Age-friendly and healthy lighting methods and systems, intelligent lighting terminals and storage media

Abstract

This application provides an age-friendly health lighting method and system, an intelligent lighting terminal, and a storage medium. The method includes: matching the motion characteristics of human targets in a monitoring space based on predefined fall behavior discrimination rules; marking human targets that meet the fall behavior discrimination rules as candidate fall instances and generating corresponding rule matching scores; inputting the motion characteristics corresponding to the candidate fall instances into a pre-trained fall recognition classification model and outputting a fall confidence score; performing a weighted decision fusion of the rule matching score and the fall confidence score to obtain a comprehensive fall judgment score; and dynamically selecting age-friendly lighting strategies from a preset strategy library for lighting based on the comprehensive fall judgment score. This application constructs a collaborative linkage mechanism between fall monitoring and lighting control, which can automatically trigger lighting intervention at critical moments when fall risks occur, providing timely, safe, and accurate lighting support.

Description

Technical Field

[0001] This application belongs to the field of intelligent lighting technology, and relates to an age-friendly healthy lighting method and system, an intelligent lighting terminal and storage medium. Background Technology

[0002] With the increasing global trend of population aging, the home safety of elderly people living alone or semi-alone is becoming increasingly prominent. Among these issues, falls, due to their high incidence and serious consequences, have become one of the leading causes of disability and even death among the elderly. If an elderly person falls and is not detected and effectively rescued within the golden hour, it can easily lead to secondary injuries, complications, and even life-threatening situations, posing a significant threat to their physical and mental health.

[0003] Studies have shown that poor indoor lighting is a significant environmental factor contributing to falls among the elderly. Problems such as excessive darkness, glare, uneven illumination, unsuitable color temperature, or delayed response when switching lights on and off can significantly reduce the elderly's ability to recognize spatial obstacles and their dynamic balance control, increasing the risk of falls during nighttime activities.

[0004] However, current mainstream lighting systems primarily focus on basic lighting functions, energy efficiency optimization, or ambiance creation. Examples include automatically switching lights on and off based on ambient light intensity, dynamically adjusting color temperature according to circadian rhythms, and achieving ambient lighting through preset scenes. Although some systems have introduced simple human body sensing mechanisms, their implementation is usually limited to a binary control logic of "lights on when someone is there, lights off when no one is there." The functional design generally does not adequately consider the physiological decline in visual perception, mobility, and balance control experienced by the elderly, and lacks safety response and proactive intervention capabilities for high-risk behaviors such as falls. Summary of the Invention

[0005] This application provides an age-friendly health lighting method and system, an intelligent lighting terminal and storage medium, which improves the safety of elderly people's home life by automatically triggering age-friendly lighting intervention at critical moments when the risk of falling occurs.

[0006] Firstly, this application provides an age-friendly health lighting method, comprising: performing rule matching on the motion features of human targets in a monitoring space based on predefined fall behavior discrimination rules, marking the human targets that satisfy the fall behavior discrimination rules as candidate fall instances, and generating corresponding rule matching scores; the motion features include the height of the human center of gravity from the ground, the vertical descent rate of the center of gravity, the proportion of the human posture, and the human movement speed; inputting the motion features corresponding to the candidate fall instances into a pre-trained fall recognition classification model, and outputting a fall confidence score; the fall recognition classification model is a deep learning model or a gradient boosting decision tree model, wherein the deep learning model includes a convolutional neural network and a recurrent neural network; performing weighted decision fusion on the rule matching score and the fall confidence score to obtain a comprehensive fall judgment score; and dynamically selecting age-friendly lighting strategies from a preset strategy library for lighting based on the comprehensive fall judgment score.

[0007] In one implementation of the first aspect, the method further includes: acquiring a TOF depth map of the monitoring space in a manned state and a TOF depth map in an unmanned state, respectively; detecting and tracking a human target in the monitoring space based on the TOF depth map in the manned state to generate a corresponding human motion trajectory; fitting a ground reference plane of the monitoring space based on the TOF depth map in the unmanned state to obtain a ground plane model; and calculating the motion characteristics of the human target based on the human motion trajectory and the ground plane model.

[0008] In one implementation of the first aspect, calculating the motion characteristics of the human target based on the human motion trajectory and the ground plane model includes: within a preset sliding time window, calculating the vertical distance from the center of mass of the human target to the ground reference plane, as the height of the human center of mass above the ground; calculating the descent velocity of the center of mass of the human target relative to the ground reference plane in the vertical direction, as the vertical descent rate of the center of mass; calculating the aspect ratio of the bounding box of the human target, as the human posture ratio; and calculating the horizontal movement speed of the human target, as the human motion speed.

[0009] In one implementation of the first aspect, the detection and trajectory tracking of human targets in the monitoring space based on the TOF depth map under the occupied state, and the generation of corresponding human motion trajectories, includes: preprocessing the TOF depth map under the occupied state; the preprocessing includes at least one or more of the following: bad pixel removal, amplitude threshold filtering, temperature compensation, nonlinear distance correction, spatial filtering, and temporal filtering; transforming the preprocessed TOF depth map to a preset indoor coordinate system through coordinate mapping to obtain a three-dimensional point cloud; performing foreground segmentation on the three-dimensional point cloud based on a foreground-background separation model to obtain a foreground point set; performing three-dimensional clustering on the foreground point set, and combining it with preset human geometric features to select several human targets; the human geometric features include voxel volume constraints, typical human height constraints, and minimum effective point threshold constraints; and generating the human motion trajectory of each human target in the monitoring space through cross-frame data association.

[0010] In one implementation of the first aspect, dynamically selecting an age-friendly lighting strategy from a preset strategy library based on the fall determination comprehensive score includes: determining the behavioral state of the human target based on the fall determination comprehensive score; the behavioral state includes a fallen state and a normal activity state; if the human target is in the fallen state, automatically increasing the illuminance of the area where the human target is located, adjusting the color temperature of the area to enhance environmental contrast, and controlling some lamps to flash at a preset frequency; if the human target is in the normal activity state, activating ground-embedded guide lights and forming a continuous light strip along a preset path.

[0011] In one implementation of the first aspect, rule matching of the motion characteristics of a human target in the monitoring space based on a predefined fall behavior discrimination rule includes: determining whether the motion characteristics of the human target satisfy any of the following conditions: the height of the human center of gravity above the ground decreases to below a first preset threshold within a short time window, and the vertical descent rate of the center of gravity exceeds a second preset threshold; the height of the human center of gravity above the ground remains below the first preset threshold within a long time window, and the human movement speed remains 0; the height of the human center of gravity above the ground slowly decreases from a preset standing height to below the first preset threshold, and the human posture proportion conforms to the characteristics of a lying down state; if so, the human target is determined to satisfy the fall behavior discrimination rule; otherwise, the human target is determined not to satisfy the fall behavior discrimination rule.

[0012] In one implementation of the first aspect, the method further includes: marking high fall risk locations as danger zones in a preset indoor coordinate system; the high fall risk locations include stairwells, toilet entrances, and locations of fixed obstacles; calculating the horizontal distance between the centroid of the human target and the danger zone; when the human target is detected entering the danger zone, activating the lights near the danger zone and switching the light color of the lights to a warning color, or controlling the lights to flash at a preset frequency.

[0013] Secondly, this application provides an intelligent lighting terminal, including a lamp body and an age-friendly health lighting device, wherein the age-friendly health lighting device is electrically connected to the lamp body; wherein the age-friendly health lighting device includes: a rule matching module, used to perform rule matching on the motion characteristics of human targets in the monitoring space based on predefined fall behavior discrimination rules, marking the human targets that meet the fall behavior discrimination rules as candidate fall instances, and generating corresponding rule matching scores; the motion characteristics include the height of the human center of gravity from the ground, the vertical descent rate of the center of gravity, the proportion of the human posture, and the human movement speed. The system includes: a fall detection module, which inputs the motion features corresponding to the candidate fall instances into a pre-trained fall detection classification model and outputs a fall confidence score; the fall detection classification model is a deep learning model or a gradient boosting decision tree model, wherein the deep learning model includes convolutional neural networks and recurrent neural networks; a decision fusion module, which performs weighted decision fusion on the rule matching score and the fall confidence score to obtain a comprehensive fall judgment score; and a lighting control module, which dynamically selects an age-appropriate lighting strategy from a preset strategy library for lighting based on the comprehensive fall judgment score.

[0014] Thirdly, this application provides an age-friendly health lighting system, comprising: multiple smart lighting terminals for performing the method described in any of the above-mentioned methods; a home gateway connected to each of the smart lighting terminals via a wired or wireless communication protocol; and a care platform communicatively connected to the home gateway for receiving alarm information reported by the smart lighting terminals through the home gateway and notifying caregivers.

[0015] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any of the preceding claims.

[0016] As described above, the age-friendly healthy lighting method and system, intelligent lighting terminal, and storage medium described in this application have the following beneficial effects:

[0017] (1) The fall detection and age-friendly lighting control strategy are deeply integrated to build a closed-loop intelligent safety system of "perception-recognition-environmental response-alarm";

[0018] (2) It can avoid the risk of privacy leakage and improve user acceptance without sacrificing the ability to recognize behavior;

[0019] (3) By introducing a fall recognition classification model, the system's ability to distinguish fall behavior is further improved while retaining the judgment based on physical rules, which significantly enhances the accuracy, robustness and generalization performance of fall recognition;

[0020] (4) By defining the danger zone, the system can accurately assess the relative spatial relationship between the human body and the high-risk location; the fall judgment achieved by this not only relies on the human body's own movement characteristics, but also integrates environmental context information, which significantly improves the foresight, scenario adaptability and practicality of the warning. Attached Figure Description

[0021] Figure 1 The flowchart shown is an embodiment of an age-friendly healthy lighting method according to this application.

[0022] Figure 2 The flowchart shown is a method for age-friendly and healthy lighting according to another embodiment of this application.

[0023] Figure 3 The flowchart shown is a process for generating a human motion trajectory according to an embodiment of this application.

[0024] Figure 4 The flowchart shown is a process for generating motion features of a human target according to an embodiment of this application.

[0025] Figure 5 The diagram shown is a rule matching flowchart according to an embodiment of this application.

[0026] Figure 6 The diagram shown is a flowchart of an embodiment of the present application for selecting an age-friendly lighting strategy.

[0027] Figure 7 The flowchart shown is a method for age-friendly and healthy lighting according to another embodiment of this application.

[0028] Figure 8 The diagram shown is a structural schematic of an intelligent lighting terminal according to an embodiment of this application.

[0029] Figure 9 The diagram shown is a structural schematic of a smart lighting terminal according to another embodiment of this application.

[0030] Figure 10 The diagram shown is a structural schematic of an age-friendly health lighting system according to an embodiment of this application. Detailed Implementation

[0031] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0032] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this application. Therefore, the drawings only show the components related to this application and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0033] Currently, existing technologies have significant shortcomings in fall detection and safety lighting solutions for the elderly, which can be summarized as follows:

[0034] (1) Wearable monitoring devices have poor compliance and are difficult to achieve reliable monitoring in an all-weather, non-intrusive manner.

[0035] Current common fall detection solutions mostly use wearable devices (such as smart bracelets, belts, or chest straps) that integrate accelerometers and gyroscopes to determine fall events by analyzing changes in posture and peak impact. However, such solutions rely heavily on long-term, continuous wear by users, which poses a serious challenge for the elderly: the devices are prone to causing discomfort, are often forgotten or removed intentionally, and require regular maintenance and charging, making them highly complex to operate.

[0036] (2) Video surveillance based on visible light cameras poses risks to privacy and robustness.

[0037] Some technical solutions utilize RGB cameras to capture video streams and combine them with human posture recognition and behavior analysis algorithms for fall detection. Although the relevant algorithms are becoming increasingly mature, they face two major bottlenecks in practical applications: First, they involve personal privacy, and deploying visible light cameras in private spaces such as bedrooms and bathrooms can easily provoke user resistance and may violate data protection regulations; second, they are sensitive to changes in ambient lighting, obstructions, or limited viewing angles, resulting in high false alarm and false negative rates and insufficient system stability.

[0038] (3) Insufficient resolution of non-imaging sensing technology.

[0039] While non-visual sensing technologies such as millimeter-wave radar and passive infrared (PIR) offer some privacy advantages and can detect human presence and rough movements without generating images, their spatial resolution is limited, making it difficult to capture the details of human three-dimensional posture. This is particularly true when distinguishing between behaviors with similar movement characteristics, such as falling, bending over, and sitting down, where the recognition accuracy is severely insufficient. Furthermore, their output is typically limited to "person / no one" or simple movement states, failing to support highly reliable determinations of fall events.

[0040] (4) The fall monitoring and lighting systems are disconnected and lack a coordinated linkage mechanism.

[0041] Current smart lighting systems primarily focus on energy efficiency management, atmosphere creation, or basic sensor control, failing to adequately consider the safety needs of the elderly. Even those systems with human perception capabilities often lack a correlation with fall risk states, hindering the dynamic adjustment of lighting strategies before, during, or after a fall. Fall detection and lighting control have long been separated between "perception" and "execution," lacking a proactive, precise, and seamless lighting linkage response mechanism tailored to the risky behaviors of the elderly.

[0042] To address the aforementioned issues, this application proposes an age-friendly healthy lighting method and system, an intelligent lighting terminal, and a storage medium. This solution, while strictly protecting user privacy, integrates high-precision, non-invasive three-dimensional human motion sensing technology to achieve fine-grained recognition of the posture and behavior of the elderly. Furthermore, it constructs a collaborative linkage mechanism between fall detection and lighting control, which can automatically trigger lighting intervention at critical moments when a fall risk occurs, providing timely, safe, and precise lighting support, effectively improving the safety and autonomy of the elderly in their home life.

[0043] The age-friendly healthy lighting method described in this application embodiment can be applied to various types of hardware devices. The hardware device is preferably a smart lamp, but can also be a computer including components such as a memory, memory controller, one or more processing units (CPUs), peripheral interfaces, RF circuits, audio circuits, speakers, microphones, input / output (I / O) subsystems, a display screen, other output or control devices, and external ports; the computer includes, but is not limited to, personal computers such as desktop computers, laptops, tablets, smartphones, smart TVs, and personal digital assistants (PDAs).

[0044] The intelligent lighting terminal can be installed on the ceiling or wall indoors, with typical deployment locations including bedrooms, bathrooms, corridors, and other areas where elderly people frequently move around or are at high risk of falls. This terminal integrates a 3D human perception module based on Time-of-Flight (TOF) technology and an age-friendly lighting control unit within a single lighting device, achieving deep integration of fall detection and intelligent lighting. By acquiring real-time depth information and posture changes of the human body in the space, the terminal can locally identify falls or abnormal behaviors and instantly trigger corresponding lighting linkage strategies. This integrated design eliminates the need for additional dedicated sensors or wearable devices, making it suitable for various age-friendly care scenarios such as home-based elderly care, community-based elderly care, nursing homes, and hospital wards, while balancing privacy protection, ease of deployment, and system reliability.

[0045] The following will describe in detail the principles and implementation methods of an age-friendly healthy lighting method and system, intelligent lighting terminal and storage medium of this embodiment, so that those skilled in the art can understand the age-friendly healthy lighting method and system, intelligent lighting terminal and storage medium of this embodiment without creative effort.

[0046] Please see Figure 1 The above is a flowchart of an embodiment of an age-friendly health lighting method according to this application.

[0047] like Figure 1 As shown, this embodiment provides an age-friendly and healthy lighting method, including the following steps S100 to S400.

[0048] In step S100, based on predefined fall behavior discrimination rules, the motion characteristics of human targets in the monitoring space are matched according to rules. The human targets that satisfy the fall behavior discrimination rules are marked as candidate fall instances, and corresponding rule matching scores are generated.

[0049] In this embodiment, the motion characteristics include the height of the human body's center of gravity above the ground, the rate of vertical descent of the center of gravity, the proportion of the human body's posture, and the speed of human movement.

[0050] In step S200, the motion features corresponding to the candidate fall instance are input into the pre-trained fall recognition classification model, and the fall confidence score is output.

[0051] In step S300, the rule matching score and the fall confidence score are weighted and fused to obtain a comprehensive fall determination score.

[0052] In step S400, based on the comprehensive score of the fall determination, an age-appropriate lighting strategy is dynamically selected from the preset strategy library for lighting.

[0053] Please see Figure 2The flowchart shown is a method for age-friendly and healthy lighting according to another embodiment of this application.

[0054] like Figure 2 As shown, this embodiment provides an age-friendly and healthy lighting method that further includes the following steps S110 to S140.

[0055] In step S110, the TOF depth map of the monitoring space under manned and unmanned conditions are obtained respectively.

[0056] Specifically, a modulation signal is generated by a main control MCU or a dedicated TOF control chip. This signal is amplified by a drive circuit and used to excite a vertical-cavity surface-emitting laser (VCSEL) or an infrared LED light source. The light emitted from these sources is processed by an optical diffusion element to form a divergent beam that can cover the entire monitoring area. To ensure that only light of specific wavelengths is used for subsequent depth information capture, a bandpass filter is used to filter out light of non-target wavelengths. The TOF sensor is responsible for receiving the reflected light and acquiring multiple phase-correlated images. Based on the phase difference or time-of-flight of light at each pixel, the depth value corresponding to each pixel can be calculated, thereby generating a corresponding depth map.

[0057] In this embodiment, the "unmanned state" refers to an empty space where no human targets are present. This application uses the depth map of the unmanned state as the empty space reference background, which can accurately eliminate static environmental interference, such as furniture and walls, thereby efficiently extracting the 3D point cloud data of dynamic human targets in the presence of people.

[0058] Furthermore, this application employs Time-of-Flight (TOF) depth maps instead of RGB color images for high-precision perception. TOF depth maps contain distance information, preserve human contours and three-dimensional geometry, and do not record sensitive visual content such as facial textures, skin tone, clothing patterns, or details of indoor objects. This design fundamentally avoids the privacy risks associated with traditional video surveillance without sacrificing behavioral recognition capabilities, significantly improving user acceptance, and is particularly suitable for sensitive home areas with extremely high privacy requirements, such as bedrooms and bathrooms.

[0059] In step S120, based on the TOF depth map of the human state, the human target in the monitoring space is detected and tracked to generate the corresponding human motion trajectory.

[0060] Please see Figure 3 The above is a flowchart illustrating the generation process of human motion trajectory according to an embodiment of this application.

[0061] like Figure 3As shown, based on the TOF depth map in the human state, the detection and trajectory tracking of human targets in the monitoring space and the generation of corresponding human motion trajectories include the following steps S121 to S125.

[0062] In step S121, the TOF depth map of the manned state is preprocessed.

[0063] In one embodiment of this application, the preprocessing includes at least one or more of the following: bad pixel removal, amplitude threshold filtering, temperature compensation, nonlinear distance correction, spatial filtering, and temporal filtering.

[0064] Specifically, the preprocessing steps include:

[0065] (1) Using the indirect TOF (iTOF) method, the modulation frequency f is set and four phases of 0°, 90°, 180° and 270° are selected for sampling to obtain the light signal intensity information under different phases.

[0066] (2) Based on the four phase data sampled, calculate the in-phase component I and the quadrature component Q respectively, where: in-phase component I = A0 - A180; quadrature component Q = A90 - A270.

[0067] (3) Based on the calculated in-phase component I and quadrature component Q, further calculate the signal amplitude Amp and phase φ, where: the signal amplitude Amp = Phase φ = atan2(Q, I).

[0068] (4) Pixels with amplitudes less than a preset threshold are considered invalid pixels and marked. This effectively eliminates inaccurate measurement results caused by noise or other interference factors.

[0069] (5) Based on the phase φ, calculate the initial distance of each pixel according to the formula d_raw = c·φ / (4πf), where c represents the speed of light.

[0070] (6) Use the factory-calibrated nonlinear correction function F(d_raw) to correct the initially calculated distance d_raw in order to compensate for the inherent nonlinear error of the hardware and obtain a more accurate distance value d_calib = F(d_raw).

[0071] (7) Considering that temperature changes may affect the accuracy of distance measurement, temperature compensation is performed according to the formula d = d_calib + k(T-T0) based on the real-time measured temperature T and the reference temperature T0, thereby adjusting the final distance d, where k is the temperature compensation coefficient.

[0072] (8) Apply spatial filtering (such as median filtering, bilateral filtering) and temporal filtering (such as exponential moving average or Kalman filtering) to the processed depth image to reduce the influence of noise and improve the spatial continuity and temporal consistency of the depth image.

[0073] In step S122, the preprocessed TOF depth map is transformed to a preset indoor coordinate system through coordinate mapping to obtain a three-dimensional point cloud.

[0074] Specifically, the coordinate mapping step includes: for each pixel (i, j) in the depth map, whose depth value is d, it can first be transformed to three-dimensional coordinates in the camera coordinate system using camera intrinsic parameters. Then, using the extrinsic parameters of the terminal's installation attitude (rotation matrix R and translation vector T), the points in the camera coordinate system are transformed to the unified indoor coordinate system.

[0075] In step S123, based on the foreground-background separation model, the three-dimensional point cloud is segmented into foreground points to obtain a foreground point set.

[0076] Specifically, a static background depth model is generated from the TOF depth map obtained in step S110 under unmanned conditions after coordinate transformation, serving as a reference benchmark for the environment. In the 3D point cloud of the current frame, the depth value of each point is compared point by point with the depth value of its corresponding position in the background model. If the absolute difference between the two exceeds a preset depth change threshold, the point is determined to be a foreground point, which may belong to a moving human body or other dynamic object.

[0077] In step S124, the foreground point set is subjected to three-dimensional clustering, and several human targets are selected by combining preset human geometric features.

[0078] Specifically, a 3D clustering algorithm based on spatial proximity can be used to group point clouds into several independent target point cloud clusters. Each cluster represents a potential dynamic entity. For each clustered target point cloud cluster, its key geometric attributes are calculated; these key geometric attributes include bounding box size, voxel volume, centroid position, and number of valid points. Based on this, preset human geometric feature constraints are introduced to filter targets in each cluster. Specific constraints include a typical human height constraint to eliminate abnormal clusters that are too high or too low; a voxel volume constraint to exclude clusters that are too small or too large; and a minimum number of valid points threshold constraint to retain targets with more than a preset threshold number of points. Through the joint screening of the above multi-dimensional geometric criteria, suspected human targets can be accurately identified from multiple clustering results.

[0079] In step S125, the human motion trajectory of each human target in the monitoring space is generated by cross-frame data association.

[0080] Specifically, for multiple suspected human targets detected in each frame, Kalman filtering or multi-target tracking algorithms are used for temporal correlation.

[0081] In step S130, based on the TOF depth map in the unmanned state, the ground reference plane of the monitoring space is fitted to obtain a ground plane model.

[0082] Specifically, the Random Sample Consensus (RANSAC) algorithm is used to perform robust plane fitting on the candidate ground point cloud.

[0083] Considering that furniture may be moved, carpets may be replaced, or terminal devices may be slightly displaced in actual use, the system supports dynamic recalibration of the ground model: when abnormal human height distribution is detected in multiple consecutive frames, or when the user actively triggers the "environment recalibration" command, the system can re-enter the empty field acquisition mode to perform a new round of unmanned state depth map acquisition and RANSAC fitting, ensuring that the ground benchmark is always consistent with the current physical environment.

[0084] In this implementation, the ground plane fitted by the TOF point cloud is used as the height reference, which can accurately determine whether the human body is in an abnormal low posture or in a fall state.

[0085] In step S140, the motion characteristics of the human target are calculated based on the human motion trajectory and the ground plane model.

[0086] Please see Figure 4 The above is a flowchart illustrating the generation of motion features of a human target according to an embodiment of this application.

[0087] like Figure 4 As shown, based on the human motion trajectory and the ground plane model, the motion characteristics of the human target are calculated as follows: within a preset sliding time window, the vertical distance from the center of mass of the human target to the ground reference plane is calculated as the height of the center of mass above the ground; the descent velocity of the center of mass of the human target relative to the ground reference plane in the vertical direction is calculated as the vertical descent rate of the center of mass; the aspect ratio of the bounding box of the human target is calculated as the human posture ratio; and the horizontal movement speed of the human target is calculated as the human motion speed.

[0088] Please see Figure 5 The above is a flowchart of rule matching according to an embodiment of this application.

[0089] like Figure 5As shown, based on predefined fall behavior discrimination rules, the rule matching of motion characteristics of human targets in the monitoring space includes: determining whether the motion characteristics of the human target meet any of the following conditions (a1) to (a3): (a1) the height of the human body's center of mass above the ground drops below a first preset threshold within a short time window, and the vertical descent rate of the center of mass exceeds a second preset threshold; (a2) the height of the human body's center of mass above the ground remains below the first preset threshold within a long time window, and the human body's motion speed remains 0; (a3) ​​the height of the human body's center of mass above the ground slowly decreases from a preset standing height to below the first preset threshold, and the human body's posture proportion conforms to the characteristics of a lying down state; if so, the human target is determined to meet the fall behavior discrimination rules; otherwise, the human target is determined not to meet the fall behavior discrimination rules.

[0090] In one embodiment of this application, the fall recognition classification model used in step S200 is a deep learning model or a gradient boosting decision tree model, wherein the deep learning model includes convolutional neural networks and recurrent neural networks.

[0091] The fall recognition classification model typically outputs a fall confidence score that is a real number between 0 and 1, with a higher value indicating a greater likelihood of a genuine fall.

[0092] In this implementation, by introducing a fall recognition classification model, the system's ability to distinguish fall behaviors is further improved through AI verification while retaining the judgment based on physical rules. This significantly enhances the accuracy, robustness, and generalization performance of fall recognition.

[0093] In one embodiment of this application, the weighted decision fusion of the rule matching score and the fall confidence score in step S300 to obtain the fall determination comprehensive score includes calculating the fall determination comprehensive score using the following formula:

[0094] Score = α·Score_rule + (1-α)·Score_AI;

[0095] In the formula, Score_rule represents the rule matching score; Score_AI represents the fall confidence score; α represents the fusion weight, 0 < α < 1; and Score represents the comprehensive score of the fall determination.

[0096] In some embodiments, the fusion weight α is a dynamic weight, which is adaptively adjusted according to at least one of the following factors: the signal quality index of the TOF depth data, the occlusion ratio of the human target, the current time information, and the spatial relationship between the human target and the preset danger zone.

[0097] In other embodiments, the threshold parameters of the fall behavior discrimination rule, the output confidence interval of the fall recognition classification model, and the fusion weights in the weighted decision fusion can be updated periodically or event-triggered based on historical fall determination results and manual confirmation results.

[0098] In other embodiments, when the proportion of effective pixels in the TOF depth map is detected to be lower than a preset threshold, the weight of the fall recognition classification model in the weighted decision fusion is reduced to improve the robustness of the determination under complex lighting or occlusion conditions.

[0099] Please see Figure 6 The diagram shows a flowchart of an embodiment of the present application for selecting an age-friendly lighting strategy.

[0100] like Figure 6 As shown, step S400, which involves dynamically selecting an age-appropriate lighting strategy from a preset strategy library based on the fall determination score, includes steps S401 to S403.

[0101] In step S401, the behavioral state of the human target is determined based on the fall determination comprehensive score; the behavioral state includes the fallen state and the normal activity state.

[0102] In step S402, if the human target is in the fallen state, the illumination of the area where the human target is located is automatically increased, the color temperature of the area is adjusted to enhance the environmental contrast, and some lights are controlled to flash at a preset frequency.

[0103] In some embodiments, when it is determined that the human target has fallen, the age-friendly lighting strategy includes the coordinated control of multiple lamps in the space where the human target is located to form a safe lighting area covering the fall location.

[0104] In step S403, if the human target is in the normal activity state, the ground-embedded guide light is activated and a continuous light strip is formed along a preset path.

[0105] In one embodiment of this application, the age-friendly healthy lighting method described in this application embodiment further includes the following steps S404 to S406.

[0106] In step S404, high fall risk locations are marked as danger zones in a preset indoor coordinate system; the high fall risk locations include stairwells, toilet entrances, and locations of fixed obstacles.

[0107] In step S405, the horizontal distance between the centroid of the human target and the danger zone is calculated.

[0108] In step S406, when the human target is detected to have entered the danger zone, the lights near the danger zone are activated and the light color of the lights is switched to a warning color, or the lights are controlled to flash at a preset frequency.

[0109] In this implementation, by defining dangerous areas, the system can accurately assess the relative spatial relationship between the human body and high-risk locations; the fall detection achieved in this way not only relies on the human body's own movement characteristics, but also integrates environmental context information, which significantly improves the foresight, scenario adaptability and practicality of the warning.

[0110] In one embodiment of this application, the age-friendly lighting strategy includes a combination of lighting parameters determined based on a model of the visual characteristics of the elderly. The combination of lighting parameters includes at least the rate of change of illuminance, the gradient of change of color temperature, and the spatial lighting distribution range, so as to suppress glare and reduce the visual adaptation burden while improving the visibility of the environment.

[0111] Please see Figure 7 The flowchart shown is a method for age-friendly and healthy lighting according to another embodiment of this application.

[0112] like Figure 7 As shown in the embodiment of this application, the age-friendly healthy lighting method further includes step S500.

[0113] In step S500, based on multi-dimensional context information, the optimal age-friendly lighting strategy is intelligently matched and executed from the preset strategy library.

[0114] Specifically, the multidimensional contextual information includes at least the fall determination score, danger zone information, current time information, and current lighting status.

[0115] The strategy library can be customized based on user habits, medical institution guidelines, or caregiver configurations, and supports OTA remote updates. By deeply integrating safety events, spatial semantics, temporal rhythms, and the current lighting environment, this application achieves a paradigm shift from "passive lighting" to "active health intervention," ensuring the safety of the elderly while also considering their physiological rhythms, visual comfort, and psychological acceptance, truly realizing age-friendly intelligent lighting that is "unobtrusive, effective, and warm."

[0116] In summary, this application deeply integrates fall detection with age-friendly lighting control strategies, constructing a closed-loop intelligent safety system encompassing "sensing—identification—environmental response—alarm." When the system detects a fall or high-risk behavior, it automatically retrieves a matching lighting scheme from a preset strategy library and adjusts the illuminance, color temperature, light color, and spatial distribution of the lamps in real time, achieving multiple safety assistance functions. This mechanism not only enables immediate environmental response to fall events but also significantly improves the initiative and timeliness of home care. Compared to traditional solutions that rely solely on sound alarms or remote notifications, this application provides an age-friendly and healthy lighting solution that combines safety, comfort, and humanistic care, while protecting privacy and eliminating the need for wearable devices.

[0117] It should be noted that the protection scope of the age-friendly healthy lighting method described in the embodiments of this application is not limited to the execution order of the steps listed in this embodiment. Any solution implemented by adding, subtracting, or replacing steps in the prior art based on the principles of this application is included within the protection scope of this application.

[0118] Please see Figure 8 The image shown is a structural schematic diagram of an intelligent lighting terminal according to an embodiment of this application.

[0119] like Figure 8 As shown in the figure, this application provides an intelligent lighting terminal, including a lamp body and an age-friendly health lighting device, wherein the age-friendly health lighting device is electrically connected to the lamp body.

[0120] Please see Figure 9 The diagram shown is a structural schematic of an intelligent lighting terminal according to another embodiment of this application.

[0121] like Figure 9 As shown, the age-friendly health lighting device includes a rule matching module, a fall recognition module, a decision fusion module, and a lighting control module.

[0122] The rule matching module is used to perform rule matching on the motion features of human targets in the monitoring space based on predefined fall behavior discrimination rules, mark the human targets that meet the fall behavior discrimination rules as candidate fall instances, and generate corresponding rule matching scores; the motion features include the height of the human body's center of gravity above the ground, the vertical descent rate of the center of gravity, the proportion of the human body's posture, and the human body's movement speed.

[0123] The fall recognition module is used to input the motion features corresponding to the candidate fall instances into a pre-trained fall recognition classification model and output a fall confidence score; the fall recognition classification model is a deep learning model or a gradient boosting decision tree model, wherein the deep learning model includes convolutional neural networks and recurrent neural networks.

[0124] The decision fusion module is used to perform weighted decision fusion on the rule matching score and the fall confidence score to obtain a comprehensive fall judgment score.

[0125] The lighting control module is used to dynamically select age-appropriate lighting strategies from a preset strategy library for lighting based on the comprehensive score of the fall determination.

[0126] It should be noted that the structure and principle of the rule matching module, fall recognition module, decision fusion module and lighting control module described in this embodiment correspond one-to-one with the steps in the above-mentioned age-friendly healthy lighting method, so they will not be repeated here.

[0127] This application integrates age-friendly health lighting devices and the lamp body into a smart lighting terminal. The integrated architecture adopted has the following technical advantages:

[0128] (1) No additional wiring or separate sensor installation is required. Traditional fall detection systems usually require separate deployment of cameras, radar or wearable base stations on walls or ceilings, while this solution embeds all sensing and execution functions into the lighting fixtures, requiring only standard lighting installation to complete the deployment.

[0129] (2) Significantly reduces system deployment costs and complexity. Users or care institutions do not need to carry out secondary construction, power supply modification or network configuration, greatly simplifying the installation process, and is especially suitable for the rapid transformation of existing residential buildings and elderly care institutions.

[0130] (3) Improve system reliability and aesthetics. The number of external connection points and independent devices is reduced, which reduces the risk of failure due to aging lines, loose interfaces or equipment disconnection; at the same time, the overall appearance is simple, with no exposed sensors or cables, which is more in line with the aesthetic requirements of home environment.

[0131] (4) Supports real-time processing and low-latency response on the device side. Sensing, recognition and lighting control are all completed locally, avoiding the latency and privacy risks caused by cloud transmission, and ensuring that lighting linkage can be triggered within hundreds of milliseconds when a fall occurs.

[0132] The age-friendly health lighting device provided in this application embodiment can realize the age-friendly health lighting method described in this application. However, the implementation device of the age-friendly health lighting method described in this application includes, but is not limited to, the structure of the age-friendly health lighting device listed in this embodiment. All structural modifications and substitutions of the prior art made based on the principles of this application are included within the protection scope of this application.

[0133] Please see Figure 10 The image shown is a schematic diagram of an age-friendly health lighting system according to an embodiment of this application.

[0134] like Figure 10As shown in the figure, this application provides an age-friendly and healthy lighting system, including:

[0135] Multiple intelligent lighting terminals are used to perform the age-friendly and healthy lighting method as described in any of the above.

[0136] A home gateway connects to each of the aforementioned smart lighting terminals via wired or wireless communication protocols.

[0137] The nursing platform is communicatively connected to the home gateway and is used to receive alarm information reported by the smart lighting terminal through the home gateway and notify the caregiver.

[0138] Specifically, the smart lighting terminal establishes a bidirectional data connection with the home gateway via Zigbee, BLE, Wi-Fi, Matter, or other IoT communication protocols to ensure efficient and stable data transmission and command response.

[0139] As the core hub of the system, the home gateway is responsible for managing communication between multiple smart lighting terminals and acts as a local data aggregation and forwarding node.

[0140] The nursing platform serves as the remote management center for the entire system, typically deployed in elderly care facilities, community service centers, or on family members' mobile devices. Its main responsibilities include: alarm information reception and processing: receiving alarm messages reported by smart lighting terminals from the home gateway and analyzing key information such as time, location, and confidence level; notifying caregivers: immediately sending alarm notifications to designated caregivers or family members upon confirmation of a fall or other emergency, which can be delivered via SMS, telephone, app push notifications, and other channels; and linking security and emergency call devices: triggering other security devices or emergency call systems according to preset strategies to assist in rapid rescue operations.

[0141] In other embodiments, the age-friendly health lighting system also includes a cloud server for centralized data management, continuous optimization of intelligent models, and remote maintenance and upgrades of the system.

[0142] It should be noted that the intelligent lighting terminal of this application possesses edge intelligence processing capabilities, enabling it to complete the entire processing chain locally, from TOF depth data acquisition, human target detection, fall behavior recognition to lighting control decisions, without relying on the cloud for real-time inference or command issuance. In this implementation, a fall event can trigger local lighting linkage within hundreds of milliseconds, significantly improving the timeliness of safety intervention; even in the event of network interruption or cloud service unavailability, core safety functions can still operate normally, avoiding monitoring failure due to communication failure; raw depth data and behavioral characteristics are processed on the terminal side, with desensitized metadata only uploaded when necessary, minimizing the risk of privacy leakage; reducing high-frequency raw data uploads optimizes network and cloud load, making it suitable for large-scale deployment scenarios.

[0143] This application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any of the preceding claims.

[0144] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, or methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of modules / units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or units 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 of apparatuses or modules or units may be electrical, mechanical, or other forms.

[0145] The modules / units described as separate components may or may not be physically separate. The components shown as modules / units may or may not be physical modules; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules / units can be selected to achieve the objectives of the embodiments of this application, depending on actual needs. For example, the functional modules / units in the various embodiments of this application may be integrated into one processing module, or each module / unit may exist physically separately, or two or more modules / units may be integrated into one module / unit.

[0146] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0147] The descriptions of the processes or structures corresponding to the above figures each have their own emphasis. For parts of a process or structure that are not described in detail, please refer to the relevant descriptions of other processes or structures.

[0148] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. An age-friendly and healthy lighting method, characterized in that, include: Based on predefined fall behavior discrimination rules, the motion characteristics of human targets in the monitoring space are matched with rules. Human targets that meet the fall behavior discrimination rules are marked as candidate fall instances, and corresponding rule matching scores are generated. The motion characteristics include the height of the human center of gravity above the ground, the vertical descent rate of the center of gravity, the proportion of the human posture, and the human movement speed. The motion features corresponding to the candidate fall instances are input into a pre-trained fall recognition and classification model, and the fall confidence score is output. The fall recognition and classification model is a deep learning model or a gradient boosting decision tree model, wherein the deep learning model includes convolutional neural networks and recurrent neural networks. The rule matching score and the fall confidence score are weighted and fused to obtain a comprehensive fall judgment score. The fusion weight is a dynamic weight, which is adaptively adjusted according to at least one of the following factors: the signal quality index of the TOF depth data, the occlusion ratio of the human target, the current time information, and the spatial relationship between the human target and the preset danger zone. Based on the comprehensive score of the fall determination, an age-appropriate lighting strategy is dynamically selected from the preset strategy library for lighting. Also includes: Under a preset indoor coordinate system, locations with high fall risk are marked as the danger zones; these locations include stairwells, restroom entrances, and the locations of fixed obstacles. Calculate the horizontal distance between the centroid of the human target and the danger zone; When a human target is detected entering the danger zone, the lights near the danger zone are activated, and the light color of the lights is switched to a warning color, or the lights are controlled to flash at a preset frequency.

2. The method according to claim 1, characterized in that, Also includes: The TOF depth maps of the monitoring space under manned and unmanned conditions were obtained respectively. Based on the TOF depth map under the human state, human targets in the monitoring space are detected and tracked to generate corresponding human motion trajectories. Based on the TOF depth map in the unmanned state, the ground reference plane of the monitoring space is fitted to obtain a ground plane model; Based on the human motion trajectory and the ground plane model, the motion characteristics of the human target are calculated.

3. The method according to claim 2, characterized in that, Based on the human motion trajectory and the ground plane model, the motion characteristics of the human target are calculated as follows: Within a preset sliding time window, the vertical distance from the center of mass of the human target to the ground reference plane is calculated and used as the height of the human center of mass above the ground. The vertical descent velocity of the human target's center of mass relative to the ground reference plane is calculated and used as the vertical descent rate of the center of mass. Calculate the aspect ratio of the bounding box of the human target as the proportion of the human pose. Calculate the horizontal speed of the human target as the human movement speed.

4. The method according to claim 2, characterized in that, Based on the TOF depth map under the presence of people, human targets in the monitoring space are detected and their trajectories are tracked to generate corresponding human motion trajectories, including: The TOF depth map under the manned state is preprocessed; the preprocessing includes at least one or more of the following: bad pixel removal, amplitude threshold filtering, temperature compensation, nonlinear distance correction, spatial filtering, and temporal filtering. The preprocessed TOF depth map is transformed to a preset indoor coordinate system through coordinate mapping to obtain a 3D point cloud; Based on the foreground-background separation model, the 3D point cloud is segmented to obtain a foreground point set; Three-dimensional clustering is performed on the foreground point set, and several human targets are selected by combining the preset human geometric features; the human geometric features include voxel volume constraints, typical human height constraints, and minimum effective point threshold constraints. By associating cross-frame data, the human motion trajectory of each human target in the monitoring space is generated.

5. The method according to claim 1, characterized in that, Based on the comprehensive score of the fall assessment, the age-appropriate lighting strategy is dynamically selected from the preset strategy library for lighting, including: Based on the fall determination score, the behavioral state of the human target is determined; the behavioral state includes the fallen state and the normal activity state. If the human target is in the fallen state, the illumination of the area where the human target is located will be automatically increased, the color temperature of the area will be adjusted to enhance the environmental contrast, and some lights will be controlled to flash at a preset frequency. If the human target is in the normal activity state, the ground-embedded guide light is activated and forms a continuous light strip along a preset path.

6. The method according to claim 1, characterized in that, Based on predefined fall behavior discrimination rules, rule matching is performed on the motion characteristics of human targets in the monitoring space, including: Determine whether the motion characteristics of the human target satisfy any of the following conditions: The height of the human body's center of gravity above the ground decreases to below a first preset threshold within a short time window, and the vertical descent rate of the center of gravity exceeds a second preset threshold. The height of the human body's center of gravity above the ground remains below a first preset threshold for an extended period of time, and the human body's movement speed remains at 0. The height of the human body's center of gravity from the ground slowly decreases from a preset standing height to below a first preset threshold, and the proportions of the human body posture conform to the characteristics of a lying position. If so, the human target is determined to satisfy the fall behavior discrimination rule; Otherwise, the human target is determined not to meet the falling behavior discrimination rule.

7. An intelligent lighting terminal, characterized in that, It includes a lamp body and an age-friendly and healthy lighting device, wherein the age-friendly and healthy lighting device is electrically connected to the lamp body; The age-friendly health lighting device mentioned above includes: The rule matching module is used to perform rule matching on the motion features of human targets in the monitoring space based on predefined fall behavior discrimination rules, mark the human targets that meet the fall behavior discrimination rules as candidate fall instances, and generate corresponding rule matching scores; the motion features include the height of the human body's center of gravity above the ground, the vertical descent rate of the center of gravity, the proportion of the human body's posture, and the human body's movement speed; The fall detection module is used to input the motion features corresponding to the candidate fall instances into a pre-trained fall detection classification model and output a fall confidence score; the fall detection classification model is a deep learning model or a gradient boosting decision tree model, wherein the deep learning model includes convolutional neural networks and recurrent neural networks; The decision fusion module is used to perform weighted decision fusion on the rule matching degree score and the fall confidence score to obtain a comprehensive fall judgment score. The fusion weight is a dynamic weight, which is adaptively adjusted according to at least one of the following factors: the signal quality index of the TOF depth data, the occlusion ratio of the human target, the current time information, and the spatial relationship between the human target and the preset danger area. The lighting control module is used to dynamically select an age-appropriate lighting strategy from a preset strategy library for lighting based on the comprehensive score of the fall determination. The age-friendly health lighting device also includes marking high fall risk locations as the danger zone in a preset indoor coordinate system; the high fall risk locations include stairwells, toilet entrances, and locations of fixed obstacles; Calculate the horizontal distance between the centroid of the human target and the danger zone; When a human target is detected entering the danger zone, the lights near the danger zone are activated, and the light color of the lights is switched to a warning color, or the lights are controlled to flash at a preset frequency.

8. An age-friendly and healthy lighting system, characterized in that, include: Multiple intelligent lighting terminals are configured to perform the method as described in any one of claims 1 to 6; A home gateway connects to each of the aforementioned smart lighting terminals via wired or wireless communication protocols; The nursing platform is communicatively connected to the home gateway and is used to receive alarm information reported by the smart lighting terminal through the home gateway and notify the caregiver.

9. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1 to 6.

Images