AI assistance-based landscape garden scheme auxiliary generation method and system

Abstract

The invention provides an AI assistance-based landscape garden scheme auxiliary generation method and system, and the method comprises the steps: firstly receiving initial topographic data and a landscape adjustment instruction containing a visual presentation requirement and an ecological function requirement, which are inputted by a user in an interactive interface, and then processing the initial topographic data through a topographic segmentation model; the method comprises the following steps: extracting gradient distribution, altitude interval and land cover features, dynamically correcting the gradient distribution features into stepped topographic features according to visual requirements, matching the land cover features according to ecological requirements to obtain vegetation distribution features, performing spatial superposition on the two features, and performing path connectivity analysis in combination with tourist flow data to determine touring path features. Finally, a landscape garden scheme containing a three-dimensional visualization model and ecological evaluation parameters is generated through multi-dimensional rendering according to the sightseeing path features and sent to a user terminal to achieve interactive feedback, so that various requirements are efficiently integrated, and scientificity and intuition of landscape garden design are improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an AI-assisted landscape gardening scheme auxiliary generation method and system. Background Art

[0002] In the field of landscape gardening design, traditional design methods face many limitations. In the past, the generation of landscape gardening plans mainly relied on the experience and manual operation of designers. When facing the design requirements put forward by users, designers often need to spend a lot of time and energy to conduct field surveys of the terrain and manually draw various sketches and design plans. This method is not only inefficient, but also difficult to fully and accurately meet the diverse needs of users due to the limitations of personal experience and cognition. For example, for some complex terrains and special visual presentation requirements, designers may need to modify the plan repeatedly, resulting in a long design cycle and increased costs.

[0003] In terms of terrain analysis, traditional methods usually only conduct simple terrain measurements and obtain preliminary data such as slope and altitude, and cannot deeply explore the various characteristics of the terrain. The comprehensive analysis of slope distribution characteristics, altitude interval characteristics, and surface cover characteristics is not comprehensive and detailed enough, making it difficult to fully integrate these characteristics into landscape design, thereby failing to give full play to the advantages of the terrain and achieving a perfect fusion of landscape and terrain. In addition, in terms of tour route planning, traditional practices are mainly determined by experience and general site layout, and rarely combined with tourist flow data for scientific analysis, which leads to the possibility that the designed tour routes may be unreasonable, such as too many tourists in some areas, while few people visit other areas, affecting the overall tourist experience. Summary of the invention

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides an AI-assisted landscape gardening scheme auxiliary generation method, the method comprising: Receiving initial terrain data and landscape adjustment instructions input by a user in an interactive interface, wherein the landscape adjustment instructions include visual presentation requirements and ecological function requirements for the initial terrain data; Performing terrain segmentation processing on the initial terrain data by using a terrain segmentation model to extract slope distribution characteristics, altitude interval characteristics and surface coverage characteristics in the initial terrain data; Dynamically modify the slope distribution characteristics according to the visual presentation requirements to generate stepped terrain characteristics that meet the height difference threshold, and perform vegetation adaptability matching on the surface cover characteristics based on the ecological function requirements to output vegetation distribution characteristics; The stepped terrain features and the vegetation distribution features are spatially superimposed to generate composite terrain vegetation features, and the composite terrain vegetation features are subjected to path connectivity analysis in combination with preset tourist flow data to determine the characteristics of the tour path; The composite terrain vegetation features are rendered in multiple dimensions according to the tour path features, a landscape gardening scheme including a three-dimensional visualization model and ecological assessment parameters is generated, and the landscape gardening scheme is sent to a user terminal for interactive feedback.

[0005] On the other hand, an embodiment of the present invention further provides an AI-assisted landscape gardening plan auxiliary generation system, including a processor and a machine-readable storage medium, the machine-readable storage medium is connected to the processor, the machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0006] Based on the above aspects, the embodiment of the present application realizes the intelligent and personalized customization of the whole process from terrain data input to the output of complete landscape gardening scheme. First of all, by integrating the initial terrain data and landscape adjustment instructions input by the user, covering the visual presentation requirements and ecological function requirements, it breaks the limitations of relying on manual processing and experience judgment in traditional landscape gardening design, and can quickly and accurately understand and respond to diverse design demands, greatly improving design efficiency and user satisfaction. The application of the terrain segmentation model deeply mines the slope distribution characteristics, altitude range characteristics and surface cover characteristics from the initial terrain data, and dynamically corrects the slope distribution characteristics according to the visual presentation requirements to generate stepped terrain characteristics that meet the height difference threshold, which not only meets the aesthetic needs, but also takes into account the feasibility of the project; based on the ecological function requirements, the surface cover characteristics are matched with vegetation adaptability to output vegetation distribution characteristics, ensuring the scientificity and sustainability of the garden ecosystem. In addition, the stepped terrain features and vegetation distribution features are spatially superimposed to generate composite terrain vegetation features, and the path connectivity analysis is combined with the preset tourist flow data to determine the tour path features, which comprehensively optimizes the spatial layout and tour experience of the landscape garden, not only creating a reasonable and efficient tour route, but also making the landscape and natural ecology harmoniously integrated, bringing tourists a unique and rich tour experience. Finally, a landscape garden scheme containing a three-dimensional visualization model and ecological assessment parameters is generated, which can not only present the design effect in an intuitive and vivid way, which is convenient for users to preview and adjust in advance, but also the ecological assessment parameters provide a quantitative basis for the long-term development and ecological balance of the garden, realizing the transformation of landscape garden design from pure artistic creation to scientific decision-making and sustainable development. The scheme is sent to the user terminal through interactive feedback, forming a closed-loop and continuously optimized design process. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 It is a schematic diagram of the execution flow of the AI-assisted landscape gardening scheme auxiliary generation method provided in an embodiment of the present invention.

[0008] Figure 2 It is a schematic diagram of the hardware architecture of the AI-assisted landscape gardening scheme auxiliary generation system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0009] The present invention will be described in detail below with reference to the accompanying drawings. Figure 1 The present invention is a flowchart of an AI-assisted landscape gardening scheme auxiliary generation method provided by an embodiment of the present invention. The AI-assisted landscape gardening scheme auxiliary generation method is introduced in detail below.

[0010] Step S110, receiving initial terrain data and landscape adjustment instructions input by the user in the interactive interface, wherein the landscape adjustment instructions include visual presentation requirements and ecological function requirements for the initial terrain data.

[0011] Taking the landscape scene of scenic spots in mountainous areas with high altitude and high drop as an example, the initial terrain data can be obtained through high-precision terrain measurement equipment. For example, a drone equipped with a laser radar can be used to scan the entire mountain area to obtain detailed terrain data including the height of the mountains, the undulations of the slopes, the depth of the valleys, etc. The above terrain data is input into the data storage system in a digital form.

[0012] The visual presentation requirements in the landscape adjustment instructions are based on the perspective of tourists' viewing experience. For example, in order to allow tourists to have a wide and rich view at different viewing locations, it may be required to set up viewing platforms at certain high points, and to have clear visual corridors between different altitude areas to avoid vegetation or terrain blocking the view. At the same time, in order to match the magnificence of the mountainous area, the height difference of the terrain may be required to have a strong visual impact.

[0013] In terms of ecological function requirements, the ecosystem in mountainous areas is relatively fragile and needs to be protected. For example, it is required to conserve water and soil and reduce the erosion of soil by slope runoff. At the same time, in order to maintain the biodiversity in mountainous areas, it is necessary to provide a suitable living environment for different types of animals and plants, such as providing forests for birds to live in and foraging and hiding places for small mammals.

[0014] Step S120, performing terrain segmentation processing on the initial terrain data through a terrain segmentation model, and extracting slope distribution characteristics, altitude interval characteristics and surface coverage characteristics in the initial terrain data.

[0015] For example, for mountainous areas with high altitudes and high drop, the initial terrain data is input into the pre-trained multi-scale terrain segmentation model. Among them, since there are often more exposed rock parts in high-altitude mountainous areas, the first convolution branch used to identify the exposed rock areas performs edge detection on the rock texture. For example, for rock surfaces with obvious ridges and rough textures, a rock distribution mask is generated. Based on the rock distribution mask, the geological stability parameters of each area can be calculated. For example, in steep cliff areas, the geological stability is poor and geological disasters such as rock slides are prone to occur.

[0016] The second convolution branch performs cluster analysis on the spectral features of vegetation in the initial terrain data. In high-altitude mountainous areas, vegetation distribution shows obvious vertical zoning. In lower altitude areas, there may be dense tree areas, such as pine trees, cypress trees and other tall trees grow luxuriantly; as the altitude increases, it transitions to the shrub transition zone, such as azalea and other shrubs are more common; in higher altitude areas, there are herbaceous plant areas, such as various cold-resistant herbaceous plants in alpine meadows. By counting the vegetation coverage density in each area, we can understand the distribution of vegetation.

[0017] The third convolution branch simulates the flow direction of the hydrological data in the initial terrain data. In mountainous areas, the flow direction of water is greatly affected by the terrain. The runoff collection area is usually located at the bottom of the valley, where streams or small lakes are easily formed. Potential erosion areas can be those with steep slopes and less vegetation cover, because water flow can easily carry away soil here. By marking the coordinates of the water area boundary, the scope of the water area can be clearly determined.

[0018] Finally, the geological stability parameters, vegetation coverage density and water boundary coordinates are integrated to generate a three-dimensional terrain grid. From this three-dimensional terrain grid, the continuous slope change curve can be extracted as the slope distribution feature, and different altitude zones can be divided as the altitude interval feature. For example, the area above 3,000 meters above sea level can be divided into high mountain areas, and 2,000-3,000 meters can be divided into medium mountain areas. At the same time, the spatial distribution of different surface types is marked as the surface coverage feature, such as clarifying which areas are rocks, which are vegetation, and which are water areas.

[0019] Step S130, dynamically modifying the slope distribution characteristics according to the visual presentation requirements, generating stepped terrain characteristics that meet the height difference threshold, and performing vegetation adaptability matching on the surface cover characteristics based on the ecological function requirements, and outputting vegetation distribution characteristics.

[0020] In the landscape of scenic spots in high-altitude mountainous areas, the viewing point setting parameters and visual corridor parameters in the visual presentation requirements are analyzed. Assuming that a viewing point is set at the top of a mountain, the maximum allowable slope value of the target area is determined according to this requirement. For example, in order to ensure the safety and comfort of tourists on the viewing platform, the slope within a certain range around the viewing platform cannot exceed 30 degrees. The height difference threshold between adjacent steps is set to 1.5 meters to facilitate tourists to climb.

[0021] Therefore, in the slope distribution characteristics, steep slope areas with a gradient of more than 30 degrees are identified, and contours of such steep slope areas are encrypted. For example, the original sparse contour spacing is 10 meters, which becomes 5 meters after encryption, generating an encrypted contour dataset. Then, the encrypted contour dataset is interpolated in a stepped manner based on the height difference threshold of 1.5 meters. A transition platform is inserted between adjacent contours, and the width of the transition platform is adjusted to 2 meters and the inclination angle is 5 degrees according to the viewing point setting parameters to ensure that tourists can stand steadily on the platform and enjoy the scenery. The interpolated contour lines are smoothed by curvature to eliminate the sharp turning points of the step edges and form a stepped surface with continuous transition. Finally, the stepped surface is edge-fused with the original terrain data to generate a stepped terrain feature containing anti-slip textures (such as anti-slip lines) and drainage grooves (groove depth of 0.1 meters and width of 0.2 meters) on the step surface.

[0022] Vegetation adaptability matching is performed on the surface cover characteristics based on the ecological function requirements. A vegetation type set is constructed based on soil and water conservation grade parameters and biodiversity parameters. For example, in high-altitude mountainous areas, in order to meet the requirements of soil and water conservation, a vegetation type set is constructed that includes a cold-resistant plant library (such as alpine azalea, snow lotus, etc.), a slope-fixing plant library (such as sea buckthorn and other plants with well-developed root systems) and an ornamental plant library (such as alpine flowers that bloom in different seasons).

[0023] In the land cover feature, shallow soil areas with soil thickness lower than the set thickness (such as 0.3 meters) are marked. These areas may be located in the rock crevices on the top of the mountain. Erosion-prone areas with slopes greater than the critical erosion angle (such as 45 degrees) are generally steep slopes. Sunny slope areas are slopes facing the direction of sufficient sunlight.

[0024] For shallow soil areas, plant species with well-developed root systems and tolerance to barrenness in the cold-resistant plant library are matched, such as alpine rhododendron, whose roots can penetrate into rock cracks and fix the soil. For erosion-prone areas, plant species with a network root structure in the slope-fixing plant library are matched, such as sea buckthorn, whose roots can interweave with each other to prevent soil from being washed away by slope runoff. For sunny slope areas, plant species with staggered flowering periods in the ornamental plant library are matched, such as primroses that bloom in spring and nasturtiums that bloom in summer, so that tourists can be provided with different ornamental flowers in different seasons. Based on the matching results of each area, a vegetation planting coordinate set is generated, and plant growth simulation parameters are added to it, such as the predicted value of the shading range of alpine rhododendron is a circular area with a radius of 1 meter, the root expansion rate is 0.2 meters per year, and the seasonal phase change data is flowering in spring and lush leaves in summer, etc., to form vegetation distribution characteristics.

[0025] Step S140, spatially superimposing the stepped terrain features and the vegetation distribution features to generate composite terrain vegetation features, and performing path connectivity analysis on the composite terrain vegetation features in combination with preset tourist flow data to determine the tour path features.

[0026] In the three-dimensional coordinate system of high-altitude mountainous areas, the vertex coordinates of the stepped terrain features and the planting coordinates of the vegetation distribution features are spatially aligned. For example, the vertex coordinates of a stepped platform of the stepped terrain are (x1, y1, z1), and the coordinates of a plant in the vegetation planting coordinate set are (x2, y2, z2). They are spatially matched. In this process, conflict points in the overlapping coordinate area are detected.

[0027] Suppose a conflict point is found in the drainage ditch area at the edge of a step. According to the priority judgment principle, the structural parameters of the stepped terrain features are retained first, and the vertical height of the vegetation planting coordinates is adjusted by 0.5 meters to avoid conflicts between vegetation and the drainage ditch structure. If a conflict point is found in the viewing area of ​​the stepped platform, the seasonal variation data of the vegetation distribution characteristics are retained first, and the local curvature of the stepped platform is modified, such as adjusting the original curvature radius from 5 meters to 4 meters to meet the growth needs of vegetation.

[0028] Perform voxel fusion on the adjusted stepped terrain features and vegetation distribution features. Obtain the surface vertex coordinate set in the adjusted stepped terrain features and the updated vegetation planting coordinate set in the vegetation distribution features. For example, extract elevation gradient data from the surface vertex coordinate set, such as the elevation of a slope from 1000 meters to 1200 meters, with a slope of 30 degrees; extract vertical height offset from the vegetation planting coordinate set, such as a plant is 0.3 meters higher than the surrounding terrain.

[0029] Based on the spatial distribution relationship between elevation gradient data and vertical height offset, the surface vertex coordinate set of the stepped terrain features is aligned with the vegetation planting coordinate set in a three-dimensional spatial grid to generate a spatial alignment data set with a unified coordinate system reference. In the spatial alignment data set, the coordinates of the drainage ditch at the edge of the stepped platform and the coordinates of the vegetation root extension range are identified to overlap, and the angle anomalies between the normal vector direction of the terrain surface and the vegetation planting direction in the coordinate overlap area are detected. For example, it is found that the normal vector direction of the bottom surface of a drainage ditch is vertically downward, while the growth direction of the vegetation root system is inclined downward by 30 degrees, forming an angle anomaly.

[0030] According to the position distribution of the angle anomaly points, the surface vertex coordinates of the stepped terrain features are displacement compensated so that the bottom surface normal vector direction of the drainage ditch forms a vertical orthogonal relationship with the growth direction of the vegetation root system. The displacement-compensated surface vertex coordinates and the vegetation planting coordinate set are imported into the preset voxel partitioning engine. The maximum elevation difference in the displacement-compensated surface vertex coordinates is 500 meters and the minimum curvature radius is 3 meters. According to the maximum elevation difference, the number of vertical layers of the voxel partitioning engine is determined to be 50 layers. The initial value of the voxel side length in the horizontal direction is calculated based on the minimum curvature radius as 1 meter, and the dynamic encryption algorithm is started in the area where the curvature change rate exceeds the threshold (such as 0.5), and the initial value of the voxel side length is reduced according to the inverse ratio of the curvature change rate. For example, in the area where the curvature change rate is 0.8, the voxel side length is reduced to 0.6 meters. Based on the number of vertical layers and the dynamically adjusted voxel side length, a non-uniform voxel grid with adaptive resolution is generated, in which the vertex coordinates of each voxel unit form a one-to-one mapping relationship with the surface vertex coordinates of the stepped terrain features.

[0031] Traverse each voxel unit in the non-uniform voxel grid. If the voxel unit contains both the surface vertex coordinates of the stepped terrain features and the vegetation planting coordinates, extract the rock texture roughness parameter, soil moisture parameter and vegetation shading range prediction value corresponding to the voxel unit. For example, the micro-surface concave-convex height data corresponding to the rock texture roughness parameter in a certain voxel unit is 0.02 meters, and the particle size distribution data in the soil diffuse reflection parameter of the adjacent voxel is 0.01 meters. According to the proportional relationship between the micro-surface concave-convex height data and the particle size distribution data, the material mixing weight coefficient is calculated to be 0.67, and the specular reflection intensity of the rock material and the diffuse reflection intensity of the soil material are linearly interpolated based on the weight coefficient. The interpolated reflection intensity data is mapped to all surface vertices of the voxel unit to generate a natural transition material effect from rock to soil.

[0032] The vegetation growth simulation parameters are injected into the non-uniform voxel grid, and the voxel units corresponding to the vegetation planting coordinate set are marked in it, and the initial root depth value and canopy coverage are assigned to each marked voxel unit. For example, the initial root depth value of the voxel unit corresponding to a certain alpine azalea is 0.1 meters, and the canopy coverage is 0.1. The daily growth of the root depth value is dynamically calculated according to the soil moisture parameter in the voxel unit. For example, when the soil moisture content is 30%, the daily growth of the root depth is 0.001 meters. At the same time, the seasonal expansion rate of the canopy coverage is calculated based on the light attenuation coefficient. For example, when the light is sufficient in summer, the seasonal expansion rate of the canopy coverage is 0.05 per month. When the canopy coverage of the adjacent voxel unit extends to the current voxel unit, the canopy gap closure algorithm is triggered to automatically adjust the leaf density to eliminate the discontinuous shadow boundary in the light simulation.

[0033] The cross-sectional geometric parameters of the drainage ditch in the stepped terrain features are extracted, such as the bottom inclination angle of 5 degrees and the roughness level of the ditch wall of level 3. The gravity acceleration component of the water flow is calculated to be 0.87 m / s² based on the inclination angle. The viscous resistance coefficient of the water boundary layer is determined to be 0.05 based on the roughness level of the ditch wall, and the dynamic equilibrium equation of the surface tension is solved in combination with the gravity acceleration component. The solution of the dynamic equilibrium equation is discretized into a water film thickness distribution map that matches the voxel unit size, such as the water film thickness is 0.01 m at the bottom of the ditch and 0.005 m at the edge of the ditch, and the intensity gradient data of the reflected light spot is generated based on the water film thickness.

[0034] Verify the physical compatibility of terrain material properties, vegetation growth properties, and light reflection properties within each voxel unit. Scan the light reflection property data of each voxel unit in the non-uniform voxel grid to detect whether the brightness value of the specular reflection highlight area exceeds the photosynthesis saturation threshold of the vegetation leaves. If a highlight area exceeding the saturation threshold is detected, the reflection intensity of the area is proportionally reduced according to the leaf density parameter in the predicted value of the vegetation shading range until it reaches the photosynthetic effective radiation range. Synchronously update the reflectivity map in the terrain material properties of the voxel unit to ensure the visual continuity of the highlight area with the surrounding material after the reflection intensity is adjusted.

[0035] Each voxel unit in the non-uniform voxel grid is encoded as a multidimensional data block containing terrain material attributes, vegetation growth attributes, and light reflection attributes. An octree index structure is constructed according to the spatial arrangement order of the multidimensional data blocks, and the voxel cluster identifiers of the material transition area are marked in the octree index structure. The octree index structure is divided into LOD levels, and the voxel data blocks of the corresponding level are dynamically called according to different observation distances to generate a composite voxel model that supports multi-resolution rendering as a composite terrain vegetation feature.

[0036] Combined with the preset tourist flow data, the path connectivity analysis of the composite terrain vegetation characteristics was performed. The peak traffic density of the historical tourist flow data was 5 people per square meter and the average stay time was 30 minutes. Combined with the viewing platform area of ​​100 square meters in the composite terrain vegetation characteristics, the maximum carrying capacity of the viewing platform was calculated to be 150 people.

[0037] Candidate routes for emergency evacuation channels are marked in the composite terrain vegetation features, such as the relatively gentle route along the bottom of the valley and the open route along the ridge. The cumulative climbing height and slope change rate of each candidate route are calculated based on the height difference data of the stepped terrain features. For example, the cumulative climbing height of the route at the bottom of the valley is 100 meters, and the slope change rate is 0.1; the cumulative climbing height of the ridge route is 200 meters, and the slope change rate is 0.2.

[0038] Diversion nodes are set for candidate routes according to the maximum carrying capacity, and virtual coordinates of indicator signs are added at the diversion nodes. For example, diversion nodes are set at the exit of the viewing platform, and the directions of different routes and the expected density of people flow are marked at the nodes. Based on the cumulative climbing height and slope change rate, a set of paths that meet the ergonomic comfort threshold (such as cumulative climbing height not exceeding 200 meters and slope change rate not exceeding 0.3) are selected.

[0039] Topological optimization is performed on each path in the path set. A node connection graph is constructed for each path in the path set, and intersection nodes with a degree greater than a set degree (such as 3) are identified in the node connection graph. Direction guidance signs are set at intersection nodes, such as at a node where three paths meet, signs are set to indicate the scenic spots to which each path leads. The connection priority of intersection nodes is adjusted according to the field of view openness parameters of adjacent paths. For example, paths with a high field of view have a higher connection priority. The curvature radius of paths with a connection priority lower than the set priority is expanded to ensure a safe distance when two-way traffic flows meet, such as expanding the curvature radius from 2 meters to 3 meters, and adding anti-slip material signs at the location where the curvature radius changes. The path topology structure is generated based on the optimized node connection graph, and the path topology structure is dynamically associated with the coordinates of the viewing platform in the composite terrain vegetation feature, and the branch guidance parameters in the tour path feature are updated.

[0040] Step S150, performing multi-dimensional rendering on the composite terrain vegetation features according to the tour path features, generating a landscape gardening scheme including a three-dimensional visualization model and ecological assessment parameters, and sending the landscape gardening scheme to a user terminal for interactive feedback.

[0041] Specifically, the voxel model in the composite terrain vegetation feature is fused with the topological structure in the tour path feature to generate a scene rendering basic data set. Rock texture maps are loaded based on the material properties in the scene rendering basic data set, such as loading realistic granite texture maps for exposed rock surfaces; dynamic vegetation growth animations are loaded, such as making the flowers of alpine azaleas change dynamically from budding to blooming in different seasons; water flow particle effects are loaded to simulate the flow effect of water flow in the streams in the valley, and the shadow projection range is calculated according to the illumination angle parameters. For example, when the sun shines from the east in the morning, the shadow range on the west side of the peak is calculated.

[0042] Real-time detection of visual occlusion areas in the 3D visualization model during the rendering process. Set multiple preset observation perspectives in the 3D visualization model of high-altitude mountainous areas, such as setting observation perspectives at the bottom of the valley, on the mountainside, and on the top of the mountain, and calculate the visible domain range under each observation perspective. Compare the visible domain range with the key display area in the landscape adjustment instruction, such as the key display area is a rare plant community on a certain peak, and identify the visual blind area that does not cover the key display area. Insert virtual observation points at the coordinate positions corresponding to the visual blind area, such as inserting virtual observation points in the area blocked by the peak. Generate a temporary connection path from the nearest tour path to the virtual observation point based on the path optimization algorithm. Adjust the rendering level of the 3D visualization model, and add a transparent terrain profile at the end point of the temporary connection path to expose the landscape details in the visual blind area, such as the distribution of rare plant communities.

[0043] Extract biodiversity parameters from vegetation distribution characteristics and soil and water conservation parameters from stepped terrain characteristics to generate an ecological benefit evaluation matrix. For example, biodiversity parameters can include the number of plant species and the number of animal habitats; soil and water conservation parameters can include the reduction ratio of slope soil loss, etc. Associate the ecological benefit evaluation matrix with the three-dimensional visualization model.

[0044] The landscape gardening scheme including the 3D visualization model and ecological assessment parameters is sent to the user terminal for interactive feedback. The rotatable thumbnail of the 3D visualization model is displayed in the user terminal interface, and the user can view the landscape from different angles by rotating the thumbnail. At the same time, a comparative radar chart of ecological assessment parameters is displayed, such as comparing the ecological benefits of the current scheme with the original terrain, intuitively showing the improvement in biodiversity protection, soil and water conservation, etc.

[0045] Receive user annotation modification instructions for the designated viewing area. For example, if the user is not satisfied with the vegetation types around a certain viewing platform, he / she may request to replace the original alpine azalea with nasturtium. Locate the target data layer in the landscape garden plan according to the annotation modification instructions, such as the vegetation distribution characteristic data layer, and call the incremental update algorithm to perform local parameter replacement on the target data layer. Compare the updated landscape garden plan with the original plan and generate a modification impact report. The modification impact report includes the change in ecological parameters, such as the impact on biodiversity after vegetation replacement; construction cost floating values, such as the difference between the planting cost of nasturtium and alpine azalea; visual effect comparison chart, such as showing the difference in visual effects between the two vegetation types on the viewing platform, and the modification impact report is superimposed and displayed on the user terminal interface so that the user can fully understand the impact of the modification.

[0046] Based on the above steps, the embodiment of the present application realizes the full process of intelligentization and personalized customization from terrain data input to the output of complete landscape gardening scheme. First, by integrating the initial terrain data and landscape adjustment instructions input by the user, covering the visual presentation requirements and ecological function requirements, it breaks the limitations of traditional landscape gardening design that relies on manual processing and experience judgment, and can quickly and accurately understand and respond to diverse design demands, greatly improving design efficiency and user satisfaction. The application of the terrain segmentation model deeply mines the slope distribution characteristics, altitude range characteristics and surface cover characteristics from the initial terrain data, and dynamically corrects the slope distribution characteristics according to the visual presentation requirements to generate stepped terrain characteristics that meet the height difference threshold, which not only meets the aesthetic needs, but also takes into account the feasibility of the project; based on the ecological function requirements, the surface cover characteristics are matched with vegetation adaptability to output vegetation distribution characteristics, ensuring the scientificity and sustainability of the garden ecosystem. In addition, the stepped terrain features and vegetation distribution features are spatially superimposed to generate composite terrain vegetation features, and the path connectivity analysis is combined with the preset tourist flow data to determine the tour path features, which comprehensively optimizes the spatial layout and tour experience of the landscape garden, not only creating a reasonable and efficient tour route, but also making the landscape and natural ecology harmoniously integrated, bringing tourists a unique and rich tour experience. Finally, a landscape garden scheme containing a three-dimensional visualization model and ecological assessment parameters is generated, which can not only present the design effect in an intuitive and vivid way, which is convenient for users to preview and adjust in advance, but also the ecological assessment parameters provide a quantitative basis for the long-term development and ecological balance of the garden, realizing the transformation of landscape garden design from pure artistic creation to scientific decision-making and sustainable development. The scheme is sent to the user terminal through interactive feedback, forming a closed-loop and continuously optimized design process.

[0047] In a possible implementation, step S120 includes: Step S121, inputting the initial terrain data into a pre-trained multi-scale terrain segmentation model, wherein the multi-scale terrain segmentation model comprises a first convolution branch for identifying rock exposed areas, a second convolution branch for identifying vegetation covered areas, and a third convolution branch for identifying water distribution areas.

[0048] First, the terrain data obtained by scanning the entire high-altitude mountain area with high-precision measurement equipment, such as drones equipped with lidar, can be input into the pre-trained multi-scale terrain segmentation model. The first convolution branch in the multi-scale terrain segmentation model is used to identify rock exposed areas. In high-altitude mountainous areas, rock exposed areas are more common. The first convolution branch performs edge detection on the rock texture in the initial terrain data, such as accurately detecting rock surfaces with obvious ridges, rough textures, and colors that are significantly different from the surrounding environment.

[0049] Step S122: performing edge detection on the rock texture in the initial terrain data through the first convolution branch to generate a rock distribution mask, and calculating the geological stability parameters of each area according to the rock distribution mask.

[0050] After the rock distribution mask is generated, the geological stability parameters of each area can be calculated based on the rock distribution mask. For example, near the top of a mountain or in a steep cliff area, the rock distribution is more concentrated and the geological structure is complex. The geological stability parameters of these areas are low, which means that geological disasters such as rock slides and landslides are more likely to occur.

[0051] Step S123, clustering analysis is performed on the vegetation spectral features in the initial terrain data through the second convolution branch to divide the tree-dense area, shrub transition area and herbaceous plant area, and the vegetation coverage density of each area is counted.

[0052] In the vertical vegetation distribution belts of high-altitude mountainous areas, different types of vegetation have different spectral characteristics. The spectral reflectance of the leaves of tall trees (such as pine and fir) in dense tree areas has a specific numerical range in certain bands. By analyzing these spectral characteristics, dense tree areas can be accurately divided. In areas with slightly higher altitudes, the spectral characteristics of shrub vegetation such as rhododendron in the shrub transition zone are different. The second convolution branch can accurately identify and divide this area. As the altitude increases further, the spectral characteristics of alpine meadow vegetation (such as fescue) in the herbaceous plant area are also identified. At the same time, the vegetation coverage density of each area is counted. For example, the vegetation coverage density in the dense tree area may reach 70%, 50% in the shrub transition area, and 30% in the herbaceous plant area, which helps to fully understand the distribution of vegetation in mountainous areas.

[0053] Step S124, performing flow direction simulation on the hydrological data in the initial terrain data through the third convolution branch, identifying the runoff convergence area and the potential erosion area, and marking the coordinates of the water area boundary.

[0054] In mountainous terrain, the direction of water flow is strongly affected by the terrain. By analyzing the elevation, slope direction and other information in the terrain data, the runoff convergence area is identified. In high-altitude mountainous areas, the bottom of the valley is often the runoff convergence area, because water will flow along the slope to the lower terrain, forming streams or small waterlogged areas here. At the same time, slopes with steeper slopes and less vegetation cover are identified as potential erosion areas, because water flows easily carry away soil in these areas during rainfall. Marking the boundary coordinates of water bodies can clarify the precise scope of water bodies such as lakes and streams. For example, the starting coordinates of a stream are (x1, y1, z1) and the end coordinates are (x2, y2, z2).

[0055] Step S125, integrating the geological stability parameters, vegetation coverage density and water boundary coordinates to generate a three-dimensional terrain grid, extracting a continuous slope change curve from the three-dimensional terrain grid as the slope distribution feature, dividing the altitude bands as the altitude interval feature, and marking the spatial distribution of different surface types as the surface coverage feature.

[0056] For example, from the generated three-dimensional terrain grid, the continuous slope change curve can be accurately extracted as the slope distribution feature. In high-altitude mountainous areas, the slope change curve may show a gradual increase from the foot of the mountain to the top of the mountain, and then there will be different degrees of ups and downs near the top of the mountain. Divide the altitude zone as the altitude interval feature. For example, the area below 3,000 meters above sea level can be divided into low-altitude areas, where the climate may be relatively mild, and the vegetation and ecosystem are relatively rich; 3,000-4,000 meters can be divided into medium-altitude areas, where the vegetation begins to become sparse and the climatic conditions are more severe; above 4,000 meters is a high-altitude area, mainly dominated by cold-resistant herbaceous plants and exposed rocks. At the same time, the spatial distribution of different surface types (such as rocks, vegetation, waters, etc.) is marked as surface cover characteristics to clarify the surface composition of each area, which is of great significance for subsequent landscape planning and ecological function design.

[0057] In a possible implementation, step S130 includes: Step S131, analyzing the viewing point setting parameters and visual corridor parameters in the visual presentation requirements, and determining the maximum allowable slope value of the target area and the height difference threshold between adjacent steps.

[0058] For example, suppose a main viewing point is set up at the top of a main peak. Starting from this viewing point, the maximum allowable slope value of the target area is determined based on the tourists' viewing experience and safety factors. For example, in order to ensure the safety of tourists during the viewing process and a comfortable visual experience, the maximum allowable slope value is set to 30 degrees within 50 meters around the viewing platform. At the same time, the height difference threshold between adjacent steps is determined to be 1.5 meters. Such a height difference is convenient for tourists to climb and can form a clear staircase effect visually.

[0059] Step S132, identifying a steep slope area exceeding the maximum allowable slope value in the slope distribution characteristics, and performing contour encryption processing on the steep slope area to generate an encrypted contour data set.

[0060] Step S133, performing stepwise interpolation on the encrypted contour data set based on the height difference threshold, inserting a transition platform between adjacent contour lines, and adjusting the width and inclination angle of the transition platform according to the viewing point setting parameters.

[0061] For example, steep slope areas with an angle of more than 30 degrees can be identified in the slope distribution characteristics. These steep slope areas may be distributed on the sides of the peaks or the steep slopes of the valleys in high-altitude mountainous areas. Contours are encrypted in these steep slope areas. For example, the original contour interval of 10 meters is encrypted to 5 meters, thereby generating an encrypted contour dataset. The encrypted contour dataset is interpolated in a stepped manner based on a height difference threshold of 1.5 meters. A transition platform is inserted between adjacent contour lines, and the width and inclination angle of the transition platform are adjusted according to the viewing point setting parameters. For example, in the area near the viewing platform, the width of the transition platform is set to 2 meters and the inclination angle is 5 degrees, so that tourists can stand steadily on the platform and easily enjoy the surrounding scenery.

[0062] Step S134, performing curvature smoothing processing on the interpolated contour lines to eliminate the sharp turns of the step edges to form a continuously transitioned stepped surface, and edge-melting the stepped surface with the original terrain data to generate stepped terrain features including anti-slip textures and drainage grooves.

[0063] In this embodiment, the curvature smoothing process is performed on the interpolated contour lines in order to eliminate the sharp turns at the edge of the steps. Under the complex terrain of high-altitude mountainous areas, the contour lines after smoothing can be better integrated with the original terrain to form a stepped surface with continuous transition. Finally, the stepped surface is edge-fused with the original terrain data to generate stepped terrain features including anti-skid textures (such as anti-skid lines engraved with a depth of 0.02 meters and a width of 0.03 meters) and drainage grooves (groove depth of 0.1 meters and width of 0.2 meters) on the step surface. The anti-skid texture can increase the friction of tourists when walking and ensure the safety of tourists, while the drainage grooves can discharge rainwater in time to prevent rainwater from accumulating on the steps and avoid dangers such as landslides.

[0064] In a possible implementation, step S130 further includes: Step S135, constructing a vegetation type set including a cold-resistant plant library, a slope-fixing plant library and an ornamental plant library according to the soil and water conservation grade parameters and the biodiversity parameters in the ecological function requirements.

[0065] For example, in high-altitude mountainous areas, due to their special ecological environment, special consideration needs to be given to the impact of vegetation on soil and water conservation and biodiversity. The constructed cold-resistant plant library includes plants such as snow lotus and alpine azalea that are adapted to the high-altitude cold environment. The slope-fixing plant library includes plants such as sea buckthorn that have strong root systems and can fix the soil. The ornamental plant library includes alpine flowers that bloom in different seasons, such as primroses that bloom in spring and nasturtiums that bloom in summer.

[0066] Step S136, marking the shallow soil layer area with soil thickness lower than the set thickness, the easily eroded area with a slope greater than the critical erosion angle, and the sunny slope area in the surface cover feature.

[0067] For example, in the cracks between rocks on the top of the mountain or in some areas with poor soil development, the soil thickness is lower than the set thickness (such as 0.3 meters), and these areas are marked as shallow soil areas. Slopes with a slope greater than the critical erosion angle (such as 45 degrees) are prone to soil erosion during rainfall and are marked as erosion-prone areas. The sunny slope area is the slope facing the direction of sufficient sunlight. These areas have better lighting conditions and relatively higher temperatures.

[0068] Step S137, matching the shallow soil area with plant species with well-developed root systems and resistance to barrenness in the cold-resistant plant library, matching the easily eroded area with plant species with a network root structure in the slope-fixing plant library, and matching the sunny slope area with plant species with staggered flowering periods in the ornamental plant library.

[0069] For shallow soil areas, we match plant species with well-developed root systems and resistance to barrenness in the cold-resistant plant library, such as alpine rhododendron, whose roots can penetrate deep into rock cracks. Not only can it grow in poor soil, but it can also fix the soil. For areas prone to erosion, we match plant species with a network root structure in the slope-fixing plant library, such as sea buckthorn, whose roots are intertwined to form a network, which can effectively prevent the soil from being washed away by slope runoff. For sunny slope areas, we match plant species with staggered flowering periods in the ornamental plant library, such as primrose and nasturtium, so that tourists can enjoy different flower landscapes in different seasons.

[0070] Step S138, generating a vegetation planting coordinate set according to the matching results of each area, and adding plant growth simulation parameters to the vegetation planting coordinate set, wherein the plant growth simulation parameters include shading range prediction value, root expansion speed and seasonal phase change data to form the vegetation distribution characteristics.

[0071] For example, the planting coordinates of a certain alpine azalea are (x3, y3, z3), and plant growth simulation parameters are added to the vegetation planting coordinate set. The predicted value of the shading range of the alpine azalea is a circular area with a radius of 1 meter, which means that during its growth, the surrounding area within 1 meter will be affected by the shading of its branches and leaves. The root system expands at a rate of 0.2 meters per year, and over time, its roots will gradually expand to the surrounding area. The seasonal phase change data are flowering in spring, lush leaves in summer, leaf color change in autumn, and dormancy in winter. These seasonal phase change data can accurately reflect the growth status of plants in different seasons, thus forming a complete vegetation distribution feature. This vegetation distribution feature provides an accurate basis for vegetation layout for garden landscape planning in high-altitude mountainous areas, which can not only meet the needs of ecological functions, but also enhance the ornamental value of the landscape.

[0072] In a possible implementation, step S140 further includes: Step S141 , spatially aligning the surface vertex coordinates of the stepped terrain feature with the planting coordinates of the vegetation distribution feature in a three-dimensional coordinate system, and detecting conflicting points in the coordinate overlapping area.

[0073] For example, the vertex coordinates of a step in the stepped terrain feature are (x1, y1, z1), (x2, y2, z2), etc., while the planting coordinates of a plant in the vegetation distribution feature are (x3, y3, z3). Through precise calculation and positioning, these planting coordinates are matched in three-dimensional space to ensure that the spatial relationship between the two is accurately reflected. In this process, it is necessary to detect conflict points in the coordinate overlapping area.

[0074] Step S142, performing priority determination on the conflict point, when the conflict point is located in the drainage ditch area at the edge of the step, the structural parameters of the stepped terrain feature are preferentially retained and the vertical height of the vegetation planting coordinates is adjusted.

[0075] For the conflict points detected, priority determination is required. For example, when the conflict point is located in the drainage ditch area at the edge of the step, the structural parameters of the stepped terrain features are retained first because the structure of the drainage ditch is crucial to the drainage function of the entire stepped terrain. For example, structural parameters such as the depth, width, and slope of the drainage ditch must remain unchanged. At this time, adjust the vertical height of the vegetation planting coordinates. Assuming that the original planting coordinates of a plant are (x4, y4, z4), and there is a conflict with the drainage ditch area, then according to the structure of the stepped terrain features and the surrounding spatial relationship, its vertical height is adjusted to (x4, y4, z5), where z5 is the new height value that can avoid conflict with the drainage ditch after calculation, while trying to ensure that the plant grows at a suitable altitude to meet its needs for environmental conditions such as light and temperature.

[0076] Step S143, when the conflict point is located in the viewing area of ​​the stepped platform, the seasonal variation data of the vegetation distribution characteristics is preferentially retained and the local curvature of the stepped platform is modified.

[0077] When the conflict point is located in the viewing area of ​​the stepped platform, the visual effect of the viewing area is mainly reflected by the seasonal changes of the vegetation, so the seasonal change data of the vegetation distribution characteristics are retained first. For example, the viewing area is planted with alpine flowers with special seasonal changes. Seasonal changes such as flowering in spring, lush leaves in summer, and color change in autumn are important components of the landscape viewing value. In order to retain this ornamental value, the local curvature of the stepped platform is modified. If the original local curvature radius of the stepped platform is r1, in order to adapt to the growth space requirements of the vegetation and not affect the viewing field, the curvature radius is adjusted to r2. This adjustment is made without destroying the stability of the entire stepped platform and the overall landscape coordination.

[0078] Step S144 , voxel-based fusion is performed on the adjusted stepped terrain features and vegetation distribution features to generate a composite voxel model including terrain material attributes, vegetation growth attributes, and light reflection attributes as the composite terrain vegetation features.

[0079] After completing the processing of the conflict points, the adjusted stepped terrain features and vegetation distribution features are voxel-fused. First, obtain the surface vertex coordinate set in the adjusted stepped terrain features and the updated vegetation planting coordinate set in the vegetation distribution features. Extract information such as elevation gradient data from the surface vertex coordinate set. Assume that the elevation of a slope from the bottom to the top changes from z6 to z7, forming a certain elevation gradient. Extract data such as vertical height offset from the vegetation planting coordinate set. For example, the vertical height offset of a plant relative to the surrounding terrain is Δz. Based on the spatial distribution relationship between these data, the surface vertex coordinate set of the stepped terrain features and the vegetation planting coordinate set are aligned in three-dimensional space grids to generate a spatial alignment data set with a unified coordinate system reference.

[0080] In the spatial alignment dataset, the coordinates of the drainage ditch at the edge of the stepped platform overlap with the coordinates of the extension range of the vegetation root system, and the angle anomaly points between the normal vector direction of the terrain surface and the planting direction of the vegetation in the coordinate overlap area are detected. For example, the normal vector direction of the bottom surface of a drainage ditch is vertically downward, while the growth direction of the vegetation root system is inclined downward at a certain angle, forming an angle anomaly point between the two. According to the position distribution of the angle anomaly point, the displacement compensation is performed on the surface vertex coordinates of the stepped terrain features. For example, the surface vertex coordinates are slightly adjusted in the x, y, and z directions so that the bottom surface normal vector direction of the drainage ditch forms a vertical orthogonal relationship with the growth direction of the vegetation root system, which helps to ensure the normal growth of the vegetation root system and the drainage function of the drainage ditch is not affected.

[0081] Import the displacement-compensated surface vertex coordinates and the vegetation planting coordinate set into the preset voxel partitioning engine. Read the parameters such as the maximum elevation difference and the minimum curvature radius in the displacement-compensated surface vertex coordinates. Assume that the maximum elevation difference is ΔZmax and the minimum curvature radius is rmin. Determine the number of vertical layers of the voxel partitioning engine according to the maximum elevation difference, for example, set the number of vertical layers to n layers, where n is the reasonable number of layers calculated according to ΔZmax. Calculate the initial value of the horizontal voxel edge length as l0 based on the minimum curvature radius, and start the dynamic encryption algorithm in the area where the curvature change rate exceeds the threshold. If the curvature change rate of a certain area is k, when k exceeds the set threshold, the initial value of the voxel edge length is reduced by the inverse ratio of the curvature change rate, such as becoming l1, thereby generating a non-uniform voxel grid with adaptive resolution, in which the vertex coordinates of each voxel unit form a one-to-one mapping relationship with the surface vertex coordinates of the stepped terrain feature.

[0082] Traverse each voxel unit in the non-uniform voxel grid. If the voxel unit contains both the surface vertex coordinates of the stepped terrain feature and the vegetation planting coordinates, extract the rock texture roughness parameter, soil moisture parameter, and vegetation shading range prediction value corresponding to the voxel unit. For example, the microsurface concave-convex height data corresponding to the rock texture roughness parameter in a voxel unit is h1, and the particle size distribution data in the soil diffuse reflection parameter of the adjacent voxel is d1. According to these data, further processing is performed to calculate the terrain material properties, vegetation growth properties, and light reflection properties. By calculating the relationship between the rock texture roughness parameter and the soil moisture content parameter, the light attenuation coefficient in the voxel unit, the osmotic pressure gradient of the vegetation root system, and other parameters are determined, thereby generating a composite voxel model containing terrain material properties (such as rock hardness, soil texture, etc.), vegetation growth properties (such as root expansion speed, canopy coverage, etc.) and light reflection properties (such as light reflectivity, scattering rate, etc.) as a composite terrain vegetation feature. This composite terrain-vegetation feature integrates multiple characteristics of terrain and vegetation, providing basic data for subsequent path connectivity analysis, multi-dimensional rendering and other operations.

[0083] In a possible implementation, step S144 includes: Step S1441, obtaining the surface vertex coordinate set in the adjusted stepped terrain feature and the updated vegetation planting coordinate set in the vegetation distribution feature, extracting the elevation gradient data in the surface vertex coordinate set and the vertical height offset in the vegetation planting coordinate set.

[0084] In this embodiment, in a high-altitude mountainous area, the surface vertex coordinate set of the stepped terrain feature contains the coordinate information of each key point that constitutes the stepped terrain. For example, the vertex coordinates of a certain step may be (x1, y1, z1), and the vertex coordinates of the next adjacent step may be (x2, y2, z2), etc. These coordinates constitute the shape of the entire stepped terrain in three-dimensional space. At the same time, the updated vegetation planting coordinate set in the vegetation distribution feature determines the planting position of various plants in the mountainous area, such as the planting coordinates of a certain alpine azalea are (x3, y3, z3). Then extract the elevation gradient data in the surface vertex coordinate set, such as a slope from the foot of the mountain to the top of the mountain, the elevation gradually rises from 1000 meters to 1500 meters, forming a certain elevation gradient. At the same time, extract the vertical height offset from the vegetation planting coordinate set, for example, a certain plant is 0.5 meters higher than the surrounding average ground height, which is the vertical height offset.

[0085] Step S1442, based on the spatial distribution relationship between the elevation gradient data and the vertical height offset, the surface vertex coordinate set of the stepped terrain feature and the vegetation planting coordinate set are aligned in three-dimensional space grid to generate a spatial alignment data set with a unified coordinate system reference.

[0086] In this process, considering the complex terrain and vegetation distribution in high-altitude mountainous areas, the relative position relationship of each coordinate point in space must be accurately calculated. For example, the grid division method at different altitudes is determined based on the elevation gradient data, and the vegetation planting coordinates are accurately placed in the corresponding terrain grid in combination with the vertical height offset of the vegetation to ensure that the spatial relationship between the two is accurate in three-dimensional space, thereby generating a spatially aligned dataset.

[0087] Step S1443, identifying the overlapping area between the drainage ditch coordinates at the edge of the stepped platform and the coordinates of the vegetation root extension range in the spatial alignment data set, and detecting abnormal points of the angles between the terrain surface normal vector direction and the vegetation planting direction in the coordinate overlapping area.

[0088] In the stepped terrain of high-altitude mountainous areas, drainage ditches are used to drain rainwater and melted snow, and their coordinate ranges are clear. For example, the starting coordinates of a drainage ditch are (x4, y4, z4), and the ending coordinates are (x5, y5, z5). At the same time, the root extension range of vegetation also has a corresponding coordinate range, and the root extension range is different for different plants. When there is a coordinate overlap area between the two, the angle abnormal point between the normal vector direction of the terrain surface and the vegetation planting direction in the coordinate overlap area is detected. For example, the normal vector direction of the bottom surface of a drainage ditch is vertically downward, while the root growth direction of a plant is inclined downward at 30 degrees, which forms an angle abnormal point.

[0089] Step S1444, based on the position distribution of the angle abnormal points, displacement compensation is performed on the surface vertex coordinates of the stepped terrain feature so that the bottom surface normal vector direction of the drainage ditch forms a vertical orthogonal relationship with the growth direction of the vegetation root system.

[0090] Under the terrain conditions of high-altitude mountainous areas, this displacement compensation is to ensure that the function of the drainage ditch is not affected and the vegetation roots can grow normally. For example, in areas where abnormal angles are found, the vertex coordinates of the stepped terrain features are moved by 0.1 meters in the x direction, 0.05 meters in the y direction, and 0.03 meters in the z direction through precise calculation, so that the bottom surface normal vector direction of the drainage ditch forms a vertical orthogonal relationship with the growth direction of the vegetation roots, which not only ensures that the drainage ditch can effectively drain water, but also provides a suitable growth space for the vegetation roots.

[0091] Step S1445, importing the displacement-compensated surface vertex coordinates and the vegetation planting coordinate set into a preset voxel partitioning engine, dynamically adjusting the voxel side length according to the curvature change rate of the stepped terrain feature, and generating a non-uniform voxel grid that matches the terrain undulation.

[0092] Step S1446, traverse each voxel unit in the non-uniform voxel grid, if the voxel unit contains both the surface vertex coordinates of the stepped terrain feature and the vegetation planting coordinates, extract the rock texture roughness parameter, soil moisture parameter and vegetation shading range prediction value corresponding to the voxel unit.

[0093] In high-altitude mountainous areas, the rock texture roughness parameter reflects the roughness of the rock surface. For example, the microsurface concave-convex height data corresponding to the rock texture roughness parameter in a certain voxel unit is 0.02 meters, which means that the rock surface has a certain degree of undulation. The soil moisture parameter is crucial for vegetation growth. Assume that the soil moisture content in a certain voxel unit is 30%. The vegetation shading range prediction value determines the shading effect of vegetation on the surrounding environment. For example, the shading range prediction value of a certain plant is a circular area with a radius of 1 meter.

[0094] Step S1447, matching the reflectivity map of the anti-slip material according to the rock texture roughness parameter, calculating the osmotic pressure gradient of the vegetation root system according to the soil moisture parameter, and determining the light attenuation coefficient within the voxel unit in combination with the shading range prediction value.

[0095] For example, on the stepped terrain of high-altitude mountainous areas, the reflectivity map of the anti-slip material has an important impact on the walking safety of tourists and the visual effect of the landscape. For example, for a rock surface with high roughness, the matching anti-slip material reflectivity map may have a lower reflectivity to reduce the visual interference caused by light reflection. The osmotic pressure gradient of the vegetation root system is calculated based on the soil moisture parameter. Assuming that the soil moisture content is 30%, the osmotic pressure gradient of the vegetation root system is 0.2 through a specific calculation formula. Combined with the predicted value of the shading range, the light attenuation coefficient within the voxel unit is determined. For example, based on the predicted value of the shading range as a circular area with a radius of 1 meter and the surrounding lighting environment, the light attenuation coefficient is calculated to be 0.3.

[0096] Step S1448, performing a material blending operation on each voxel unit, weightedly superimposing the reflectivity map of the anti-slip material and the soil diffuse reflection parameters of adjacent voxels, and generating a continuous transition effect of the terrain material attributes.

[0097] Step S1449, injecting vegetation growth simulation parameters into the non-uniform voxel grid, dynamically adjusting the expansion speed of the vegetation root system according to the osmotic pressure gradient, and correcting the pigment deposition rate of the vegetation leaves based on the light attenuation coefficient.

[0098] Step S14410, extracting the cross-sectional geometric parameters of the drainage ditch in the stepped terrain feature, calculating the surface tension distribution of water flowing through the ditch, and generating water flow reflection spot data matching the voxel unit size.

[0099] Step S14411, superimposing the water flow reflection spot data and the continuous transition effect of the terrain material attribute by optical path, marking the specular reflection highlight area and the diffuse reflection shadow area in the non-uniform voxel grid.

[0100] Under the complex terrain and lighting conditions of high-altitude mountainous areas, light path superposition is a complex process. Through accurate light propagation calculations, the water reflection spot data is combined with the continuous transition effect of the terrain material properties. For example, in a certain voxel unit, the water reflection spot data causes a certain area to have a highlight reflection, while the continuous transition effect of the terrain material determines the surrounding diffuse reflection, thereby accurately marking the specular reflection highlight area and the diffuse reflection shadow area to simulate the real lighting effect.

[0101] Step S14412, verify the physical compatibility of terrain material attributes, vegetation growth attributes and light reflection attributes in each voxel unit, and when it is detected that there is a spatial overlap between the specular reflection highlight area and the predicted value of the vegetation shading range, perform attenuation filtering on the reflection intensity of the highlight area.

[0102] Step S14413, updating the attribute parameters in the non-uniform voxel grid according to the verification result, and outputting a composite voxel model containing multi-attribute fusion data.

[0103] For example, in a possible implementation, step S1445 includes: The maximum elevation difference and the minimum curvature radius in the displacement-compensated surface vertex coordinates are read, and the number of vertical layers of the voxel partitioning engine is determined according to the maximum elevation difference.

[0104] The initial value of the voxel side length in the horizontal direction is calculated based on the minimum curvature radius, and a dynamic encryption algorithm is started in the area where the curvature change rate exceeds a threshold value to reduce the initial value of the voxel side length in proportion to the inverse of the curvature change rate.

[0105] Based on the number of vertical layers and the dynamically adjusted voxel side length, a non-uniform voxel grid with adaptive resolution is generated, wherein the vertex coordinates of each voxel unit form a one-to-one mapping relationship with the surface vertex coordinates of the stepped terrain feature.

[0106] In this embodiment, in a high-altitude mountainous area, the maximum elevation difference may reach 500 meters, which reflects the undulation of the entire terrain; the minimum curvature radius may be 3 meters, which reflects the minimum radius of the terrain curvature. The number of vertical layers of the voxel partitioning engine is determined according to the maximum elevation difference. For example, according to the maximum elevation difference of 500 meters, the number of vertical layers is set to 50 layers, so that each layer corresponds to an elevation range of approximately 10 meters. The initial value of the voxel side length in the horizontal direction is calculated based on the minimum curvature radius. Assuming that the minimum curvature radius is 3 meters, the initial value of the voxel side length in the horizontal direction is calculated to be 1 meter. And the dynamic encryption algorithm is started in the area where the curvature change rate exceeds the threshold (such as 0.5). If the curvature change rate of a certain area is 0.8, the initial value of the voxel side length is reduced by the inverse ratio of the curvature change rate, that is, it becomes 0.625 meters. Based on the number of vertical layers and the dynamically adjusted voxel side length, a non-uniform voxel grid with adaptive resolution is generated, in which the vertex coordinates of each voxel unit form a one-to-one mapping relationship with the surface vertex coordinates of the stepped terrain feature. In this way, in the complex terrain of high-altitude mountainous areas, a suitable voxel grid can be generated according to the actual terrain conditions, which can not only accurately represent the ups and downs of the terrain, but also reasonably distribute the vegetation information.

[0107] For example, in a possible implementation, step S1448 includes: The micro-surface concave-convex height data corresponding to the rock texture roughness parameters in the voxel unit and the particle size distribution data in the soil diffuse reflection parameters of the adjacent voxels are obtained.

[0108] According to the proportional relationship between the micro-surface concave-convex height data and the particle size distribution data, the material mixing weight coefficient is calculated, and based on the weight coefficient, the specular reflection intensity of the rock material and the diffuse reflection intensity of the soil material are linearly interpolated.

[0109] The interpolated reflection intensity data is mapped to all surface vertices of the voxel unit to generate a natural transition material effect from rock to soil.

[0110] In the voxel unit of the high-altitude mountainous area, the micro-surface concave-convex height data (such as 0.02 meters) corresponding to the rock texture roughness parameter in the voxel unit and the particle size distribution data (such as 0.01 meters) in the soil diffuse reflection parameter of the adjacent voxel are obtained. The material mixing weight coefficient is calculated according to the proportional relationship between the micro-surface concave-convex height data and the particle size distribution data. Assuming that the proportional relationship is 2:1, the material mixing weight coefficient is calculated to be 0.67. Based on this weight coefficient, the specular reflection intensity of the rock material and the diffuse reflection intensity of the soil material are linearly interpolated. For example, the specular reflection intensity of the rock material is 0.4, and the diffuse reflection intensity of the soil material is 0.6. After linear interpolation, the new reflection intensity data is obtained. The interpolated reflection intensity data is mapped to all surface vertices of the voxel unit to generate a natural transition material effect from rock to soil, so that in the terrain of the high-altitude mountainous area, the transition between different materials is more natural and in line with the actual visual and physical effects.

[0111] For example, in a possible implementation, step S1449 includes: The voxel units corresponding to the vegetation planting coordinate set are marked in the non-uniform voxel grid, and an initial root depth value and a canopy coverage rate are assigned to each marked voxel unit.

[0112] The daily growth of the root depth value is dynamically calculated according to the soil moisture parameter in the voxel unit, and the seasonal expansion rate of the canopy coverage is calculated based on the light attenuation coefficient.

[0113] When the canopy coverage of the adjacent voxel unit extends to the current voxel unit, the canopy gap closing algorithm is triggered to automatically adjust the leaf density to eliminate the discontinuous shadow boundaries in the lighting simulation.

[0114] In this process, the voxel units corresponding to the vegetation planting coordinate set are first annotated in the non-uniform voxel grid. For example, the voxel units corresponding to a certain alpine rhododendron are accurately annotated. Then, the initial root depth value and canopy coverage are assigned to each annotated voxel unit. Assume that the initial root depth value of a certain plant is 0.1 meters and the canopy coverage is 0.1. The daily growth of the root depth value is dynamically calculated according to the soil moisture parameter in the voxel unit. When the soil moisture content is 30%, the daily growth of the root depth value is calculated to be 0.001 meters through a specific growth model. At the same time, the seasonal expansion rate of the canopy coverage is calculated based on the light attenuation coefficient. For example, in the case of sufficient light in summer and a light attenuation coefficient of 0.3, the seasonal expansion rate of the canopy coverage is calculated to be 0.05 per month. When the canopy coverage of the adjacent voxel unit extends to the current voxel unit, the canopy gap closure algorithm is triggered to automatically adjust the leaf density to eliminate the discontinuous shadow boundaries in the light simulation. In high-altitude mountainous areas, the growth of vegetation is affected by many factors. This simulation can accurately reflect the growth status of vegetation under different environmental conditions.

[0115] For example, in a possible implementation, step S14410 includes: The groove bottom inclination angle and the groove wall roughness level are extracted from the drainage groove cross-sectional geometric parameters, and the gravity acceleration component of the water flow is calculated according to the inclination angle.

[0116] The viscous resistance coefficient of the water boundary layer is determined based on the groove wall roughness level, and the dynamic equilibrium equation of the surface tension is solved in combination with the gravity acceleration component.

[0117] The solution result of the dynamic balance equation is discretized into a water film thickness distribution map matching the voxel unit size, and intensity gradient data of the reflected light spot is generated according to the water film thickness.

[0118] In high-altitude mountainous areas, the cross-sectional geometric parameters of drainage ditches include the bottom inclination angle and the roughness level of the groove wall. For example, the bottom inclination angle of a drainage ditch is 5 degrees, and the roughness level of the groove wall is level 3. The gravity acceleration component of the water flow is calculated according to the inclination angle, and the gravity acceleration component is calculated by the physical formula to be 0.87 m / s². Based on the roughness level of the groove wall, the viscous resistance coefficient of the water flow boundary layer is determined to be 0.05, and the dynamic equilibrium equation of the surface tension is solved in combination with the gravity acceleration component. The solution of the dynamic equilibrium equation is discretized into a water film thickness distribution map that matches the voxel unit size. For example, when the voxel unit size is 0.1 m × 0.1 m × 0.1 m, the water film thickness is 0.01 m at the bottom of the groove and 0.005 m at the edge of the groove. The intensity gradient data of the reflected light spot is generated according to the water film thickness, which can accurately simulate the surface tension of the water flow and the light reflection effect in the drainage ditch in high-altitude mountainous areas.

[0119] For example, in a possible implementation, step S14412 includes: The illumination reflection attribute data of each voxel unit is scanned in the non-uniform voxel grid to detect whether the brightness value of the specular reflection highlight area exceeds the photosynthesis saturation threshold of the vegetation leaves.

[0120] If a high light area exceeding the saturation threshold is detected, the reflection intensity of the area is proportionally reduced according to the leaf density parameter in the predicted value of the vegetation shading range until the photosynthetic active radiation range is reached.

[0121] The reflectivity map in the terrain material properties of the voxel unit is synchronously updated to ensure visual continuity between the highlight area after the reflection intensity is adjusted and the surrounding material.

[0122] For example, the brightness value of the specular reflection highlight area in a certain voxel unit is 1000 lux, while the photosynthesis saturation threshold of the vegetation leaves in this area is 800 lux. If a highlight area exceeding the saturation threshold is detected, the reflection intensity of the area is proportionally reduced until the photosynthetic effective radiation range is reached according to the leaf density parameter in the predicted value of the vegetation shading range. Assuming that the leaf density parameter is 0.5, the reflection intensity is reduced to 0.8 times the original. The reflectivity map in the terrain material properties of the voxel unit is updated synchronously to ensure the visual continuity of the highlight area after the reflection intensity is adjusted with the surrounding materials, so that in the landscape of high-altitude mountainous areas, the lighting effect can meet the needs of vegetation growth and have a good visual effect.

[0123] For example, in a possible implementation, step S14413 includes: Each voxel unit in the non-uniform voxel grid is encoded into a multi-dimensional data block including terrain material attributes, vegetation growth attributes and light reflection attributes.

[0124] An octree index structure is constructed according to the spatial arrangement order of the multi-dimensional data blocks, and voxel cluster identifiers of material transition areas are marked in the octree index structure.

[0125] The octree index structure is divided into LOD levels, and voxel data blocks of corresponding levels are dynamically called according to different observation distances to generate a composite voxel model supporting multi-resolution rendering.

[0126] For example, in a multidimensional data block of a voxel unit, the terrain material attributes include information such as rock hardness and soil texture, the vegetation growth attributes include data such as root depth and canopy coverage, and the light reflection attributes include values ​​such as reflectivity and scattering rate. The octree index structure is constructed according to the spatial arrangement order of the multidimensional data blocks, and the voxel clustering identifiers of the material transition area are marked in the octree index structure. The octree index structure is divided into LOD levels, and the voxel data blocks of the corresponding level are dynamically called according to different observation distances. In the landscape browsing of high-altitude mountainous areas, when the observation distance is far, low-resolution voxel data blocks are called to improve rendering efficiency; when the observation distance is close, high-resolution voxel data blocks are called to display more details, thereby generating a composite voxel model that supports multi-resolution rendering. This composite voxel model can accurately reflect the complex terrain, vegetation growth, and lighting effects of the garden landscape of high-altitude and high-drop mountainous scenic spots.

[0127] In a possible implementation, step S140 further includes: Step S145, obtaining the peak-hour passenger flow density and average stay time in the historical passenger flow data, and calculating the maximum carrying capacity of each area in combination with the viewing platform area in the composite terrain vegetation feature.

[0128] Assume that through the statistical analysis of tourist data over the past years, the peak flow density is 5 people per square meter, and the average stay time is 30 minutes. At the same time, the area of ​​the viewing platform in the composite terrain vegetation feature has been accurately measured to be 100 square meters. Based on these data, the maximum carrying capacity of the viewing platform is calculated as follows: Maximum carrying capacity = viewing platform area × peak flow density × (average stay time ÷ total stay time). Assuming the total stay time is 60 minutes, the maximum carrying capacity of this viewing platform = 100 × 5 × (30 ÷ 60) = 250 people. The maximum carrying capacity of different viewing platforms in the entire high-altitude mountainous area and other tourist gathering areas is calculated in this way. This data will provide an important basis for subsequent route planning and tourist diversion.

[0129] Step S146, marking candidate routes of emergency evacuation passages in the composite terrain vegetation feature, and calculating the cumulative climbing height and slope change rate of each candidate route based on the height difference data of the stepped terrain feature.

[0130] In the complex terrain of high-altitude mountainous areas, candidate routes for emergency evacuation channels need to consider a variety of factors. For example, a relatively gentle route along the bottom of the valley may be marked as a candidate route because the bottom of the valley is low and relatively flat, which is conducive to the rapid evacuation of personnel. At the same time, an open route along the ridge may also be a candidate route. The ridge has a wide field of view, which is convenient for guiding tourists to evacuate and is not easily affected by geological disasters such as landslides. For these candidate routes, their relevant parameters are calculated based on the height difference data of the stepped terrain characteristics. Assuming that a candidate route along the bottom of the valley has a cumulative climbing height of 100 meters from the starting point to the end point, the height difference data shows that the cumulative climbing height is 100 meters. By measuring and calculating the slope of different sections on the route, the slope change rate is 0.1. Another candidate route along the ridge has a cumulative climbing height of 200 meters and a slope change rate of 0.2. The accurate calculation of these data is crucial to evaluating the difficulty and safety of the candidate routes.

[0131] Step S147, setting diversion nodes for candidate routes according to the maximum carrying capacity, adding virtual coordinates of indicator marks at the diversion nodes, and filtering out a set of paths that meet the ergonomic comfort threshold based on the cumulative climbing height and slope change rate.

[0132] Taking the previously calculated maximum carrying capacity of the viewing platform as 250 people as an example, diversion nodes are set on each candidate route leading to the viewing platform. For example, a diversion node is set at a certain distance (such as 50 meters) from the viewing platform, the location of the node is determined according to the terrain and tourist flow of the area, and the virtual coordinates (x6, y6, z6) of the indicator are added to the diversion node. This virtual coordinate will play a role in the subsequent navigation and sign settings. The path set is filtered based on the cumulative climbing height and slope change rate, and the ergonomic comfort threshold is set to a cumulative climbing height of no more than 200 meters and a slope change rate of no more than 0.3. Among the candidate routes calculated previously, the cumulative climbing height of the candidate route at the bottom of the valley is 100 meters and the slope change rate is 0.1, which meets this threshold requirement; the cumulative climbing height of the candidate route on the ridge is 200 meters and the slope change rate is 0.2, which also meets the requirements. These routes that meet the requirements are filtered out to form a path set.

[0133] Step S148, topological optimization is performed on each path in the path set, infinite loop sections are eliminated and the density of interlacing viewing points is increased, and a tour path feature including path width parameters, handrail setting points and rest area coordinates is generated.

[0134] Wherein, step S148 includes: Step S1481, constructing a node connection graph of each path in the path set, and identifying cross nodes in the node connection graph whose degree is greater than a set degree.

[0135] Step S1482, setting a direction guidance mark at the intersection node, and adjusting the connection priority of the intersection node according to the field of view openness parameter of the adjacent path.

[0136] In the path network of high-altitude mountainous areas, the intersections and turning points of each path are regarded as nodes. For example, a path forms nodes at locations such as viewing platforms, rest areas, and forks in the road. These nodes and the connection relationships between the nodes are constructed into a node connection graph. In this node connection graph, the intersection nodes whose degrees are greater than the set degree (such as 3) are identified. For example, at a place where three paths intersect, the degree of this intersection node is 3 because it connects three different paths. Direction guidance signs are set at such intersection nodes. The direction guidance signs can be clear road signs or electronic navigation signs to indicate the direction tourists should go. At the same time, the connection priority of the intersection node is adjusted according to the field of view openness parameters of adjacent paths. In high-altitude mountainous areas, field of view openness is very important for tourists' travel experience and safety. For example, a path leads to an open valley, where you can see distant peaks and streams, and its field of view openness is high; while another path may be blocked by dense vegetation and has a low field of view openness. For paths with high field of view openness, a higher connection priority is given at the intersection node because tourists are more inclined to choose routes with good views.

[0137] Step S1483, the curvature radius of the paths whose connection priority is lower than the set priority is enlarged to ensure a safe distance when two-way traffic flows meet, and anti-slip material marks are added at the places where the curvature radius changes.

[0138] Step S1484, generating a path topology structure according to the optimized node connection graph, dynamically associating the path topology structure with the viewing platform coordinates in the composite terrain vegetation feature, and updating the branch guidance parameters in the tour path feature.

[0139] In detail, if the curvature radius of a path with a low connection priority is small, for example, it is originally 2 meters, it is easy to cause congestion and danger when people flow. The curvature radius is expanded to 3 meters, which can increase the safe distance when two-way people flow converges. At the same time, anti-slip material signs are added where the curvature radius changes, such as the transition area from 2 meters to 3 meters. This is because in high-altitude mountainous areas, the weather is changeable, there may be rain or snow, and anti-slip signs can remind tourists to pay attention to their feet and avoid slipping. A path topology structure is generated based on the optimized node connection graph. This path topology structure clearly represents the connection relationship between each path, the location of the node, and the priority of the path. Then this path topology structure is dynamically associated with the coordinates of the viewing platform in the composite terrain vegetation feature. For example, when tourists approach the viewing platform, the path topology structure can accurately guide tourists to the best viewing position of the viewing platform according to the coordinate information of the viewing platform. At the same time, the branch guidance parameters in the tour path feature are updated. These branch guidance parameters can include the content of the guidance signs on different branch paths, the guidance frequency, etc., so as to better guide tourists in the tour process of high-altitude mountainous areas.

[0140] Based on the above steps, in the garden landscape of high-altitude and high-drop mountain scenic spots, the path connectivity analysis of the composite terrain vegetation characteristics was combined with the preset tourist flow data, and the tour path characteristics were finally determined. These tour path characteristics not only take into account the flow distribution and ergonomic comfort of tourists, but also take into account the complexity of the terrain and the needs of viewing, providing tourists with a safe, comfortable and good viewing experience for the tour of high-altitude mountainous areas.

[0141] In a possible implementation, step S150 may include: Step S151 , fusing the voxel model in the composite terrain vegetation feature with the topological structure in the tour path feature to generate a scene rendering basic data set.

[0142] For example, the voxel model in the composite terrain vegetation feature contains various information about the terrain, such as the elevation, slope, rock distribution, vegetation distribution, etc., which exist in the form of voxels. The topological structure in the tour path feature describes in detail the connection relationship of the path, the node location, the width of the path, the handrail setting point and the coordinates of the rest area. In the data fusion process, each voxel in the voxel model needs to be associated with the corresponding element in the topological structure. For example, the voxels in the voxel model located near the tour path need to be associated with the path elements in the topological structure to clarify their positional relationship around the path. At the same time, for the voxels located near the viewing platform, they need to be accurately associated with the viewing platform elements in the topological structure to determine their orientation and distance relative to the viewing platform. Through such association and fusion, a basic data set for scene rendering is generated. This data set provides a comprehensive data foundation for subsequent rendering, including information on terrain, vegetation, and tour paths.

[0143] Step S152, loading rock texture maps, vegetation dynamic growth animations and water flow particle effects based on the material properties in the scene rendering basic data set, and calculating the shadow projection range according to the illumination angle parameters.

[0144] For example, the material attributes in the scene rendering basic data set include characteristic information of rocks, vegetation, and soil. For rocks, rock texture maps are loaded according to their material attributes. For example, if the rocks in a certain area are granite, then a texture map with granite characteristics is loaded, such as a map with a grainy texture, grayish white color, and irregular texture, so that it presents a real granite appearance in the rendered model. For vegetation, dynamic vegetation growth animation is loaded according to the type and growth stage of vegetation. In high-altitude mountainous areas, different vegetation has different growth cycles and characteristics. For example, alpine rhododendrons will gradually grow buds in spring, flowers bloom in summer, leaves change color in autumn, and enter a dormant period in winter. By loading dynamic vegetation growth animation, the growth changes of vegetation over time can be truly shown in the rendered 3D visualization model. For water flow in mountainous areas, water flow particle effects are loaded according to the relevant data of water flow. In high-altitude mountainous areas, streams in valleys or slope runoff after rain can be simulated by water flow particle effects, making the water flow look more real and natural. At the same time, the shadow projection range is calculated according to the illumination angle parameter. In high-altitude mountainous areas, the position and angle of the sun will change at different times, so the illumination angle parameters will be set according to the local longitude and latitude, season, and time. For example, in the morning, the sun rises from the east and the light is oblique, which will produce a long shadow on the west side of the mountain; while at noon, the sun is directly shining and the shadow range is relatively small. By accurately calculating the illumination angle parameters, the shadow projection range can be realistically presented in the rendered model, enhancing the realism of the model.

[0145] Step S153, detecting the visually blocked area in the three-dimensional visualization model in real time during the rendering process, and adaptively adjusting the viewing angle of the visually blocked area to add an auxiliary observation viewpoint.

[0146] Step S154, extracting biodiversity parameters from the vegetation distribution characteristics and soil and water conservation parameters from the stepped terrain characteristics, generating an ecological benefit evaluation matrix, and associating and annotating the ecological benefit evaluation matrix with the three-dimensional visualization model.

[0147] In this embodiment, the biodiversity parameters include information such as the number of plant species, the distribution ratio of different plants, and the structure of plant communities. For example, statistics show that there are many plants such as alpine rhododendron, fir, and alpine meadow in high-altitude mountainous areas, among which the distribution area of ​​alpine rhododendron accounts for 20% of the total vegetation area, fir accounts for 30%, and alpine meadow accounts for 50%, and the vertical distribution structure of these plant communities is analyzed, such as alpine rhododendron is mainly distributed in the area with an altitude of 3000-3500 meters, fir is distributed in the area with an altitude of 2500-3000 meters, and alpine meadow is distributed in the area with an altitude of 3500-4000 meters. The soil and water conservation parameters in the stepped terrain features include the slope of the terrain, the impact of vegetation coverage on soil erosion, etc. For example, in areas with a gentle slope and high vegetation coverage, the degree of soil erosion is low; while in areas with a steep slope and low vegetation coverage, the degree of soil erosion is high. Generate an ecological benefit evaluation matrix based on the above biodiversity parameters and soil and water conservation parameters. The elements in the ecological benefit evaluation matrix include the contribution of plant species diversity to ecosystem stability, the effect of vegetation coverage on soil and water conservation, etc. The ecological benefit evaluation matrix is ​​associated with the three-dimensional visualization model. In the three-dimensional visualization model, for areas with rich plant species, specific colors or logos can be used to indicate their high contribution to ecosystem stability; for areas with good soil and water conservation effects, corresponding logos can also be used to display them. For example, in areas with high vegetation coverage and low soil erosion, green logos are used to indicate good soil and water conservation effects. At the same time, biodiversity information such as the richness of plant species in the area is displayed in the model, so that tourists can intuitively understand the ecological benefits of high-altitude mountainous areas when browsing the three-dimensional visualization model.

[0148] Wherein, step S153 includes: Step S1531, setting a plurality of preset observation angles in the three-dimensional visualization model, and calculating the visible range under each observation angle.

[0149] Step S1532, comparing the visible area range with the key display area in the landscape adjustment instruction, and identifying a visual blind area that is not covered by the key display area.

[0150] Step S1533, inserting a virtual observation point at the coordinate position corresponding to the visual blind spot, and generating a temporary connection path from the nearest tour path to the virtual observation point based on a path optimization algorithm.

[0151] Step S1534, adjusting the rendering level of the three-dimensional visualization model, adding a transparent terrain profile at the end point of the temporary connection path, and exposing the landscape details in the visual blind spot.

[0152] Due to the undulation of the terrain and the distribution of vegetation, there may be visually blocked areas. First, set multiple preset observation perspectives in the 3D visualization model. These preset observation perspectives are set according to the terrain characteristics of high-altitude mountainous areas and the key landscape areas. For example, setting an observation perspective at the top of the mountain allows you to overlook the entire mountainous area; setting an observation perspective at the bottom of the valley allows you to observe the streams and vegetation in the valley; setting an observation perspective at the viewing platform on the mountainside allows you to enjoy the vegetation layers at different altitudes and the peaks in the distance. For each preset observation perspective, calculate its visible range. The calculation of the visible range needs to take into account factors such as the undulation of the terrain, the height of the vegetation, and the buildings (if any). For example, from the observation perspective of the viewing platform on the mountainside, due to the obstruction of a mountain in front, the visible range may be the area from the valley on the left to the ridge on the right. Compare the visible range with the key display area in the landscape adjustment instruction to identify the visual blind spots that do not cover the key display area. Assume that the key display area in the landscape adjustment instruction is the valley where a rare plant community is located. However, due to the occlusion of the surrounding peaks and vegetation, the visible range from the existing observation perspective cannot completely cover the valley, and the valley becomes a visual blind spot. Insert a virtual observation point at the coordinate position corresponding to the visual blind spot. For example, insert a virtual observation point above the valley where the rare plant community is located. The coordinates of this virtual observation point (x7, y7, z7) are determined according to the location of the valley and the surrounding terrain. Then generate a temporary connection path from the nearest tour path to the virtual observation point based on the path optimization algorithm. The path optimization algorithm considers the slope, distance and existing tour path conditions of the terrain to find the shortest and safest path from the nearest tour path to the virtual observation point. For example, along a relatively gentle path on the hillside, after several turning points, you will finally reach the virtual observation point. Finally, adjust the rendering level of the 3D visualization model, add a transparent terrain profile at the end of the temporary connection path, and expose the landscape details in the visual blind spot. In high-altitude mountainous areas, by adjusting the rendering level, the rare plant communities in the visual blind spot can be displayed more clearly. For example, the terrain profile at the end of the temporary connection path is set to a semi-transparent state, so that tourists can see the distribution and growth status of the rare plant communities in the valley through the terrain profile.

[0153] Through the above steps, the composite terrain and vegetation characteristics are rendered in multiple dimensions according to the characteristics of the tour paths, and a landscape garden plan including a three-dimensional visualization model and ecological assessment parameters is generated. This can comprehensively and accurately display the terrain, vegetation, tour paths, ecological benefits and other information of the garden landscape in high-altitude and high-drop mountainous areas, providing an important reference for the planning, construction of scenic spots and tourists' visits.

[0154] In a possible implementation, step S150 further includes: Step S155: displaying a rotatable thumbnail of the three-dimensional visualization model and a comparative radar chart of the ecological assessment parameters in a user terminal interface.

[0155] In this embodiment, on the user terminal interface, the rotatable thumbnails of the three-dimensional visualization model enable users to view the garden landscape of high-altitude mountainous areas from different angles. For example, the user can rotate the thumbnail to a perspective overlooking the entire mountain area from the top of the mountain by touching the screen or operating the mouse, and view the layout of the stepped terrain, vegetation distribution, and tour paths. At the same time, the comparative radar chart of ecological assessment parameters intuitively presents the situation of multiple ecological assessment indicators. For example, one axis of the radar chart may represent biodiversity, another axis represents soil and water conservation effects, and another axis represents vegetation coverage, etc. By comparing the radar charts, users can clearly see the comparison of ecological indicators of the current landscape gardening scheme with other reference schemes or original states, and quickly understand the performance of the scheme in terms of ecological benefits.

[0156] Step S156, receiving a user's instruction to modify the annotation of the designated viewing area, wherein the instruction to modify the annotation includes a request to replace a vegetation type, a request to adjust a path curvature, or a request to modify the number of terrain steps.

[0157] For example, for a vegetation type replacement request, it may be because the alpine azalea currently planted in a certain viewing area is not good in a certain season, and the user hopes to replace it with a nasturtium with a longer flowering period and brighter colors. For a path curvature adjustment request, it may be because a certain section of the tour path is close to the edge of a cliff and has a large curvature, which poses a safety hazard. The user hopes to adjust the curvature of this section of the path to make it smoother. For a terrain step number modification request, it may be that there are too many existing terrain steps near a certain viewing platform, making it difficult for tourists to climb, and the user hopes to reduce the number of steps.

[0158] Step S157, locating the target data layer in the landscape gardening scheme according to the annotation modification instruction, and calling the incremental update algorithm to replace local parameters of the target data layer.

[0159] For example, if it is a vegetation type replacement request, it is necessary to locate the vegetation distribution feature data layer. In the vegetation distribution feature data layer, find the relevant parameters of the alpine azalea in the specified viewing area, including its planting coordinates, growth simulation parameters, etc. Then call the incremental update algorithm to replace the relevant parameters of the alpine azalea with the parameters of the nasturtium. For example, the planting coordinates of the nasturtium remain unchanged, but its shading range prediction value, root expansion speed, and seasonal phase change data are replaced with the corresponding parameters of the nasturtium. If it is a path curvature adjustment request, locate the tour path feature data layer, find the curvature parameter of the target path segment, and adjust the parameter through the incremental update algorithm to change the curvature of the path. For the request to modify the number of terrain steps, locate the stepped terrain feature data layer, and adjust the parameters related to the number of steps according to the modification instructions, such as the spacing of contour lines, the height difference of the steps, etc.

[0160] Step S158, compare the updated landscape gardening plan with the original plan, generate a modification impact report, the modification impact report includes the ecological parameter change, the construction cost floating value and the visual effect comparison chart, and the modification impact report is superimposed and displayed on the user terminal interface.

[0161] In detail, in the process of difference comparison, for the case of replacement of vegetation types, the change in ecological parameters is calculated. For example, the contribution of nasturtium and alpine azalea to biodiversity may be different, and nasturtium may attract different types of insects, thus affecting biodiversity indicators. At the same time, in terms of construction cost fluctuation, there may be differences in the planting cost and maintenance cost of nasturtium and alpine azalea. For example, the seed cost of nasturtium may be higher than that of alpine azalea, but the maintenance difficulty is lower, so the cost fluctuation is calculated. The visual effect comparison chart intuitively shows the different visual effects of alpine azalea and nasturtium when viewed from the designated viewing area, such as color contrast during flowering, plant height and density contrast, etc. For the request for path curvature adjustment, the change in ecological parameters may be reflected in the impact on the surrounding vegetation. If the path adjustment leads to changes in earthwork, it may affect soil stability and thus affect vegetation growth. The construction cost fluctuation is related to the engineering input required for path adjustment, such as cost changes caused by changes in the workload of excavation or filling. The visual effect comparison chart shows the changes in the convenience and aesthetics of the tour before and after the path adjustment. For requests to modify the number of terrain steps, the change in ecological parameters takes into account the impact on soil and water conservation. A reduction in the number of steps may change the slope runoff, thereby affecting the degree of soil erosion. The construction cost floating value is related to the workload of step construction or removal. The visual effect comparison chart shows the overall landscape effect change after the number of terrain steps changes. These modification impact reports containing the change in ecological parameters, construction cost floating values, and visual effect comparison charts are superimposed and displayed on the user terminal interface, allowing users to fully understand the impact of the modification and make more reasonable decisions.

[0162] Figure 2 A schematic diagram of exemplary hardware and software components of an AI-assisted landscape gardening scheme assisted generation system 100 that can implement the concept of the present application is shown in some embodiments of the present application. For example, the processor 120 can be used in the AI-assisted landscape gardening scheme assisted generation system 100 and used to perform the functions in the present application.

[0163] The AI-assisted landscape gardening scheme auxiliary generation system 100 can be a general server or a special-purpose server, both of which can be used to implement the AI-assisted landscape gardening scheme auxiliary generation method of the present application. Although only one server is shown in the present application, for convenience, the functions described in the present application can be implemented in a distributed manner on multiple similar platforms to balance the processing load.

[0164] For example, the AI-assisted landscape gardening scheme assisted generation system 100 may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and storage media 140 in different forms, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the AI-assisted landscape gardening scheme assisted generation system 100 may also include program instructions stored in ROM, RAM, or other types of non-temporary storage media, or any combination thereof. The method of the present application can be implemented according to these program instructions. The AI-assisted landscape gardening scheme assisted generation system 100 also includes an input / output (I / O) interface 150 between the computer and other input / output devices.

[0165] For ease of explanation, only one processor is described in the AI-assisted landscape gardening scheme auxiliary generation system 100. However, it should be noted that the AI-assisted landscape gardening scheme auxiliary generation system 100 in the present application may also include multiple processors, so the steps performed by one processor described in the present application may also be performed jointly or individually by multiple processors. For example, if the processor of the AI-assisted landscape gardening scheme auxiliary generation system 100 executes step A and step B, it should be understood that step A and step B may also be performed jointly by two different processors or individually in one processor. For example, the first processor executes step A, the second processor executes step B, or the first processor and the second processor execute steps A and B together.

[0166] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer executable instructions are preset. When a processor executes the computer executable instructions, the above-mentioned AI-assisted landscape gardening scheme auxiliary generation method is implemented.

[0167] It should be noted that in order to simplify the description of the present invention and thus help understand one or more embodiments of the invention, in the foregoing description of the embodiments of the present invention, various features are sometimes combined into one embodiment, drawing or description thereof.

Claims

1. A method for generating landscape gardening schemes based on AI, characterized in that: The method comprises: Receiving initial terrain data and landscape adjustment instructions input by a user in an interactive interface, wherein the landscape adjustment instructions include visual presentation requirements and ecological function requirements for the initial terrain data; Performing terrain segmentation processing on the initial terrain data by using a terrain segmentation model to extract slope distribution characteristics, altitude interval characteristics and surface coverage characteristics in the initial terrain data; Dynamically modify the slope distribution characteristics according to the visual presentation requirements to generate stepped terrain characteristics that meet the height difference threshold, and perform vegetation adaptability matching on the surface cover characteristics based on the ecological function requirements to output vegetation distribution characteristics; The stepped terrain features and the vegetation distribution features are spatially superimposed to generate composite terrain vegetation features, and the composite terrain vegetation features are subjected to path connectivity analysis in combination with preset tourist flow data to determine the characteristics of the tour path; The composite terrain vegetation features are rendered in multiple dimensions according to the tour path features, a landscape gardening scheme including a three-dimensional visualization model and ecological assessment parameters is generated, and the landscape gardening scheme is sent to a user terminal for interactive feedback.

2. The AI-assisted landscape gardening scheme generation method according to claim 1 is characterized in that: The performing terrain segmentation processing on the initial terrain data by using a terrain segmentation model to extract slope distribution characteristics, altitude interval characteristics and surface coverage characteristics in the initial terrain data includes: Inputting the initial terrain data into a pre-trained multi-scale terrain segmentation model, wherein the multi-scale terrain segmentation model comprises a first convolution branch for identifying rock exposed areas, a second convolution branch for identifying vegetation covered areas, and a third convolution branch for identifying water distribution areas; Performing edge detection on rock texture in the initial terrain data through the first convolution branch to generate a rock distribution mask, and calculating geological stability parameters of each region according to the rock distribution mask; Performing cluster analysis on the vegetation spectral features in the initial terrain data through the second convolution branch, dividing the tree-dense area, the shrub transition area and the herbaceous plant area, and counting the vegetation coverage density of each area; The third convolution branch is used to simulate the flow direction of the hydrological data in the initial terrain data, identify the runoff convergence area and the potential erosion area, and mark the coordinates of the water area boundary; The geological stability parameters, vegetation coverage density and water boundary coordinates are integrated to generate a three-dimensional terrain grid, a continuous slope change curve is extracted from the three-dimensional terrain grid as the slope distribution feature, altitude zones are divided as the altitude interval feature, and the spatial distribution of different surface types is marked as the surface cover feature.

3. The AI-assisted landscape gardening scheme generation method according to claim 1 is characterized in that: The dynamically modifying the slope distribution characteristics according to the visual presentation requirements to generate stepped terrain characteristics that meet the height difference threshold includes: Analyze the viewing point setting parameters and visual corridor parameters in the visual presentation requirements, and determine the maximum allowable slope value of the target area and the height difference threshold between adjacent steps; Identifying a steep slope area exceeding the maximum allowable slope value in the slope distribution characteristics, and performing contour encryption processing on the steep slope area to generate an encrypted contour data set; Performing stepwise interpolation on the encrypted contour data set based on the height difference threshold, inserting a transition platform between adjacent contour lines, and adjusting the width and inclination angle of the transition platform according to the viewing point setting parameters; The interpolated contour lines are subjected to curvature smoothing to eliminate sharp turns at the step edges, thereby forming a continuously transitioned stepped surface. The stepped surface is then edge-fused with the original terrain data to generate stepped terrain features including anti-slip textures and drainage grooves.

4. The AI-assisted landscape gardening scheme generation method according to claim 1 is characterized in that: The performing vegetation adaptability matching on the surface cover characteristics based on the ecological function requirements and outputting vegetation distribution characteristics includes: According to the soil and water conservation grade parameters and biodiversity parameters in the ecological function requirements, a vegetation type collection including a cold-resistant plant library, a slope-fixing plant library and an ornamental plant library is constructed; In the surface cover features, shallow soil areas with soil thickness lower than a set thickness, erosion-prone areas with slopes greater than a critical erosion angle, and sunny slope areas are marked; The shallow soil layer area is matched with plant species with well-developed root systems and resistance to barrenness in the cold-resistant plant library, the easily eroded area is matched with plant species with a network root structure in the slope-fixing plant library, and the sunny slope area is matched with plant species with staggered flowering periods in the ornamental plant library; A vegetation planting coordinate set is generated according to the matching results of each area, and plant growth simulation parameters are added to the vegetation planting coordinate set. The plant growth simulation parameters include shading range prediction value, root expansion speed and seasonal phase change data to form the vegetation distribution characteristics.

5. The AI-assisted landscape gardening scheme generation method according to claim 1 is characterized in that: The step of spatially superimposing the stepped terrain features and the vegetation distribution features to generate composite terrain vegetation features includes: spatially aligning the surface vertex coordinates of the stepped terrain feature with the planting coordinates of the vegetation distribution feature in a three-dimensional coordinate system, and detecting conflicting points in the coordinate overlapping area; Prioritize the conflict point, and when the conflict point is located in the drainage ditch area at the edge of the step, give priority to retaining the structural parameters of the stepped terrain features and adjust the vertical height of the vegetation planting coordinates; When the conflict point is located in the viewing area of ​​the stepped platform, the seasonal variation data of the vegetation distribution characteristics is preferentially retained and the local curvature of the stepped platform is modified; The adjusted stepped terrain features and vegetation distribution features are voxel-fused to generate a composite voxel model including terrain material attributes, vegetation growth attributes, and light reflection attributes as the composite terrain vegetation features.

6. The AI-assisted landscape gardening scheme generation method according to claim 5 is characterized in that: The step of voxelizing the adjusted stepped terrain features and vegetation distribution features to generate a composite voxel model including terrain material attributes, vegetation growth attributes, and light reflection attributes includes: Acquire the surface vertex coordinate set in the adjusted stepped terrain feature and the updated vegetation planting coordinate set in the vegetation distribution feature, extract the elevation gradient data in the surface vertex coordinate set and the vertical height offset in the vegetation planting coordinate set; Based on the spatial distribution relationship between the elevation gradient data and the vertical height offset, the surface vertex coordinate set of the stepped terrain feature and the vegetation planting coordinate set are aligned in three-dimensional space grids to generate a spatial alignment data set with a unified coordinate system reference; Identify the overlapping area between the coordinates of the drainage ditch at the edge of the step platform and the coordinates of the extension range of the vegetation root system in the spatial alignment data set, and detect the abnormal angle between the normal vector direction of the terrain surface and the vegetation planting direction in the overlapping area of ​​the coordinates; According to the position distribution of the angle abnormal points, displacement compensation is performed on the coordinates of the curved surface vertices of the stepped terrain features, so that the bottom surface normal vector direction of the drainage ditch forms a vertical orthogonal relationship with the growth direction of the vegetation root system; The displacement-compensated surface vertex coordinates and the vegetation planting coordinate set are imported into a preset voxel partitioning engine, and the voxel side length is dynamically adjusted according to the curvature change rate of the stepped terrain feature to generate a non-uniform voxel grid that matches the terrain undulation; Traversing each voxel unit in the non-uniform voxel grid, if the voxel unit contains both the surface vertex coordinates of the stepped terrain feature and the vegetation planting coordinates, extracting the rock texture roughness parameter, soil moisture parameter and vegetation shading range prediction value corresponding to the voxel unit; Matching the reflectivity map of the anti-slip material according to the rock texture roughness parameter, calculating the osmotic pressure gradient of the vegetation root system according to the soil moisture content parameter, and determining the light attenuation coefficient within the voxel unit in combination with the shade range prediction value; Performing a material blending operation on each voxel unit, weightedly superimposing the reflectivity map of the anti-slip material with the soil diffuse reflection parameters of adjacent voxels, and generating a continuous transition effect of the terrain material attributes; Injecting vegetation growth simulation parameters into the non-uniform voxel grid, dynamically adjusting the expansion speed of vegetation roots according to the osmotic pressure gradient, and correcting the pigment deposition rate of vegetation leaves based on the light attenuation coefficient; Extracting cross-sectional geometric parameters of the drainage ditch in the stepped terrain feature, calculating the surface tension distribution of water flowing through the ditch, and generating water flow reflection spot data matching the voxel unit size; Superimposing the water flow reflection spot data and the continuous transition effect of the terrain material attribute on the optical path, marking the specular reflection highlight area and the diffuse reflection shadow area in the non-uniform voxel grid; Verify the physical compatibility of terrain material properties, vegetation growth properties, and light reflection properties within each voxel unit, and when it is detected that there is a spatial overlap between the specular reflection highlight area and the predicted value of the vegetation shading range, perform attenuation filtering on the reflection intensity of the highlight area; The attribute parameters in the non-uniform voxel grid are updated according to the verification result, and a composite voxel model containing multi-attribute fusion data is output.

7. The AI-assisted landscape gardening scheme generation method according to claim 1 is characterized in that: The method of performing path connectivity analysis on the composite terrain vegetation features in combination with preset tourist flow data to determine the characteristics of the tour path includes: Obtain the peak-hour passenger flow density and average length of stay in historical tourist flow data, and calculate the maximum carrying capacity of each area in combination with the viewing platform area in the composite terrain vegetation feature; Marking candidate routes of emergency evacuation passages in the composite terrain vegetation feature, and calculating the cumulative climbing height and slope change rate of each candidate route based on the height difference data of the stepped terrain feature; Setting diversion nodes for candidate routes according to the maximum carrying capacity, adding virtual coordinates of indicator marks at the diversion nodes, and selecting a set of paths that meet the ergonomic comfort threshold based on the cumulative climbing height and slope change rate; Topologically optimize each path in the path set, eliminate infinite loop sections and increase the density of interlaced viewing points, and generate a tour path feature including path width parameters, handrail setting points and rest area coordinates; The performing topology optimization on each path in the path set includes: Constructing a node connection graph of each path in the path set, and identifying cross nodes in the node connection graph whose degree is greater than a set degree; Setting a direction guidance mark at the intersection node, and adjusting the connection priority of the intersection node according to the field of view openness parameter of the adjacent path; For paths with a connection priority lower than the set priority, the curvature radius is enlarged to ensure a safe distance when two-way traffic flows meet, and anti-slip material markings are added where the curvature radius changes; A path topology structure is generated according to the optimized node connection graph, and the path topology structure is dynamically associated with the viewing platform coordinates in the composite terrain vegetation feature, and the branch guidance parameters in the tour path feature are updated.

8. The AI-assisted landscape gardening scheme generation method according to claim 1 is characterized in that: The multi-dimensional rendering of the composite terrain vegetation features according to the tour path features to generate a landscape gardening scheme including a three-dimensional visualization model and ecological assessment parameters includes: Performing data fusion on the voxel model in the composite terrain vegetation feature and the topological structure in the tour path feature to generate a scene rendering basic data set; Load rock texture maps, vegetation dynamic growth animations, and water flow particle effects based on the material properties in the scene rendering basic data set, and calculate the shadow projection range according to the illumination angle parameters; During the rendering process, the visual occlusion area in the three-dimensional visualization model is detected in real time, and the viewing angle of the visual occlusion area is adaptively adjusted to increase auxiliary observation viewpoints; Extracting biodiversity parameters from the vegetation distribution characteristics and soil and water conservation parameters from the stepped terrain characteristics, generating an ecological benefit evaluation matrix, and associating and annotating the ecological benefit evaluation matrix with the three-dimensional visualization model; The step of detecting the visually blocked area in the three-dimensional visualization model in real time during the rendering process and adaptively adjusting the viewing angle of the visually blocked area includes: Setting a plurality of preset observation angles in the three-dimensional visualization model, and calculating the visible range under each observation angle; Comparing the visual range with the key display area in the landscape adjustment instruction, and identifying a visual blind area that does not cover the key display area; Inserting a virtual observation point at the coordinate position corresponding to the visual blind spot, and generating a temporary connection path from the nearest tour path to the virtual observation point based on a path optimization algorithm; The rendering level of the three-dimensional visualization model is adjusted, and a transparent terrain section is added at the end point of the temporary connection path to expose the landscape details in the visual blind spot.

9. The AI-assisted landscape gardening scheme generation method according to claim 1, characterized in that: The sending of the landscape gardening scheme to the user terminal for interactive feedback includes: Displaying a rotatable thumbnail of the three-dimensional visualization model and a comparative radar chart of the ecological assessment parameters in a user terminal interface; Receiving a user's instruction to modify a mark of a designated viewing area, wherein the instruction to modify a mark includes a request to replace a vegetation type, a request to adjust a path curvature, or a request to modify the number of terrain steps; Locating the target data layer in the landscape gardening scheme according to the annotation modification instruction, and calling the incremental update algorithm to replace local parameters of the target data layer; The updated landscape gardening plan is compared with the original plan to generate a modification impact report, which includes the change in ecological parameters, the floating value of construction costs and a visual effect comparison chart, and the modification impact report is superimposed and displayed on the user terminal interface.

10. An AI-assisted landscape gardening scheme generation system, characterized in that: The AI-assisted landscape gardening scheme assisted generation system includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the AI-assisted landscape gardening scheme assisted generation method according to any one of claims 1 to 9 above.

Images