Toilet falling detection method based on millimeter wave radar

Abstract

The invention discloses a method for detecting falling in a toilet scene based on a millimeter wave radar, and relates to the technical field of millimeter wave radar monitoring. According to the method, distance, angle and speed information of a target is obtained through the millimeter wave radar, and according to the difference of water flow and human body point cloud in spatial distribution when a person falls down, the distance, angle and speed information of the target is obtained; a radar point cloud is divided into two clusters in a spatial dimension based on a clustering algorithm. Then, by analyzing the difference of the two clusters in the speed dimension, the average speed parameter in the vertical direction in the clusters is introduced to recognize a target source and retain the target source; and on the basis, the falling trend of the height of the center of mass of the reserved cluster is monitored, so that the falling action is accurately detected. Based on the multi-dimensional clustering algorithm, the invention designs a set of bathroom fall detection system which is low in complexity and does not need data training.

Description

Technical Field

[0001] This invention relates to the field of radar detection, and more particularly to a method for detecting falls in a bathroom scene based on millimeter-wave radar. Background Technology

[0002] By recognizing falls and other special human activities, timely medical assistance can be provided to prevent secondary injuries. Falls are especially likely to occur in small and damp bathrooms, making fall detection essential.

[0003] In a bathroom environment, camera-based human activity recognition solutions are difficult to deploy due to user privacy restrictions in the shower setting. While wearable devices and sensors can function stably in most cases, they suffer from high costs and uncontrollable wear behavior. Radar, on the other hand, is highly robust to various lighting and weather conditions, allowing it to operate normally in diverse environments (such as a bathroom). Furthermore, radar protects personal privacy because it receives three-dimensional spatial information extracted from reflected echo signals, rather than visual image information.

[0004] However, numerous interference sources exist in bathroom scenarios, especially in shower settings. Millimeter-wave radar signals are easily affected by water flow, causing point cloud data to inaccurately represent human targets. During a shower, millimeter-wave radar collects point cloud data of both the person and the water flow. The height of the point cloud generated by the water flow typically exceeds the height of the human body's point cloud. Therefore, even if a fall or squatting motion occurs, the height information obtained from the radar remains unchanged, making traditional radar algorithms ineffective in detecting height changes caused by falls.

[0005] In a shower environment, the radar echo signal generated by flowing water can strongly interfere with or even mask the radar echo signal of human targets, making it unreliable to detect abnormal human activity based solely on height features. While deep learning models can be trained on large amounts of data to learn human fall behavior, the inherent randomness and instability of flowing water make it difficult to guarantee the robustness of the model in practical applications. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of existing technologies by proposing a fall detection method based on millimeter-wave radar in a bathroom setting.

[0007] The objective of this invention is achieved through the following technical solution: a fall detection method for a bathroom scene based on millimeter-wave radar, the method comprising:

[0008] S1. Install a millimeter-wave radar at a position higher than the shower head, and convert the point cloud information obtained by the millimeter-wave radar into a Cartesian coordinate system;

[0009] S2. Based on the difference in spatial distribution between the water flow and the human body point cloud when a person falls, a clustering algorithm is used to separate the point cloud into human body point cloud clusters and water flow point cloud clusters; during the clustering process, weights are set to reduce the influence of the vertical direction on the distance parameter;

[0010] S3. Calculate the average vertical velocity of each cluster. Based on the difference in vertical velocity between the human point cloud and the water flow point cloud, filter out the water point cloud and retain the human point cloud cluster.

[0011] S4. Select multiple points with the highest coordinate values ​​in the Z-axis direction from the point cloud data retained in the current frame, and calculate their average value as the instantaneous height of the target in this frame. Use the exponential moving average method to filter the instantaneous height to obtain a more representative target representation height. Detect the fall action based on the target representation height.

[0012] S5. Introduce a fixed-capacity circular buffer to store the target representation height calculated over multiple consecutive frames. The header of this buffer stores the earliest recorded valid reference height within a preset time window. Compare the target height calculated in the current frame with this reference height and calculate the height drop rate between them. If the height drop rate is not lower than a set threshold, the subsequent judgment process is not triggered, and the fall judgment counter remains in its initial state. If the height drop rate is lower than the threshold, proceed to the next judgment step.

[0013] S6. Compare the target representation height of the current frame with the preset absolute height threshold. If the current frame height is continuously detected to be lower than the absolute height threshold, confirm that the target has fallen.

[0014] On the other hand, this specification also provides a system for implementing the method, comprising:

[0015] The signal processing module is used to generate point cloud data containing information on distance, angle of arrival, radial velocity, and signal-to-noise ratio.

[0016] The clustering and separation module is used to perform preliminary separation of point cloud data using clustering algorithms;

[0017] The velocity discrimination module is used to introduce velocity features as discrimination criteria to accurately identify and filter out water point clouds;

[0018] The application module extracts human activity information based on the retained point cloud data to achieve fall detection.

[0019] On the other hand, the present invention also provides a fall detection device for a bathroom scene based on millimeter-wave radar, including a memory and one or more processors. The memory stores executable code, and when the processor executes the executable code, it implements the fall detection method for a bathroom scene based on millimeter-wave radar.

[0020] On the other hand, the present invention also provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the aforementioned method for detecting falls in a bathroom scene based on millimeter-wave radar.

[0021] The beneficial effects of this invention are:

[0022] (1) This invention uses millimeter-wave radar installed in the bathroom to monitor users entering the bathroom in real time. To address the interference of water flow point clouds in the shower environment on human body recognition, this invention proposes a multi-dimensional point cloud filtering algorithm based on spatial distribution features and velocity features, which can effectively distinguish and filter out water flow point clouds, significantly improving the purity and recognition accuracy of human body point cloud data.

[0023] (2) The retained point cloud information can be directly used for subsequent fall detection, realizing fall detection in the shower environment, providing users with a safer and more reliable bathroom environment, and providing a technical foundation for the functional expansion of smart home systems. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating the fall detection method for a bathroom scene based on millimeter-wave radar provided in an embodiment of the present invention.

[0025] Figure 2 This is a schematic diagram of a millimeter-wave radar detection scenario provided in an embodiment of the present invention.

[0026] Figure 3 This is a schematic diagram of the preliminary spatial clustering results provided in an embodiment of the present invention.

[0027] Figure 4 This is a schematic diagram of the multidimensional clustering results provided in an embodiment of the present invention.

[0028] Figure 5 This is a schematic diagram illustrating the change in target height before and after multidimensional clustering processing in a shower scene where the person falls, as provided in an embodiment of the present invention.

[0029] Figure 6 This invention provides a fall detection method for bathroom scenarios based on millimeter-wave radar. Detailed Implementation

[0030] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0031] like Figure 1 , Figure 2 As shown in the embodiment, a fall detection method for a bathroom scene based on millimeter-wave radar is provided, including the following steps:

[0032] 1. Scene setup:

[0033] A ceiling-mounted millimeter-wave radar is installed on the bathroom ceiling, above the shower head, to ensure radar coverage of the entire shower area. In this embodiment, the millimeter-wave radar is installed at a height of 2.1m.

[0034] The shower head uses a conventional wall-mounted installation, which suits the daily showering habits of most people.

[0035] To make the difference between humans and water more obvious in the speed dimension, this embodiment sets the tilt angle ω of the radar relative to the ground to 90°, and the installation angle of the millimeter-wave radar to be parallel to the ground. The measured radial velocity of the point cloud directly represents the velocity in the z-axis direction. Under this condition, the vertical velocity difference between water and human target reaches its maximum value.

[0036] 2. Signal reception and point cloud generation

[0037] Millimeter-wave radar can obtain sparse point clouds from raw signals through a series of operations. Each point cloud contains information such as range, angle of arrival (azimuth and elevation), Doppler shift (radial velocity), and signal-to-noise ratio. For ease of subsequent processing, the point cloud information in spherical coordinates is converted to Cartesian coordinates using the following formula:

[0038] x = r.cos(φ).cos(θ)

[0039] y = r.cos(φ).sin(θ)

[0040] z = r.sin(φ)

[0041] Where r is the distance from the point cloud to the radar (radial distance), θ is the azimuth angle (horizontal angle, usually with the radar's front as the reference), and φ is the elevation angle (vertical angle, usually with the horizontal plane as the reference).

[0042] 3. Point cloud clustering and separation

[0043] When a person falls, the point cloud of the flowing water exhibits a significant clustering characteristic in its spatial distribution, mainly concentrated in the area around the nozzle; in contrast, the point cloud of the human body, due to the physical state of the person falling, is mainly distributed in the area near the ground. Thus, the point clouds of the flowing water and the human body form a clear and distinguishable spatial distribution difference. Since the 3D point clouds of the same person or object have a high correlation in the horizontal plane (xy plane) but a sparser distribution in the vertical direction (z-axis), an optimized distance parameter formula is used in the clustering process:

[0044] D(p i ,p j )=(x i -xj ) 2 +(y i -y j ) 2 +α·(z i -z j ) 2

[0045] Where, p i Let represent the i-th point cloud, and α be the weighting coefficient used to reduce the influence of the z-axis on the distance parameter. In this example, α is set to 0.25.

[0046] After clustering, the point cloud data was initially separated into human body point cloud clusters and water flow point cloud clusters.

[0047] 4. Water point cloud filtering based on velocity characteristics

[0048] Since the vertical movement speed of a human body is less than the downward velocity of water, velocity characteristics are introduced as a discrimination criterion to calculate the average vertical velocity of each cluster:

[0049]

[0050] in, For cluster C k The average velocity in the vertical direction, N k For cluster C k The number of points in the cloud, v i ω is the radial velocity obtained from the Doppler frequency shift of the i-th point cloud, and ω is the installation tilt angle of the radar.

[0051] The human body point cloud cluster is preserved using the following formula:

[0052]

[0053] The retained human point cloud will be used for subsequent fall detection, while other point clouds will be filtered out.

[0054] 5. Estimating target height based on preserved point cloud data.

[0055] In the retained point cloud data of the current frame, select multiple points with the highest coordinate values ​​in the Z-axis direction, calculate their average value, and use it as the instantaneous height of the target in this frame. Then, use the exponential moving average method to smooth the instantaneous height to obtain the target's representative height, thereby realizing the detection of fall actions. The expression is as follows:

[0056]

[0057] Where t represents the frame index, The height represents the instantaneous height of the target in the current frame. tα represents the target height of the current frame, and α is the weighting coefficient of the height of the previous frame when calculating the height in the current frame.

[0058] 6. Detect fall movements based on estimated target height.

[0059] To implement fall detection, a fixed-capacity circular buffer is introduced to store the target height calculated over multiple consecutive frames. In this embodiment, the target height of the most recent 2.5 seconds is stored. The historical height data stored at the top of this buffer represents the earliest recorded valid reference height within a preset time window. By comparing the target height calculated in the current frame with this reference height, the height drop rate between the two is calculated and defined as the ratio of the current height to the reference height.

[0060] If the rate of descent is not lower than the set threshold, it is considered that the target does not show a significant trend of height change, and the subsequent judgment process is not triggered, and the fall judgment counter remains in its initial state; if the rate of descent is lower than the threshold (ΔH), it is considered that the target does not show a significant trend of height change, and the subsequent judgment process is not triggered, and the fall judgment counter remains in its initial state; th If the target height of the current frame is found to be lower than the set absolute height threshold, then the next step is to compare whether the target height of the current frame is lower than the set absolute height threshold. th When the current frame height exceeds this threshold, the fall detection counter increments by 1. If the current frame height is higher than this threshold, it is considered that no continuous fall has occurred, and the counter is reset to 1. Only when the counter accumulates to a preset trigger threshold does the system confirm that the target has fallen and activate the alarm mechanism. By introducing a three-level judgment mechanism of "height descent rate - absolute height - continuity judgment", the accuracy and stability of fall detection are effectively improved, avoiding misjudgments caused by brief posture changes or environmental interference. In this embodiment, ΔH th Set to 0.4, H th Set it to 0.4m.

[0061] According to the settings of this embodiment, such as Figure 3 As shown, after a person falls, the point cloud of the human body and the point cloud of the water flow form obvious clusters in spatial distribution after spatial feature clustering.

[0062] like Figure 4 As shown, in Figure 3 Based on this, the average vertical velocities of the two clusters were calculated to be 3.93 m / s and 0.16 m / s, respectively. The point cloud cluster with the smaller average vertical velocity was retained based on the velocity characteristics. The average z-axis value of the two highest points was adjusted from 1.52 m to 0.2 m. The retained point cloud clusters can be used for fall detection.

[0063] Figure 5For a slip-and-fall event in a shower environment, the original target height trajectory (green curve) cannot reflect the fall due to the interference of shower water flow. In contrast, the blue curve is the target height change curve after processing by the multi-dimensional point cloud clustering algorithm proposed in this invention. The target fall detection algorithm proposed in this application can effectively detect human fall activities based on the target height processed by the multi-dimensional point cloud clustering algorithm.

[0064] Corresponding to the aforementioned embodiment of a fall detection method for a bathroom scene based on millimeter-wave radar, the present invention also provides an embodiment of a fall detection device for a bathroom scene based on millimeter-wave radar, comprising:

[0065] The signal processing module is used to generate point cloud data containing information on distance, angle of arrival, radial velocity, and signal-to-noise ratio.

[0066] The clustering and separation module is used to perform preliminary separation of point cloud data using clustering algorithms;

[0067] The velocity discrimination module is used to introduce velocity features as discrimination criteria to accurately identify and filter out water point clouds;

[0068] The application module extracts human activity information based on the retained point cloud data to achieve fall detection.

[0069] Corresponding to the aforementioned embodiment of a fall detection method for a bathroom scene based on millimeter-wave radar, the present invention also provides an embodiment of a fall detection device for a bathroom scene based on millimeter-wave radar.

[0070] See Figure 6 The present invention provides a fall detection device for a bathroom scene based on millimeter-wave radar, comprising a memory and one or more processors. The memory stores executable code, and when the processor executes the executable code, it is used to implement a fall detection method for a bathroom scene based on millimeter-wave radar as described in the above embodiment.

[0071] The present invention provides an embodiment of a fall detection device for a bathroom scene based on millimeter-wave radar, which can be applied to any device with data processing capabilities, such as a computer. The device embodiment can be implemented through software, hardware, or a combination of both. Taking software implementation as an example, as a logical device, it is formed by the processor of any data processing device loading the corresponding computer program instructions from non-volatile memory into memory for execution. From a hardware perspective, such as... Figure 6The diagram shown is a hardware structure diagram of any device with data processing capabilities, including a fall detection device for a bathroom scene based on millimeter-wave radar provided by the present invention. (Except for...) Figure 6 In addition to the processor, memory, network interface, and non-volatile memory shown, any data processing device in the embodiment may also include other hardware depending on the actual function of the data processing device, which will not be described in detail here.

[0072] The specific implementation process of the functions and roles of each unit in the above device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0073] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of the present invention according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0074] This invention also provides a computer-readable storage medium storing a program thereon, which, when executed by a processor, implements a fall detection method for a bathroom scene based on millimeter-wave radar as described in the above embodiments.

[0075] The computer-readable storage medium can be an internal storage unit of any data processing device as described in any of the foregoing embodiments, such as a hard disk or memory. The computer-readable storage medium can also be an external storage device of any data processing device, such as a plug-in hard disk, smart media card (SMC), SD card, flash card, etc., equipped on the device. Furthermore, the computer-readable storage medium can include both internal storage units and external storage devices of any data processing device. The computer-readable storage medium is used to store the computer program and other programs and data required by the data processing device, and can also be used to temporarily store data that has been output or will be output.

[0076] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the aforementioned method for detecting falls in a bathroom scene based on millimeter-wave radar.

[0077] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the disclosure herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this application are indicated by the claims.

[0078] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this application. This application is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A fall detection method for a bathroom scene based on millimeter-wave radar, characterized in that, The method includes: S1. Install a millimeter-wave radar at a position higher than the shower head, and convert the point cloud information obtained by the millimeter-wave radar into a Cartesian coordinate system; S2. Based on the difference in spatial distribution between the water flow and the human body point cloud when a person falls, a clustering algorithm is used to separate the point cloud into human body point cloud clusters and water flow point cloud clusters; during the clustering process, weights are set to reduce the influence of the vertical direction on the distance parameter; S3. Calculate the average vertical velocity of each cluster. Based on the difference in vertical velocity between the human point cloud and the water flow point cloud, filter out the water point cloud and retain the human point cloud cluster. S4. Select multiple points with the highest coordinate values ​​in the Z-axis direction from the point cloud data retained in the current frame, and calculate their average value as the instantaneous height of the target in this frame. Use the exponential moving average method to filter the instantaneous height to obtain a more representative target representation height. Detect the fall action based on the target representation height. S5. Introduce a fixed-capacity circular buffer to store the target representation height calculated over multiple consecutive frames. The header of this buffer stores the earliest recorded valid reference height within a preset time window. Compare the target height calculated in the current frame with this reference height and calculate the height drop rate between them. If the height drop rate is not lower than a set threshold, the subsequent judgment process is not triggered, and the fall judgment counter remains in its initial state. If the height drop rate is lower than the threshold, proceed to the next judgment step. S6. Compare the target representation height of the current frame with the preset absolute height threshold. If the current frame height is continuously detected to be lower than the absolute height threshold, confirm that the target has fallen.

2. The fall detection method in a bathroom scene based on millimeter-wave radar according to claim 1, characterized in that, The specific steps of converting the point cloud information obtained from millimeter-wave radar into a Cartesian coordinate system include: x = r.cos(φ).cos(θ) y = r.cos(φ).sin(θ) z = r.sin(φ) Where r is the radial distance from the point cloud to the radar, θ is the azimuth angle, and φ is the elevation angle.

3. The fall detection method in a bathroom scene based on millimeter-wave radar according to claim 1, characterized in that, The method of reducing the influence of the vertical direction on the distance parameter by setting weights includes: D(p) i ,p j )=(x i -x j ) 2 +(y i -y j ) 2 +α·(z i -z j ) 2 Here, α is a weighting coefficient used to reduce the influence of the z-axis on the distance parameter.

4. The fall detection method in a bathroom scene based on millimeter-wave radar according to claim 1, characterized in that, In S3, the average vertical velocity of each cluster is calculated: in, For cluster C k The average velocity in the vertical direction, N k For cluster C k The number of points in the cloud, v i Let ω be the radial velocity of the i-th point cloud, and ω be the tilt angle at which the radar is mounted. The human body point cloud cluster is preserved using the following formula: The preserved human point cloud data will be used for subsequent fall detection.

5. A fall detection method for a bathroom scene based on millimeter-wave radar according to claim 1, characterized in that, The instantaneous height is filtered using the exponential moving average method to obtain a more representative target height. Where t represents the frame index, This represents the average height of the n highest point clouds in the current frame, i.e., the instantaneous height of the current frame. t β represents the target representation height of the current frame, while β is the weighting coefficient of the target representation height of the previous frame in the calculation of the current frame.

6. The method for detecting falls in a bathroom scene based on millimeter-wave radar according to claim 1, characterized in that, The altitude descent rate is defined as the ratio of the current altitude to the reference altitude.

7. A fall detection method for a bathroom scene based on millimeter-wave radar according to claim 1, characterized in that, The comparison of the target representation height of the current frame with a preset absolute height threshold specifically includes: When the current frame height is detected to be lower than the absolute height threshold, the fall detection counter is incremented by 1; if the current frame height is higher than the threshold, it is considered that no continuous fall behavior has occurred, and the counter is reset to 1. Only when the counter accumulates to the preset trigger threshold, the system confirms that the target has fallen and activates the alarm mechanism.

8. A system for implementing the method according to any one of claims 1-7, characterized in that, include: The signal processing module is used to generate point cloud data containing information on distance, angle of arrival, radial velocity, and signal-to-noise ratio. The clustering and separation module is used to perform preliminary separation of point cloud data using clustering algorithms; The velocity discrimination module is used to introduce velocity features as discrimination criteria to accurately identify and filter out water point clouds; The application module extracts human activity information based on the retained point cloud data to achieve fall detection.

9. A fall detection device for a bathroom scene based on millimeter-wave radar, comprising a memory and one or more processors, wherein the memory stores executable code, characterized in that, When the processor executes the executable code, it implements a fall detection method for a bathroom scene based on millimeter-wave radar as described in any one of claims 1-7.

10. A computer-readable storage medium having a program stored thereon, characterized in that, When the program is executed by the processor, it implements a fall detection method for a bathroom scene based on millimeter-wave radar as described in any one of claims 1-7.

Images