An old person home light environment evaluation method based on mobile phone visual perception

Abstract

The application discloses a kind of based on the visual perception of mobile phone old person home light environment evaluation method, comprising: obtaining the multi-view image data of target indoor space by mobile terminal;Multi-view image data is handled to visual perception, obtain the space characteristic information for light environment evaluation and existing lighting equipment information, according to space characteristic information, identify functional attribute area, according to ground illuminance distribution information, space characteristic information and existing lighting equipment information, correct light utilization coefficient, calculate the light illumination parameter required to be supplemented to reach preset standard of old age illumination;According to the light illumination parameter required to be supplemented, generate light supplement optimization scheme and feedback to user end.The application improves the consistency between the evaluation result of old person home light environment and actual space structure, activity area and existing lamp condition, effectively reduces the cost of light environment evaluation, realizes the accurate customization of old age light supplement scheme.

Description

Technical Field

[0001] This invention relates to the technical field of optimizing the home lighting environment for the elderly, and more particularly to a method for assessing the home lighting environment for the elderly based on mobile phone visual perception. Background Technology

[0002] As a crucial component of the home environment, the lighting environment directly impacts the visual health, activity safety, and psychological comfort of the elderly. An age-friendly lighting environment requires sufficient, uniform, and glare-free illumination, especially in key areas such as corridors, bathrooms, and bedside areas, where adequate illuminance must be ensured to prevent accidents such as falls. However, currently, most residential homes for the elderly generally suffer from insufficient illuminance and uneven lighting distribution.

[0003] Currently, home lighting environment assessment and renovation services typically rely on professionals using specialized equipment such as lux meters to conduct on-site measurements and propose renovation plans based on their design experience. However, this method is costly and unaffordable for ordinary families; it requires scheduling professional visits, which is cumbersome; and the assessment results heavily depend on the experience of the personnel, making standardization and widespread adoption difficult. With the widespread use of smartphones and the development of computer vision technology, some simple light measurement applications based on mobile phone sensors have emerged, allowing users to measure ambient illuminance using the built-in light sensors in their phones. However, existing image recognition-based indoor analysis technologies mostly focus on room layout and object recognition, and have not yet been deeply integrated with the specific needs of lighting environment assessment, especially the optimization of age-friendly lighting environments.

[0004] Therefore, there is a lack of existing technologies that can conveniently, cost-effectively and accurately assess the home lighting environment of the elderly using ordinary smartphones, and automatically generate personalized and executable modification plans. Summary of the Invention

[0005] This invention overcomes the shortcomings of the prior art and provides a method for assessing the home lighting environment of the elderly based on mobile phone visual perception.

[0006] To achieve the above objectives, the technical solution adopted by this invention is: a method for assessing the home lighting environment for the elderly based on mobile phone visual perception, comprising the following steps: S100: Acquire multi-view image data of the target indoor space via a mobile terminal; S200: Perform visual perception processing on the multi-view image data to obtain spatial feature information and existing lighting equipment information for light environment assessment; wherein, the spatial feature information includes geometric structure information, as well as spatial distribution characteristics of ground area, wall area and area related to elderly activities, room size parameters, and at least one of surface reflection characteristics and shading layout characteristics that affect indoor light utilization. S300: Identify functional attribute areas in the elderly's home environment based on the spatial feature information, and determine ground illuminance measurement points and corresponding measurement weights based on the fall risk level and activity frequency of the functional attribute areas. Provide visual guidance for the ground illuminance measurement points through a mobile terminal, and form ground illuminance distribution information based on the illuminance values ​​collected by the user. S400: Based on the ground illuminance distribution information, spatial characteristic information and existing lighting equipment information, the light utilization coefficient is corrected, the effective light flux reaching the ground area is determined based on the ground illuminance distribution information, and the total output light flux of existing light sources in the current environment is estimated based on the corrected light utilization coefficient, and then the supplementary lighting parameters required to reach the preset age-appropriate illuminance standard are calculated. S500: Generate a supplementary lighting optimization scheme based on the supplementary lighting parameters required to achieve the preset age-appropriate illuminance standard, and feed the supplementary lighting optimization scheme back to the user terminal.

[0007] In a preferred embodiment of the present invention, step S200 includes: Camera correction, pose estimation, and scale normalization are performed on the multi-view image data to obtain an indoor image sequence; the indoor image sequence is input into a visual perception model to obtain a reflectance component map, an illumination component map, and a spatial semantic segmentation map; Based on the spatial semantic segmentation map and combined with the geometric relationships between different perspectives in the multi-view image data, room size parameters are obtained; the lighting areas in the multi-view image data are identified to obtain information on existing lighting equipment. The existing lighting equipment information includes at least one of the following: existing lamp type, installation location, luminous surface size, rated power range, luminous efficacy value, and luminous flux estimate.

[0008] In a preferred embodiment of the present invention, the visual perception model includes an intrinsic image decomposition network and a multi-task joint learning network; the intrinsic image decomposition network is used to separate reflectance component map and illumination component map from an indoor image sequence; The multi-task joint learning network performs feature splicing and fusion of reflectivity component map and illumination component map in the channel dimension, and performs pixel-level semantic segmentation through encoder-decoder structure to generate spatial semantic segmentation map.

[0009] In a preferred embodiment of the present invention, the room size parameters are obtained in the following manner: The spatial semantic segmentation map is back-projected onto the 3D sparse point cloud generated by visual SLAM to construct a semantic point cloud; Three-dimensional plane fitting is performed on the ground points, wall points, and ceiling points in the semantic point cloud to construct a simplified indoor structured three-dimensional model; Detect a known-sized reference object placed by the user in the target indoor space and calculate the size factor; The room size parameters are obtained by quantizing and scaling the structured 3D model of the interior according to the size factor.

[0010] In a preferred embodiment of the present invention, determining the ground illuminance measurement points and their corresponding measurement weights includes: The ground region is determined based on the spatial feature information, and the ground region is converted into a two-dimensional plane coordinate system; Candidate measurement points are generated in the two-dimensional plane coordinate system; measurement weights are assigned to the candidate measurement points based on at least one of the following: passage area, bedside area, doorway area, rest and activity area, blind spot area, area around fixed furniture, and area near obstacles. Key measurement points are selected from the candidate measurement points according to the measurement weights; the location markers of the key measurement points are overlaid on the real-time preview screen of the mobile terminal through an augmented reality guidance interface, and the user is guided to collect illuminance values ​​at the key measurement points. Based on the collected illuminance values ​​and corresponding measurement weights, the weighted average illuminance, minimum illuminance, and illuminance uniformity of the ground area are calculated and compared with the preset age-appropriate illuminance standard to output a comprehensive judgment result.

[0011] In a preferred embodiment of the present invention, the calculation of the supplementary illumination parameters required to achieve the preset age-appropriate illumination standard includes: Based on the ground area and the weighted average illuminance, the effective luminous flux reaching the ground area is calculated; combined with the existing lighting equipment information and the corrected light utilization coefficient, the total output luminous flux of existing light sources in the current environment is estimated. Based on the target illuminance value of the preset age-friendly illuminance standard, the area of ​​the ground region, and the corrected light utilization coefficient, calculate the target total luminous flux required to achieve the preset age-friendly illuminance standard; calculate the additional luminous flux required based on the difference between the target total luminous flux and the total output luminous flux of existing light sources in the current environment. Based on the luminous efficacy value of the lamp type selected by the user or recommended by the system, the required luminous flux is converted into the required lamp power, and a supplementary lighting optimization scheme including the range of lamp power is generated.

[0012] In a preferred embodiment of the present invention, the correction of the light utilization coefficient includes: Based on the wall reflectivity characteristics, furniture layout density, and the deviation between the measured illuminance and the luminous flux estimated by existing lighting equipment, the preset initial light utilization coefficient is weighted and corrected. When the wall reflectivity characteristics indicate that the target interior space is dominated by light-colored walls or has a low furniture density, the light utilization coefficient should be increased. When the wall reflectivity characteristics indicate that the target interior space is dominated by dark walls or has a high density of furniture, the light utilization coefficient should be reduced.

[0013] In a preferred embodiment of the present invention, the method for generating the power range of the lamp includes: Calculate the supplementary power required to reach the target lower limit and target upper limit of the preset age-friendly illuminance standard, respectively, to obtain the theoretical power range; Based on the uncertainty of the corrected light utilization coefficient, a margin correction is made to the theoretical power range; Align the theoretical power range, corrected for margin, with the common power specifications of commercially available lighting fixtures to obtain the actual selectable range of lighting fixture power adjustment.

[0014] In a preferred embodiment of the present invention, the supplementary lighting optimization scheme further includes: Based on geometric structure information and spatial semantic segmentation maps, we recommend the installation location and layout of lighting fixtures; It offers a variety of optional lighting configuration options, each including the specific number and wattage of the lighting fixtures; Simulate and demonstrate the expected illuminance distribution effect after adopting the supplementary lighting optimization scheme.

[0015] A home lighting environment assessment system for the elderly based on mobile phone visual perception includes a user terminal device and a cloud server device. The user-end device includes an image acquisition module for acquiring multi-view image data of the target indoor space; Cloud server equipment includes: The light environment information recognition module is used to identify the spatial semantic segmentation map, geometric structure information and existing lighting equipment information of the target indoor space based on the multi-view image data through visual perception processing. The illuminance discrimination module is used to plan a ground illuminance measurement scheme based on the spatial semantic segmentation map and geometric structure information to guide the user's measurement, and to obtain the illuminance distribution information of the ground area based on the measurement data fed back by the user terminal device, and to make a comprehensive judgment with the preset age-friendly illuminance standard. The supplementary lighting optimization scheme generation module is used to calculate the supplementary lighting parameters required to achieve the preset age-friendly illuminance standard based on the determination result of the preset age-friendly illuminance standard, the illuminance distribution information of the ground area and the information of the existing lighting equipment, and generate a personalized supplementary lighting optimization scheme that includes specific power range and layout suggestions. The user terminal device also includes an information feedback module, which is used to receive and display the ground illuminance measurement guidance, comprehensive judgment results and supplementary lighting optimization scheme issued by the cloud server device.

[0016] This invention addresses the shortcomings of the prior art and has the following beneficial effects: This invention provides a convenient overall solution for assessing and optimizing the home lighting environment for the elderly. It automates the entire process—from environmental sensing and intelligent diagnosis to the generation of personalized modification plans—through a regular smartphone, effectively lowering the barrier to entry for age-friendly lighting environment modifications.

[0017] This invention employs a visual perception model composed of a cascaded intrinsic image decomposition network and a multi-task joint learning network. The intrinsic image decomposition network separates the reflectance component map and the illumination component map, extracting illumination-invariant reflectance features of the ground and walls. The reflectance and illumination component maps are then feature-stitched and fused before being input into the multi-task joint learning network to generate a spatial semantic segmentation map. Visual SLAM technology is combined to generate a 3D sparse point cloud. The semantic segmentation map is then back-projected to construct a semantic point cloud. After 3D plane fitting and quantization scaling with a known-size reference object, a structured 3D model with true dimensions is obtained. This addresses the problems of existing indoor visual perception technologies being severely affected by shadows, highlights, and uneven lighting, resulting in low accuracy in ground and wall recognition. This architecture improves the accuracy of key structure recognition, reduces room size parameter measurement errors, and the extracted wall reflectance features are simultaneously used for semantic segmentation and light utilization coefficient correction, improving the system's data utilization and computational efficiency.

[0018] This invention generates a weighted sampling grid based on spatial semantic segmentation maps and geometric structure information, assigning higher-than-average measurement weights to key movement paths, blind spots, and high-frequency activity areas for the elderly; it generates an augmented reality guidance interface corresponding to the weighted sampling grid, overlaying and displaying precise location markers of key grid points in the real-time preview on the user's terminal; it solves the problem that existing weighted sampling technologies only allocate weights based on illumination uniformity, without considering the activity characteristics and safety needs of the elderly, ensuring that the assessment results accurately reflect the illumination conditions of safety-critical areas.

[0019] This invention dynamically calculates the light utilization coefficient (LVC) based on room dimensions, measured illuminance, and existing luminous efficacy parameters. The LVC is weighted and corrected according to wall reflectivity characteristics and furniture density. Based on ground illuminance distribution information, the effective luminous flux reaching the ground area is determined. The total output luminous flux of existing light sources is estimated based on the corrected LVC. A supplementary lighting optimization scheme is generated by comparing this scheme with a preset age-appropriate illuminance standard. This scheme includes a power adjustment range aligned with commercially available luminous fixtures, suggested luminous fixture installation layouts, and a simulation of the illuminance distribution after supplementary lighting. This addresses the problem of existing light environment assessment technologies commonly using fixed empirical values ​​to calculate the LVC, failing to consider differences in wall color and furniture layout across different rooms, resulting in high calculation errors in supplementary lighting.

[0020] This invention adopts an edge-cloud collaborative architecture, deploying deep learning inference, 3D reconstruction, and lighting simulation tasks in the cloud, while the mobile device is responsible for image acquisition, augmented reality guidance, and result display. Existing professional lighting environment design software requires high-performance computers and professional skills, and cannot run on ordinary mobile phones. The architecture in this invention reduces the performance requirements of mobile phones, and the cloud can update the preset age-appropriate illuminance standard library, lamp parameter library, and algorithm model in real time, ensuring the timeliness and accuracy of the evaluation results. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Figure 1 This is a flowchart of the method for assessing the home lighting environment of the elderly based on mobile phone visual perception, which is the basis of this invention. Figure 2 This is a schematic diagram of the module structure of the home lighting environment assessment system for the elderly based on mobile phone visual perception, according to the present invention. Figure 3 This is a schematic diagram of the structure of the electronic device of the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein. Therefore, the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0024] Example 1: As Figure 1 As shown, a method for assessing the home lighting environment for the elderly based on mobile phone visual perception includes the following steps: S100: Acquire multi-view image data of the target indoor space via a mobile terminal; S200: Perform visual perception processing on multi-view image data to obtain spatial feature information and existing lighting equipment information for light environment assessment; wherein, the spatial feature information includes geometric structure information, as well as spatial distribution characteristics of ground area, wall area and area related to elderly activities, room size parameters, and at least one of surface reflection characteristics and shading layout characteristics that affect indoor light utilization. S300: Identifies functional attribute areas in the home environment of the elderly based on spatial feature information, and determines the ground illuminance measurement points and corresponding measurement weights based on the fall risk level and activity frequency of the functional attribute areas. It provides visual guidance for the ground illuminance measurement points through mobile terminals and forms ground illuminance distribution information based on the illuminance values ​​collected by users. S400: Based on ground illuminance distribution information, spatial characteristic information and existing lighting equipment information, the light utilization coefficient is corrected, the effective luminous flux reaching the ground area is determined based on the ground illuminance distribution information, and the total output luminous flux of existing light sources in the current environment is estimated based on the corrected light utilization coefficient, and then the supplementary lighting parameters required to reach the preset age-appropriate illuminance standard are calculated. S500: Generates a supplementary lighting optimization plan based on the supplementary lighting parameters required to achieve the preset age-appropriate illuminance standard, and feeds the supplementary lighting optimization plan back to the user.

[0025] This invention faces several challenges in its application: It requires the stable and accurate separation of mixed information of light and material reflection from single or small numbers of mobile phone photos in complex home environments, overcoming interference from shadows, highlights, and uneven lighting; how to invert the true illuminance distribution in three-dimensional space based solely on the brightness information of two-dimensional images obtained from a mobile phone camera, and establish a reliable calibration model that integrates with ambient light sensor readings; how to transform the identified spatial structure, existing light sources, and other discrete information into a calculable physical model of light propagation; and how to combine theoretical calculation results with the specifications of commercially available lighting fixtures, installation conditions, and the actual movement patterns of the elderly when generating practical supplementary lighting solutions.

[0026] The following will describe in detail the method for assessing the home lighting environment for the elderly based on mobile phone visual perception, in conjunction with the accompanying drawings.

[0027] Step S100: Acquire multi-view image data of the target indoor space through a mobile terminal.

[0028] The target indoor space is a home room or continuous activity area that requires a light environment assessment, including at least one of the following: bedroom, living room, corridor, bathroom entrance area, and kitchen / dining area. The multi-view image data is a sequence of images acquired by the mobile terminal from different shooting positions and directions. The image sequence includes RGB images covering the ground, walls, ceiling, furniture, doors and windows, lighting fixtures, and areas related to elderly activity.

[0029] When acquiring multi-view image data, the mobile terminal records image frames, shooting time, exposure parameters, focal length parameters, and inertial sensor data. The multi-view image data is obtained by extracting frames from continuous video or is composed of multiple photos. There are overlapping areas between adjacent view images. Corner points, edge points, or texture points in the overlapping areas are used for subsequent pose estimation and 3D point cloud generation.

[0030] In one embodiment, the mobile terminal application displays a coverage prompt in the real-time preview screen. The prompt range includes the four corners of the room, the ground boundary, the doorway, the bedside, the area around the sofa, the area around the dining table, the window position, and the position of the light fixture. The mobile terminal application provides prompts for blurry images, overexposed images, underexposed images, and images with insufficient overlap, and retains image frames that meet the sharpness threshold and overlap threshold as the original data of the indoor image sequence.

[0031] Step S200: Perform visual perception processing on the multi-view image data to obtain spatial feature information and existing lighting equipment information for light environment assessment.

[0032] Spatial feature information includes geometric structure information, as well as at least one of the following: spatial distribution features of ground area, wall area and area related to elderly activities, room size parameters, surface reflection features and occlusion layout features. The spatial semantic segmentation map is a pixel-level expression of spatial distribution features. The category labels in the spatial semantic segmentation map include at least several of the following: ground, wall, ceiling, door, window, bed, sofa, table, chair, cabinet, lamp, passage area and obstacle.

[0033] Camera calibration, pose estimation, and scale normalization are performed on multi-view image data to obtain an indoor image sequence. Camera calibration includes intrinsic parameter calibration and distortion correction. Intrinsic parameter calibration yields focal length, principal point coordinates, and distortion parameters. Distortion correction maps the original image to a calibrated image that satisfies the pinhole camera model. Pose estimation is based on feature point matching results in adjacent view images, calculating the camera rotation matrix and translation vector corresponding to each image frame. Scale normalization converts images with different shooting distances, image resolutions, and exposure conditions to a uniform input size and a uniform numerical range.

[0034] The geometric relationships between different viewpoints in multi-view image data include the correspondence between feature points in adjacent image frames, epipolar constraints, camera pose transformations, and correspondence between points in three-dimensional space. For the t-th frame, let the camera intrinsic parameter matrix be K and the rotation matrix be R. t The translation vector is t t Space point X j Projected onto image coordinates p tj The following equation must be satisfied: p tj =K·[Rt |t t ]·X j ; Where, p tj Let R be the homogeneous pixel coordinates of the j-th spatial point in the t-th frame, K be the camera intrinsic parameter matrix, and R be the pixel coordinates of the j-th spatial point. t Let t be the camera rotation matrix for frame t. t Let X be the camera translation vector for frame t. j These are the homogeneous coordinates of a point in three-dimensional space; geometric relationships are used to unify the ground area, wall area, and lighting area identified from different perspectives into the same indoor spatial coordinate system.

[0035] After the indoor image sequence is input into the visual perception model, reflectance component maps, illumination component maps, and spatial semantic segmentation maps are obtained. The visual perception model includes an intrinsic image decomposition network and a multi-task joint learning network. The input to the intrinsic image decomposition network is the normalized RGB image I. t The output is a reflectivity component map R. t And illumination component diagram L t The reflectance component map represents the inherent color and texture of surfaces such as floors, walls, and furniture; the illumination component map represents the intensity of illumination, shadows, and local dark areas in the image.

[0036] The intrinsic image decomposition network comprises a decomposition encoder, a reflectance decoder, and an illumination decoder. The decomposition encoder includes multiple levels of convolutional layers and downsampling layers to extract edge, texture, and brightness variation features from indoor image sequences. The reflectance decoder outputs a reflectance component map with the same spatial size as the input image. The illumination decoder outputs a single-channel or three-channel illumination component map. In a linear color space, the input image, reflectance component map, and illumination component map satisfy the following equation: I t (x,y,c)=R t (x,y,c)·L t (x,y)+ε; Among them, I t (x,y,c) represents the pixel value of the t-th frame image located at pixel coordinates x, y and color channel c in the linear color space, R t (x,y,c) represents the reflectance components, L t (x,y) represents the illumination component, and ε is a preset small value to prevent the value from being zero.

[0037] The multi-task joint learning network performs feature splicing and fusion of reflectance component map and illumination component map in the channel dimension. The fused features are input into the encoder-decoder structure. The encoder extracts semantic features through convolutional layer, normalization layer and non-linear activation layer. The decoder restores spatial resolution through upsampling layer and skip connection. The segmentation head outputs the classification score of each pixel to each semantic category through 1×1 convolution. After normalization processing, the spatial semantic segmentation map is obtained.

[0038] The spatial semantic segmentation map is a label map or probability map whose size corresponds to the input image. Each pixel in the label map corresponds to a semantic category, and each pixel in the probability map corresponds to the probability value of multiple semantic categories. The spatial semantic segmentation map is used to determine the spatial distribution characteristics of ground areas, wall areas, ceiling areas, door and window areas, furniture areas, lighting areas, and areas related to the activities of the elderly.

[0039] The training data for the visual perception model includes publicly available indoor scene image datasets, authorized and anonymized home interior images, and synthetic images rendered from indoor 3D scenes. Before training, the authorized home interior images have faces, ID cards, photos, screen content, and personally identifiable information removed. The training data does not contain biometric labels used to identify individuals. The annotations in the training data include pixel-level labels for the ground, walls, ceiling, doors, windows, beds, sofas, tables, chairs, cabinets, lamps, passageways, and obstacles, as well as lamp area boxes, lamp type labels, and labels for reference areas of known dimensions.

[0040] During the training of the visual perception model, the input images are uniformly scaled to 512×512 pixels or 768×768 pixels, and the RGB values ​​are normalized to 0 to 1. The training process uses random cropping, brightness perturbation, color temperature perturbation, shadow enhancement, motion blur and perspective transformation to augment the data, so as to cover indoor low light, local strong light and furniture occlusion scenes.

[0041] When the intrinsic image decomposition network is trained using ground truth or weakly supervised methods without reflectance, the loss function includes reconstruction loss, illumination smoothing loss, and reflectance consistency loss, satisfying the following equation: L de =λ1·L rec +λ2·L s h+λ3·L ref ; Among them, L de L represents the total loss of the intrinsic image decomposition network. rec L1 loss is the difference between the input image and the reconstructed image based on the reflectance component map and the illumination component map. s For edge-aware smoothing loss of illumination component maps, L refThe consistency loss between reflectance component maps obtained under different brightness perturbations for the same indoor scene; λ1, λ2 and λ3 are loss weights, which are 1.0, 0.2 and 0.5 respectively in one embodiment.

[0042] The multi-task joint learning network employs pixel-level semantic segmentation training. Its loss function includes cross-entropy loss, Dice loss, and boundary loss, satisfying the following equation: L seg =L ce +0.5·L d +0.1·L b ; Among them, L seg For semantic segmentation loss, L ce For cross-entropy loss, L d To handle class imbalance in Dice loss, L b The Adam optimizer is used to train the boundary loss mechanism, which constrains the boundaries of the ground, walls, furniture, and lighting fixtures, with a learning rate of 1×10⁻⁶. -4 Up to 3×10 -4 Batch size is 8 to 16, training epochs are 80 to 150, and weight decay coefficient is 1×10. -4 .

[0043] In one embodiment, the intrinsic image decomposition network adopts a U-Net-type encoder-decoder structure. The decomposition encoder includes four convolutional modules, each outputting 64, 128, 256, and 512 channel feature maps sequentially. Each convolutional module includes two 3×3 convolutional layers, a normalization layer, and a ReLU activation layer, and is passed to the next level through a 2x downsampling. The reflectance decoder and illumination decoder recover spatial resolution step by step through bilinear upsampling or deconvolution, and are connected to the corresponding coding layer features in a skip connection, respectively outputting a three-channel reflectance component map and a single-channel or three-channel illumination component map. Specifically, the visual perception model adopts a cascaded architecture, in which the intrinsic image decomposition network serves as the front-end module, and the output reflectance features are fed to the feature fusion layer of the multi-task joint learning network. The feature fusion layer uses an attention mechanism to weight the spatial correlation between the reflectance component and the illumination component, enhancing the semantic recognition accuracy in shadow areas.

[0044] The multi-task joint learning network concatenates the reflectance component map and the illumination component map in the channel dimension and inputs them into the semantic segmentation encoder. The semantic segmentation encoder can adopt U-Net, DeepLabv3+ or its equivalent encoder-decoder structure. The segmentation head outputs pixel-level classification results of the ground, walls, ceiling, doors and windows, furniture, lamps, passage areas and obstacles through 1×1 convolution.

[0045] Lighting fixture region recognition can be obtained from the lighting fixture category segmentation results in a multi-task joint learning network, or from a separate lighting fixture detection branch. The lighting fixture detection branch takes candidate region image patches from multi-view images as input and outputs the lighting fixture region bounding box, lighting fixture type, estimated luminous surface size, and recognition confidence. Lighting fixture types include several types from ceiling lights, downlights, spotlights, light strips, table lamps, floor lamps, and wall lamps. The lighting fixture detection branch is trained using lighting fixture region bounding box regression loss, lighting fixture type classification loss, and luminous surface size regression loss.

[0046] Existing lighting equipment information is obtained through luminaire area recognition results and a luminaire parameter database. The luminaire parameter database records the rated power range, luminous efficacy range, luminous surface size range, and typical luminous flux range corresponding to different luminaire types. If the confidence level of luminaire image recognition is higher than a preset confidence threshold, the system queries the luminaire parameter database for the estimated rated power range, luminous efficacy, and luminous flux based on the luminaire type and luminous surface size. If the recognition confidence level is lower than a preset confidence threshold, the system displays candidate luminaire types on the user terminal device for user confirmation. The confirmation result and the visual recognition result together determine the existing lighting equipment information.

[0047] The room size parameters are obtained based on the spatial semantic segmentation map and the geometric relationships between different views in the multi-view image data. Visual SLAM generates a 3D sparse point cloud based on the feature point matching results and pose estimation results. The spatial semantic segmentation map is back-projected onto the 3D sparse point cloud according to the camera pose and camera intrinsic parameters to form a semantic point cloud containing 3D coordinates and semantic labels.

[0048] Ground points, wall points, and ceiling points in the semantic point cloud are used as candidate points for structural plane fitting. The system uses a random sampling consensus algorithm to fit the three-dimensional plane and uses semantic labels to constrain the sampling range, excluding points corresponding to beds, sofas, tables, chairs, cabinets, and other obstacles. The fitted ground plane, wall plane, and ceiling plane constitute a simplified indoor structured three-dimensional model.

[0049] The user places a known-size reference object in the target indoor space to calculate the size factor. This known-size reference object can be a calibration board with known side lengths, A4 paper, a card with preset side lengths, or a mobile terminal screen. The system detects the region of the known-size reference object in the image and combines it with its reconstructed length L in the structured 3D model. model and the actual length L real Calculate the size factor s, satisfying the following formula: s=L real / L model ; Where s is the size factor, L real The actual length of a reference object of known size, in meters (m), is L. modelThe length of a known-size reference object in an unscaled, structured 3D indoor model is given. The system applies the size factor s to the 3D coordinates of the structured 3D indoor model to obtain a scaled 3D indoor model.

[0050] The specific process for calculating the size factor s is as follows: First, the system uses the Canny operator to extract the edge pixels of a reference object of known size in the image, obtaining its vertex coordinates on the two-dimensional image plane. Combining this with the camera intrinsic parameter matrix and the camera pose estimated in real-time by visual SLAM, the pixel region of the reference object is back-projected onto the SLAM-reconstructed ground plane using homography transformation. By calculating the ratio of the virtual length after back-projection to the preset physical length of the reference object, the scale recovery coefficient is obtained, eliminating perspective distortion errors caused by the tilt of the mobile phone shooting angle and ensuring the absolute accuracy of room size measurement.

[0051] The room size parameters include at least one of the following: length, width, height, floor area, main passage width, bedside area width, doorway area width, and installation height of light fixtures from the floor; the length, width, and height are obtained from the distance between the quantified structural planes, and the floor area is obtained from the area of ​​the boundary polygon on the floor plane.

[0052] Surface reflectance features are obtained jointly from the reflectance component map and the spatial semantic segmentation map. For the wall area, the system obtains the wall pixel set from the spatial semantic segmentation map and calculates the average reflectance feature ρ of the wall area in the reflectance component map. wall The following equation is satisfied: ρ wall =mean(R t (x,y,c)),(x,y)∈M wall ; Where, ρ wall M represents the reflectivity characteristics of the wall surface. wall R is the set of pixels in the wall area. t (x,y,c) represents the reflectivity components corresponding to the pixels on the wall. ρ wall The value of ρ is normalized to between 0 and 1. wall When the reflectivity is greater than 0.65, the wall reflectivity characteristic indicates that the target interior space is mainly composed of light-colored walls; ρ wall When the reflectance is less than 0.35, the wall reflectance characteristic indicates that the target interior space is dominated by dark-colored walls; ρ wall When the reflectance is between 0.35 and 0.65, the wall reflectance characteristics indicate that the target indoor space is mainly composed of medium reflective walls.

[0053] The occlusion layout characteristics are derived from the spatial relationships between furniture areas, obstacle areas, and the ground area. The system adds up the projected areas of beds, sofas, tables, chairs, cabinets, and other obstacles on the ground plane and calculates the ratio with the ground area to obtain the furniture layout density D, satisfying the following formula: D=A o / A floor ; Where D is the furniture layout density, A o Let A be the projected area of ​​the furniture and obstacles on the ground plane. floor The area represents the ground area, and the unit is m². The furniture layout density D is used to characterize the impact of furniture blocking and limited passage space on the propagation of indoor light.

[0054] Step S300: Identify functional attribute areas in the elderly's home environment based on spatial feature information, and determine ground illuminance measurement points and corresponding measurement weights based on the fall risk level and activity frequency of the functional attribute areas. Provide visual guidance for the ground illuminance measurement points through a mobile terminal, and form ground illuminance distribution information based on the illuminance values ​​collected by the user.

[0055] The functional attribute areas include at least one of the following: passage area, bedside area, doorway area, rest and activity area, lighting dead zone area, area around fixed furniture, and area near obstacles. The passage area is defined by the passable strip area connecting the doorway, bedside, sofa, dining table, and bathroom entrance in the ground area; the bedside area is obtained by expanding the bed semantic area outward by 0.4m to 0.8m; the doorway area is obtained by expanding the door semantic area inward and outward by 0.5m to 1.0m; the rest and activity area is obtained by the area around the sofa, dining table, desk, and reading chair; the lighting dead zone is jointly determined by the low brightness area in the illumination component diagram, the area far from the light fixture, and the area blocked by furniture. When the measured illuminance at the key measurement point corresponding to the lighting dead zone is lower than the preset minimum illuminance threshold, the lighting dead zone is judged as a local dark area; the local dark area is used for subsequent local supplementary luminous flux calculation.

[0056] The system determines the ground area based on spatial feature information and converts the ground area into a two-dimensional plane coordinate system. The two-dimensional plane coordinate system takes the ground plane as the reference plane, takes one side of the ground boundary polygon or the main direction of the room as the x-axis, and takes the direction perpendicular to the x-axis as the y-axis. Each three-dimensional point in the ground area is projected onto the two-dimensional plane coordinate system to obtain plane coordinates.

[0057] Candidate measurement points are generated in a two-dimensional plane coordinate system. The spacing between candidate measurement points can be from 0.5m to 1.0m, or adaptively determined according to the area of ​​the ground region. When the ground region area is less than 8m², the number of candidate measurement points is no less than 4; when the ground region area is between 8m² and 20m², the number of candidate measurement points is no less than 6; when the ground region area is greater than 20m², the number of candidate measurement points is no less than 9. When a candidate measurement point falls into the furniture projection area or an impassable area, the system moves it to an adjacent accessible ground area or deletes the candidate measurement point.

[0058] Each candidate measurement point is assigned a measurement weight. The measurement weight is determined by the fall risk level, activity frequency level, and blind spot status of the functional attribute area where the candidate measurement point is located. The fall risk level (r) ranges from 1 to 3, with higher values ​​indicating a higher fall risk. The activity frequency level (f) ranges from 1 to 3, with higher values ​​indicating a higher activity frequency. The blind spot marker (z) is either 0 or 1, with z = 1 indicating that the candidate measurement point is located in a blind spot. Measurement weight (w) i Satisfy the following formula: w i =1+0.25·r i +0.15·f i +0.30·z i ; Among them, w i Let r be the measurement weight of the i-th candidate measurement point. i f represents the fall risk level corresponding to the i-th candidate measurement point. i z represents the activity frequency level corresponding to the i-th candidate measurement point. i Mark the lighting blind spot corresponding to the i-th candidate measurement point. The fall risk level of passage areas, bedside areas, and doorway areas is higher than that of ordinary floor areas; the activity frequency level of staying and active areas is higher than that of inactive areas; areas whose brightness in the illumination component map is lower than the 20th percentile of the brightness of the floor area in the same room can be marked as lighting blind spots.

[0059] The system determines the corresponding fall risk level and activity frequency level based on the functional attribute area to which the candidate measurement point belongs. Specifically, when a candidate measurement point is located in a passable strip area connecting a doorway, bedside, sofa, dining table, or bathroom entrance, it is identified as a passable area, and the fall risk level r is determined to be 3, and the activity frequency level f is determined to be 3. When a candidate measurement point is located in the ground area extending 0.4m to 0.8m outward from the bed semantic area, it is identified as a bedside area, and the fall risk level r is determined to be 3, and the activity frequency level f is determined to be 2. When a candidate measurement point is located in the ground area extending 0.5m to 1.0m inward from the door semantic area, it is identified as a doorway area, and the fall risk level r is determined to be 3, and the activity frequency level f is determined to be 2. The fall risk level r is set to 3, and the activity frequency level f is set to 2. When the candidate measurement point is located in the area around a sofa, dining table, desk, or reading chair, it is identified as a resting activity area, and the fall risk level r is set to 2, and the activity frequency level f is set to 3. When the candidate measurement point is located in a normal ground area that is walkable and does not fall into the above areas, it is identified as a normal walkable ground area, and the fall risk level r is set to 1, and the activity frequency level f is set to 1. When the candidate measurement point is located in a walkable area less than a preset distance from the furniture projection boundary, it is identified as the area around fixed furniture or near an obstacle, and the fall risk level r is set to 2, and the activity frequency level f is set to 1.

[0060] The system selects key measurement points from candidate measurement points according to measurement weights. The selection rules include weight threshold selection and spatial coverage constraints. Weight threshold selection retains candidate measurement points whose measurement weights are greater than the preset weight threshold. Spatial coverage constraints ensure that key measurement points cover different orientations of the ground area and include at least one measurement point in a normal ground area and one measurement point in a high-risk functional attribute area.

[0061] The mobile terminal provides visual guidance for key measurement points through an augmented reality interface. The system converts key measurement points in the two-dimensional plane coordinate system into three-dimensional ground coordinates and projects them onto the real-time preview screen based on the current camera pose of the mobile terminal to form positioning marks. The user terminal device overlays the positioning marks, distance prompts, and acquisition status prompts on the real-time preview screen. After the user moves the mobile terminal to the ground position corresponding to the positioning mark, the illumination value is collected.

[0062] Illuminance values ​​can be read by the ambient light sensor of the mobile terminal or by an external illuminance acquisition device connected to the mobile terminal. When the ambient light sensor of the mobile terminal acquires illuminance values, the system records the sensor reading, acquisition time, device posture, and key measurement point number. When the device posture deviates from a horizontal state by more than a preset angle threshold, the system prompts for re-acquisition. When the external illuminance acquisition device acquires illuminance values, the system records the illuminance value returned by the device and the key measurement point number.

[0063] In one embodiment, when a user collects illuminance values ​​at a key measurement point, the mobile terminal is placed at a distance of 0m to 0.05m from the ground, or an external illuminance acquisition device connected to the mobile terminal is placed at the ground position corresponding to the key measurement point. When using a mobile terminal ambient light sensor, the sensor's light-receiving surface faces upwards indoors, and the angle between the mobile terminal's posture and the horizontal plane of the ground is no greater than 10°. The system continuously collects at least three ambient light sensor readings, discards abnormal readings that deviate from the median value by more than 20%, and takes the average of the remaining readings as the original illuminance value.

[0064] Based on the collected illuminance values ​​and corresponding measurement weights, the system calculates the weighted average illuminance E of the ground area. w Minimum Illumination E min The illuminance uniformity U satisfies the following formula: E w =Σ(w i ·E i ) / Σw i E min =min(E i ); U=E min / E w Among them, E i Let represent the illuminance value at the i-th key measurement point, in lx and w. i E represents the measurement weight of the i-th key measurement point. w This is a weighted average illuminance, expressed in lx (E). min The minimum illuminance is expressed in lx, and U represents the illuminance uniformity.

[0065] A preset age-friendly illuminance standard library stores the target illuminance lower limit, target illuminance upper limit, minimum illuminance threshold, and illuminance uniformity threshold corresponding to different room types and functional attribute areas. In one embodiment, the target illuminance range for ordinary activity floor areas is 100lx to 150lx, the minimum illuminance threshold for passage areas and bedside areas is 50lx, and the illuminance uniformity threshold is 0.4. The system compares the weighted average illuminance, minimum illuminance, and illuminance uniformity with the preset age-friendly illuminance standards and outputs a comprehensive judgment result of at least one of the following: meeting the standard, insufficient illuminance, excessive illuminance, insufficient uniformity, and local dark areas.

[0066] Step S400: Based on the ground illuminance distribution information, spatial characteristic information and existing lighting equipment information, the light utilization coefficient is corrected, the effective luminous flux reaching the ground area is determined based on the ground illuminance distribution information, and the total output luminous flux of existing light sources in the current environment is estimated based on the corrected light utilization coefficient, and then the supplementary lighting parameters required to reach the preset age-appropriate illuminance standard are calculated.

[0067] The light utilization coefficient k represents the proportion of the luminous flux output by the existing light source or supplementary light source that ultimately reaches the ground area and participates in the formation of ground illuminance. k is a dimensionless parameter with a value range of 0 to 1. The preset initial light utilization coefficient k0 is obtained by querying the parameter library based on the room type, lamp installation method, and room index. In one embodiment, k0 for ordinary ceiling lights in bedrooms and living rooms is 0.45 to 0.65, and k0 for wall lamps or local table lamps is 0.25 to 0.45.

[0068] The system performs a weighted correction on the preset initial light utilization coefficient based on the wall reflectivity characteristics, furniture layout density, and the deviation between the measured illuminance and the luminous flux estimated by the existing lighting equipment; the luminous flux Φ estimated by the existing lighting equipment... lamp The estimated rated power P of each existing luminaire j and luminous efficacy η j We obtain the following equation: Φ lamp =Σ(P j ·η j ); Where, Φ lamp Estimate the luminous flux of existing lighting equipment, in lm and P. j The rated power estimate for the j-th existing luminaire is given in W, η. j Let be the luminous efficacy value of the j-th existing luminaire, in lm / W.

[0069] Current effective luminous flux Φ reaching the ground area eff It is obtained from the ground area and the weighted average illuminance, and satisfies the following formula: Φ eff =E w ·A floor ; Where, Φ eff The current effective luminous flux is expressed in lm (E). w This is a weighted average illuminance, expressed in lx (A). floor This represents the area of ​​the ground region, expressed in m².

[0070] Actual measured conversion light utilization coefficient k meas The luminous flux is estimated from the current effective luminous flux and existing lighting equipment, satisfying the following formula: k meas =Φ eff / Φ lamp ; When Φ lamp When the threshold is less than the preset lower limit or the confidence level of existing lighting equipment is lower than the preset confidence threshold, the system reduces k. meas The weights are being adjusted, and users are required to confirm the existing lighting parameters.

[0071] Corrected light utilization coefficient kcorr Satisfy the following formula: k corr =clamp(k0·[1+a·(ρ wall -0.5)-b·(D-0.3)]+c·(k meas -k0),k min ,k max ); Where, k corr ρ is the corrected light utilization coefficient. wall Here, D represents the wall reflectivity characteristic, a is the furniture layout density, b is the wall reflectivity correction factor, c is the measured deviation correction factor, and k is the wall reflectivity correction factor. min and k max These represent the lower and upper limits of the light utilization factor, respectively; `clamp` indicates that the calculated value is limited between the lower and upper limits; in one embodiment, `a` is 0.20 to 0.35, typically 0.28; `b` is 0.15 to 0.30, typically 0.22; `c` is 0.30 to 0.50, typically 0.40; `k` min Take 0.20, k max Take 0.85; the wall reflectivity characteristic characterizes light-colored walls or low furniture density, k corr Compared to k0, it increases; the wall reflectivity characteristics indicate that when the wall is dark or the furniture layout density is high, k... corr It decreases relative to k0.

[0072] The values ​​of correction coefficients a, b, and c were determined as follows: 100 standard home scene models were pre-constructed using lighting simulation software, each containing different wall reflectivities ranging from 0.1 to 0.9 and furniture layout densities ranging from 0 to 0.8. The theoretical utilization coefficients for each scene were calculated. The measured data from mobile phones were compared with the simulated values, and multiple linear regression fitting was performed using the least squares method to determine the optimal empirical values ​​for a, b, and c. a is used to compensate for the contribution of secondary reflection from the wall, and b is used to compensate for the illuminance loss caused by furniture blocking, so that the empirical model can adapt to complex real physical environments.

[0073] Based on existing lighting equipment information and the corrected light utilization coefficient, the system estimates the total output luminous flux Φ of existing light sources in the current environment. src The total output luminous flux of existing light sources can be estimated using existing lighting equipment. lamp Calculated output luminous flux Φ eff / k corr The weighted average yields the following equation: Φ src =β·Φ lamp +(1-β)·Φ eff / kcorr ; Where, Φ src The total output luminous flux of existing light sources in the current environment is expressed in lm. β is the confidence weight for luminaire identification, ranging from 0 to 1. When the confidence of luminaire type, rated power range, and luminous efficacy value identification is high, β is 0.6 to 0.8. When the confidence of luminaire identification is low, β is 0.3 to 0.5.

[0074] Target total luminous flux Φ target The target illuminance value E of the preset age-appropriate illuminance standard target Ground area A floor and the corrected light utilization coefficient k corr We obtain the following equation: Φ target =E target ·A floor / k corr ; Where, Φ target The target total luminous flux required to achieve the preset age-appropriate illuminance standard is expressed in lm (E). target The target illuminance value is expressed in lx; when the preset age-appropriate illuminance standard gives a lower limit E of the target illuminance. low When the target illuminance upper limit Ehigh is reached, the system calculates the corresponding target total luminous flux lower limit and target total luminous flux upper limit, respectively.

[0075] The required additional luminous flux Φ sup The total luminous flux of the target light source is obtained by the difference between the total output luminous flux of the light source in the current environment and the total output luminous flux of the light source in the current environment, satisfying the following formula: Φ sup =max(0,Φ target -Φ src ); Where, Φ sup The required luminous flux is expressed in lm. When Φ target Less than or equal to Φ src And when the overall judgment result does not include local dark areas, Φ sup Set to 0; when the comprehensive judgment result includes local dark areas, the system calculates the local supplementary luminous flux according to the difference between the area corresponding to the local dark area and the target illuminance.

[0076] The luminous efficacy value η is based on the type of lamp selected by the user or recommended by the system. new The required luminous flux is converted into the required lamp power P. sup The following equation is satisfied: P sup =Φ sup / η new ; Among them, P supThe required additional wattage for the lighting fixtures is expressed in watts (W) and in η. new To supplement the luminous efficacy value of the luminaire, the unit is lm / W. The supplementary luminous efficacy value of the luminaire is provided by the luminaire parameter library or determined by the luminaire parameters entered by the user.

[0077] The method for generating the luminaire power range includes calculating the supplementary power required to achieve the target lower limit and target upper limit of the preset age-appropriate illuminance standard, respectively, to obtain the theoretical power range; the lower limit P of the theoretical power range. low and upper limit P high Satisfy the following formula: P low =max(0,E low ·A floor / k corr -Φ src ) / η new ; P high =max(0,E high ·A floor / k corr -Φ src ) / η new ; Among them, E low To preset the lower limit of the age-appropriate illuminance standard, E high The upper limit of the illuminance standard for aging is preset, and the unit is lx. The system makes a margin correction to the theoretical power range based on the uncertainty of the corrected light utilization coefficient. The uncertainty of the light utilization coefficient comes from semantic segmentation error, three-dimensional scale error, lamp recognition confidence and illuminance acquisition error.

[0078] In one embodiment, when the furniture layout density D is less than 0.20 and the lamp recognition confidence level is higher than 0.80, the margin coefficient μ is 0.05 to 0.10; when the furniture layout density D is 0.20 to 0.50 or the lamp recognition confidence level is 0.60 to 0.80, the margin coefficient μ is 0.10 to 0.15; when the furniture layout density D is greater than 0.50 or the lamp recognition confidence level is lower than 0.60, the margin coefficient μ is 0.15 to 0.20; the power range after margin correction satisfies the following formula: P range =[P low ·(1-μ),P high ·(1+μ)]; The system aligns the theoretical power range, corrected for margin, with common power specifications of commercially available lighting fixtures to obtain the actual selectable lighting fixture power adjustment range. Common power specifications can include several from 3W, 5W, 7W, 9W, 12W, 18W, 24W, 36W, and 48W. During alignment, the system selects a power range no lower than P. rangeThe smaller adjacent specification of the lower limit is used as the actual power lower limit, and it is selected to be no less than P. range The larger of the adjacent specifications is used as the actual power limit; when P range When the lower limit is less than the minimum common power specification, the actual power lower limit shall be the minimum common power specification.

[0079] Common light source types are shown in Table 1: Table 1 - Light Source Types Incandescent lamp 10-15 500-700 Inefficient and energy-intensive, not recommended for use in large spaces. halogen lamp 17-33 227-441 It has good color rendering, but still consumes a lot of power, making it suitable for small areas. Fluorescent lamp (CFL) 50-70 107-150 Medium energy efficiency, suitable for general lighting. High-frequency HID 90-120 63-83 Highly efficient, suitable for industrial or outdoor use, but more expensive. LED lights 100-150 50-75 High efficiency, energy saving, and long lifespan make it the preferred solution. Step S500: Generate a supplementary lighting optimization plan based on the supplementary lighting parameters required to achieve the preset age-appropriate illuminance standard, and feed the supplementary lighting optimization plan back to the user terminal.

[0080] The supplementary lighting optimization scheme includes at least one of the following: lamp power range, number of lamps, lamp type, installation location, layout, and expected illuminance distribution effect after adopting the supplementary lighting optimization scheme. The system determines the installable and unsuitable areas based on geometric structure information and spatial semantic segmentation map. Installable areas include areas with unobstructed ceilings, areas on walls where lamps can be placed, and areas on furniture surfaces where lamps can be placed. Unsuitable areas include window areas, areas where doors swing, areas where cabinet doors open, areas above low obstacles, and areas that are too close to existing lamps.

[0081] Recommended lighting installation locations target functional areas with high weight and insufficient illumination. For passageways and doorways, the system prioritizes linear lighting fixtures or multiple point lighting fixtures along the passageway. For bedside areas, the system prioritizes low-glare wall lamps, bedside lamps, or indirect lighting fixtures. For areas where people stay or move around, the system prioritizes local lighting fixtures. For blind spots, the system recommends locations near the edge of the dark area that are not obstructed by furniture.

[0082] Multiple optional lighting configuration schemes are generated based on the same supplemental luminous flux target. Each configuration scheme includes the specific number of lighting fixtures, single lamp power, total power, recommended installation location and layout. The system generates a small number of high-power lighting fixture schemes, a multi-point low-power lighting fixture scheme, and a local supplemental lighting fixture scheme, and matches each scheme with the actual selectable lighting fixture power adjustment range.

[0083] The expected illuminance distribution effect is obtained through a simplified lighting model or an offline lighting simulation model. In one embodiment, the system establishes an illuminance calculation grid in a two-dimensional plane coordinate system of the ground area, calculates the illuminance contribution of existing lamps and supplementary lamps to each grid point, and superimposes them to form the expected illuminance distribution map after supplementary lighting; for the q-th grid point, the illuminance contribution E generated by the supplementary lamp is... q Satisfy the following formula: E q =Σ[Φ m ·k corr·g(θ qm ) / (4π·d qm 2 )]; Among them, E q The illuminance contribution of the q-th grid point, in lx, Φ m The luminous flux of the m-th supplementary luminaire is expressed in lm or d. qm The distance from the m-th supplementary light fixture to the q-th grid point is in meters, g(θ) qm ) is the emission angle correction function, θ qm Let g(θ) be the angle between the principal optical axis direction of the m-th supplementary luminaire and the vector pointing from that luminaire to the q-th grid point; for diffuse luminaires, g(θ) qm ) can be taken as 1; for directional luminaires, g(θ) qm The angle attenuation coefficient is determined according to the light distribution curve of the luminaire or the parameter library.

[0084] The user terminal device's information feedback module displays ground illuminance measurement guidance, comprehensive judgment results, and supplementary lighting optimization schemes. The comprehensive judgment results can be displayed in the form of text, numerical values, and ground area heat maps. The supplementary lighting optimization schemes are displayed in the form of a list of luminaires, power range, installation location diagram, layout diagram, and expected illuminance distribution map. After receiving the user's selection of luminaire type, budget range, and installation conditions, the user terminal device requests the cloud server device to recalculate the supplementary lighting optimization scheme.

[0085] Example 2: Figure 2 As shown, the home lighting environment assessment system for the elderly based on mobile phone visual perception includes a user terminal device and a cloud server device. The user terminal device includes an image acquisition module and an information feedback module.

[0086] The image acquisition module is used to call the mobile terminal camera to acquire multi-view image data of the target indoor space and record the time, exposure parameters and device posture corresponding to the image acquisition; the information feedback module is used to display the augmented reality guidance interface, comprehensive judgment results and supplementary lighting optimization scheme.

[0087] The cloud server device includes a light environment information recognition module, an illuminance discrimination module, and a supplementary lighting optimization scheme generation module. The light environment information recognition module performs camera calibration, pose estimation, scale normalization, visual perception model inference, room size parameter calculation, and existing lighting equipment information recognition in step S200. The illuminance discrimination module performs functional attribute area recognition, candidate measurement point generation, measurement weight configuration, key measurement point screening, and comprehensive judgment in step S300. The supplementary lighting optimization scheme generation module performs light utilization coefficient correction, calculation of required supplementary lighting parameters, generation of luminaire power range, recommendation of luminaire installation location, and simulation of expected illuminance distribution effect in steps S400 and S500.

[0088] In a specific application scenario, the user terminal device acquires multi-view image data of the bedroom. The light environment information recognition module identifies the bed, door, window, floor, wall and existing ceiling light based on the spatial semantic segmentation map, and generates a structured 3D model of the bedroom based on visual SLAM. After the system detects the known size reference objects placed by the user, it calculates the size factor to obtain the floor area of ​​the bedroom and the installation height of the light fixtures.

[0089] Example 3: Figure 3 A schematic diagram of an electronic device that can be used to implement embodiments of the present invention is shown. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0090] like Figure 3 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., which is communicatively connected to the at least one processor 11. The memory stores a computer program that can be executed by the at least one processor 11, and the computer program is executed by the at least one processor 11 to enable the at least one processor 11 to perform the method provided by the present invention.

[0091] The processor 11 can perform various appropriate actions and processes based on a computer program stored in the read-only memory (ROM) 12 or a computer program loaded from the storage unit 18 into the random access memory (RAM) 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. The input / output (I / O) interface 15 is also connected to the bus 14.

[0092] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0093] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as a method for assessing the home lighting environment for the elderly based on mobile phone visual perception.

[0094] In some embodiments, the method for assessing the home lighting environment of the elderly based on mobile phone visual perception can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the method for assessing the home lighting environment of the elderly based on mobile phone visual perception described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the method for assessing the home lighting environment of the elderly based on mobile phone visual perception by any other suitable means (e.g., by means of firmware).

[0095] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard parts (ASSPs), systems-on-chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0096] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0097] In the context of this invention, a computer-readable storage medium stores computer instructions that, when executed by a processor, implement the mobile phone-based visual perception-based home lighting environment assessment method for the elderly provided by this invention. The computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. The computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0098] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device for displaying information to the user (e.g., a cathode ray tube (CRT) or a liquid crystal display (LCD monitor)); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0099] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0100] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system. It addresses the shortcomings of traditional physical hosts and Virtual Private Server (VPS) services, such as high management difficulty and weak business scalability.

[0101] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0102] The illuminance discrimination module identifies the bedside area, doorway area, and passageway area as functional attribute areas and assigns higher measurement weights to these areas. The user terminal device guides the user to collect illuminance values ​​at key measurement points through an augmented reality interface; the supplementary lighting optimization scheme generation module corrects the light utilization coefficient and calculates the required supplementary luminous flux based on the weighted average illuminance, minimum illuminance, illuminance uniformity, wall reflectivity characteristics, furniture layout density, and estimated luminous flux of existing lamps; the information feedback module displays the power range of the supplementary lamps, the bedside supplementary lighting location, and the expected illuminance distribution effect after supplementary lighting.

[0103] The preset threshold, weighting coefficient, light utilization coefficient range, lamp power specifications, and preset age-appropriate illuminance standards in the above embodiments can be stored in a parameter library and adjusted according to room type, user age group, usage habits, and updated standard requirements. The adjustment of the parameter library does not change the technical concept of this invention, which obtains spatial feature information and existing lighting equipment information through visual perception processing, and generates measurement guidance, light utilization coefficient correction, and supplementary lighting optimization schemes based on the spatial feature information.

[0104] Based on the preferred embodiments of the present invention described above, those skilled in the art can make various changes and modifications without departing from the inventive concept. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for assessing the home lighting environment for the elderly based on mobile phone visual perception, characterized in that, Includes the following steps: S100: Acquire multi-view image data of the target indoor space via a mobile terminal; S200: Perform visual perception processing on the multi-view image data to obtain spatial feature information and existing lighting equipment information for light environment assessment; wherein, the spatial feature information includes geometric structure information, as well as spatial distribution characteristics of ground area, wall area and area related to elderly activities, room size parameters, and at least one of surface reflection characteristics and shading layout characteristics that affect indoor light utilization. S300: Identify functional attribute areas in the elderly's home environment based on the spatial feature information, and determine ground illuminance measurement points and corresponding measurement weights based on the fall risk level and activity frequency of the functional attribute areas. Provide visual guidance for the ground illuminance measurement points through a mobile terminal, and form ground illuminance distribution information based on the illuminance values ​​collected by the user. S400: Based on the ground illuminance distribution information, spatial characteristic information and existing lighting equipment information, the light utilization coefficient is corrected, the effective light flux reaching the ground area is determined based on the ground illuminance distribution information, and the total output light flux of existing light sources in the current environment is estimated based on the corrected light utilization coefficient, and then the supplementary lighting parameters required to reach the preset age-appropriate illuminance standard are calculated. S500: Generate a supplementary lighting optimization scheme based on the supplementary lighting parameters required to achieve the preset age-appropriate illuminance standard, and feed the supplementary lighting optimization scheme back to the user terminal.

2. The method for assessing the home lighting environment for the elderly based on mobile phone visual perception according to claim 1, characterized in that: S200 includes: Camera correction, pose estimation, and scale normalization are performed on the multi-view image data to obtain an indoor image sequence; the indoor image sequence is input into a visual perception model to obtain a reflectance component map, an illumination component map, and a spatial semantic segmentation map; Based on the spatial semantic segmentation map and combined with the geometric relationships between different perspectives in the multi-view image data, room size parameters are obtained; the lighting areas in the multi-view image data are identified to obtain information on existing lighting equipment. The existing lighting equipment information includes at least one of the following: existing lamp type, installation location, luminous surface size, rated power range, luminous efficacy value, and luminous flux estimate.

3. The method for assessing the home lighting environment for the elderly based on mobile phone visual perception according to claim 2, characterized in that: The visual perception model includes an intrinsic image decomposition network and a multi-task joint learning network; the intrinsic image decomposition network is used to separate the reflectance component map and the illumination component map from an indoor image sequence. The multi-task joint learning network performs feature splicing and fusion of reflectivity component map and illumination component map in the channel dimension, and performs pixel-level semantic segmentation through encoder-decoder structure to generate spatial semantic segmentation map.

4. The method for assessing the home lighting environment for the elderly based on mobile phone visual perception according to claim 2, characterized in that: The room size parameters were obtained in the following manner: The spatial semantic segmentation map is back-projected onto the 3D sparse point cloud generated by visual SLAM to construct a semantic point cloud; Three-dimensional plane fitting is performed on the ground points, wall points, and ceiling points in the semantic point cloud to construct a simplified indoor structured three-dimensional model; Detect a known-sized reference object placed by the user in the target indoor space and calculate the size factor; The room size parameters are obtained by quantizing and scaling the structured 3D model of the interior according to the size factor.

5. The method for assessing the home lighting environment for the elderly based on mobile phone visual perception according to claim 1, characterized in that: The determination of ground illuminance measurement points and corresponding measurement weights includes: The ground region is determined based on the spatial feature information, and the ground region is converted into a two-dimensional plane coordinate system; Candidate measurement points are generated in the two-dimensional plane coordinate system; measurement weights are assigned to the candidate measurement points based on at least one of the following: passage area, bedside area, doorway area, rest and activity area, blind spot area, area around fixed furniture, and area near obstacles. Key measurement points are selected from the candidate measurement points according to the measurement weights; the location markers of the key measurement points are overlaid on the real-time preview screen of the mobile terminal through an augmented reality guidance interface, and the user is guided to collect illuminance values ​​at the key measurement points. Based on the collected illuminance values ​​and corresponding measurement weights, the weighted average illuminance, minimum illuminance, and illuminance uniformity of the ground area are calculated and compared with the preset age-appropriate illuminance standard to output a comprehensive judgment result.

6. The method for assessing the home lighting environment for the elderly based on mobile phone visual perception according to claim 5, characterized in that: The calculation of the supplementary lighting parameters required to achieve the preset age-appropriate illuminance standard includes: Based on the ground area and the weighted average illuminance, the effective luminous flux reaching the ground area is calculated; combined with the existing lighting equipment information and the corrected light utilization coefficient, the total output luminous flux of existing light sources in the current environment is estimated. Based on the target illuminance value of the preset age-friendly illuminance standard, the area of ​​the ground region, and the corrected light utilization coefficient, calculate the target total luminous flux required to achieve the preset age-friendly illuminance standard; calculate the additional luminous flux required based on the difference between the target total luminous flux and the total output luminous flux of existing light sources in the current environment. Based on the luminous efficacy value of the lamp type selected by the user or recommended by the system, the required luminous flux is converted into the required lamp power, and a supplementary lighting optimization scheme including the range of lamp power is generated.

7. A method for assessing the home lighting environment for the elderly based on mobile phone visual perception according to claim 6, characterized in that: The correction of the light utilization coefficient includes: Based on the wall reflectivity characteristics, furniture layout density, and the deviation between the measured illuminance and the luminous flux estimated by existing lighting equipment, the preset initial light utilization coefficient is weighted and corrected. When the wall reflectivity characteristics indicate that the target interior space is dominated by light-colored walls or has a low furniture density, the light utilization coefficient should be increased. When the wall reflectivity characteristics indicate that the target interior space is dominated by dark walls or has a high density of furniture, the light utilization coefficient should be reduced.

8. A method for assessing the home lighting environment for the elderly based on mobile phone visual perception according to claim 6, characterized in that: The method for generating the power range of the lamps includes: Calculate the supplementary power required to reach the target lower limit and target upper limit of the preset age-friendly illuminance standard, respectively, to obtain the theoretical power range; Based on the uncertainty of the corrected light utilization coefficient, a margin correction is made to the theoretical power range; Align the theoretical power range, corrected for margin, with the common power specifications of commercially available lighting fixtures to obtain the actual selectable range of lighting fixture power adjustment.

9. A method for assessing the home lighting environment for the elderly based on mobile phone visual perception according to claim 2, characterized in that: The supplementary lighting optimization scheme also includes: Based on geometric structure information and spatial semantic segmentation maps, we recommend the installation location and layout of lighting fixtures; It offers a variety of optional lighting configuration options, each including the specific number and wattage of the lighting fixtures; Simulate and demonstrate the expected illuminance distribution effect after adopting the supplementary lighting optimization scheme.

10. A home lighting environment assessment system for the elderly based on mobile phone visual perception, used to implement the home lighting environment assessment method for the elderly based on mobile phone visual perception as described in any one of claims 1-9, characterized in that: Including user-end devices and cloud server devices, The user-end device includes an image acquisition module for acquiring multi-view image data of the target indoor space; Cloud server equipment includes: The light environment information recognition module is used to identify the spatial semantic segmentation map, geometric structure information and existing lighting equipment information of the target indoor space based on the multi-view image data through visual perception processing. The illuminance discrimination module is used to plan a ground illuminance measurement scheme based on the spatial semantic segmentation map and geometric structure information to guide the user's measurement, and to obtain the illuminance distribution information of the ground area based on the measurement data fed back by the user terminal device, and to make a comprehensive judgment with the preset age-friendly illuminance standard. The supplementary lighting optimization scheme generation module is used to calculate the supplementary lighting parameters required to achieve the preset age-friendly illuminance standard based on the determination result of the preset age-friendly illuminance standard, the illuminance distribution information of the ground area and the information of the existing lighting equipment, and generate a personalized supplementary lighting optimization scheme that includes specific power range and layout suggestions. The user terminal device also includes an information feedback module, which is used to receive and display the ground illuminance measurement guidance, comprehensive judgment results and supplementary lighting optimization scheme issued by the cloud server device.

Images