Garden landscape simulation method and system based on virtual reality technology

Abstract

The invention relates to the technical field of image analysis, in particular to a garden landscape simulation method and system based on a virtual reality technology. The technical problem that scene switching is stiff due to rendering performance bottleneck in landscape simulation of an existing method is solved. The method comprises the following steps: acquiring historical session data and building module position information; based on the historical session data and the position information of the building module, determining the attention degrees and attention positions of the plurality of historical users to the building module, and based on the target session data, the attention degrees and the attention positions, determining the rendering priority degree of the building module; and performing hierarchical rendering scheduling on the virtual scene based on the rendering priority of each building module in the virtual scene. The method is used for the virtual garden three-dimensional image optimization scene.

Description

Technical Field

[0001] This invention relates to the field of image analysis technology, and specifically to a method and system for simulating garden landscapes based on virtual reality technology. Background Technology

[0002] Virtual reality (VR) technology has been widely applied in the field of landscape simulation, enabling immersive experiences to showcase spatial structures and material details, thus aiding in design optimization and communication. Traditional landscape display methods, such as 2D drawings or static renderings, struggle to realistically reproduce dynamic lighting and spatial layers. VR technology, however, provides designers, owners, and construction teams with intuitive spatial perception through 3D interaction. Existing VR systems typically employ rendering strategies based on the user's real-time field of view, dynamically loading scene content within the field of view to control computational load. However, landscape scenes often involve complex elements such as vegetation and water bodies, resulting in massive amounts of data. Existing rendering methods are prone to frame rate drops and latency issues during real-time processing, leading to abrupt scene transitions and impacting the user's immersion. Summary of the Invention

[0003] To address the technical problem that existing methods easily suffer from frame rate drops and latency during real-time processing, resulting in abrupt scene transitions that negatively impact user immersion, the present invention aims to provide a garden landscape simulation method and system based on virtual reality technology. The specific technical solution adopted is as follows: In a first aspect, the present invention provides a method for simulating a garden landscape based on virtual reality technology. The method includes: acquiring historical session data and building module location information; the historical session data includes: footprint points and gaze vectors of multiple historical users at multiple moments in the virtual scene; footprint points are used to represent the coordinates of historical users in the virtual scene; gaze vectors are used to represent the direction and angle of historical users' gaze at building modules; the virtual scene includes multiple building modules; building module location information is used to represent the coordinates of building modules in the virtual scene; based on the historical session data and building module location information, determining the degree of attention and attention position of multiple historical users towards building modules; the attention position is the location of the footprint point when a historical user pays attention to a building module; based on target session data, degree of attention, and attention position, determining the rendering priority of building modules; the target session data is the session data of the current user; and performing hierarchical rendering scheduling of the virtual scene based on the rendering priority of each building module in the virtual scene.

[0004] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: determining multiple sets of first reference users for the target historical user at multiple times; the target historical user being any one of the multiple historical users; the first reference users in the first set of first reference users being other historical users located within a preset range of the target historical user at the corresponding times; determining the degree of peering for the target historical user at multiple times based on the multiple sets of first reference users; the degree of peering is used to characterize the similarity of the movement paths of the target historical user and the reference users; determining the degree of peering convergence for the target historical user based on the changing trend of the degree of peering at multiple times; the degree of peering convergence is used to characterize the probability that multiple users are attracted to the same spatial region; and determining the degree of attention and the location of attention for the building module based on the degree of peering convergence.

[0005] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: clustering the gaze vectors of the first reference user set at each time step across multiple time steps based on the degree of peer convergence, and determining at least one peripheral attention vector; the peripheral attention vector is used to characterize the spatial region direction that attracts the target historical user; determining the initial attention level of the building module for each peripheral attention vector based on the cluster size corresponding to the peripheral attention vector and the degree of directional deviation of the building module relative to the peripheral attention vector; determining the maximum value of the initial attention level of the building module for each peripheral attention vector as the overall attention level of the building module towards the target historical user; determining the location of the target historical user's footprint point at the time when the overall attention level is greater than a first preset threshold as the attention location of the target historical user.

[0006] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: using a preset clustering algorithm to cluster the gaze vectors at each time step across multiple time steps; the clustering distance in the clustering algorithm is measured by: the degree of convergence of the gaze vectors in the same row and the directional angle between gaze vectors.

[0007] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: determining a second set of reference users for the current user at the current moment; the second reference users in the second set are historical users whose absolute value of the difference between their peer status and that of the current user is less than a second preset threshold; determining a rendering priority index for the building module for the second reference users based on the distance between the recently viewed position of the second reference user and the current user's position, and the level of attention corresponding to the recently viewed position; and weighting and summing the rendering priority indexes using the peer status of the current user and each second reference user as weights to determine the rendering priority of the building module.

[0008] In conjunction with the first aspect mentioned above, in one possible implementation, the rendering priority index is positively correlated with the degree of attention corresponding to the most recently viewed position; the rendering priority index is negatively correlated with the distance between the most recently viewed position of the second reference user and the current user's position.

[0009] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: determining the rendering priority of each building module in the virtual scene; determining the level of detail of each building module based on the rendering priority of each building module; and dynamically adjusting the level of detail of the building modules based on the real-time system load during the rendering process.

[0010] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes monitoring the frame rate and graphics processor load of the virtual reality device; when the frame rate is lower than a preset frame rate threshold or the graphics processor load is higher than a preset load threshold, reducing the rendering details of medium-priority and low-priority building modules.

[0011] In conjunction with the first aspect mentioned above, in one possible implementation, the method further includes: recording the footprint points and gaze vectors of each historical user based on a preset period; and cleaning the footprint points and gaze vectors of each historical user to eliminate noisy data.

[0012] Secondly, the present invention provides a garden landscape simulation system based on virtual reality technology. The system includes: a data acquisition unit for acquiring historical session data and building module location information; the historical session data includes footprint points and gaze vectors of multiple historical users at multiple moments in the virtual scene; footprint points represent the coordinates of historical users in the virtual scene; gaze vectors represent the direction and angle of historical users' gaze at building modules; building module location information represents the coordinates of building modules in the virtual scene; an analysis unit for determining the degree of attention and attention position of multiple historical users towards building modules based on the historical session data and building module location information; the attention position is the location of the footprint points when a historical user pays attention to a building module; a priority calculation unit for determining the rendering priority of building modules based on target session data, degree of attention, and attention position; the target session data is the session data of the current user; and a rendering scheduling unit for performing hierarchical rendering scheduling of the virtual scene based on the rendering priority of each building module in the virtual scene.

[0013] The present invention has the following beneficial effects: The method provided by this invention predicts the current user's region of interest by analyzing historical user behavior data. The system can intelligently preload and schedule the rendering of building modules in the virtual scene based on rendering priority before the user's gaze reaches them, thus significantly reducing computational latency and frame rate fluctuations during scene transitions. This fundamentally solves the rendering stuttering problem caused by the massive amount of data for elements such as vegetation and water in existing technologies, greatly enhancing the user's immersion in the virtual garden. While maintaining high-detail rendering, it ensures the smoothness and continuity of the visual experience in complex garden scenes, thereby resolving the technical problem of frame rate drops and latency issues that easily occur in real-time processing of existing rendering methods, leading to abrupt scene transitions that negatively impact user immersion. Attached Figure Description

[0014] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart illustrating a garden landscape simulation method based on virtual reality technology, provided as an embodiment of the present invention. Figure 2 A schematic flowchart illustrating another method for simulating garden landscapes based on virtual reality technology, provided in one embodiment of the present invention; Figure 3 A schematic flowchart illustrating another method for simulating garden landscapes based on virtual reality technology, provided in one embodiment of the present invention; Figure 4 A schematic flowchart illustrating another method for simulating garden landscapes based on virtual reality technology, provided in one embodiment of the present invention; Figure 5 A schematic flowchart illustrating another method for simulating garden landscapes based on virtual reality technology, provided in one embodiment of the present invention; Figure 6 A schematic flowchart illustrating another method for simulating garden landscapes based on virtual reality technology, provided in one embodiment of the present invention; Figure 7 A schematic flowchart illustrating another method for simulating garden landscapes based on virtual reality technology, provided in one embodiment of the present invention; Figure 8 A schematic flowchart illustrating another method for simulating garden landscapes based on virtual reality technology, provided in one embodiment of the present invention; Figure 9This is a schematic diagram of the system architecture of a garden landscape simulation system based on virtual reality technology, provided as an embodiment of the present invention. Detailed Implementation

[0016] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a garden landscape simulation method based on virtual reality technology proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0018] The following description, in conjunction with the accompanying drawings, details a specific scheme for a garden landscape simulation method based on virtual reality technology provided by the present invention.

[0019] Please see Figure 1 The diagram illustrates a flowchart of a garden landscape simulation method based on virtual reality technology provided by an embodiment of the present invention. The method includes the following steps S101-S104, which will be described in detail below.

[0020] S101. Obtain historical session data and building module location information.

[0021] One possible implementation involves acquiring time-series data containing historical user footprints and gaze vectors as the basis for analysis, while simultaneously obtaining the spatial coordinates of building modules within the virtual scene. A multi-dimensional data acquisition strategy is employed, recording users' movement trajectories and gaze directions within the virtual scene to establish a user behavior database. Once the system has collected a sufficient quantity and quality of user behavior data, it can provide reliable data support for subsequent intelligent predictions.

[0022] Understandably, footprint data can effectively reflect the spatial movement characteristics of users in virtual scenes, gaze vector data reflects the degree of visual attention users pay to different building modules in the scene, and building module location information provides the spatial location information of building modules in the virtual scene. Together, the three constitute the foundation for complete virtual scene data analysis.

[0023] S102. Based on historical session data and building module location information, determine the level of attention and location of attention of multiple historical users to building modules.

[0024] Among them, the focus location refers to the location of the footprint points when historical users focus on building modules.

[0025] In one possible implementation, the potential attention areas of historical users are identified by analyzing the convergence characteristics of user movement trajectories, and then the degree of attention and attention location of multiple historical users to building modules are determined by combining the cluster analysis of visual gaze direction.

[0026] S103. Based on the target session data, level of attention, and location of attention, determine the rendering priority of the building module.

[0027] The target session data is the session data of the current user.

[0028] One possible implementation involves acquiring multi-source data, including target session data, attention level, and attention location, as the processing object. By comprehensively considering user behavior similarity, spatial proximity, and historical attention, a quantitative evaluation system for rendering priority is established. When a specific building module achieves the optimal comprehensive evaluation result across the three dimensions of behavior matching, spatial proximity, and historical attention, that building module can be determined to have the highest rendering priority.

[0029] In one possible implementation, a second reference user set is established by calculating the similarity of behavioral characteristics between the current user and historical users; then, based on the attention position and attention level of each historical user in the set, combined with the real-time position of the current user, the rendering priority index of each building module is calculated; finally, the final rendering priority is obtained through weighted fusion.

[0030] S104. Perform hierarchical rendering scheduling of the virtual scene based on the rendering priority of each building module in the virtual scene.

[0031] One possible implementation involves a three-tiered rendering mechanism. By setting different priority thresholds, building modules are divided into high, medium, and low rendering levels. Different rendering schemes are then implemented for each level of building modules based on a preset resource allocation strategy.

[0032] Understandably, the high priority level ensures the visual fidelity of key building modules within the user's current and recent field of view, the medium priority level ensures the smoothness of scene transitions, and the low priority level effectively controls the overall rendering load of the system. The three work together to achieve the best balance between rendering quality and system performance.

[0033] The technical solution provided by the above embodiments can bring at least the following beneficial effects: This embodiment combines historical user behavior data with the current user state to achieve dynamic pre-allocation of rendering resources. The system predicts the area of ​​interest of the current user based on the group behavior pattern and performs global hierarchical rendering scheduling according to the rendering priority. This fundamentally solves the technical problem that existing methods are prone to frame rate drop and latency issues during real-time processing, resulting in abrupt scene transitions that affect the user's immersion.

[0034] Please see Figure 2 The diagram illustrates a flowchart of a garden landscape simulation method based on virtual reality technology according to an embodiment of the present invention. The method includes the following steps S201-S204, which will be described in detail below.

[0035] S201. Determine the set of multiple first reference users for the target historical user at multiple times.

[0036] The target historical user is any one of multiple historical users; the first reference user in the first reference user set is another historical user who is within the preset range of the target historical user at the corresponding time.

[0037] One possible implementation involves establishing spatial filtering conditions. A spherical spatial range with a radius of 5 meters is constructed, centered on the footprint point of the target historical user at each moment. Other historical users within this range are included in a preliminary filtering set. For historical users in the preliminary filtering set, the spatial continuity of their positions at the current moment and adjacent moments is verified, excluding unrelated users who accidentally entered the range. Users who simultaneously meet the requirements of spatial proximity and temporal continuity are included in the final first reference user set, and a corresponding reference user record is established for each target historical user at each independent moment.

[0038] Understandably, spatial filtering conditions ensure the geographical relevance of reference users, while temporal continuity verification guarantees the correlation of user behavior. Together, they constitute an accurate and reliable method for determining the first set of reference users.

[0039] S202. Based on multiple sets of first reference users, determine the degree of peer interaction of the target historical user at multiple times.

[0040] Among them, the "alignment degree" is used to quantify the consistency between the target user and the reference user on their movement paths. Its calculation is based on the temporal trend of the alignment distance between the two users and the consistency of their movement directions. When the distance tends to stabilize or decrease and the movement directions are consistent, the alignment degree is high. The alignment distance is used to quantify the spatial proximity between the target user and the reference user at the same time. This parameter is a scalar value, representing the straight-line distance between the two users' footprint points, and the distance unit is a unit length in the virtual scene coordinate system.

[0041] In one possible implementation, the degree of deviation between the target historical user and each first reference user at each time step is calculated. This parameter is quantified by combining the following two factors: a relative distance change factor, which assesses the tendency for the two to move closer or further apart in spatial location based on the difference in the distance between the current time and the previous time step; and a movement direction difference factor, which assesses the consistency of their movement paths based on the angle between the target historical user's current movement direction and the reference user's movement direction at the corresponding point in the same path. After obtaining the degree of deviation, it is converted into the degree of same-path alignment through a preset mapping relationship, where the degree of deviation and the degree of same-path alignment are negatively correlated.

[0042] For example, the target historical user in the first The moment and the first Peer deviation of the first reference user Satisfy the following formula: ] in, For target historical users in the first The moment and the first The peer distance between the first reference users represents the spatial proximity of the two users' locations. For target historical users in the first The moment and the first The peer distance of the first reference user; Let be the radian corresponding to the footprint direction of the target historical user at time b and the footprint direction of the first reference user at time a at the same point; As a normalization function, the minimum-maximum normalization method is used to map the input value to the interval [0,1]. Pi is a constant used to normalize angles.

[0043] in, Based on the deviation factor, the absolute parallel distance is normalized to quantify the spatial separation between two users. Even if their movement trends are the same, a large distance means deviation. This is a distance change penalty term used to quantify whether two users are moving closer or further apart; the greater the change, the higher the degree of deviation. This is a directional difference penalty term used to quantify the consistency of the movement directions of two users; the larger the angle, the greater the degree of deviation. To balance the weighting coefficients, used to adjust the relative importance of distance variation and direction difference in the assessment, for example, It was set to 0.5. The specific value is set based on historical experience and can be optimized and modified based on actual results. This invention does not limit this.

[0044] Understandably, for parameters such as peer compatibility and peer deviation, which require historical time-series data for trend analysis, the calculation starts from the second time point (…). =2) It officially begins, because the system only possesses the previous timeframe from the second timeframe onwards. The data (=1) is used as the calculation basis.

[0045] For example, the target historical user in the first The moment and the first The peer level of the first reference user Satisfy the following formula: in, For target historical users in the first The moment and the first The peer level of the first reference user; For target historical users in the first The moment and the first The degree of deviation of each reference user from their peers; The larger the value, the more likely the target historical user is in the [number]th [period]. The moment and the first The more reference users use it, the more likely they are to have the same path.

[0046] S203. Based on the changing trend of peer convergence at multiple time points, determine the peer convergence degree of the target historical user.

[0047] Among them, the degree of convergence among peers is used to characterize the probability that multiple users are attracted to the same spatial region.

[0048] One possible implementation involves establishing a peer trend analysis model. By analyzing the changes in the peer ratio between the target historical user and each first reference user over a continuous time period, the convergence of the user group's movement trend can be identified. When it is detected that the peer ratio between the target historical user and multiple first reference users shows a synchronous upward trend within a specific time period, it can be determined that the user group is converging towards the same spatial area.

[0049] For example, the target historical user in the first The moment and the first The rise of peers of reference users Satisfy the following formula: in, For target historical users in the first The moment and the first The rate of increase of peers of each reference user; For target historical users in the first The moment and the first The peer level of each reference user; For target historical users in the first The moment and the first The peer level of each reference user.

[0050] The above formula uses the difference between the degree of parallelism at adjacent times to reflect the trend of parallelism over time. Positive values ​​indicate that the degree of parallelism is increasing, while negative values ​​indicate that the degree of parallelism is decreasing. The changing trend of the similarity of the movement paths of target historical users and reference users provides a temporal basis for identifying common concerns of user groups.

[0051] For example, the target historical user in the first Peer convergence at time 1 Satisfy the following formula: in, For target historical users in the first The number of first reference users at any given moment, when When it is 0, The value is 0; For target historical users in the first Within 5 moments before the first moment and the second moment The formula above calculates the average increase of peers within a time window to eliminate the impact of individual abnormal fluctuations, and then normalizes the result by dividing by the number of first reference users.

[0052] Understandably, the first The time window of 5 moments before the given moment can be optimized and modified based on actual simulation requirements. The 5 moments are only used as an example here, and this invention does not limit it.

[0053] This value represents the degree of consistency between the target user's historical movement trends and those of the surrounding user group. The higher the value, the more users are moving towards the same area.

[0054] S204. Determine the attention level and attention position of the building module based on the degree of convergence among peers.

[0055] Among them, the degree of attention is used to quantify the degree of attention a building module receives from users, and the location of attention is used to determine the spatial coordinates when a user generates an attention behavior.

[0056] In one possible implementation, for each building module and each surrounding attention vector, the initial attention level of the building module in the attention direction is determined by taking into account the number of users and the directional deviation angle of the building module relative to the attention direction through coupled calculation.

[0057] In one possible implementation, for each building module, the maximum value of the initial attention level obtained on different surrounding attention vectors is taken as the overall attention level, and the historical target user location corresponding to the moment when the overall attention level exceeds a preset threshold is recorded as the attention location.

[0058] The technical solution provided by the above embodiments can bring at least the following beneficial effects: By introducing the concepts of "first reference user set" and "peer engagement," this embodiment identifies historical user groups with similar movement paths from massive amounts of historical user trajectory data. This usually means that the group is attracted by the same landscape element. Based on a data-driven discovery mechanism, it replaces the traditional method of subjectively pre-setting hotspot areas, thereby enabling more intelligent and accurate location of truly attractive building modules in the virtual scene, laying a reliable data foundation for subsequent rendering priority calculations.

[0059] Please see Figure 3 The diagram illustrates a flowchart of a garden landscape simulation method based on virtual reality technology provided by an embodiment of the present invention. The method includes the following steps S301-S304, which will be described in detail below.

[0060] S301. Based on the convergence degree of peers, cluster the gaze vectors of the first reference user set at each time step in multiple time steps to determine at least one surrounding attention vector.

[0061] Among them, the surrounding attention vector is used to represent the spatial region direction that attracts the target historical users.

[0062] One possible implementation employs a weighted optimization-based clustering analysis method. By using peer convergence as a weighting factor in the clustering process, a gaze direction clustering model considering the salience of user behavior is established. When the gaze vectors of multiple first reference users exhibit a clustered distribution in space, and these first reference users have a high degree of peer convergence, peripheral attention vectors representing the common direction of attention can be generated.

[0063] For example, the significance threshold for peer convergence is determined to be 0.6 based on historical data analysis. This value is determined based on historical data analysis and can be optimized and corrected according to the actual analysis results. For each first reference user, when its peer convergence reaches or exceeds this threshold, its gaze vector is considered a valid sample to participate in subsequent cluster analysis. Then, the valid gaze vectors are grouped by the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm. The cluster distance is measured by the directional radians between gaze vectors and the minimum peer convergence corresponding to the gaze vector. Finally, for each formed valid cluster, the weighted average direction of all gaze vectors in the cluster is calculated, and the weight of each gaze vector is its corresponding peer convergence. Finally, the peripheral attention vector in unit vector form is obtained through normalization. S302, Based on the cluster size corresponding to the peripheral attention vector and the directional deviation of the building module from the peripheral attention vector, the initial attention level of the building module for each peripheral attention vector is determined.

[0064] The initial level of attention is used to quantify the potential for a building module to receive attention in a specific area of ​​interest.

[0065] In one possible implementation, for each cluster corresponding to a surrounding attention vector, the number of attention vectors it contains is counted, and a scale factor is obtained by processing it through a normalization function. This factor reflects the basic number of users focusing on that direction; a larger value indicates that the demand represented by that direction is more prevalent. The angle between the direction of the line connecting the centroid position of each building module and the current position of the historical target user, and the surrounding attention vector, is calculated, and the angle value is converted into an alignment factor using an exponential decay function. This factor characterizes the spatial matching degree between the building module and the group's attention direction; a smaller angle indicates a higher alignment. Finally, the scale factor and the alignment factor are multiplied to obtain the initial attention level of the building module to the surrounding attention vector.

[0066] For example, the initial level of attention to building modules Satisfy the following formula: in, The level of attention given to this building module; This represents the number of gaze vectors in the cluster corresponding to the surrounding attention vector; Let be the radian of the building module in relation to the surrounding attention vector; The constant pi (π) of 3.14 is used to normalize the deviation angle. For the normalization function, the maximum and minimum value normalization method is used; It is a natural exponential function, which ensures that the effect of the deviation angle changes smoothly and avoids abrupt changes.

[0067] This reflects the scale of users interested in this area; the larger the value, the higher the level of interest. Its normalized baseline value... This represents the number of gaze vectors in the largest cluster ever observed. This represents the degree of matching between the building module's orientation and the direction of interest; the smaller the deviation, the larger the value. The product of the two values ​​reflects the combined impact of "user scale" and "orientation matching degree". This value, which characterizes the degree of attention a building module receives in a specific direction of focus, takes into account both the size of the user group and the accuracy of the visual direction, providing a quantitative basis for identifying truly attention-grabbing building modules.

[0068] S303. In determining the initial attention level of the building module for each surrounding attention vector, the maximum value of the initial attention level is the overall attention level of the building module to the target historical user.

[0069] Among them, overall attention is used to characterize the highest level of attention a building module receives across all possible areas of attention.

[0070] One possible implementation involves a maximum value selection strategy. By comparing the initial attention levels of a building module across various surrounding attention vectors, the maximum value is selected as the final overall attention level for that building module. When a building module achieves a high initial attention level in any surrounding attention vector direction, it can be determined that the building module has a high overall attention value.

[0071] S304. When the overall attention level is greater than the first preset threshold, the corresponding footprint point of the target historical user is the attention position of the target historical user.

[0072] Among them, the attention location is used to record the spatial location information when the target historical user has a substantial interest in the building module.

[0073] In one possible implementation, by setting a reasonable threshold standard, moments and locations with significant attention value are selected, and a correlation mapping between building modules and user spatial locations is established. When the overall attention of a building module exceeds a preset threshold at a certain moment, it can be determined that the target historical user has paid attention to the building module at that moment, and the location of the target historical user's footprint at that moment is recorded as the location of attention.

[0074] For example, the first preset threshold is 0.6. The first preset threshold is the significance threshold of the overall attention verified by experiments, and can be optimized and modified based on the experimental results.

[0075] The technical solution provided in the above embodiments can bring at least the following beneficial effects: This embodiment integrates discrete user gaze vectors into representative peripheral attention vectors through cluster analysis. This not only reveals the user's location but also determines the direction the user is looking, thereby accurately inferring the spatial area that attracts them. By calculating the initial attention level of building modules relative to these attention vectors, the system can quantify the attention level of each building module and record key attention locations in conjunction with spatial location. This allows the visual focus in the virtual scene to be digitized and modeled, greatly improving the accuracy of preloading decisions.

[0076] Please see Figure 4 The diagram illustrates a flowchart of a garden landscape simulation method based on virtual reality technology provided by an embodiment of the present invention. The method includes the following steps S401, which will be described in detail below.

[0077] S401. Use a preset clustering algorithm to cluster the gaze vectors at each time step across multiple time steps.

[0078] For example, the target historical user is in the first DBSCAN clustering is performed on the gaze vectors corresponding to the peer points of all first reference users at each time step, and the cluster distance is calculated. Satisfy the following formula: in, For the first The gaze vector and the first Clustering distance of each gaze vector; For the first The degree of convergence of the peer points corresponding to each gaze vector; For the first The degree of convergence of the peer points corresponding to each gaze vector; For the first The gaze vector and the first The angle between the gaze vectors; The function is for finding the minimum value; For the natural exponential function, Perform normalization processing; Pi is a constant.

[0079] The smaller of the convergence values ​​of two peers is used as the weight; This is an exponentially decaying term for the effect of the included angle; the smaller the included angle, the larger the value. This value represents the degree of dissimilarity between two gaze vectors during clustering. The smaller the value, the more likely the two gaze vectors should be clustered into one class. The above formula improves the effectiveness of clustering results by combining behavioral convergence and visual directionality.

[0080] Understandable, As weights, if the peer convergence of two gaze vectors is high, it indicates that the corresponding user is likely to be attracted. However, if a user's peer convergence is low, it suggests that the user may be indifferent to the corresponding spatial area or landscape elements. In this case, it is necessary to increase the distance between these users and interested users. ;exist Under relatively large circumstances, The larger the value, the more likely the users corresponding to the two gaze vectors are interested in the same spatial area or landscape element. Therefore, they need to be grouped into one category to obtain the location of the same spatial area or landscape element that they are interested in, providing a basis for preloading.

[0081] The technical solution provided in the above embodiments can bring at least the following beneficial effects: This embodiment optimizes the clustering algorithm itself by incorporating the peer convergence degree, a key indicator reflecting the attractiveness of a group, into the distance metric. Gaze vectors emitted by users with high peer convergence are given higher weight in clustering, making it easier to form effective surrounding attention vectors. It effectively filters out random, aimless gaze interference, ensuring that the generated attention directions truly represent the real intentions of the user group, thereby improving the robustness and accuracy of hotspot area judgment.

[0082] Please see Figure 5 The diagram illustrates a flowchart of a garden landscape simulation method based on virtual reality technology provided by an embodiment of the present invention. The method includes the following steps S501-S503, which will be described in detail below.

[0083] S501. Determine the second set of reference users for the current user at the current moment.

[0084] In this context, the second reference user in the second reference user set is a historical user whose absolute value of the difference between the current user and the current user's peer level is less than the second preset threshold.

[0085] One possible implementation involves using a behavioral similarity matching method for filtering. This involves calculating the similarity index of motion characteristics between the current user and each historical user to establish a reference user filtering mechanism based on behavioral patterns.

[0086] For example, the second preset threshold is 0.2. The second reference user in the second reference user set is a historical user whose absolute value of the difference between the current user and the current user is less than 0.2. It can be understood that the second preset threshold of 0.2 is determined based on historical experience data. The second preset threshold can be optimized and corrected in the future based on the experimental results of rendering priority.

[0087] S502. Based on the distance between the second reference user's most recently viewed location and the current user's location, and the degree of attention corresponding to the most recently viewed location, determine the rendering priority index of the building module for the second reference user.

[0088] In one possible implementation, the Euclidean distance between the nearest attention location to the second reference user and the current user's location is obtained and normalized using an inverse proportional function to obtain a distance factor. This factor is negatively correlated with the distance value; the smaller the distance, the larger the factor value. Then, the attention level value corresponding to the nearest attention location is obtained. This value has been normalized and ranges from [0, 1]. Finally, a product calculation model is established, multiplying the distance factor and the attention level factor to obtain the rendering priority index of the building module for the second reference user.

[0089] It is understandable that the rendering priority index is positively correlated with the level of attention corresponding to the most recently viewed position; the rendering priority index is negatively correlated with the distance between the most recently viewed position of the second reference user and the current user's position.

[0090] For example, the first The building module is for the current user's [number]. Rendering priority index of a reference user Satisfy the following formula: in, For the first The building module is for the current user's [number]. The rendering priority index of the second reference user; For the first The building module is for the current user's [number]. The distance between the nearest followed location of the second reference user and the current user's real-time location; For the first The building module is for the current user's [number]. The level of attention corresponding to the most recent followed position of the second reference user. If the first The building module is for the current user's [number]. If a reference user does not have an overall level of attention at any given moment, then... Recorded as 0; The function is a normalization function, and the maximum-minimum value normalization method is used.

[0091] The distance factor is normalized, with a larger value for smaller distances; This directly reflects the degree of historical attention; the multiplication of the two items reflects a comprehensive consideration of "spatial proximity" and "historical attention".

[0092] This value represents the rendering priority of building modules based on the perspective of a single reference user. The larger the value, the higher the priority of rendering the building module, providing basic data for subsequent comprehensive evaluation.

[0093] S503. Using the degree of peer relationship between the current user and each second reference user as the weight, the rendering priority index is weighted and summed to determine the rendering priority of the building module.

[0094] One possible implementation involves normalizing the weight coefficients, and then applying a weighting method to the peer compatibility values ​​for all second reference users. The function is normalized to obtain the weight coefficient of each second reference user. Then, the rendering priority index of each second reference user is multiplied by its corresponding weight coefficient, and the weighted results of all second reference users are summed to obtain the initial rendering priority of the building module. Finally, the initial rendering priority is normalized to the maximum and minimum values, and the result is mapped to the [0,1] interval to determine the rendering priority of the building module. The larger the value, the more priority the building module needs to be rendered at the current moment.

[0095] For example, the rendering priority of building modules Satisfy the following formula: In the formula, For the first The rendering priority of each building module for the current user; The number of reference users for the current user; For the current user at the current moment and the first The peer level of the second reference user; For the first The building module is for the current user's [number]. The rendering priority index of the second reference user; To normalize the exponential function, the input vector is converted into a probability distribution, ensuring a reasonable distribution of weight coefficients and preventing individual abnormal users from having an excessive impact on the results. The larger the value, the more likely the current user is to be connected to the first user at the current moment. If the travel routes of the second reference user are similar, then the first... The more reliable the prediction of the current user is from the second reference user, the better; The normalization function performs maximum and minimum value normalization on the initial rendering priority, maps the result to the interval [0, 1], and determines the rendering priority of the building module. The larger the value, the more priority the building module needs to be rendered at the current moment.

[0096] Characterizing the final rendering priority of building modules, this value integrates historical behavior data from all historically similar users, improving the accuracy and robustness of predictions through weighted fusion.

[0097] Understandably, this involves calculating the rendering priority of building modules. At that time, the cumulative operation starts from =1 starts, covering all second reference users ( =1 to p), By ensuring that the sum of the weights of each second reference user is 1, the weighted sum naturally eliminates the impact of differences in the number of users.

[0098] The technical solution provided by the above embodiments can bring at least the following beneficial effects: This embodiment filters out historical users (i.e., the "second reference user set") whose behavior is similar to that of the current user. By comprehensively calculating the distance, historical attention and weighting by peer degree, the final rendering priority combines spatial proximity, historical attention and behavioral similarity, making the preloading strategy forward-looking and targeted, and effectively avoiding the waste of resources in areas that users are not interested in at all.

[0099] Please see Figure 6 The diagram illustrates a flowchart of a garden landscape simulation method based on virtual reality technology provided by an embodiment of the present invention. The method includes the following steps S601-S603, which will be described in detail below.

[0100] S601. Determine the rendering priority of each building module in the virtual scene.

[0101] In one possible implementation, each building module in the virtual scene is traversed, and the rendering priority index and corresponding weight coefficient of all second reference users corresponding to each building module are obtained. The rendering priority of each building module is calculated one by one. The method for calculating the rendering priority has been described in embodiments S501-S503 and will not be repeated here.

[0102] S602. Determine the detailed level of each building module based on its priority.

[0103] In one possible implementation, building modules with a rendering priority higher than a third preset threshold are assigned high-priority rendering resources; high-priority rendering resources include loading high-precision models and complete textures; building modules with a rendering priority lower than the third preset threshold but higher than a fourth preset threshold are assigned medium-priority rendering resources; medium-priority rendering resources include preloading medium-precision detail models; building modules with a rendering priority lower than the fourth preset threshold are assigned low-priority rendering resources; low-priority rendering resources include loading low-resolution placeholder models.

[0104] For example, building modules with a rendering priority of ≥0.7 prioritize loading high-precision models and complete textures in the user's field of view and prediction path to achieve real-time high-fidelity display; building modules with a rendering priority of ≥0.4 preload or use medium-detail models in areas that the user may reach to ensure a smooth transition when entering the field of view; building modules with a rendering priority of ≥0.4 are low-priority modules, retaining only low-resolution or placeholder models, and gradually refining the rendering as the user approaches or gazes at them.

[0105] It is understood that the aforementioned third preset threshold of 0.7 and fourth preset threshold of 0.4 are empirical values ​​determined based on historical rendering priority experiments, and can be optimized and corrected based on actual conditions. This invention does not limit these values.

[0106] S603. During the rendering process, the detail level of the building module is dynamically adjusted based on the real-time system load.

[0107] One possible implementation employs an adaptive detail level adjustment mechanism. By monitoring system performance metrics in real time, a strategy for balancing rendering quality and system load is established. When the system load exceeds a preset threshold, the rendering detail of non-critical building modules is automatically reduced; when system resources are sufficient, the rendering quality of each building module is gradually restored.

[0108] The technical solution provided by the above embodiments can bring at least the following beneficial effects: This embodiment achieves refined and differentiated allocation of computing resources by establishing a clear hierarchical rendering scheduling mechanism. By setting different priority thresholds, the system can ensure that high-value areas within the user's field of vision and those about to be seen are presented in real time with high fidelity.

[0109] Please see Figure 7 The diagram illustrates a flowchart of a garden landscape simulation method based on virtual reality technology provided by an embodiment of the present invention. The method includes the following steps S701-S702, which will be described in detail below.

[0110] S701 monitors the frame rate and graphics processor load of virtual reality devices.

[0111] In one possible implementation, a real-time load assessment system is established based on the collection of system performance indicators at a preset period. When the system starts running the virtual scene, the performance monitoring process is automatically started to obtain frame rate and graphics processor usage data at a fixed sampling frequency, providing a basis for decision-making for subsequent rendering detail adjustments.

[0112] S702. When the frame rate is lower than the preset frame rate threshold or the graphics processor load is higher than the preset load threshold, reduce the rendering details of medium-priority and low-priority building modules.

[0113] One possible implementation involves a tiered dynamic degradation mechanism. By monitoring system performance metrics in real time, corresponding rendering optimization measures are immediately initiated when a performance bottleneck is detected.

[0114] For example, when the system frame rate is below 90 frames per second or the graphics processor load exceeds 85%, the system automatically executes the following degradation process: First, low-priority building modules are degraded: the number of model faces is reduced to 30% of the original number, the texture resolution is reduced to 512×512 pixels, dynamic lighting effects are turned off, and basic material textures are retained. Second, medium-priority building modules are degraded: the number of model faces is reduced to 60% of the original number, the texture resolution is reduced to 1024×1024 pixels, the lighting calculation model is simplified, and the main shadow effects are retained.

[0115] In one possible implementation, when the frame rate recovers to above a preset frame rate threshold and the graphics processor load is below a preset load threshold, the rendering details of the building module are gradually restored.

[0116] It is understandable that the preset frame rate threshold and preset load threshold are empirical values ​​determined based on experience in using graphics processors, and can be optimized and adjusted according to actual experimental conditions.

[0117] The technical solution provided by the above embodiments can bring at least the following beneficial effects: This embodiment endows the rendering system with dynamic adaptive capabilities, enabling the system to automatically adjust the rendering quality according to real-time performance indicators (frame rate, load), thereby ensuring the stability of the user experience. When the system load is too high, it can intelligently reduce the details of non-critical areas to maintain smoothness; when resources are sufficient, it can gradually restore the image quality to provide a better visual effect.

[0118] Please see Figure 8 The diagram illustrates a flowchart of a garden landscape simulation method based on virtual reality technology provided by an embodiment of the present invention. The method includes the following steps S801-S802, which will be described in detail below.

[0119] S801 records the footprint points and gaze vectors of each historical user based on a preset period.

[0120] In one possible implementation, a timer is used to trigger the data recording process, synchronously collecting historical user location and line-of-sight data in the virtual scene at fixed time intervals. When a historical user begins experiencing the virtual scene, the system automatically starts the data collection service to ensure the integrity and continuity of historical user behavior data.

[0121] For example, a data acquisition event is triggered based on a preset period of 0.1 seconds. The position coordinates of the user's head are obtained as footprint point data through the positioning system of the virtual reality device. At the same time, the user's gaze direction vector is obtained through the eye tracking module. It can be understood that the preset period of 0.1 seconds is determined based on the actual data acquisition needs and can be optimized and corrected according to the data acquisition needs. This disclosure does not limit this.

[0122] S802. Perform data cleaning on the footprint points and gaze vectors of each historical user to eliminate noisy data.

[0123] In one possible implementation, after acquiring historical session data, the footprint points and gaze vectors are preprocessed as necessary to remove noisy data and improve data quality.

[0124] The technical solution provided by the above embodiments can bring at least the following beneficial effects: This embodiment provides a high-quality data foundation for the entire system through a standardized data preprocessing process.

[0125] Please see Figure 9 The diagram illustrates the architecture of a virtual reality-based garden landscape simulation system 900 according to an embodiment of the present invention. The system includes: a data acquisition unit 901, used to acquire historical session data and building module location information; the historical session data includes footprint points and gaze vectors of multiple historical users at multiple moments in the virtual scene; the footprint points represent the coordinates of the historical users in the virtual scene; the gaze vectors represent the direction and angle of the historical users' gaze at the building modules; and the building module location information represents the coordinates of the building modules in the virtual scene. Analysis unit 902 is used to determine the degree of attention and attention location of multiple historical users to building modules based on historical session data and building module location information; the attention location is the location of the footprint point when the historical user pays attention to the building module; Priority calculation unit 903 is used to determine the rendering priority of building modules based on target session data, attention level, and attention location; the target session data is the current user's session data; The rendering scheduling unit 904 is used to perform hierarchical rendering scheduling of the virtual scene based on the rendering priority of each building module in the virtual scene.

[0126] The technical solution provided by the above embodiments can bring at least the following beneficial effects: This embodiment, by constructing a system architecture with clear functions and modular coupling, solidifies the innovation at the method level into a specific physical implementation, provides system-level protection that is completely corresponding to the method claims, and ensures that the inventive idea can be implemented in the form of a product combining software and hardware.

[0127] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0128] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for simulating garden landscapes based on virtual reality technology, characterized in that, The method includes: Acquire historical session data and building module location information; the historical session data includes: footprint points and gaze vectors of multiple historical users at multiple moments in the virtual scene; the footprint points are used to represent the coordinates of historical users in the virtual scene; the gaze vectors are used to represent the direction and angle of historical users' gaze at the building modules; the virtual scene includes multiple building modules; the building module location information is used to represent the coordinates of the building modules in the virtual scene; Determine multiple sets of first reference users for the target historical user at multiple times; the target historical user is any one of the multiple historical users. Based on the multiple sets of first reference users, the degree of companionship of the target historical user at multiple times is determined; the degree of companionship is used to characterize the similarity of the movement paths of the target historical user and the reference users. Based on the changing trends of peer convergence at multiple time points, the peer convergence degree of the target historical user is determined; the peer convergence degree is used to characterize the probability that multiple users are attracted to the same spatial region. The degree of attention and the location of attention for the building module are determined based on the degree of convergence among peers. The rendering priority of the building module is determined based on the target session data, the level of attention, and the location of attention; the target session data is the session data of the current user. The virtual scene is rendered in a hierarchical manner based on the rendering priority of each building module in the virtual scene.

2. The garden landscape simulation method based on virtual reality technology according to claim 1, characterized in that, The determination of the attention level and attention position of the building module based on the peer convergence degree includes: Based on the peer convergence degree, the gaze vectors of the first reference user set at each of the multiple time points are clustered to determine at least one peripheral attention vector; the peripheral attention vector is used to characterize the spatial region direction that attracts the target historical user. Based on the cluster size corresponding to the surrounding attention vectors and the degree of directional deviation of the building module relative to the surrounding attention vectors, the initial attention level of the building module for each surrounding attention vector is determined; the initial attention level is used to quantify the attention potential of the building module in a specific attention direction. In determining the initial attention level of the building module for each surrounding attention vector, the maximum value of the initial attention level is the overall attention level of the building module to the target historical user; the overall attention level is used to characterize the highest level of attention the building module receives in all possible attention directions. When the overall attention level is determined to be greater than a first preset threshold, the corresponding footprint point location of the target historical user is the attention location of the target historical user.

3. The garden landscape simulation method based on virtual reality technology according to claim 2, characterized in that, The step of clustering the gaze vectors of the first reference user set at each of the multiple time points based on the peer convergence degree to determine at least one surrounding attention vector includes: The gaze vectors at each of the multiple time points are clustered using a preset clustering algorithm; the clustering distance in the clustering algorithm is measured by the degree of convergence of the gaze vectors in the same row and the directional angle between the gaze vectors.

4. The garden landscape simulation method based on virtual reality technology according to claim 1, characterized in that, Determining the rendering priority of the building module based on the target session data, the level of attention, and the location of attention includes: Determine the second reference user set for the current user at the current time; the second reference user in the second reference user set is a historical user whose absolute value of the difference between the current user and the current user's peer level is less than a second preset threshold. Based on the distance between the most recently viewed location of the second reference user and the current user's location, and the degree of attention corresponding to the most recently viewed location, the rendering priority index of the building module for the second reference user is determined; The rendering priority of the building module is determined by weighting the rendering priority index by using the degree of peer relationship between the current user and each second reference user as the weight.

5. The garden landscape simulation method based on virtual reality technology according to claim 4, characterized in that, The rendering priority index is positively correlated with the degree of attention corresponding to the most recently viewed position; the rendering priority index is negatively correlated with the distance between the most recently viewed position of the second reference user and the position of the current user.

6. The garden landscape simulation method based on virtual reality technology according to claim 1, characterized in that, The hierarchical rendering scheduling of the virtual scene based on the rendering priority of each building module in the virtual scene includes: Determine the rendering priority of each building module in the virtual scene; The level of detail for each building module is determined based on its rendering priority. During the rendering process, the level of detail of the building module is dynamically adjusted based on the real-time system load.

7. The garden landscape simulation method based on virtual reality technology according to claim 6, characterized in that, The level of detail includes high priority, medium priority, and low priority. During the rendering process, dynamically adjusting the level of detail of the building module based on real-time system load includes: Monitor the frame rate and graphics processor load of virtual reality devices; When the frame rate is lower than the preset frame rate threshold or the graphics processor load is higher than the preset load threshold, reduce the rendering details of medium-priority and low-priority building modules.

8. The garden landscape simulation method based on virtual reality technology according to claim 1, characterized in that, The method further includes: Record the footprints and gaze vectors of each historical user based on a preset period; For each historical user's footprint points and gaze vector, data cleaning is performed to remove noisy data.

9. A garden landscape simulation system based on virtual reality technology, characterized in that, include: The data acquisition unit is used to acquire historical session data and building module location information; The historical session data includes footprint points and gaze vectors of multiple historical users at multiple moments in the virtual scene; the footprint points are used to represent the coordinates of historical users in the virtual scene; the gaze vectors are used to represent the direction and angle of historical users' gaze at the building modules; and the building module location information is used to represent the coordinates of the building modules in the virtual scene. The analysis unit is used to determine the degree of attention and attention location of the multiple historical users to the building module based on the historical session data and the building module location information; the attention location is the location of the footprint point when the historical user pays attention to the building module; The priority calculation unit is used to determine the rendering priority of the building module based on the target session data, the level of attention, and the attention location; the target session data is the session data of the current user; The rendering scheduling unit is used to perform hierarchical rendering scheduling of the virtual scene based on the rendering priority of each building module in the virtual scene.

Images