A residential sight line design method, system, device and medium based on space vector analysis

Abstract

The application relates to the technical field of intelligent building optimization design. A residential line-of-sight design method, system, equipment and medium based on space vector analysis are provided, wherein the method comprises the following steps: performing space vector modeling based on a preset residential building model to obtain a three-dimensional line-of-sight vector group; performing privacy exposure parameter calculation on the three-dimensional line-of-sight vector group to generate a privacy intervention degree distribution map; performing perception calibration on a preset privacy intervention degree calculation model by using the privacy intervention degree distribution map to generate a calibrated privacy intervention degree calculation model; and performing design intervention on an over-threshold exposure area by using the calibrated privacy intervention degree calculation model to obtain an optimized residential building model, so that the three-dimensional non-linear line-of-sight path recognition accuracy is improved, the privacy quantification model prediction accuracy is enhanced, and a man-machine collaborative design optimization closed loop is constructed.

Description

Technical Field

[0001] This invention relates to the field of intelligent building optimization design technology, and in particular to a residential sightline design method, system, device and medium based on spatial vector analysis. Background Technology

[0002] With the deep integration of artificial intelligence and building information modeling, the importance of privacy protection in smart home design is becoming increasingly prominent. Visual privacy design, as a core indicator of residential comfort, directly impacts residential safety and psychological experience.

[0003] Traditional technologies employ two-dimensional planar line-of-sight analysis for spatial privacy analysis. However, this approach fails to capture non-linear line-of-sight paths in three-dimensional space, making it difficult to identify indirect exposure risks. Existing privacy quantification models are constructed using physical distance as the core parameter, neglecting the influence of spatial function weights and temporal pedestrian flow patterns, leading to distorted model predictions. Furthermore, spatial privacy design optimization relies solely on human experience for adjustments, lacking closed-loop technologies such as quantitative feedback, human-machine collaborative calibration, and dynamic optimization, thus failing to effectively verify privacy enhancement effects and achieve precise optimization. Summary of the Invention

[0004] Therefore, it is necessary to provide a residential sightline design method, system, device and medium based on spatial vector analysis to address the above-mentioned technical problems, so as to improve the accuracy of three-dimensional non-linear sightline path recognition, enhance the prediction accuracy of privacy quantification model, and build a human-machine collaborative design optimization closed loop.

[0005] Firstly, this application provides a residential sightline design method based on spatial vector analysis, the method comprising:

[0006] Spatial vector modeling is performed based on a pre-defined residential building model to obtain a three-dimensional line-of-sight vector set.

[0007] Calculate privacy exposure parameters for the three-dimensional line-of-sight vector group and generate a privacy intervention degree distribution map;

[0008] The privacy intervention degree distribution map is used to perform perceptual calibration on the preset privacy intervention degree calculation model to generate a calibrated privacy intervention degree calculation model;

[0009] By using a calibrated privacy intervention calculation model to design interventions in areas exposed beyond the threshold, an optimized residential building model is obtained.

[0010] In one embodiment, a pre-defined privacy intervention degree calculation model is perceptually calibrated using a privacy intervention degree distribution map to generate a calibrated privacy intervention degree calculation model, including:

[0011] Transform a pre-defined residential building model into a virtual reality scene;

[0012] By fusing privacy intervention distribution maps with virtual reality scenes, a perception verification environment can be generated;

[0013] In a perception verification environment, user eye movement trajectories are collected to obtain the actual exposure feature set;

[0014] Simultaneously acquire users' subjective ratings of the exposure events to obtain a perceptual bias dataset;

[0015] By comparing the geometric parameters of the actual exposure feature set and the privacy intervention degree distribution map, preliminary bias parameters are generated;

[0016] By combining the perception bias dataset, the initial bias parameters are corrected, and model calibration parameters are generated.

[0017] The mapping function of the privacy intervention calculation model is adjusted based on the model calibration parameters to generate a calibrated privacy intervention calculation model.

[0018] In one embodiment, preliminary bias parameters are generated by comparing the geometric parameters of the actual exposure feature set and the privacy intervention degree distribution map, including:

[0019] Transform the spatial coordinates of the actual exposed feature set to the spatial coordinate system of the privacy intervention degree distribution map to generate a unified coordinate mapping relationship;

[0020] Based on the unified coordinate mapping relationship, the geometric deviation between the exposure intensity parameter of the actual exposure feature set and the predicted exposure probability at the corresponding location is calculated to obtain the local deviation vector set;

[0021] Density clustering is performed on local deviation vector sets to identify spatial clusters whose deviation density exceeds a preset threshold.

[0022] Extract the average deviation and spatial distribution dispersion index of the spatial clusters to generate preliminary deviation parameters.

[0023] In one embodiment, users' subjective ratings of exposure events are acquired synchronously to obtain a perceptual bias dataset, including:

[0024] When an exposure event is detected in the perception verification environment, a real-time scoring interface is triggered;

[0025] The system collects user ratings on the degree of privacy violation of the current exposure event through a real-time rating interface, generating raw rating data.

[0026] The raw scoring data is correlated and encoded with the spatiotemporal characteristics of the exposure event to generate structured scoring units;

[0027] By aggregating structured scoring units from multiple exposure events, a perceptual bias dataset is obtained.

[0028] In one embodiment, a privacy intervention degree distribution map and a virtual reality scene are fused to generate a perception verification environment, including:

[0029] The coordinate system of the privacy intervention degree distribution map is registered with the coordinate system of the virtual reality scene to generate spatial alignment parameters;

[0030] Based on the spatial alignment parameter, the intervention degree value of the privacy intervention degree distribution map is mapped to the visual attribute parameter;

[0031] Generate dynamic heat maps in virtual reality scenes based on visual attribute parameters;

[0032] The dynamic heat map layer is fused with the geometric model of the virtual reality scene at the pixel level to generate a perception verification environment.

[0033] In one embodiment, privacy exposure parameters are calculated for the three-dimensional gaze vector set to generate a privacy intervention degree distribution map, including:

[0034] A line-of-sight penetration analysis was performed on the three-dimensional line-of-sight vector set to obtain the spatial exposure probability matrix;

[0035] Construct a line-of-sight influence network based on the spatial topological relationship of three-dimensional line-of-sight vector groups;

[0036] By combining the time dimension of people's flow activity patterns, weights are assigned to the line-of-sight influence network to generate a weighted line-of-sight influence network;

[0037] By fusing the spatial exposure probability matrix and the weighted line-of-sight influence network, a privacy intervention degree distribution map is generated.

[0038] In one embodiment, spatial vector modeling is performed based on a preset residential building model to obtain a three-dimensional line-of-sight vector set, including:

[0039] Analyze the spatial topology and functional attributes of a pre-defined residential building model to generate a spatial semantic map;

[0040] Identify privacy-critical regions in the spatial semantic graph to obtain a set of privacy-sensitive nodes;

[0041] Based on the privacy-sensitive node set and spatial topology, potential line-of-sight paths are calculated, and a line-of-sight path network is generated.

[0042] Each path in the line-of-sight path network is parameterized into a direction vector and a distance parameter to obtain a three-dimensional line-of-sight vector set.

[0043] Secondly, this application also provides a residential sightline design system based on spatial vector analysis, the system comprising:

[0044] The spatial vector modeling module is used to perform spatial vector modeling based on a preset residential building model to obtain a three-dimensional line-of-sight vector set.

[0045] The privacy computing module is used to calculate privacy exposure parameters of the three-dimensional gaze vector group and generate a privacy intervention degree distribution map;

[0046] The model calibration module is used to perform perceptual calibration on the preset privacy intervention degree calculation model using the privacy intervention degree distribution map, and generate a calibrated privacy intervention degree calculation model;

[0047] The design intervention module is used to perform design interventions on areas exposed beyond the threshold by using a calibrated privacy intervention degree calculation model, resulting in an optimized residential building model.

[0048] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of any of the methods in the first aspect of this application.

[0049] This application provides a residential sightline design method, system, device, and medium based on spatial vector analysis. The method involves spatial vector modeling based on a pre-set residential building model to obtain a three-dimensional sightline vector set, which helps improve the recognition accuracy of three-dimensional non-linear sightline paths. On this basis, privacy exposure parameters are calculated on the three-dimensional sightline vector set to generate a privacy intervention degree distribution map. Based on this distribution map, a pre-set privacy intervention degree calculation model is perceptually calibrated, which can enhance the prediction accuracy of the privacy quantification model. Subsequently, the calibrated privacy intervention degree calculation model is used to implement design interventions on areas with exposure exceeding the threshold and obtain an optimized residential building model, thereby constructing a human-machine collaborative design optimization closed loop. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 A flowchart illustrating a residential sightline design method based on spatial vector analysis in one embodiment of the present invention;

[0052] Figure 2 A flowchart for generating preliminary deviation parameters by comparing the geometric parameters of the actual exposure feature set and the privacy intervention degree distribution map in one embodiment of the present invention;

[0053] Figure 3 This is a structural diagram of a residential sightline design system based on spatial vector analysis, according to one embodiment of the present invention. Detailed Implementation

[0054] To make the above-mentioned objects, features, and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0055] This application provides a residential sightline design method, system, device, and medium based on spatial vector analysis. This solution is applicable to scenarios such as privacy optimization design of new residential space layout, privacy renovation and upgrading of existing residential sightlines, and sightline planning of functional areas in smart residential spaces.

[0056] For example, the above solutions can also be applied to scenarios such as optimizing the privacy performance of affordable housing, customizing visual privacy design for high-end residences, and controlling visual interference in mixed-use living spaces. The above are merely illustrative examples and do not limit the specific application scenarios of the solutions.

[0057] like Figure 1 As shown, this application provides a residential sightline design method based on spatial vector analysis, the method comprising:

[0058] S101: Based on the preset residential building model, perform spatial vector modeling to obtain a three-dimensional line-of-sight vector group.

[0059] For example, the architectural design terminal performs structural analysis on a preset residential building model, extracts spatial topology and regional functional attribute information, and integrates them to generate a spatial semantic map. Then, it identifies privacy-critical areas in the spatial semantic map and collects the spatial points of each privacy-critical area to obtain a set of privacy-sensitive nodes.

[0060] Combining a set of privacy-sensitive nodes and spatial topology, the architectural design terminal deduces and verifies valid potential line-of-sight paths. After integrating and generating a line-of-sight path network, each path in the network is further characterized in three dimensions using parametric representation. The vector definition is completed through a spatial vector construction formula, specifically:

[0061]

[0062] In the formula, For a single 3D line of sight vector, Privacy-sensitive nodes The three-dimensional x-coordinate, Privacy-sensitive nodes The three-dimensional ordinate, Privacy-sensitive nodes The three-dimensional vertical coordinates, Privacy-sensitive nodes The three-dimensional x-coordinate, Privacy-sensitive nodes The three-dimensional ordinate, Privacy-sensitive nodes The three-dimensional vertical coordinates, Privacy-sensitive nodes With privacy-sensitive nodes The line-of-sight path length coefficient.

[0063] The architectural design terminal collects all parametrically processed 3D line-of-sight vectors to obtain a 3D line-of-sight vector set.

[0064] S102: Calculate privacy exposure parameters for the three-dimensional gaze vector group and generate a privacy intervention degree distribution map.

[0065] For example, the architectural design terminal performs a vector-by-vector line-of-sight penetration analysis on the three-dimensional line-of-sight vector group. Combined with the spatial distribution information of obstructions in the preset residential building model, it determines the occlusion status of each three-dimensional line-of-sight vector on the propagation path, calculates the line-of-sight penetration coefficient corresponding to each three-dimensional line-of-sight vector, and then associates the line-of-sight path length coefficient of the three-dimensional line-of-sight vector. The spatial exposure probability of each spatial point is obtained through the spatial exposure probability calculation formula. The spatial exposure probabilities of all spatial points are integrated in coordinate order to obtain the spatial exposure probability matrix.

[0066] Based on the spatial topological relationships of each vector in the three-dimensional line-of-sight vector group, the architectural design terminal sets all spatial points in the preset residential building model as nodes and sets the spatial influence relationships corresponding to each three-dimensional line-of-sight vector as edges, thus building a basic line-of-sight influence network.

[0067] The architectural design terminal imports preset time-dimensional pedestrian activity pattern data, and assigns corresponding visual influence weights to each edge in the visual influence network based on the pedestrian density level at different times and the pedestrian gathering characteristics of different spatial areas, thus completing the generation of the weighted visual influence network.

[0068] To achieve data fusion computation, the architectural design terminal constructs a privacy intervention degree fusion formula, combining probability values ​​from the spatial exposure probability matrix and weight values ​​from the weighted line-of-sight influence network. The specific formula is as follows:

[0069]

[0070] In the formula, For the first Privacy intervention in each spatial area To achieve the fusion balance coefficient, For the first The spatial exposure probability of a spatial point. To weight the influence of gaze on the 3D gaze vector in the network The corresponding line of sight affects the weight. A unique identifier for a spatial region. A unique identifier for a spatial location. This refers to a single 3D view vector within a 3D view vector group.

[0071] Space Exposure Probability The calculation is based on the following formula:

[0072]

[0073] In the formula, Three-dimensional view vector Line penetration coefficient, Three-dimensional view vector The line-of-sight path length coefficient, To cover the first The total number of three-dimensional line-of-sight vectors for each spatial point. It is a unique identifier for spatial points, used to uniquely distinguish each independent spatial point within the preset residential building model.

[0074] The architectural design terminal calculates the privacy intervention degree of each spatial area using the above formula, maps the privacy intervention degree of all spatial areas to the spatial coordinates of the preset residential building model, and generates a privacy intervention degree distribution map.

[0075] S103: Use the privacy intervention degree distribution map to perform perceptual calibration on the preset privacy intervention degree calculation model, and generate a calibrated privacy intervention degree calculation model.

[0076] For example, the architectural design terminal converts a preset residential building model into a virtual reality scene. Through spatial registration and visual fusion, it integrates the privacy intervention degree distribution map and the virtual reality scene to build a perception verification environment. Within this environment, it simultaneously collects the user's eye movement trajectory to form an actual exposure feature set and collects subjective scores of exposure events to form a perception bias dataset.

[0077] After completing the basic data collection, the architectural design terminal performs coordinate unification processing and geometric deviation calculation on the actual exposure feature set and privacy intervention degree distribution map. It extracts deviation features based on density clustering and generates preliminary deviation parameters. The deviation parameters are then comprehensively corrected through deviation fusion.

[0078] Based on the aforementioned corrected deviation parameters, the architectural design terminal integrates and generates model calibration parameters. These parameters are used to adjust the mapping function of the preset privacy intervention calculation model, thereby generating a calibrated privacy intervention calculation model.

[0079] S104: Design interventions are performed on the over-threshold exposure area using a calibrated privacy intervention degree calculation model to obtain an optimized residential building model.

[0080] For example, the architectural design terminal uses a calibrated privacy intervention calculation model to compare the privacy intervention values ​​of each spatial area in the privacy intervention distribution map, quantifies the privacy intervention of each spatial area with a preset intervention threshold, and then locates the over-threshold exposure area where the privacy intervention exceeds the preset intervention threshold.

[0081] For the identified over-threshold exposure areas, the architectural design terminal combines the spatial distribution characteristics of the area and the three-dimensional line-of-sight vector propagation path to match the corresponding spatial separation and line-of-sight occlusion design schemes. Based on the spatial topology of the preset residential building model, it determines the layout of optimized components and completes the quantitative evaluation of the scheme effect through intervention effect verification.

[0082] After the scheme verification is completed, the architectural design terminal adjusts the spatial layout and shading structure of the preset residential building model according to the qualified design intervention scheme, updates the spatial topology parameters of the model, and obtains the optimized residential building model.

[0083] One embodiment of this application provides a residential sightline design method based on spatial vector analysis. By performing spatial vector modeling based on a preset residential building model to obtain a three-dimensional sightline vector set, it helps to improve the recognition accuracy of three-dimensional non-linear sightline paths. On this basis, privacy exposure parameters are calculated on the three-dimensional sightline vector set to generate a privacy intervention degree distribution map. Based on this distribution map, the preset privacy intervention degree calculation model is perceptually calibrated, which can enhance the prediction accuracy of the privacy quantification model. Subsequently, the calibrated privacy intervention degree calculation model is used to implement design interventions on areas with exposure exceeding the threshold and obtain an optimized residential building model, thereby constructing a human-machine collaborative design optimization closed loop.

[0084] In one embodiment, a pre-defined privacy intervention degree calculation model is perceptually calibrated using a privacy intervention degree distribution map to generate a calibrated privacy intervention degree calculation model, including:

[0085] (1) Convert the preset residential building model into a virtual reality scene.

[0086] For example, the architectural design terminal reads complete data from a preset residential building model, including spatial topology, functional area division, component size parameters, and material properties. It then reconstructs the three-dimensional spatial relationships of the preset residential building model using a 3D modeling engine, restoring the spatial scale and relative positions of each functional area, thus transforming the preset residential building model into a virtual reality scene with immersive interactive capabilities. This virtual reality scene fully preserves the core spatial features of the preset residential building model, allowing users to move freely within the scene and observe the line-of-sight propagation in each spatial area in real time.

[0087] The complete data of the preset residential building model includes spatial topology data, functional area division data, component size parameter data, component material attribute data, and relative positional relationship data of each area.

[0088] (2) The privacy intervention degree distribution map and virtual reality scene are fused to generate a perception verification environment.

[0089] For example, the architectural design terminal extracts the spatial coordinate system and intervention value matrix of the privacy intervention degree distribution map, registers and aligns the coordinate system of the privacy intervention degree distribution map with the three-dimensional coordinate system of the virtual reality scene, realizes a one-to-one mapping between the intervention degree values ​​and the spatial points of the virtual reality scene through standardized data format conversion, and converts the intervention degree values ​​into corresponding visual rendering parameters, so that each spatial area in the virtual reality scene presents differentiated visual effects according to the intervention degree values, completes the data fusion of the privacy intervention degree distribution map and the virtual reality scene, and generates a perception verification environment.

[0090] The privacy intervention distribution map's intervention value matrix includes unique identifiers for all spatial points within the preset residential building model, their corresponding spatial coordinates, and quantified privacy intervention values ​​for each point. Visual rendering parameters include color gradient, transparency, and brightness parameters, used to distinguish spatial areas with different privacy intervention levels through visual differences. The perception verification environment possesses both the spatial interaction characteristics of a virtual reality scene and the quantitative presentation capability of a privacy intervention distribution map, allowing for an intuitive display of the privacy intervention distribution status of each spatial area.

[0091] (3) Collect the user's eye movement trajectory in the perception verification environment to obtain the actual exposure feature set.

[0092] For example, the architectural design terminal deploys a professional eye-tracking unit in the perception verification environment. When the user enters the perception verification environment and performs scene browsing operations, the eye-tracking unit captures the user's eye movement data in real time, extracts core indicators such as the three-dimensional spatial coordinates of the user's gaze point, the duration of gaze, and changes in gaze trajectory, integrates the above core indicators in a time series, and associates them with the spatial area of ​​the perception verification environment corresponding to the gaze point to obtain an actual exposure feature set containing the user's actual gaze attention characteristics.

[0093] The core indicators of the actual exposure feature set include the three-dimensional spatial coordinates of the user's gaze point, the duration of gaze, the rate of change of gaze trajectory, the frequency of gaze point dwell, and the association information between each indicator and the spatial area of ​​the perception verification environment.

[0094] (4) Simultaneously obtain users' subjective ratings of the exposure events to obtain a perception bias dataset.

[0095] For example, the architectural design terminal sets up an exposure event triggering mechanism in the perception verification environment. When the user's gaze point stays in a certain spatial area for a set period of time, the architectural design terminal automatically determines that the spatial area is the area where the exposure event occurs and triggers the scoring interaction interface to guide the user to quantify the degree of privacy infringement of the exposure event. At the same time, it records the spatial coordinates, duration and other spatiotemporal characteristics of the exposure event, associates and stores the user's score with the spatiotemporal characteristics of the corresponding exposure event, and integrates all the associated data to obtain the perception deviation dataset.

[0096] The perception bias dataset contains the correspondence between users' subjective ratings and objective exposure features, reflecting the deviation between the prediction results of the preset privacy intervention degree calculation model and the actual user perception. The spatiotemporal features of the exposure event include the three-dimensional spatial coordinates of the exposure event, the duration of the exposure event, the time period of the exposure event, the frequency of the user's gaze at the area, and the corresponding privacy intervention degree prediction value for that area.

[0097] (5) Compare the geometric parameters of the actual exposure feature set and the privacy intervention degree distribution map to generate preliminary deviation parameters.

[0098] For example, the architectural design terminal extracts geometric parameters such as the spatial coordinates of the gaze point and the exposure intensity from the actual exposure feature set, and simultaneously extracts the predicted exposure probability parameters of the corresponding spatial coordinates from the privacy intervention degree distribution map. The difference between the two types of parameters is compared using the geometric deviation calculation formula, which is as follows:

[0099]

[0100] In the formula, For the first Preliminary deviation parameters for each spatial cluster, For the first The total number of spatial points contained within a spatial cluster For the actual exposure feature set, the first The actual exposure intensity at each spatial location The first in the privacy intervention distribution map Predicted exposure probability for each spatial location This is a unique identifier for the spatial cluster. It is a unique identifier for a spatial point.

[0101] The actual exposure intensity includes the first The weighted values ​​of gaze duration and gaze frequency for each spatial point, as well as the visual salience weight of that point in the perception verification environment; the predicted exposure probability includes the line-of-sight exposure probability of that point obtained based on the preset privacy intervention degree calculation model, the occlusion influence correction coefficient, and the spatial topology association correction value.

[0102] (6) Correct the initial deviation parameters by combining the perception deviation dataset and generate model calibration parameters.

[0103] For example, the architectural design terminal preprocesses the perception bias dataset through a subjective rating processing unit, removes abnormal rating data, and calculates the first... The arithmetic mean of the subjective scores of all exposure events within the spatial cluster is used to obtain the _th _ ... Average subjective score corresponding to each spatial cluster Then, based on the pre-defined model calibration requirements and considering the relative importance of objective geometric deviation and subjective perception deviation, the weighting coefficient for objective deviation is determined. The value of .

[0104] The architectural design terminal activates the deviation correction unit, which adjusts the initial deviation parameters for each spatial cluster. Average subjective rating and objective deviation weighting coefficient Substituting the values ​​into the bias correction formula, the model calibration parameters for each spatial cluster are calculated. The parameter integration unit is invoked to sort and associate all model calibration parameters according to the spatial coordinate distribution of the spatial cluster, thereby obtaining global model calibration parameters covering the entire spatial range of the preset residential building model.

[0105]

[0106] In the formula, For the first Model calibration parameters for each spatial cluster, This is the objective deviation weighting coefficient. For the first Preliminary deviation parameters for each spatial cluster, For the first The average subjective rating of the perceptual bias dataset corresponding to each spatial cluster. It is a unique identifier for the spatial cluster.

[0107] Among them, the average subjective rating of the perceptual bias dataset includes the first The score for the degree of user privacy violation, the score consistency coefficient, and the weight of user attention to each type of exposure event within a spatial cluster.

[0108] (7) Adjust the mapping function of the privacy intervention degree calculation model based on the model calibration parameters to generate the calibrated privacy intervention degree calculation model.

[0109] For example, the architectural design terminal invokes the model retrieval unit to extract the core mapping function of the preset privacy intervention degree calculation model, clarifying the input parameters (spatial feature parameters), output parameters (privacy intervention degree values), and intermediate calculation process of the mapping function. The parameter adjustment unit is then activated, substituting the global model calibration parameters into the parameter adjustment equation of the mapping function, and successively correcting the coefficient matrix and constraints of the mapping function using a gradient descent iterative optimization algorithm.

[0110] During the iteration process, the architectural design terminal calculates the model prediction deviation after each iteration, and stops iterating when the deviation value is lower than the preset convergence threshold. The corrected mapping function can fully compensate for objective geometric deviations and subjective perception deviations, adapt to the spatial characteristics of the preset residential building model and the actual perception needs of users, and generate a calibrated privacy intervention calculation model with significantly improved prediction accuracy.

[0111] The core mapping function of the preset privacy intervention degree calculation model includes a spatial feature parameter extraction unit, an intervention degree prediction coefficient matrix, an occlusion impact calculation unit, and a spatiotemporal feature association unit, which are used to realize the conversion from spatial features to privacy intervention degree values.

[0112] like Figure 2 As shown, by comparing the geometric parameters of the actual exposure feature set and the privacy intervention degree distribution map, preliminary bias parameters are generated, including:

[0113] S201: Transform the spatial coordinates of the actual exposed feature set to the spatial coordinate system of the privacy intervention degree distribution map to generate a unified coordinate mapping relationship.

[0114] For example, the architectural design terminal extracts the three-dimensional spatial coordinates of all spatial points in the actual exposure feature set, and simultaneously extracts the spatial coordinate system parameters of the privacy intervention degree distribution map, including core information such as the coordinate origin position, coordinate axis direction, and coordinate scale unit. By comparing the differences between the two spatial coordinate systems through the coordinate system analysis unit, the rotation matrix, translation vector, and scaling factor required for coordinate transformation are determined, and a preliminary coordinate transformation model is established.

[0115] The spatial coordinates of the actual exposure feature set include the three-dimensional Cartesian coordinates corresponding to the user's gaze point, and the spatial coordinate system of the privacy intervention degree distribution map is the global spatial coordinate system of the preset residential building model. Both are established based on a unified spatial reference surface.

[0116] Using a coordinate transformation algorithm, the architectural design terminal transforms the coordinates of each spatial point in the actual exposed feature set point by point. The transformed coordinates are then matched and verified with the corresponding coordinates on the privacy intervention degree distribution map, and the transformation error value is calculated. If the error value exceeds a preset accuracy threshold, the rotation matrix, translation vector, and scaling factor are iteratively optimized until the transformation error meets the accuracy requirements, generating a unified coordinate mapping relationship covering all spatial points.

[0117] The unified coordinate mapping relationship includes a one-to-one correspondence between the coordinates of the actual exposed feature set and the coordinates of the privacy intervention degree distribution map, a coordinate transformation parameter set, and a transformation accuracy verification report, which are used to ensure that the two types of data are compared under the same spatial reference.

[0118] S202: Based on the unified coordinate mapping relationship, calculate the geometric deviation between the exposure intensity parameter of the actual exposure feature set and the predicted exposure probability of the corresponding location, and obtain the local deviation vector set.

[0119] For example, based on the established unified coordinate mapping relationship, the architectural design terminal associates the exposure intensity parameter corresponding to each spatial point in the actual exposure feature set with the predicted exposure probability corresponding to the same spatial coordinate in the privacy intervention degree distribution map. The calculation method for geometric deviation is determined through a deviation calculation unit, and absolute deviation, relative deviation, or standardized deviation is selected as the core calculation indicator based on data characteristics.

[0120] The exposure intensity parameters of the actual exposure feature set include the weighted value of gaze duration, the weighted value of gaze frequency, and the visual salience weight, which are fused together according to a preset ratio to obtain a quantitative value; the predicted exposure probability of the privacy intervention degree distribution map includes the comprehensive calculation results of the line-of-sight penetration coefficient correction value, the line-of-sight path length correction value, and the spatial topological association correction value.

[0121] According to the preset spatial coordinate sorting rules, the architectural design terminal performs point-by-point geometric deviation calculations on the associated exposure intensity parameters and predicted exposure probabilities. The deviation value of each spatial point is combined with its corresponding unified coordinates to obtain a vector unit containing coordinate information and deviation values. All vector units are sorted and integrated according to spatial regions to construct a complete local deviation vector set, ensuring that each vector unit corresponds one-to-one with a spatial point on the privacy intervention degree distribution map.

[0122] The local deviation vector set includes the total number of vector units, the spatial coordinate identifier of each vector unit, the deviation value, and the deviation calculation method identifier. The structure of the vector units is uniform and facilitates subsequent cluster analysis.

[0123] S203: Perform density clustering on the local deviation vector set to identify spatial clusters whose deviation density exceeds a preset threshold.

[0124] For example, the architectural design terminal performs data preprocessing on the local deviation vector set, removing abnormal vector units remaining due to coordinate transformation (such as vectors with deviation values ​​exceeding a reasonable range), and initially sorting the remaining vector units according to spatial coordinates to improve the computational efficiency of the clustering algorithm. Based on the requirements of privacy intervention degree analysis, core parameters for density clustering are set, including the neighborhood radius (used to define the adjacent range of vector units) and the minimum number of points threshold (used to determine the number of core vector units).

[0125] Among them, the neighborhood radius of density clustering is set based on the spatial grid size of the privacy intervention degree distribution map, and the minimum number of points threshold is determined according to the total number of local deviation vector sets and spatial distribution density. The two together ensure the rationality of the clustering results.

[0126] After initiating the density clustering algorithm, the architectural design terminal traverses all vector units in the local deviation vector set. Using each vector unit as the center, it searches for neighboring vector units by their neighborhood radius, counting the number of neighboring vector units. When the number of neighboring vector units for a given vector unit reaches a minimum threshold, it is identified as a core vector unit, and the core vector unit and its neighboring vector units are grouped into the same initial cluster. This process is repeated until all vector units are clustered. The deviation density (the ratio of the number of vector units within a cluster to the spatial volume occupied by the cluster) of each initial cluster is calculated. Initial clusters with deviation densities exceeding a preset threshold are selected and identified as the target spatial clusters.

[0127] Among them, a spatial cluster is a set of vector units that are spatially adjacent and have similar deviation values. Each spatial cluster corresponds to a continuous spatial region in the privacy intervention degree distribution map, and the deviation density reflects the degree of concentration of deviation values ​​in that region.

[0128] S204: Extract the average deviation and spatial distribution dispersion index of the spatial clusters to generate preliminary deviation parameters.

[0129] For example, for each selected target spatial cluster, the architectural design terminal extracts the deviation values ​​of all vector units within the cluster and assigns weight coefficients based on the importance of the spatial region corresponding to the vector unit (vector units in the core spatial region have higher weights). The weighted average of the deviation values ​​within the cluster is calculated using a weighted average algorithm. If there is no difference in the importance of the spatial regions, an arithmetic average algorithm is used to calculate the average deviation value, ensuring that the result reflects the overall deviation level within the cluster.

[0130] Among them, the average deviation is a comprehensive quantitative index of the deviation values ​​of all vector units within the target spatial cluster, and the weight coefficients of the weighted average algorithm are determined based on the importance level of the functional areas of the preset residential building model.

[0131] The architectural design terminal calculates the spatial distribution dispersion index for each target spatial cluster through statistical analysis, using variance or standard deviation as the core calculation index. The variance is obtained by averaging the squared differences between each deviation value and the average deviation value, and the standard deviation is obtained by taking the square root of the variance. The average deviation value of each target spatial cluster is correlated with the spatial distribution dispersion index, and the clusters are sorted according to their unique identifiers. This integration yields preliminary deviation parameters that encompass the core deviation characteristics of all target spatial clusters.

[0132] Among them, the spatial distribution dispersion index is used to reflect the dispersion of the deviation values ​​within the target spatial cluster. The larger the value, the more dispersed the deviation distribution within the cluster, and the smaller the value, the more concentrated the deviation distribution within the cluster. The preliminary deviation parameters include the identification number of each target spatial cluster, the average deviation, the spatial distribution dispersion index, and the corresponding spatial region range information.

[0133] In one embodiment, users' subjective ratings of exposure events are acquired synchronously to obtain a perceptual bias dataset, including:

[0134] (1) When an exposure event is detected in the perception verification environment, the real-time scoring interface is triggered.

[0135] For example, the architectural design terminal relies on the event recognition unit built into the perception and verification environment to continuously capture the user's dynamic gaze data within the scene at a preset sampling frequency, including core parameters such as eye rotation angle, three-dimensional coordinates of the gaze point, gaze duration, and gaze movement rate. After the architectural design terminal removes interfering data such as blinking and head shaking through the data preprocessing unit and retains the valid gaze trajectory, the event recognition unit compares the processed gaze parameters with the exposure event judgment conditions in real time.

[0136] When a user's sustained gaze at a specific spatial area reaches a preset threshold, the rate of gaze movement falls below a stability threshold, and the privacy sensitivity level of that area meets the triggering requirements, the architectural design terminal immediately generates an exposure event confirmation signal and sends an interface activation command to the scene interaction unit via the internal communication bus. Upon receiving the command, the scene interaction unit invokes the interface rendering engine to load a preset real-time scoring interface template, simultaneously reads the spatial area information corresponding to the current exposure event, the predicted privacy intervention value, and other auxiliary display content, and completes the rendering of interactive components such as the scoring scale, submit button, and return button. This ensures that the interface is displayed non-intrusively in the perception verification environment, without obstructing the user's observation of the scene.

[0137] The criteria for determining exposure events include a threshold for gaze duration, a gaze focus stability index, and a threshold for the privacy sensitivity level of the corresponding spatial area. The real-time scoring interface is a dedicated subjective scoring interaction platform within the perception verification environment. The scoring scale adopts an equidistant hierarchical design, covering a complete evaluation dimension from non-intrusion to serious infringement.

[0138] (2) Collect users’ ratings on the degree of privacy violation of the current exposure event through the real-time rating interface and generate raw rating data.

[0139] For example, the architectural design terminal presents users with clear operation instructions and explanations of the privacy infringement rating dimensions through a real-time scoring interface, supporting users to submit ratings through various interaction methods such as handheld clicks, voice input, or eye-tracking selection. The interface interaction unit of the architectural design terminal monitors user input behavior in real time, performs format and logic validation on the input content, and filters out invalid inputs such as empty values, out-of-range values, and duplicate submissions.

[0140] The architectural design terminal stores user-generated raw scores that conform to the specifications by associating them with timestamps and event identifiers, generating raw score data. To eliminate potential scale differences between different scoring interfaces, the architectural design terminal introduces standardized calculations for raw scores, normalizing the raw score data to a fixed numerical range to facilitate subsequent cross-event comparisons and calculations.

[0141] The original rating data is a direct quantitative result of the user's subjective perception. The storage format adopts a key-value pair structure, which includes fields such as event identifier, user identifier, rating value, and submission timestamp. The standardized subjective rating eliminates scale differences through linear transformation to ensure that all rating data are on a unified comparison benchmark. Invalid inputs include values ​​that exceed the set rating range and the same value submitted three times consecutively.

[0142] (3) The original scoring data is associated with the spatiotemporal characteristics of the exposure event and encoded to generate structured scoring units.

[0143] For example, the architectural design terminal initiates the spatiotemporal feature extraction unit to retrieve the complete spatiotemporal information of the q-th exposure event from the log database of the perception verification environment. The temporal features include the event trigger timestamp, gaze duration, and event end timestamp, which the architectural design terminal converts into fixed-range temporal feature quantization values ​​using a time quantization algorithm. The spatial features include the three-dimensional spatial coordinates of the event, the privacy sensitivity level of the corresponding spatial area, and the predicted exposure probability of the area in the privacy intervention degree distribution map. The architectural design terminal fuses the multi-dimensional spatial information into a single quantification index using a spatial coding algorithm.

[0144] The data encoding unit of the architectural design terminal determines the weight coefficients of each feature based on the entropy weight method. The weight coefficients for scoring features reflect the importance of standardized subjective scores, the weight coefficients for time features reflect the influence of time features, and the weight coefficients for spatial features reflect the correlation weight of spatial features. The sum of these three weight coefficients is 1. The architectural design terminal substitutes the standardized subjective scores, the quantified values ​​of time features, and the quantified values ​​of spatial features into the weighted correlation encoding logic to generate a fixed-format encoding structure containing feature weights, original data references, and quantification results—that is, a structured scoring unit. The weighted correlation encoding formula is as follows:

[0145]

[0146] In the formula, For the first A structured scoring unit for each exposure event. These are the weighting coefficients for the scoring features. These are the time feature weighting coefficients. These are the spatial feature weighting coefficients. For the first Standardized subjective ratings of individual exposure events. For the first Quantitative value of the temporal characteristics of an exposure event For the first Quantitative values ​​of spatial characteristics of an exposure event.

[0147] (4) Aggregate the structured scoring units of multiple exposure events to obtain the perception bias dataset.

[0148] For example, the architectural design terminal establishes a real-time collection task through a data aggregation unit, extracting structured scoring units one by one according to the order of occurrence of exposed events and storing them in a temporary data buffer. The aggregation unit of the architectural design terminal performs integrity verification on each structured scoring unit, checking whether core fields such as event identifier, standardized score, spatiotemporal feature quantification value, and weight coefficient are missing; at the same time, it performs rationality verification, identifying abnormal values ​​that exceed the interquartile range using the box plot method, determining them as invalid units, and marking the cause of the abnormality.

[0149] The architectural design terminal sorts valid, structured scoring units that have passed verification in ascending order by the unique identifier of the exposure event. Simultaneously, it establishes a secondary index based on spatial region divisions, grouping units within the same spatial region for storage. The architectural design terminal optimizes the storage structure through data compression algorithms, eliminating redundant fields and duplicate information, and integrating all valid units into an ordered, traceable, and easily analyzed perceptual bias dataset. The data aggregation formula is as follows:

[0150]

[0151] In the formula, For the perceptual bias dataset, For the first A structured scoring unit for each exposure event. The total number of events that were effectively exposed.

[0152] Among them, the anomaly detection threshold of the box plot method is determined based on the statistical distribution of all valid units. The data compression algorithm adopts lossless compression to ensure that the original information is not lost. The storage format of the perceptual deviation dataset supports fast query by event identifier and spatial region. It includes two parts: dataset description information and core data units. The dataset description information covers the collection time, number of valid units, data source and other content.

[0153] In one embodiment, a privacy intervention degree distribution map and a virtual reality scene are fused to generate a perception verification environment, including:

[0154] (1) Register the coordinate system of the privacy intervention degree distribution map with the coordinate system of the virtual reality scene to generate spatial alignment parameters.

[0155] For example, the architectural design terminal initiates a coordinate system parameter extraction program to read the core parameters of the two-dimensional coordinate system from the privacy intervention degree distribution map. Specifically, this includes key information such as the planar position of the coordinate origin, the direction of the X and Y axes, the unit of measurement of the coordinate scale, and the projection method (such as orthographic projection or perspective projection). Simultaneously, it extracts the three-dimensional Cartesian coordinate system parameters of the virtual reality scene, including the three-dimensional coordinates of the scene's global origin, the spatial direction of the X, Y, and Z axes, the conversion standard of the unit scale, and the reference plane of the spatial topology. Through a coordinate system difference analysis program, it compares the differences between the two types of coordinate systems in terms of origin position, axis direction, unit of measurement, and projection logic.

[0156] The architectural design terminal employs an iterative nearest-point algorithm to select a set of feature points (including corner points of building components, wall boundary points, door and window outline points, etc.) from the 3D geometric data of a pre-defined residential building model. These feature points correspond to planar projection feature points in the privacy intervention distribution map. With the goal of minimizing the spatial distance error between feature point pairs, the terminal calculates the rotation matrix, translation vector, and scaling factor required for coordinate system registration. The parameter values ​​are adjusted through multiple iterations. After each iteration, the overall registration error is calculated. Iteration stops when the error value is below a pre-defined accuracy threshold (e.g., millimeter-level error). The terminal then integrates the rotation matrix, translation vector, scaling factor, and registration accuracy report to generate complete spatial alignment parameters. The registration error calculation formula is as follows:

[0157]

[0158] In the formula, For overall registration error, This represents the total number of feature point pairs. A unique identifier for each feature point pair. For the first in virtual reality scene Feature points Three-dimensional coordinates of axes For the first in virtual reality scene Feature points Three-dimensional coordinates of axes For the first in virtual reality scene Feature points Three-dimensional coordinates of axes The first in the privacy intervention distribution map Feature points Two-dimensional coordinates of axes The first in the privacy intervention distribution map Feature points Two-dimensional coordinates of axes The first in the privacy intervention distribution map Feature points Two-dimensional coordinates of axes.

[0159] Among them, the coordinate system of the privacy intervention degree distribution map is a two-dimensional plane coordinate system established based on the ground projection surface of the preset residential building model, and the coordinate system of the virtual reality scene is a global three-dimensional Cartesian coordinate system set in the modeling stage; the spatial alignment parameters include rotation matrix, translation vector, scaling factor, and registration accuracy report, which are used to achieve accurate mapping of the two types of coordinate systems in spatial position, scale, and direction.

[0160] (2) Based on the spatial alignment parameter, the intervention value of the privacy intervention degree distribution map is mapped to the visual attribute parameter.

[0161] For example, the architectural design terminal traverses all data nodes of the privacy intervention degree distribution map, extracts the intervention degree value corresponding to each spatial point, divides the gradient level according to the value range of the intervention degree value (such as four levels: low risk, medium risk, high risk, and extremely high risk), determines the value range of visual attributes corresponding to each level in combination with the characteristics of human visual perception, and clarifies the core dimensions of visual attribute parameters, including color channel parameters (RGB three color components), transparency parameters (the degree of opacity in the 0-1 range), and brightness parameters (light reflection intensity coefficient).

[0162] The architectural design terminal establishes a linear mapping function between intervention values ​​and visual attribute parameters. Independent mapping formulas are constructed for color, transparency, and brightness. The intervention value of each spatial point is substituted into the corresponding formula to calculate the specific RGB values, transparency coefficient, and brightness coefficient of that point. This ensures that the higher the intervention value, the more significant the difference in visual attributes (e.g., a high intervention value corresponds to deep red, low transparency, and high brightness). All parameters are integrated to generate a structured set of visual attribute parameters for each spatial point.

[0163]

[0164] In the formula, The RGB parameter value of the color at a certain spatial point. This represents the degree of intervention at that spatial location. The minimum intervention value in the privacy intervention degree distribution map. This represents the maximum intervention value in the privacy intervention degree distribution map. This represents the maximum value of the red color channel. This represents the maximum value of the green color channel. This represents the maximum value of the blue color channel. This represents the minimum value for the red color channel. This represents the minimum value for the green color channel. This is the minimum value for the blue color channel.

[0165] Among them, the intervention degree value is a spatial location privacy exposure risk indicator quantified based on preset privacy assessment standards, and the value is positively correlated with the risk level; the visual attribute parameter set is structured data containing color RGB values, transparency coefficient, and brightness coefficient, and each parameter is distributed according to the intervention degree value gradient to ensure the sense of hierarchy in visual presentation.

[0166] (3) Generate dynamic heat maps in virtual reality scenes based on visual attribute parameters.

[0167] For example, the architectural design terminal constructs a heat map data matrix with the same grid precision as the spatial grid division result of the virtual reality scene. The number of rows and columns of the matrix correspond to the number of horizontal and vertical grids in the scene space, respectively. Each matrix element corresponds one-to-one with a single spatial grid point in the virtual reality scene, storing the complete set of visual attribute parameters (color RGB value, transparency coefficient, brightness coefficient) of the point. At the same time, the update rules of the dynamic heat map layer are set, specifying the update frequency (such as updating once per second) and response trigger conditions (including real-time changes in privacy intervention value, user movement interaction in the scene, and scene perspective switching).

[0168] The architectural design terminal calls the professional layer rendering engine of the virtual reality scene, loads the constructed heat map data matrix, and renders layer by layer in the corresponding spatial area of ​​the virtual reality scene according to the coordinate system mapping relationship determined by the spatial alignment parameters. It renders the underlying basic color and then overlays transparency and brightness effects to ensure that the heat map layer matches the position of the scene space, generating a dynamic heat map layer with real-time dynamic update capability. This layer adopts a semi-transparent design to avoid obscuring the original spatial details of the virtual reality scene (such as the texture of building components and the structure of doors and windows).

[0169] The grid precision of the heatmap data matrix is ​​consistent with the visualization precision of the virtual reality scene, ensuring that the visual attributes of each spatial point can be presented. The dynamic update rule is used to maintain the synchronization between the heatmap layer and the scene state. When the trigger condition is met, the visual attribute parameters of the relevant area are automatically recalculated and the layer is updated. The dynamic heatmap layer is an independent overlay layer with interactive functions such as being able to be turned on / off and having its transparency adjusted, without affecting the core geometric structure of the virtual reality scene.

[0170] (4) The dynamic heat map layer is fused with the geometric model of the virtual reality scene at the pixel level to generate a perception verification environment.

[0171] For example, the architectural design terminal acquires pixel-level data of the dynamic heat map layer and the virtual reality scene geometric model through a pixel data acquisition program: the pixel data of the dynamic heat map layer includes the RGB color value, transparency value, and spatial coordinate index of each pixel; the pixel data of the virtual reality scene geometric model covers the texture pixel value, light reflection pixel value, and spatial position pixel coordinate of the model surface. Through a pixel coordinate calibration program, the coordinate offset of the pixels in the dynamic heat map layer is corrected based on the pixel coordinate of the geometric model of the virtual reality scene, so as to ensure that the two types of pixel data are aligned in spatial position.

[0172] The architectural design terminal employs a weighted fusion algorithm, setting a fusion weight coefficient for the dynamic heatmap layer. This coefficient (ranging from 0 to 1) is determined based on the balance between visual readability and smooth scene interaction. Pixel data from the dynamic heatmap layer and pixel data from the virtual reality scene's geometric model are calculated pixel-by-pixel according to their weights. A pixel fusion formula is used to comprehensively integrate color, transparency, and brightness, generating a fused pixel matrix. This matrix is ​​then loaded into the real-time rendering pipeline of the virtual reality scene. After anti-aliasing optimization and lighting adaptation adjustments, a perceptual verification environment is generated. The pixel fusion formula is as follows:

[0173]

[0174] In the formula, The red channel value of the merged pixel. This represents the green channel value of the merged pixel. This refers to the blue channel value of the merged pixel. This represents the transparency value of the merged pixels. A specific blending weight coefficient for dynamic heatmap layers. The red channel value of the pixels in the dynamic heatsink layer. This represents the green channel value for the pixels in the dynamic heatsink layer. This represents the blue channel value of the pixels in the dynamic heatsink layer. This represents the transparency value of the pixels in the dynamic heatsink layer. The red channel value for pixels in the geometric model of a virtual reality scene. This represents the green channel value for pixels in the geometric model of a virtual reality scene. This represents the blue channel value of the pixels in the geometric model of a virtual reality scene. This represents the transparency value of pixels in the geometric model of a virtual reality scene.

[0175] Among them, the value of the fusion weight coefficient needs to take into account both the visualization effect of privacy intervention and the original details of the virtual reality scene; pixel-level fusion ensures that the privacy risk visualization information of the dynamic heat map layer is seamlessly connected with the three-dimensional geometric information of the virtual reality scene, without obvious splicing marks; the perception verification environment has both three-dimensional spatial interaction capabilities and privacy risk visualization presentation characteristics, providing an interactive carrier for subsequent data collection and model calibration.

[0176] In one embodiment, privacy exposure parameters are calculated for the three-dimensional gaze vector set to generate a privacy intervention degree distribution map, including:

[0177] (1) Perform line penetration analysis on the three-dimensional line vector group to obtain the spatial exposure probability matrix.

[0178] For example, the architectural design terminal performs a refined line-of-sight penetration analysis on the three-dimensional line-of-sight vector group. It sequentially extracts the spatial starting coordinates, spatial propagation path, path occlusion features, and line-of-sight endpoint coordinates of each line-of-sight vector within the three-dimensional line-of-sight vector group. Through occlusion attenuation calculation logic, it determines the degree of weakening of line-of-sight propagation by various occlusion components. It calculates the probability value of direct line-of-sight penetration for each spatial grid point of the preset residential building model. The probability values ​​of all spatial grid points are arranged and integrated in coordinate order to form a complete two-dimensional numerical matrix, thus obtaining the spatial exposure probability matrix.

[0179] Among them, the path occlusion features of the three-dimensional line-of-sight vector group include solid wall occlusion, glass component occlusion, and hollow partition occlusion. Different occlusion types correspond to different line-of-sight attenuation coefficients. Each element of the spatial exposure probability matrix corresponds to a single spatial grid point of the preset residential building model, and the value is positively correlated with the line-of-sight exposure probability of that point.

[0180] (2) Construct a line-of-sight influence network based on the spatial topological relationship of the three-dimensional line-of-sight vector group.

[0181] For example, the architectural design terminal performs network construction operations based on the spatial topological relationship of the three-dimensional line-of-sight vector group. All spatial grid points covered by line-of-sight vectors in the preset residential building model are set as network nodes, and the spatial intersection, overlapping, and adjacency transmission relationships between different line-of-sight vectors are set as network edges. The corresponding binding relationship between nodes and network edges is established through the topological association algorithm to form a structured network that can reflect the propagation and transmission law of line-of-sight, thus completing the construction of the line-of-sight influence network.

[0182] The spatial topological relationship of the three-dimensional line-of-sight vector group includes the intersection points of the line-of-sight vectors, the overlapping areas of the coverage, and the spatial hierarchical distribution characteristics; the line-of-sight influence network is presented in the form of a directed network, and the connectivity of the network edges represents the propagation influence relationship of the line of sight between different spatial points.

[0183] (3) Weight the line-of-sight influence network by combining the time dimension of the pedestrian activity pattern to generate a weighted line-of-sight influence network.

[0184] For example, the architectural design terminal introduces data related to pedestrian activity patterns in the time dimension, extracts quantitative indicators such as pedestrian density, frequency of spatial activities, and duration of stay in different time intervals, and transforms these indicators into dimensionless weighted values ​​through indicator normalization. According to the intensity of pedestrian activity corresponding to spatial locations, corresponding weight values ​​are applied to all network nodes and network edges of the line-of-sight influence network to strengthen the proportion of line-of-sight influence in areas with high pedestrian activity, thereby generating a weighted line-of-sight influence network.

[0185] Among them, the time-dimensional pedestrian activity pattern is divided into time period characteristics based on the daily usage patterns of residential buildings, and the weighted value increases positively with the increase of pedestrian activity intensity; the weighted line-of-sight influence network adds the time-dimensional activity influence on the original topology, which is more in line with the privacy exposure status in actual use scenarios.

[0186] (4) The spatial exposure probability matrix and the weighted line-of-sight influence network are integrated to generate a privacy intervention degree distribution map.

[0187] For example, the architectural design terminal initiates numerical fusion calculation, couples the point values ​​of the spatial exposure probability matrix with the corresponding weight values ​​of the weighted line-of-sight influence network, uses linear fusion to complete the integration calculation of the two types of data, and then maps the fused comprehensive values ​​to the planar projection space of the preset residential building model, arranges them according to coordinates to form a visual numerical distribution layer, and generates a privacy intervention degree distribution map.

[0188] Among them, the spatial exposure probability weighting coefficient is used to balance the contribution ratio of spatial exposure probability and weighted line of sight influence, and the value range is from zero to one; the privacy intervention degree distribution map is a two-dimensional planar visualization chart that comprehensively reflects the distribution of privacy exposure risk under the combined effect of spatial exposure probability and time dimension of human flow activities.

[0189] In one embodiment, spatial vector modeling is performed based on a preset residential building model to obtain a three-dimensional line-of-sight vector set, including:

[0190] (1) Analyze the spatial topology and functional attributes of the preset residential building model to generate a spatial semantic map.

[0191] For example, the architectural design terminal retrieves the complete three-dimensional geometric data and functional zoning data of the preset residential building model, and disassembles the spatial topology information such as the distribution of walls, the location of doors and windows, and the spatial connectivity of the building layer by layer. At the same time, it extracts the functional attribute tags of areas such as bedrooms, bathrooms, and dressing rooms, and clarifies the usage scenarios and privacy association levels of each space.

[0192] The architectural design terminal structurally associates the disassembled spatial topology information and functional attribute tags, assigns a unique semantic identifier to each independent spatial unit, establishes a mapping logic for the adjacency, connectivity and hierarchical relationships between spatial units, and generates a spatial semantic map that can represent the characteristics of architectural space after integrating all associated information and mapping relationships.

[0193] The pre-defined residential building model's spatial topology includes wall partitions, spatial connectivity, and component obstruction layouts. Functional attributes include the usage, privacy level, and activity characteristics of each space. The spatial semantic graph is a structured graph integrating geometric structure and functional semantics, comprising three core components: spatial nodes, relational edges, and attribute labels.

[0194] (2) Identify privacy-critical regions in the spatial semantic graph to obtain a set of privacy-sensitive nodes.

[0195] For example, the architectural design terminal traverses all spatial nodes and attribute labels in the spatial semantic map, and filters out spatial units with high privacy requirements based on preset privacy level judgment criteria, while eliminating low privacy requirements spatial units of public activity type, thus completing the preliminary positioning of privacy-critical areas.

[0196] The architectural design terminal transforms the initially identified privacy-critical areas into discrete nodes in a spatial semantic graph. Each node is labeled with its corresponding privacy level weight and spatial coordinate information. Invalid nodes with duplicate coordinates or missing attributes are removed. After integrating all valid nodes, a set of privacy-sensitive nodes is obtained.

[0197] Among them, the privacy-critical areas are independent spaces in the pre-defined residential building model where people have high privacy needs. The privacy level determination criteria are based on the space's function and the needs of people's private activities. The privacy-sensitive node set is a set of nodes transformed from the privacy-critical areas in the spatial semantic map. Each node has independent spatial coordinates and privacy attributes.

[0198] (3) Calculate potential line-of-sight paths based on privacy-sensitive node set and spatial topology, and generate line-of-sight path network.

[0199] For example, the architectural design terminal takes the set of privacy-sensitive nodes as the core analysis object, combines the spatial topology carried by the spatial semantic graph, and calculates the possibility of visibility between different privacy-sensitive nodes and between privacy-sensitive nodes and public space nodes through the line-of-sight analysis algorithm, eliminating invalid paths that are completely blocked by walls or fixed components.

[0200] The architectural design terminal connects and integrates all spatial paths corresponding to all possible sightlines, constructs a path association system covering all privacy-sensitive nodes, clarifies the starting point, ending point, and intermediate crossing areas of each path, forms a complete path topology, and generates a sightline path network.

[0201] The line-of-sight analysis algorithm is used to determine whether there is an unobstructed visual propagation path between two spatial nodes. The occlusion coverage rate is calculated as the ratio of the length of the physical occlusion component on the path to the total path length. The line-of-sight path network is a topological network composed of multiple effective line-of-sight paths, reflecting the line-of-sight relationships between privacy-sensitive areas and other spaces.

[0202] (4) Parameterize each path in the line-of-sight path network into a direction vector and a distance parameter to obtain a three-dimensional line-of-sight vector group.

[0203] For example, the architectural design terminal extracts the starting point three-dimensional coordinates, ending point three-dimensional coordinates, and path extension trend of each independent sight path in the sight path network. Through vector operations, each path is transformed into a direction vector with spatial orientation. At the same time, the spatial distance between the starting point and the ending point of the path is calculated to obtain the corresponding distance parameters.

[0204] The architectural design terminal pairs and binds the direction vectors and distance parameters corresponding to all sight paths to form a standardized three-dimensional vector structure. Abnormal vector data with abnormal directions or missing distances are eliminated, and after integrating all compliant vector structures, a three-dimensional sight vector group is obtained.

[0205] The direction vector represents the spatial propagation direction of the line-of-sight path, and the distance parameter represents the spatial propagation length of the line-of-sight path. A three-dimensional line-of-sight vector set is a collection of multiple independent three-dimensional line-of-sight vectors, each corresponding to a valid line-of-sight path in the line-of-sight path network.

[0206] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0207] In one embodiment, such as Figure 3 As shown, this application also provides a residential sightline design system 300 based on spatial vector analysis, the system 300 comprising:

[0208] The spatial vector modeling module 301 is used to perform spatial vector modeling based on a preset residential building model to obtain a three-dimensional line-of-sight vector group.

[0209] Privacy computing module 302 is used to calculate privacy exposure parameters of three-dimensional gaze vector group and generate privacy intervention degree distribution map;

[0210] The model calibration module 303 is used to perform perceptual calibration on the preset privacy intervention degree calculation model using the privacy intervention degree distribution map, and generate a calibrated privacy intervention degree calculation model.

[0211] The design intervention module 304 is used to perform design intervention on the over-threshold exposure area through the calibrated privacy intervention degree calculation model to obtain the optimized residential building model.

[0212] Specifically, the architectural design terminal includes a spatial vector modeling module 301, a privacy computing module 302, a model calibration module 303, and a design intervention module 304.

[0213] The spatial vector modeling module retrieves complete data from a preset residential building model, analyzes the spatial topological relationships and regional functional attributes within the building, and constructs a spatial semantic graph that integrates geometric structure and functional semantics. Then, it selects areas with high privacy requirements, transforms them into discrete nodes with spatial coordinates and privacy attributes, and integrates them to obtain a set of privacy-sensitive nodes. It infers unobstructed potential sight paths between nodes and constructs a sight path network. Each path is parameterized as a direction vector and a length attribute is added to generate a three-dimensional sight vector group.

[0214] The core function of the spatial vector modeling module is to transform architectural spatial features into analyzable vector data. The complete data of the pre-set residential building model includes spatial parameters, component relationships, functional definitions, etc. The privacy-sensitive node set is used to locate the key points of privacy protection, the line-of-sight path network intuitively presents the line-of-sight propagation logic, and the three-dimensional line-of-sight vector group provides the core analysis object for subsequent privacy calculations.

[0215] The privacy computing module receives a set of three-dimensional line-of-sight vectors, analyzes the line-of-sight penetration capability, and combines it with spatial grid division to generate a spatial exposure probability matrix. At the same time, it constructs a line-of-sight influence network that reflects the impact of line-of-sight propagation, introduces pedestrian activity feature data in the time dimension for weighting, and forms a weighted line-of-sight influence network. By integrating the two types of data, the comprehensive risk value is mapped to the building plan to generate a privacy intervention degree distribution map.

[0216] Among them, the core function of the privacy computing module is to quantify the risk of privacy exposure. The spatial exposure probability matrix reflects the basic risk level of each region, the weighted line-of-sight influence network fits the actual use scenario, and the privacy intervention degree distribution map provides the core basis for subsequent model calibration and design intervention.

[0217] The model calibration module uses the privacy intervention degree distribution map as a test reference, compares the predicted data of the preset privacy intervention degree calculation model with the test data, and identifies the source of deviation. It adjusts the model parameters and logic in combination with actual perception feedback data to make up for the perception difference between the theoretical model and the actual scenario. After multiple rounds of verification, a calibrated privacy intervention degree calculation model is generated.

[0218] The core function of the model calibration module is to improve the accuracy of calculations. The preset privacy intervention degree calculation model is the initial theoretical model. The actual perception feedback data includes user subjective evaluations and objective monitoring results. After calibration, the model has higher accuracy and scenario adaptability.

[0219] The design intervention module uses a calibrated privacy intervention degree calculation model to scan the preset residential building model, detect the privacy exposure risk level of each area one by one, and locate the areas with exposure exceeding the threshold. It then formulates design optimization plans based on the regional characteristics and risk causes, applies the plans to the corresponding areas of the model to complete the adjustment and reconstruction, and generates an optimized residential building model.

[0220] The core function of the design intervention module is to reduce the risk of privacy exposure. Areas exceeding the exposure threshold are the focus of design optimization. The design scheme takes into account both privacy protection and functionality. The optimized residential building model achieves a dual improvement in privacy protection and the rationality of sightline design.

[0221] Model calibration module 303 is also used for:

[0222] Transform a pre-defined residential building model into a virtual reality scene;

[0223] By fusing privacy intervention distribution maps with virtual reality scenes, a perception verification environment can be generated;

[0224] In a perception verification environment, user eye movement trajectories are collected to obtain the actual exposure feature set;

[0225] Simultaneously acquire users' subjective ratings of the exposure events to obtain a perceptual bias dataset;

[0226] By comparing the geometric parameters of the actual exposure feature set and the privacy intervention degree distribution map, preliminary bias parameters are generated;

[0227] By combining the perception bias dataset, the initial bias parameters are corrected, and model calibration parameters are generated.

[0228] The mapping function of the privacy intervention calculation model is adjusted based on the model calibration parameters to generate a calibrated privacy intervention calculation model.

[0229] Model calibration module 303 is also used for:

[0230] Transform the spatial coordinates of the actual exposed feature set to the spatial coordinate system of the privacy intervention degree distribution map to generate a unified coordinate mapping relationship;

[0231] Based on the unified coordinate mapping relationship, the geometric deviation between the exposure intensity parameter of the actual exposure feature set and the predicted exposure probability at the corresponding location is calculated to obtain the local deviation vector set;

[0232] Density clustering is performed on local deviation vector sets to identify spatial clusters whose deviation density exceeds a preset threshold.

[0233] Extract the average deviation and spatial distribution dispersion index of the spatial clusters to generate preliminary deviation parameters.

[0234] Model calibration module 303 is also used for:

[0235] When an exposure event is detected in the perception verification environment, a real-time scoring interface is triggered;

[0236] The system collects user ratings on the degree of privacy violation of the current exposure event through a real-time rating interface, generating raw rating data.

[0237] The raw scoring data is correlated and encoded with the spatiotemporal characteristics of the exposure event to generate structured scoring units;

[0238] By aggregating structured scoring units from multiple exposure events, a perceptual bias dataset is obtained.

[0239] Model calibration module 303 is also used for:

[0240] The coordinate system of the privacy intervention degree distribution map is registered with the coordinate system of the virtual reality scene to generate spatial alignment parameters;

[0241] Based on the spatial alignment parameter, the intervention degree value of the privacy intervention degree distribution map is mapped to the visual attribute parameter;

[0242] Generate dynamic heat maps in virtual reality scenes based on visual attribute parameters;

[0243] The dynamic heat map layer is fused with the geometric model of the virtual reality scene at the pixel level to generate a perception verification environment.

[0244] Privacy computing module 302 is also used for:

[0245] A line-of-sight penetration analysis was performed on the three-dimensional line-of-sight vector set to obtain the spatial exposure probability matrix;

[0246] Construct a line-of-sight influence network based on the spatial topological relationship of three-dimensional line-of-sight vector groups;

[0247] By combining the time dimension of people's flow activity patterns, weights are assigned to the line-of-sight influence network to generate a weighted line-of-sight influence network;

[0248] By fusing the spatial exposure probability matrix and the weighted line-of-sight influence network, a privacy intervention degree distribution map is generated.

[0249] The spatial vector modeling module 301 is also used for:

[0250] Analyze the spatial topology and functional attributes of a pre-defined residential building model to generate a spatial semantic map;

[0251] Identify privacy-critical regions in the spatial semantic graph to obtain a set of privacy-sensitive nodes;

[0252] Based on the privacy-sensitive node set and spatial topology, potential line-of-sight paths are calculated, and a line-of-sight path network is generated.

[0253] Each path in the line-of-sight path network is parameterized into a direction vector and a distance parameter to obtain a three-dimensional line-of-sight vector set.

[0254] In one embodiment, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0255] In one embodiment, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the above-described method embodiments.

[0256] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0257] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A residential sightline design method based on spatial vector analysis, characterized in that, The method includes: Spatial vector modeling is performed based on a pre-defined residential building model to obtain a three-dimensional line-of-sight vector set. The privacy exposure parameters of the three-dimensional line-of-sight vector group are calculated to generate a privacy intervention degree distribution map; The privacy intervention degree distribution map is used to perform perceptual calibration on the preset privacy intervention degree calculation model to generate a calibrated privacy intervention degree calculation model; The optimized residential building model is obtained by designing interventions in the over-threshold exposure area through the calibrated privacy intervention degree calculation model.

2. The residential sightline design method based on spatial vector analysis according to claim 1, characterized in that, The step of using the privacy intervention degree distribution map to perform perceptual calibration on the preset privacy intervention degree calculation model to generate a calibrated privacy intervention degree calculation model includes: The preset residential building model is transformed into a virtual reality scene; The privacy intervention degree distribution map and the virtual reality scene are fused to generate a perception verification environment; The user's eye movement trajectory is collected in the perception verification environment to obtain the actual exposure feature set; Simultaneously acquire users' subjective ratings of the exposure events to obtain a perceptual bias dataset; By comparing the geometric parameters of the actual exposure feature set and the privacy intervention degree distribution map, preliminary deviation parameters are generated; The initial deviation parameters are corrected using the perception deviation dataset to generate model calibration parameters; The mapping function of the privacy intervention degree calculation model is adjusted based on the model calibration parameters to generate the calibrated privacy intervention degree calculation model.

3. The residential sightline design method based on spatial vector analysis according to claim 2, characterized in that, The geometric parameters of the comparison between the actual exposure feature set and the privacy intervention degree distribution map are used to generate preliminary bias parameters, including: Transform the spatial coordinates of the actual exposed feature set to the spatial coordinate system of the privacy intervention degree distribution map to generate a unified coordinate mapping relationship; Based on the unified coordinate mapping relationship, the geometric deviation between the exposure intensity parameter of the actual exposure feature set and the predicted exposure probability at the corresponding location is calculated to obtain a local deviation vector set; Density clustering is performed on the local deviation vector set to identify spatial clusters whose deviation density exceeds a preset threshold; The average deviation and spatial distribution dispersion index of the spatial cluster are extracted to generate the preliminary deviation parameters.

4. The residential sightline design method based on spatial vector analysis according to claim 2, characterized in that, The synchronous acquisition of users' subjective ratings of exposure events yields a perceptual bias dataset, including: When an exposure event is detected in the perception verification environment, a real-time scoring interface is triggered; The real-time scoring interface collects user ratings on the degree of privacy violation of the current exposure event, generating raw scoring data. The original scoring data is correlated and encoded with the spatiotemporal characteristics of the exposure event to generate structured scoring units; The structured scoring units from multiple exposure events are aggregated to obtain the perceptual bias dataset.

5. A residential sightline design method based on spatial vector analysis according to claim 2, characterized in that, The step of fusing the privacy intervention degree distribution map and the virtual reality scene to generate a perception verification environment includes: The coordinate system of the privacy intervention degree distribution map is registered with the coordinate system of the virtual reality scene to generate spatial alignment parameters; Based on the spatial alignment parameters, the intervention degree values ​​of the privacy intervention degree distribution map are mapped to visual attribute parameters; Generate a dynamic heat map in the virtual reality scene based on the aforementioned visual attribute parameters; The dynamic heat map layer is then fused pixel-level with the geometric model of the virtual reality scene to generate the perception verification environment.

6. The residential sightline design method based on spatial vector analysis according to claim 1, characterized in that, The step of calculating privacy exposure parameters for the three-dimensional gaze vector group and generating a privacy intervention degree distribution map includes: A line-of-sight penetration analysis is performed on the three-dimensional line-of-sight vector group to obtain the spatial exposure probability matrix; A line-of-sight influence network is constructed based on the spatial topological relationship of the three-dimensional line-of-sight vector group. The line-of-sight influence network is weighted by combining the time dimension of people's activity patterns, thus generating a weighted line-of-sight influence network. The spatial exposure probability matrix and the weighted line-of-sight influence network are fused to generate the privacy intervention degree distribution map.

7. The residential sightline design method based on spatial vector analysis according to claim 1, characterized in that, The spatial vector modeling based on a preset residential building model yields a three-dimensional line-of-sight vector set, including: The spatial topology and functional attributes of the preset residential building model are analyzed to generate a spatial semantic map. Privacy-critical regions are identified in the spatial semantic graph to obtain a set of privacy-sensitive nodes; Based on the privacy-sensitive node set and spatial topology, potential line-of-sight paths are calculated, and a line-of-sight path network is generated. Each path in the line-of-sight path network is parameterized into a direction vector and a distance parameter to obtain the three-dimensional line-of-sight vector group.

8. A residential sightline design system based on spatial vector analysis, characterized in that, The system includes: The spatial vector modeling module is used to perform spatial vector modeling based on a preset residential building model to obtain a three-dimensional line-of-sight vector set. The privacy computing module is used to calculate the privacy exposure parameters of the three-dimensional gaze vector group and generate a privacy intervention degree distribution map; The model calibration module is used to perform perceptual calibration on the preset privacy intervention degree calculation model using the privacy intervention degree distribution map, and generate a calibrated privacy intervention degree calculation model. The design intervention module is used to perform design intervention on the over-threshold exposure area through the calibrated privacy intervention degree calculation model to obtain an optimized residential building model.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the residential sightline design method based on spatial vector analysis according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the residential sightline design method based on spatial vector analysis according to any one of claims 1 to 7.

Images