected regions, resulting in a labeled range-Doppler spectrum. Since the showerhead water flow target in a shower state has a wide distribution on the range-Doppler spectrum, the connected region with the largest area represents the interference distribution of the showerhead water flow. ,like Figure 8 As shown; (4) Transfer the interference distribution information corresponding to the shower water flow from After removing the middle part, the target distance-Doppler signal containing only human target information is obtained. ; (5) Based on the target range-Doppler signal, the Digital Beamforming (DBF) algorithm is used to estimate the angle of the detected target unit; after coordinate transformation, a 4D point cloud image containing rich human physical information is obtained. ,in, Represents the three-dimensional Cartesian coordinate system Axis coordinates Represents the three-dimensional Cartesian coordinate system Axis coordinates Represents the three-dimensional Cartesian coordinate system Axis coordinates Indicates speed, It represents energy.

[0036] In one embodiment of this example, the step of determining the detection result of a human fall event based on a 4D point cloud image includes: determining whether the user has made a specific posture adjustment within a preset time based on the 4D point cloud image; wherein, the specific posture adjustment process is the process of changing from a high posture to a low posture; if it is determined that the user has made a specific posture adjustment within the preset time, then determining the target point cloud image corresponding to the user's specific posture adjustment process; extracting the width and height information of the point cloud distribution in each frame of the target point cloud image; and combining all the width and height information to determine the detection result of the human fall event.

[0037] Specifically, the process of specific posture adjustment can refer to the process by which a human target changes from a high posture such as standing or walking to a low posture such as falling or squatting. In the process of a human falling, the most obvious change is in height. In common human movements, such as sitting down, squatting down, and bending over to pick something up, there are similar characteristics. Therefore, before recognizing the falling action, we can first detect and judge the specific posture adjustment (that is, the change from a high specific posture to a low posture).

[0038] Further, in one embodiment of this example, the step of determining whether a user has made a specific posture adjustment within a preset time based on the 4D point cloud image includes: determining the associated point cloud quantity sequence and centroid height sequence of the 4D point cloud image; performing sliding window detection on the 4D point cloud image based on the associated point cloud quantity sequence and centroid height sequence; determining that the user has made a specific posture adjustment within the preset time when the associated point cloud quantity and centroid height corresponding to the target area within the time window meet the second preset condition; wherein, the target area includes a first area and a second area, and the first area is later than the second area in time sequence; the second preset condition includes: the proportion of data frames with a centroid height lower than a preset height threshold in the first area of ​​the time window is greater than a first proportion threshold, the proportion of data frames with an associated point cloud quantity higher than a preset point quantity threshold in the first area is greater than a second proportion threshold, and the proportion of data frames with a centroid height higher than a preset height threshold in the second area of ​​the time window is greater than a third proportion threshold.

[0039] Specifically, the first region can refer to the latter part of the time sequence window (e.g., the second half of the time sequence window) or the former part of the time sequence window (e.g., the first half of the time sequence window). In the specific implementation process, it can be first determined whether the number of data frames in the latter part of the time sequence window whose centroid height is lower than the height threshold (i.e., the aforementioned preset height threshold) exceeds the first proportional threshold. If so, the human target is considered to be in a low posture. Next, it can be determined whether the number of data frames in the latter part of the time sequence window whose associated point cloud number is higher than the point number threshold (i.e., the aforementioned preset point number threshold) exceeds the second proportional threshold. If so, the human target is considered to be present; otherwise, the human target is considered to have left the detection area. Finally, it can be determined whether the number of data frames in the former part of the time sequence window whose centroid height is higher than the height threshold exceeds the third proportional threshold. If so, the human target is considered to be in a high posture. When all three determinations pass, it is considered that the human target has performed a high posture to low posture transition.

[0040] Furthermore, in one embodiment of this example, the step of determining the detection result of a human fall event by combining all width and height information includes: determining the frame-level features of the target point cloud image by combining the width and height information; wherein the frame-level features include the width distribution of the point cloud on the x-axis, the width distribution on the y-axis, the maximum height on the z-axis, and the average height of the point cloud; extracting extended features and height features from the frame-level features; wherein the extended features are used to characterize the distribution changes of the human projection in the horizontal direction, and the height features are used to characterize the distribution changes of the human projection in the vertical direction; and inputting the extended features and height features into a preset classifier to output the detection result of the human fall event.

[0041] Specifically, for example, the distribution width and height information of each frame of point cloud can be represented as:

[0042]

[0043]

[0044]

[0045] in, Indicates the first Frame cloud edge The width of the axis distribution, Indicates the first Frame cloud edge The width of the axis distribution, Indicates the first Maximum height of the frame point cloud Indicates the first The average height of the frame point cloud, For frame index, , For the duration of data used for fall detection, This represents the number of points in a single frame of the point cloud. .

[0046] When a person falls and lies on the ground, their height remains at a relatively low level for an extended period, and the horizontal distribution of the torso and limbs increases significantly. In contrast, while actions such as squatting, bending over, and picking up objects also cause a decrease in height, the height after the fall is still higher than the height of the fallen state, and the torso and limbs do not show a significant horizontal expansion.

[0047] Therefore, it can be seen from , , Extract extended features , , , :

[0048]

[0049]

[0050]

[0051] And, from and Extracting height features and :

[0052]

[0053] in, express The normalized ratio of axis width to maximum height. express The normalized ratio of axis width to maximum height. This indicates the distribution range of the horizontal plane projection (Euclidean distance). This represents the normalized ratio of the horizontal distribution range to the maximum height. This represents the median of the maximum height. This represents the median of the average height.

[0054] Based on the extracted extended features and height features, an SVM classifier is used to identify human fall states. The SVM classifier is suitable for binary classification tasks with small sample sizes, can effectively handle nonlinear feature relationships, and improve the accuracy of human fall state judgment.

[0055] Based on the technical solution of the above-described embodiments of this application, Doppler spectral width features are extracted from the range-Doppler spectrum corresponding to the bathroom radar echo signal. The Doppler spectral width features are used to characterize the breadth of the velocity distribution of the detected target. Based on the Doppler spectral width features, it is determined whether the shower equipment in the bathroom is currently in operation. If the shower equipment is currently in operation, a 4D point cloud image representing human motion information is generated based on the range-Doppler spectrum. Based on the 4D point cloud image, the detection result of a human fall event is determined. Through the implementation of this application's solution, the on / off state of the showerhead in the bathroom environment is identified on the range-Doppler spectrum of the radar echo signal. When it is determined that the showerhead is in operation, i.e., when there is continuous water flow from the showerhead in the bathroom, 4D human point cloud imaging technology is used to identify the specific posture adjustment of the human body. This effectively suppresses the interference of the showerhead's water flow on human motion recognition, greatly improving the accuracy of human fall detection.

[0056] Figure 9 The method described in the second embodiment of this application is a refined method for detecting human falls in a bathroom environment, which includes: Step 901: Perform range-Doppler imaging on the radar echo signal of the bathroom environment to obtain the range-Doppler spectrum.

[0057] Step 902: Extract the Doppler spectral width feature from the distance-Doppler spectrum.

[0058] Step 903: Based on the Doppler spectral width characteristics, perform sliding window detection on the range-Doppler spectrum.

[0059] Step 904: When the number of target data frames that meet the first preset condition in the time sequence window is greater than the preset frame number threshold, it is determined that the shower equipment in the current bathroom is in working state.

[0060] Specifically, in this embodiment, the target data is distance-Doppler spectrum data, and the first preset condition is that the Doppler spectrum width feature corresponding to the target data is greater than a preset feature threshold.

[0061] Step 905: Detect the interference distribution information corresponding to the shower head water flow in the range-Doppler spectrum, and remove the interference distribution information from the range-Doppler spectrum to obtain the target range-Doppler signal.

[0062] Step 906: Process the target distance-Doppler signal based on the digital beamforming algorithm to obtain a 4D point cloud image representing human motion information.

[0063] Step 907: Determine whether the user makes a specific posture adjustment within a preset time based on the 4D point cloud image.

[0064] Specifically, the process of adjusting a particular posture is the process of changing from a high posture to a low posture.

[0065] Step 908: If it is determined that the user made a specific posture adjustment within a preset time, then determine the target point cloud image corresponding to the user's specific posture adjustment process.

[0066] Step 909: Based on the width and height information of the point cloud distribution in the target point cloud image, determine the detection result of the human fall event.

[0067] Specifically, frame-level features of the target point cloud image can be determined by combining width and height information. These frame-level features include the width distribution of the point cloud on the x-axis, the width distribution on the y-axis, the maximum height on the z-axis, and the average height of the point cloud. Extended features and height features are then extracted from the frame-level features. The extended features characterize the distribution variation of the human body projection in the horizontal direction, while the height features characterize the distribution variation of the human body projection in the vertical direction. The extended features and height features are then input into a pre-defined classifier to output the detection results of human fall events.

[0068] In this embodiment, range-Doppler imaging is performed on the radar echo signal of the bathroom environment, and the shower state is detected; after determining that the device is in working condition, the human target and water flow target are separated on the range-Doppler spectrum, and 4D human point cloud imaging is performed; based on the 4D point cloud image, the specific posture adjustment of the user in the bathroom environment is detected, and it is determined whether the human body is in a low posture state, thus efficiently completing the human fall recognition. Through the implementation of this embodiment, the following beneficial effects can be obtained: (1) The user does not need to contact the sensor device or wear any device; it does not infringe on personal privacy; it is not sensitive to light, dust, smoke and temperature; the device can work all day and all day; (2) It can work in a variety of complex bathroom shower environments and can penetrate bathroom glass to achieve robust detection of different fall postures; (3) It has low computational complexity, can run on a general ARM processor, and is easy to edge computing and perception.

[0069] It should be understood that the sequence number of each step in this embodiment does not imply the order in which the steps are executed. The execution order of each step should be determined by its function and internal logic, and should not constitute a unique limitation on the implementation process of this application embodiment.

[0070] Figure 10 An electronic device is provided in the third embodiment of this application. This electronic device can be used to implement the human fall detection method in a bathroom environment as described in the foregoing embodiments, and mainly includes: The system includes a memory 1001, a processor 1002, and a computer program 1003 stored on the memory 1001 and executable on the processor 1002. The memory 1001 and the processor 1002 are connected via communication. When the processor 1002 executes the computer program 1003, it implements the method described in Embodiment 1 or 2 above. The number of processors can be one or more.

[0071] The memory 1001 can be a high-speed random access memory (RAM) or a non-volatile memory, such as a disk storage device. The memory 1001 is used to store executable program code, and the processor 1002 is coupled to the memory 1001.

[0072] Furthermore, embodiments of this application also provide a computer-readable storage medium, which may be disposed in the aforementioned electronic device, and the computer-readable storage medium may be as described above. Figure 10 The memory in the illustrated embodiment.

[0073] The computer-readable storage medium stores a computer program that, when executed by a processor, implements the human fall detection method in the bathroom environment described in the foregoing embodiments. Furthermore, the computer-readable storage medium can also be a USB flash drive, external hard drive, read-only memory (ROM), RAM, magnetic disk, or optical disk, or any other medium capable of storing program code.

[0074] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0075] The modules described as separate components may or may not be physically separate. Similarly, the components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0076] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0077] If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a readable 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 of the various embodiments of this application. The aforementioned readable storage medium includes various media capable of storing program code, such as USB flash drives, external hard drives, ROM, RAM, magnetic disks, or optical disks.

[0078] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0079] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0080] The above is a description of the method, device and readable storage medium for detecting human falls in a bathroom environment provided in this application. For those skilled in the art, based on the ideas of the embodiments of this application, there will be changes in the specific implementation and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for detecting human falls in a bathroom environment, characterized in that, include: Doppler spectral width features are extracted from the range-Doppler spectrum corresponding to the bathroom radar echo signal; wherein, the Doppler spectral width features are used to characterize the breadth of the velocity distribution of the detected target; Based on the Doppler spectral width characteristics, determine whether the shower equipment in the bathroom is currently in operation; If the shower equipment is currently in operation, a 4D point cloud image representing human motion information is generated based on the distance-Doppler spectrum. Based on the 4D point cloud image, the detection result of the human fall event is determined; The Doppler spectral width feature is represented as follows: in, This indicates the Doppler spectral width characteristic. This represents the distance-Doppler spectrum. Indicates the index of the Doppler cell. Indicates the index of the distance cell. Represents the average coefficient. Indicates the number of Doppler elements. Indicates the number of distance units; Determining whether the shower equipment in the bathroom is currently in operation based on the Doppler spectral width characteristics includes: Based on the Doppler spectral width characteristics, sliding window detection is performed on the distance-Doppler spectrum; When the number of target data frames that meet the first preset condition in the timing window is greater than the preset frame number threshold, it is determined that the shower equipment in the bathroom is currently in working state; wherein, the target data is distance-Doppler spectrum data, and the first preset condition is: the Doppler spectrum width feature corresponding to the target data is greater than the preset feature threshold.

2. The method for detecting human falls according to claim 1, characterized in that, The generation of 4D point cloud images representing human motion information based on the distance-Doppler spectrum includes: The interference distribution information corresponding to the shower head water flow in the distance-Doppler spectrum is detected, and the interference distribution information is removed from the distance-Doppler spectrum to obtain the target distance-Doppler signal; The target range-Doppler signal is processed using a digital beamforming algorithm to obtain a 4D point cloud image representing human motion information.

3. The method for detecting human falls according to claim 2, characterized in that, The detection of interference distribution information corresponding to the shower head water flow in the distance-Doppler spectrum includes: Create a statistical matrix corresponding to the distance-Doppler spectrum; wherein the statistical matrix has the same size as the distance-Doppler spectrum; Based on the statistical matrix, interference targets are detected in each frame of the range-Doppler spectrum to determine all target locations in the range-Doppler spectrum where interference targets appear and the frequency of interference targets at each target location. The range-Doppler spectrum is labeled by combining the target location and the frequency to obtain a labeled range-Doppler spectrum; Based on the marker information in the marked distance-Doppler spectrum, the interference distribution information corresponding to the shower head water flow is determined.

4. The method for detecting human falls according to claim 1, characterized in that, The step of determining the detection result of a human fall event based on the 4D point cloud image includes: Based on the 4D point cloud image, it is determined whether the user performs a specific posture adjustment within a preset time; wherein, the process of specific posture adjustment is the process of changing from a high posture to a low posture; If it is determined that the user made a specific posture adjustment within a preset time, then the target point cloud image corresponding to the user's specific posture adjustment process is determined. The width and height information of the point cloud distribution in each frame of the target point cloud image are extracted respectively; By combining all the width and height information, the detection result of the human fall event is determined.

5. The method for detecting human falls according to claim 4, characterized in that, The step of determining whether the user makes a specific posture adjustment within a preset time based on the 4D point cloud image includes: Determine the associated point cloud quantity sequence and centroid height sequence of the 4D point cloud image; Based on the associated point cloud quantity sequence and the centroid height sequence, sliding window detection is performed on the 4D point cloud image; When the number of associated point clouds and the centroid height of the target area within the time sequence window meet the second preset condition, it is determined that the user has made a specific posture adjustment within a preset time period; wherein, the target area includes a first area and a second area, and the first area is later than the second area in time sequence; the second preset condition includes: the proportion of data frames in the first area of ​​the time sequence window whose centroid height is lower than a preset height threshold is greater than a first proportion threshold, the proportion of data frames in the first area whose number of associated point clouds is higher than a preset number of points threshold is greater than a second proportion threshold, and the proportion of data frames in the second area of ​​the time sequence window whose centroid height is higher than a preset height threshold is greater than a third proportion threshold.

6. The method for detecting human falls according to claim 4, characterized in that, The step of combining all the width information and the height information to determine the detection result of a human fall event includes: By combining the width information and the height information, the frame-level features of the target point cloud image are determined; wherein, the frame-level features include the width distribution of the point cloud on the x-axis, the width distribution on the y-axis, the maximum height on the z-axis, and the average height of the point cloud; Extended features and height features are extracted from the frame-level features; wherein, the extended features are used to characterize the distribution variation of human body projection in the horizontal direction, and the height features are used to characterize the distribution variation of human body projection in the vertical direction. The extended features and the height features are input into a preset classifier to output the detection results of human fall events.

7. An electronic device, characterized in that, Includes memory and processor, of which: The processor is used to execute computer programs stored in the memory; When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images