Indoor Green Landscape Space Suitability Determination Method and System

Abstract

This application provides a method and system for determining the spatial adaptation of indoor green landscape units. The method involves acquiring spatial environment images of the target indoor space and images of the green landscape units to be adapted; extracting a spatial enclosure distribution map of the target indoor space from the spatial environment images; determining the visual permeability coefficient corresponding to the green landscape unit to be adapted based on the green landscape unit images; determining the target enclosure range that adapts to the green landscape unit to be adapted by using the visual permeability coefficient and a preset enclosure threshold range; identifying candidate embedding regions; constructing an adaptation index for each candidate embedding region based on the visual focus weight value of each candidate embedding region in the spatial environment image and the visual permeability coefficient; and outputting the candidate embedding region with the highest adaptation index as the optimal embedding point. The technical solution provided by this application can solve the problem of the lack of dynamic adaptability between the visual embedding of green landscape units in indoor spaces and the perceived spatial enclosure.

Description

Technical Field

[0001] This application relates to the field of spatial image analysis technology, and more specifically, to a method and system for determining the suitability of indoor green landscape spaces. Background Technology

[0002] In indoor green landscapes, traditional spatial image analysis relies on manual on-site surveys, which suffers from low efficiency, strong subjectivity, and difficulty in large-scale monitoring. With the rapid development of deep learning technology, image semantic segmentation, target detection, and 3D reconstruction methods based on convolutional neural networks provide efficient and accurate technical means for structural analysis, vegetation identification, green volume estimation, and spatial quality assessment of green spaces. These methods can achieve automated and quantitative evaluation of indoor green landscapes, providing data support for environmental monitoring, user experience optimization, and smart building management.

[0003] In existing spatial image analysis, the process typically begins with image preprocessing to eliminate physical acquisition errors. This is followed by feature extraction, using algorithms such as convolutional neural networks to identify geometric contours, textures, and semantic objects in the image, constructing a mapping relationship between low-level visual features and high-level spatial elements. Based on this, visual reconstruction techniques are used to estimate scene depth, object scale, and relative positions, mapping pixel planes to 3D point clouds or voxel spaces. Finally, by combining semantic segmentation results with spatial topological relationships, quantitative indicators of the space are analyzed, enabling computational evaluation of spatial quality and functional attributes. However, in determining the suitability of indoor green landscape spaces, existing technologies often rely on human experience or are based on spatial dimensions. Static rules for greening arrangements often overlook the interaction between the physical form of the greening units (such as canopy density, branch height, and leaf area distribution) and the spatial enclosure. Specifically, placing highly obstructive greening units in areas with high enclosure exacerbates the feeling of confinement and visual obstruction, disrupting the hierarchical perception of space. Conversely, placing low-obstructive greening units in areas with low enclosure fails to effectively define the desired territorial division. Consequently, the placement of greening elements fails to balance the need for spatial privacy with the optimization of visual permeability. Therefore, addressing the lack of dynamic adaptability between the visual embedding of greening elements in indoor spaces and the perception of spatial enclosure has become a challenge for the industry. Summary of the Invention

[0004] This application provides a method and system for determining the spatial adaptability of indoor green landscape units, which can solve the problem of the lack of dynamic adaptability between the visual embedding of green landscape units in indoor spaces and the feeling of spatial enclosure.

[0005] Firstly, this application provides a method for determining the suitability of indoor green landscape spaces, comprising the following steps: Acquire spatial environment images of the target indoor space and images of the green landscape units to be adapted; Visual enclosure features are extracted from the spatial environment image to obtain a spatial enclosure distribution map that characterizes the enclosure intensity of each area in the target indoor space; The plant canopy structure of the greening unit image is analyzed to obtain the visual permeability coefficient, which characterizes the degree of visual obstruction of the greening landscape unit to be adapted. The visual permeability coefficient is compared with the preset enclosure threshold range to determine the target enclosure range that is compatible with the green landscape unit to be adapted. The continuous areas in the spatial enclosure distribution map that fall into the target enclosure range are marked as candidate embedding areas, and the visual focus weight value of each candidate embedding area in the spatial environment image is determined. Based on the visual permeability coefficient and the visual focus weight value, an adaptation index is constructed for each candidate embedding region. The candidate embedding region with the highest adaptation index is output as the optimal embedding point of the green landscape unit to be adapted.

[0006] In some embodiments, extracting visual enclosure features from the spatial environment image to obtain a spatial enclosure distribution map characterizing the enclosure intensity of each region in the target indoor space specifically includes: Edge detection and plane segmentation are performed on the spatial environment image to extract vertical plane elements that are perpendicular to the horizontal plane in the spatial environment image, and spatial coordinate parameters are assigned to each vertical plane element. The spatial environment image is mapped onto a two-dimensional top-view projection plane based on the spatial coordinate parameters of each vertical plane element, and a sampling grid is established in the two-dimensional top-view projection plane with a preset step size; Calculate the horizontal distance between each sampling point in the sampling grid and each vertical plane element, and determine the enclosure strength value of each sampling point based on the corresponding horizontal distance and the view occlusion angle of each vertical plane element; The sampling grid is filled with the enclosure strength value of each sampling point to generate a spatial enclosure distribution map.

[0007] In some embodiments, performing plant canopy structure analysis on the image of the greening unit to obtain a visual permeability coefficient characterizing the degree of visual obstruction by the greening landscape unit to be adapted specifically includes: The image of the greening unit is segmented into plant subjects to extract the canopy outline region of the greening landscape unit to be adapted. Within the canopy outline region, the leaf and branch pixels and the gap pixels are distinguished, and the proportion of leaf and branch pixels within the canopy outline region is calculated to obtain the canopy leaf and branch coverage rate. Morphological skeleton extraction is performed on the canopy contour region to obtain the distribution of branches and main veins inside the canopy, and the structural void distribution index inside the canopy is determined based on the density of the distribution of branches and main veins. The canopy foliage coverage rate and the structural void distribution index are weighted and fused to generate the visual permeability coefficient of the green landscape unit to be adapted to the degree of visual obstruction.

[0008] In some embodiments, comparing the visual permeability coefficient with a preset enclosure threshold range to determine the target enclosure range that matches the green landscape unit to be adapted specifically includes: Retrieve a pre-stored green landscape adaptation mapping library, which stores the correlation between multiple enclosure degree threshold intervals and corresponding plant canopy structure features; Using the visual permeability coefficient as the retrieval key, query the enclosedness threshold range that matches the visual permeability coefficient in the green landscape adaptation mapping library; The range of enclosedness thresholds obtained from the query is determined as the target enclosedness range.

[0009] In some embodiments, marking continuous regions in the spatial enclosure distribution map that fall within the target enclosure interval as candidate embedding regions, and determining the visual focus weight value of each candidate embedding region in the spatial environment image specifically includes: Perform connected component analysis on the spatial enclosure degree distribution map, extract all continuous regions in the spatial enclosure degree distribution map whose enclosure intensity value falls into the target enclosure degree interval, and mark each extracted continuous region as a candidate embedding region. Visual saliency detection is performed on the spatial environment image to obtain a visual focus distribution map, where the pixel value of each pixel in the visual focus distribution map represents the visual saliency of that pixel. For each candidate embedding region, based on the mapping position of the candidate embedding region in the spatial environment image, the visual saliency feature value corresponding to the candidate embedding region is extracted from the visual focus distribution map, and the visual saliency feature value is used as the visual focus weight value of the corresponding candidate embedding region.

[0010] In some embodiments, outputting the candidate embedding region with the highest adaptation index as the optimal embedding point of the green landscape unit to be adapted specifically includes: Extract the maximum value from the adaptation index of each candidate embedding region, and determine the candidate embedding region corresponding to the maximum value as the target embedding region; Extract the set of pixel coordinates of the target embedding region in the spatial environment image. Based on the spatial mapping relationship between the spatial environment image and the target indoor space, convert the set of pixel coordinates into a set of physical coordinates in the target indoor space. Output the geometric center point of the set of physical coordinates as the optimal embedding point.

[0011] In some embodiments, an image acquisition device with a panoramic lens is used to acquire spatial environment images of the target indoor space.

[0012] Secondly, this application provides an indoor green landscape space adaptation determination system, including: The acquisition module is used to acquire spatial environment images of the target indoor space and images of the greening units to be adapted to the green landscape units. The processing module is used to extract visual enclosure features from the spatial environment image to obtain a spatial enclosure distribution map that characterizes the enclosure intensity of each area in the target indoor space. The processing module is also used to perform plant canopy structure analysis on the greening unit image to obtain the visual permeability coefficient, which characterizes the degree of visual obstruction by the greening landscape unit to be adapted. The processing module is further configured to compare the visual field permeability coefficient with a preset enclosure threshold range, determine the target enclosure range that is compatible with the green landscape unit to be adapted, mark the continuous areas in the spatial enclosure distribution map that fall into the target enclosure range as candidate embedding areas, and determine the visual focus weight value of each candidate embedding area in the spatial environment image. The execution module is used to construct the adaptation index of each candidate embedding region based on the visual field transparency coefficient and the visual focus weight value, and output the candidate embedding region with the highest adaptation index as the optimal embedding point of the green landscape unit to be adapted.

[0013] Thirdly, this application provides a computer device, the computer device including a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the above-described indoor green landscape space adaptation determination method.

[0014] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for determining the adaptation of indoor green landscape space.

[0015] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects: The indoor green landscape space adaptation determination method and system provided in this application first acquires the spatial environment image of the target indoor space and the green unit image of the green landscape unit to be adapted; secondly, visual enclosure features are extracted from the spatial environment image to obtain a spatial enclosure degree distribution map representing the enclosure intensity of each area in the target indoor space; further, plant canopy structure analysis is performed on the green unit image to obtain a visual permeability coefficient representing the degree of visual obstruction by the green landscape unit to be adapted; then, the visual permeability coefficient is compared with a preset enclosure degree threshold interval to determine the target enclosure degree interval that is compatible with the green landscape unit to be adapted, and the continuous areas in the spatial enclosure degree distribution map that fall into the target enclosure degree interval are marked as candidate embedding areas, and the visual focus weight value of each candidate embedding area in the spatial environment image is determined; finally, an adaptation index is constructed for each candidate embedding area based on the visual permeability coefficient and the visual focus weight value, and the candidate embedding area with the highest adaptation index is output as the optimal embedding point of the green landscape unit to be adapted.

[0016] Therefore, this application can solve the problem of the lack of dynamic adaptability between the visual embedding and spatial enclosure perception of green landscape units in indoor spaces. First, by acquiring images of the spatial environment and green units separately, the indoor space structure and the morphology of green landscape units are collected separately at the image level, providing independent raw data sources for subsequent quantitative analysis based on visual features, avoiding mutual interference between spatial structure and plant features during information extraction. Second, visual enclosure features are extracted from the spatial environment images to obtain a spatial enclosure degree distribution map, transforming the enclosure perception intensity, which is difficult to quantify directly in indoor spaces, into a continuously distributed numerical grid map. This provides a calculable and comparable quantitative basis for the degree of enclosure in each area of ​​the space, providing a precise structural dimension reference for the spatial matching of green landscape units. Furthermore, plant canopy structure analysis is performed on the green unit images to obtain a visual permeability coefficient, integrating the density of the plant canopy branches and leaves with the branch and trunk gap structure into a single quantitative index. This transforms the subjective visual perception of plant occlusion into an objectively representable numerical value, providing a basis for the spatial matching of green landscape units and the spatial enclosure of indoor spaces. The adaptation of the environment establishes a quantitative basis for the plant dimension. Then, by comparing the visual permeability coefficient with the preset enclosure threshold range, the target enclosure range is determined, and continuous areas falling into this range are selected as candidate embedding areas. At the same time, the visual focus weight value of each candidate embedding area is determined, realizing the first matching and screening of plant visual characteristics and spatial structure features, and introducing visual attention as an additional evaluation dimension of spatial location. Finally, an adaptation index is constructed based on the visual permeability coefficient and the visual focus weight value, and the candidate embedding point with the highest adaptation index is output. The plant's own permeability characteristics and the visual attention of the candidate position are weighted and integrated to realize a comprehensive quantitative evaluation of spatial enclosure adaptation and visual display effect. Finally, a unique optimal embedding point is output, providing an objective and scientific decision-making basis for the spatial layout of indoor green landscapes, so as to achieve the synergistic optimization of the embedding point of green landscapes in consideration of the needs of spatial privacy and visual permeability. In summary, the technical solution provided by this application can solve the problem of the lack of dynamic adaptation between the visual embedding and spatial enclosure feeling of green landscape units in indoor spaces. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of an application scenario architecture for an indoor green landscape space adaptation determination method according to some embodiments of this application; Figure 2 This is an exemplary flowchart of an indoor green landscape space adaptation determination method according to some embodiments of this application; Figure 3 This is an exemplary flowchart illustrating the determination of a spatial enclosure distribution map according to some embodiments of this application; Figure 4This is a structural schematic diagram of an indoor green landscape space adaptation determination system according to some embodiments of this application; Figure 5 This is a schematic diagram of the structure of a computer device for implementing an indoor green landscape space adaptation determination method according to some embodiments of this application. Detailed Implementation

[0018] To better understand the technical solution of this application, the technical solution of this application will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] refer to Figure 1 This figure is a schematic diagram of an application scenario architecture for the indoor green landscape space adaptation determination method according to some embodiments of this application. The application scenario architecture includes a data acquisition terminal, a communication network, and a server. The data acquisition terminal and the server are directly or indirectly connected through the communication network. The data acquisition terminal acquires spatial environment images of the target indoor space and images of the green landscape units to be adapted, and uploads them to the server. The server extracts visual enclosure features from the spatial environment images to obtain a spatial enclosure degree distribution map characterizing the enclosure intensity of each area in the target indoor space; and performs plant canopy structure analysis on the green unit images to obtain a distribution map characterizing the enclosure intensity of each area in the target indoor space. The method describes the visual permeability coefficient of the green landscape unit to be adapted to determine the degree of visual obstruction; compares the visual permeability coefficient with a preset enclosure threshold range to determine the target enclosure range that is compatible with the green landscape unit to be adapted; marks the continuous areas in the spatial enclosure distribution map that fall into the target enclosure range as candidate embedding areas; and determines the visual focus weight value of each candidate embedding area in the spatial environment image; constructs an adaptation index for each candidate embedding area based on the visual permeability coefficient and the visual focus weight value; and outputs the candidate embedding area with the highest adaptation index as the optimal embedding point of the green landscape unit to be adapted.

[0020] refer to Figure 2 The figure is an exemplary flowchart of an indoor green landscape space adaptation determination method according to some embodiments of this application. The indoor green landscape space adaptation determination method mainly includes the following steps: In step 101, the spatial environment image of the target indoor space and the greening unit image of the greening landscape unit to be adapted are acquired.

[0021] In specific implementation, an image acquisition device with a panoramic lens is used to acquire spatial environment images of the target indoor space. The image acquisition device can be a DSLR camera, a camera with a 3D laser scanner, or a smartphone; there are no specific limitations. Specifically, the panoramic image acquisition device performs a panoramic scan of the target indoor space to acquire spatial environment images that reflect spatial depth, interface enclosure relationships, and visual structure. At the same time, a handheld hyperspectral imaging device is used to acquire images of the greening units to be adapted to the target indoor space. Specifically, the handheld hyperspectral imaging device acquires the appearance and outline information of the plant canopy structure of the greening units to be adapted from different viewing angles from 0° to 90° horizontally, forming greening unit images that include branch and leaf density, pore distribution, and hierarchical structure. The greening units to be adapted refer to greening combinations with independent forms that are to be embedded in the target indoor space. The greening units to be adapted can be single potted plants or combined vertical greening modules or landscape features; further details will not be elaborated here.

[0022] It should be noted that, in this application, spatial environment images refer to image data reflecting the geometric structure and visual characteristics of the target indoor space. The spatial environment images can completely record the layout relationship of the enclosing interfaces such as walls, columns, furniture, and partitions in the space. In this application, greening unit images refer to image data reflecting the three-dimensional morphology of the canopy and the density distribution of branches and leaves of the greening landscape unit to be adapted. Greening unit images can effectively record the porosity, depth, and leaf density of the canopy outline, providing basic data support for subsequent quantification of the degree of occlusion of the unit on the line of sight.

[0023] In step 102, visual enclosure features are extracted from the spatial environment image to obtain a spatial enclosure distribution map that characterizes the enclosure intensity of each area in the target indoor space.

[0024] In some embodiments, reference Figure 3 As shown, this figure is an exemplary flowchart of determining a spatial enclosure distribution map according to some embodiments of this application. In this embodiment, visual enclosure feature extraction is performed on the spatial environment image to obtain a spatial enclosure distribution map characterizing the enclosure intensity of each area in the target indoor space. This can be achieved by the following steps: In step 1021, edge detection and plane segmentation are performed on the spatial environment image to extract vertical plane elements that are perpendicular to the horizontal plane in the spatial environment image, and spatial coordinate parameters are assigned to each vertical plane element. In step 1022, the spatial environment image is mapped to a two-dimensional top-view projection plane according to the spatial coordinate parameters of each vertical plane element, and a sampling grid is established in the two-dimensional top-view projection plane with a preset step size. In step 1023, the horizontal distance between each sampling point in the sampling grid and each vertical plane element is calculated, and the enclosure strength value of each sampling point is determined according to the corresponding horizontal distance and the view occlusion angle of each vertical plane element. In step 1024, the sampling grid is filled with the enclosure strength value of each sampling point to generate a spatial enclosure distribution map.

[0025] In specific implementation, firstly, the Canny edge detection operator in image processing is used to extract edge contours from the acquired spatial environment image. These edge contours are the set of locations in the spatial environment image where pixel grayscale values ​​undergo a step change. Then, combined with a conditional random field-based plane segmentation algorithm within a graph cut-based plane segmentation algorithm, pixel regions with similar texture and color features are clustered into independent planar regions. From the segmentation results, planar regions whose normal vector direction is perpendicular to the gravity direction are selected and marked as vertical plane elements. Vertical plane elements refer to planar regions representing vertical structures in the target indoor space. Each vertical plane element is assigned spatial coordinate parameters, including the coordinates of the vertical plane element in three-dimensional space. The spatial coordinates of two endpoints and the plane orientation angle are used. For example, the spatial coordinate parameters of a vertical plane element can be represented as endpoint A (2.5m, 1.0m), endpoint B (2.5m, 4.0m), and orientation angle 90°, indicating that the element is located at x=2.5m, extends 3.0m along the y-axis, and the plane normal points in the positive x-axis direction. Secondly, based on the spatial coordinate parameters of each vertical plane element, the spatial environment image is mapped to a two-dimensional top-view projection plane through inverse perspective projection transformation. The two-dimensional top-view projection plane is a horizontal projection view with the indoor ground as the reference plane. A sampling grid is established within this two-dimensional top-view projection plane with a preset step size of 0.2m. The sampling grid is a discrete set of sampling points composed of row and column point arrays. For example, in a 5m×5m indoor space... A total of 625 sampling points (25×25) are formed. Then, each sampling point in the sampling grid is traversed, and the horizontal distance between the sampling point and each vertical plane element is calculated. If the horizontal distance between the sampling point and a certain vertical plane element is 0, the horizontal distance is assigned the minimum effective physical distance of 0.01m. The horizontal distance refers to the shortest vertical distance from the sampling point to the line segment of the vertical plane element. At the same time, the view occlusion angle of the vertical plane element relative to the sampling point is calculated. The view occlusion angle is the angle between two rays drawn from the sampling point as the vertex to the two endpoints of the vertical plane element, expressed in radians. Then, the enclosure strength value is calculated based on the horizontal distance and the view occlusion angle. The enclosure strength value is calculated by: dividing each vertical plane element... The view occlusion angle is divided by the square of the horizontal distance between the vertical plane element and the sampling point, and then the calculation results of all vertical plane elements are summed to obtain the enclosure strength value of each sampling point. This calculation method serves as a custom enclosure strength quantification index, and its value is used to compare the degree to which each sampling point is surrounded by the surrounding vertical plane elements. No unified dimension is required. The enclosure strength value is used to characterize the degree of closure of the sampling point's location by the surrounding vertical plane elements. Finally, the enclosure strength values ​​of each sampling point are used to fill the corresponding point matrix in the sampling grid to generate a spatial enclosure degree distribution map. The spatial enclosure degree distribution map is a two-dimensional numerical matrix of the same size as the sampling grid, where the value of each matrix element represents the spatial enclosure strength at that location.

[0026] It should be noted that the spatial enclosure degree distribution map in this application is used to characterize the continuous distribution of enclosure intensity in various areas of the target indoor space. Specifically, the spatial enclosure degree distribution map is a gridded data map of the same size as the two-dimensional top-view projection plane. The spatial adaptation of indoor green landscapes not only depends on the morphological characteristics of the green landscape units themselves, but also requires accurate visual perception of the enclosure degree of different locations in the target indoor space. This is because the visual coordination of the same green landscape unit in open space and enclosed space is significantly different. If only the apparent characteristics of the spatial environment image or simple physical size measurement are used, it is impossible to accurately reflect the real perception of spatial enclosure intensity by the human eye from multiple perspectives. By determining the spatial enclosure degree distribution map, the complex indoor space structure can be transformed into a continuous and quantitative two-dimensional enclosure intensity distribution field, realizing the precise coordination between green landscape units and indoor space structure at the visual perception level.

[0027] In step 103, the plant canopy structure of the greening unit image is analyzed to obtain the visual permeability coefficient, which characterizes the degree of visual obstruction by the greening landscape unit to be adapted.

[0028] In some embodiments, the visual permeability coefficient, which characterizes the degree of visual obstruction by the greening unit image, is obtained by performing plant canopy structure analysis on the greening unit image using the following steps: The image of the greening unit is segmented into plant subjects to extract the canopy outline region of the greening landscape unit to be adapted. Within the canopy outline region, the leaf and branch pixels and the gap pixels are distinguished, and the proportion of leaf and branch pixels within the canopy outline region is calculated to obtain the canopy leaf and branch coverage rate. Morphological skeleton extraction is performed on the canopy contour region to obtain the distribution of branches and main veins inside the canopy, and the structural void distribution index inside the canopy is determined based on the density of the distribution of branches and main veins. The canopy foliage coverage rate and the structural void distribution index are weighted and fused to generate the visual permeability coefficient of the green landscape unit to be adapted to the degree of visual obstruction.

[0029] In specific implementation, firstly, the acquired greening unit images are segmented into plant subjects using a semantic segmentation network based on the U-Net architecture. The U-Net network consists of an encoder, a decoder, and skip connections. The encoder contains four downsampling modules, each consisting of two 3×3 convolutional layers, a batch normalization layer, a ReLU activation layer, and a 2×2 max pooling layer stacked sequentially, used to extract high-level semantic features of the image step by step. The decoder contains four upsampling modules, each consisting of a 2×2 deconvolutional layer and the feature map of the corresponding skip connection of the encoder layer concatenated sequentially, and then refined by two 3×3 convolutional layers to gradually restore the spatial resolution of the image. The network input is an RGB three-channel matrix of a greening unit image of size 512×512 pixels, and the output is a 512×512 pixel binary segmentation mask. The set of pixels corresponding to the value 1 in the segmentation mask constitutes the canopy contour region of the greening landscape unit to be adapted. The canopy contour region refers to the closed region formed by the continuous set of pixels occupied by the plant canopy in the image. Subsequently, a method combining superpixel segmentation and random forest classifier was used to distinguish between leaf pixels and gap pixels within the canopy contour region. First, a simple linear iterative clustering algorithm was used to divide the canopy contour region into uniformly sized superpixel blocks, each containing approximately 50 to 100 pixels. Then, the mean and standard deviation in the RGB color space, the histogram statistics of the H component in the HSV color space, and the texture feature values ​​extracted using the local binary mode operator were extracted for each superpixel block, totaling 36-dimensional feature vectors. These feature vectors were input into a trained random forest classifier, which contains 100 decision trees. Each tree splits nodes based on a randomly selected subset of features. The classifier outputs the category label for each superpixel block as either a leaf block or a gap block. All pixels within a superpixel block are labeled as leaf pixels. The percentage of leaf pixels in the canopy contour region is calculated to represent the canopy leaf coverage rate, which is used to characterize the density of the canopy structure.Then, the Zhang-Sun parallel thinning algorithm is used to extract the morphological skeleton from the binary mask of the canopy contour region. This algorithm iteratively traverses each foreground pixel in the binary mask, determining whether the pixel is a deletable boundary point based on the pixel distribution pattern in its 3×3 neighborhood. After multiple iterations, it stops when no pixels can be deleted. Finally, the remaining foreground pixels form a single-pixel-width connected network, which is the distribution of branches and veins inside the canopy. The total number of skeleton pixels in the branch and vein distribution is N. A branch point detection algorithm is used to traverse the skeleton pixels, counting the number of connections of skeleton pixels in their 8-neighborhood, and connecting the connections... Pixels with a number greater than 2 are marked as branch points, and the number of branch points is denoted as M. The skeleton pixel density is determined by the ratio of N to the area of ​​the canopy contour region, and the branch point density is determined by the ratio of M to the area of ​​the canopy contour region. The skeleton pixel density and the branch point density are then weighted and summed according to a preset weighting coefficient to obtain the structural void distribution index. Specifically, the skeleton pixel density is first reverse normalized, that is, normalized to the 0-1 interval and then 1 is taken and subtracted from the value. Then, the reverse normalized value of the skeleton pixel density and the branch point density are weighted and summed according to a preset weighting coefficient to obtain the structural void distribution index. The weighting coefficients are set to 0.6 and 0.4 respectively, without limitation. The structural void distribution index is used to characterize the complexity and density of the void space defined by the branch network inside the canopy. Finally, the canopy foliage coverage and the structural void distribution index are weighted and fused according to the preset weighting coefficients. When the canopy foliage coverage is taken as a positive value, the higher the value, the greater the contribution to the coefficient and the stronger the shading effect. The structural void distribution index is taken as a reverse normalized value, that is, it is first normalized to the 0-1 interval and then 1 is subtracted from the value. The higher the value, the smaller the reverse normalized value, the smaller the contribution to the coefficient, and the stronger the penetration. The weighting coefficients were obtained by performing a multiple linear regression analysis on green landscape samples with known permeability levels. The regression model used the inversely normalized values ​​of canopy foliage coverage and structural void distribution index as independent variables, and manually labeled visual permeability levels as dependent variables. The regression coefficients of the two independent variables were fitted and used as weighted fusion coefficients. The fused values ​​were then mapped to the 0-1 interval using the Sigmoid function and output as the visual permeability coefficients, ensuring that the fusion results perfectly matched the coefficient definitions: lower values ​​indicated stronger visual penetration and weaker occlusion, while higher values ​​indicated weaker visual penetration and stronger occlusion.

[0030] It should be noted that the visual permeability coefficient in this application is used to characterize the degree of visual occlusion of the background scenery by the green landscape unit to be adapted. The lower the value, the stronger the visual penetration and the weaker the occlusion; the higher the value, the weaker the visual penetration and the stronger the occlusion. Since the spatial adaptation of indoor green landscapes cannot rely solely on the apparent parameters such as the type and size of the green landscape unit, the core issue is to quantify the degree of visual impact of the green unit on the sense of spatial permeability. This is because the differences in the canopy structure and branch density of different plants will significantly change the observer's ability to perceive the space behind through the green unit. This visual occlusion characteristic directly determines the appropriate level of enclosure in the indoor space where the green unit is suitable to be embedded. By determining the visual permeability coefficient, the density of the plant canopy and the complexity of its internal void structure can be effectively reflected, providing a unified comparison benchmark for establishing a precise match between the green landscape unit and the target indoor space with a specific level of enclosure.

[0031] In step 104, the visual permeability coefficient is compared with a preset enclosure threshold range to determine the target enclosure range that is compatible with the green landscape unit to be adapted. The continuous areas in the spatial enclosure distribution map that fall into the target enclosure range are marked as candidate embedding areas, and the visual focus weight value of each candidate embedding area in the spatial environment image is determined.

[0032] In some embodiments, comparing the visual permeability coefficient with a preset enclosure threshold range to determine the target enclosure range that matches the green landscape unit to be adapted is achieved through the following steps: Retrieve a pre-stored green landscape adaptation mapping library, which stores the correlation between multiple enclosure degree threshold intervals and corresponding plant canopy structure features; Using the visual permeability coefficient as the retrieval key, query the enclosedness threshold range that matches the visual permeability coefficient in the green landscape adaptation mapping library; The range of enclosedness thresholds obtained from the query is determined as the target enclosedness range.

[0033] In practice, firstly, a pre-built green landscape adaptation mapping library is retrieved from local storage or a cloud database. This library stores the associations between multiple enclosure threshold intervals and corresponding plant canopy structure features. The library is organized in the form of a relational data table, where each row corresponds to a plant type or a typical canopy structure category, containing two core fields: one is the plant canopy structure feature field, which describes the quantitative parameter combinations of this type of plant in terms of canopy foliage density, branch density, and openness; the other is the enclosure threshold interval field. This mapping library is used to store the enclosure threshold ranges of indoor spaces suitable for this type of plant. The construction of this library is based on the collection and calibration of a large number of experimental samples. Specifically, it involves selecting green landscape samples covering different canopy morphologies, calculating the visual permeability coefficient of each sample using the aforementioned steps, and having multiple indoor environmental design experts conduct adaptation evaluations in simulated spatial environments. The appropriate enclosure threshold ranges for each sample are then marked. The visual permeability coefficient and the corresponding enclosure threshold range are linked as key-value pairs and stored in the mapping library. The enclosure threshold ranges are stored in the form of numerical ranges, such as a lower limit and a lower limit. The upper limits represent the minimum and maximum enclosure strength that the green landscape unit can adapt to, respectively. Then, using the visual permeability coefficient calculated in the aforementioned steps for the current green landscape unit to be adapted as the search key, an interval matching query is performed in the green landscape adaptation mapping library. The specific implementation of the interval matching query is as follows: traverse the enclosure threshold interval field of all records in the green landscape adaptation mapping library, and determine whether the current visual permeability coefficient falls within the interval range. Since the visual permeability coefficient is a continuous value, while the enclosure threshold interval corresponding to each record in the green landscape adaptation mapping library may be... Since there are overlaps or gaps, a nearest neighbor matching strategy is adopted. That is, the difference between the visual permeability coefficient and the median of each enclosure threshold interval is calculated, and the interval with the smallest distance is selected as the matching result. If multiple intervals contain the same visual permeability coefficient, the interval with the smaller interval width is selected as the matching enclosure threshold interval. The enclosure threshold interval refers to the reasonable fluctuation range allowed to characterize the degree to which a local area in an indoor space is visually surrounded by surrounding vertical plane elements. Finally, the enclosure threshold interval obtained from the matching query is determined as the target enclosure interval of the green landscape unit to be adapted.

[0034] It should be noted that the target enclosure degree interval in this application refers to the enclosure degree threshold interval that is compatible with the green landscape unit to be adapted. Since the visual coordination and spatial integration of the same plant in spaces with different enclosure degrees are significantly different, it is necessary to establish a quantitative mapping mechanism from the plant's own attributes to spatial adaptation conditions. This transforms a single visual permeability coefficient into a spatial enclosure intensity requirement range with upper and lower boundaries. The determination of the target enclosure degree interval can convert the originally isolated plant permeability quantification result into a matching condition that can be used for spatial screening, so that subsequent steps can perform precise interval screening operations on the spatial enclosure degree distribution map based on this interval.

[0035] In some embodiments, continuous regions falling within the target enclosure interval in the spatial enclosure distribution map are marked as candidate embedding regions, and the visual focus weight value of each candidate embedding region in the spatial environment image is determined by the following steps: Perform connected component analysis on the spatial enclosure degree distribution map, extract all continuous regions in the spatial enclosure degree distribution map whose enclosure intensity value falls into the target enclosure degree interval, and mark each extracted continuous region as a candidate embedding region. Visual saliency detection is performed on the spatial environment image to obtain a visual focus distribution map, where the pixel value of each pixel in the visual focus distribution map represents the visual saliency of that pixel. For each candidate embedding region, based on the mapping position of the candidate embedding region in the spatial environment image, the visual saliency feature value corresponding to the candidate embedding region is extracted from the visual focus distribution map, and the visual saliency feature value is used as the visual focus weight value of the corresponding candidate embedding region.

[0036] In practice, the spatial enclosure distribution map is first converted into a binary mask. The binary mask is a matrix of the same size as the spatial enclosure distribution map. Grid cells in the spatial enclosure distribution map whose enclosure strength values ​​fall within the target enclosure interval are assigned a value of 1, while other grid cells are assigned a value of 0. This results in the binary mask. A connected component labeling algorithm based on 8 neighborhoods is then used to analyze the connected components of this binary mask. This algorithm iterates through all grid cells in the binary mask that are assigned a value of 1, grouping adjacent grid cells into the same connected region and assigning a unique region number to each connected region. After extracting all connected regions, each connected region is projected onto a two-dimensional top-view plane. Each set of continuous grid cells occupied by a greening landscape unit is marked as a candidate embedding region. A candidate embedding region refers to a continuous spatial range whose spatial enclosure intensity meets the adaptation requirements of the greening landscape unit. Then, the original acquired spatial environment image is subjected to visual saliency detection using either the graph theory-based visual saliency detection algorithm GBVS or the frequency domain-based spectral residual algorithm SR. Taking GBVS as an example, the GBVS algorithm first extracts three low-level features of the spatial environment image: color, brightness, and orientation, and constructs corresponding feature maps. For each feature map, a Markov chain model is used to calculate the differences and scarcity of pixel distributions. After iterative convergence through a random walk mechanism, the result is obtained. The saliency maps for each feature dimension are generated, and then the saliency maps are weighted and fused to generate the final visual focus distribution map. The visual focus distribution map is a grayscale image of the same size as the spatial environment image, where the grayscale value of each pixel is between 0 and 255. The higher the grayscale value, the higher the visual saliency of the pixel, that is, the easier it is to attract the observer's attention. The visual focus distribution map refers to the image that characterizes the visual saliency of pixels in the spatial environment image. The higher the grayscale value, the greater the difference between the location and the surrounding area in terms of color, brightness, texture, or orientation features. Finally, for each candidate embedding region, the saliency distribution map is generated based on the two-dimensional top-view projection plane of the candidate embedding region. The grid positions in the image are used to calculate the set of pixel coordinates corresponding to the candidate embedding region in the original spatial environment image through spatial mapping. The gray values ​​of all pixels in the set of pixel coordinates are extracted from the visual focus distribution map. The arithmetic mean of these gray values ​​is calculated and the calculated mean is used as the visual saliency feature value of the candidate embedding region. The visual saliency feature value is also used as the visual focus weight value of the candidate embedding region. In addition, the threshold of the visual focus weight value is set to 1 / 5 of the gray mean of the visual focus distribution map. Initial candidate embedding regions with visual focus weight values ​​lower than the threshold are eliminated, and the remaining regions are determined as the final candidate embedding regions.

[0037] It should be noted that the visual focus weight value in this application refers to the degree of visual attention that can be obtained when placing green landscape units in the candidate embedding area. The higher the weight value, the easier it is for the location to become a visual focus, thus giving it a higher priority in the subsequent comprehensive adaptation evaluation. The visual focus weight can quantify the difference in attention of candidate embedding areas at the visual perception level, providing an objective basis for the visual dimension weight for the subsequent comprehensive adaptation evaluation, so that the optimal embedding point output in the end meets the adaptation requirements of spatial enclosure strength and can obtain the best display effect at the visual level.

[0038] In step 105, an adaptation index for each candidate embedding region is constructed based on the visual field transparency coefficient and the visual focus weight value. The candidate embedding region with the highest adaptation index is output as the optimal embedding point of the green landscape unit to be adapted.

[0039] In some embodiments, the adaptation index of each candidate embedding region is constructed based on the visual field transparency coefficient and the visual focus weight value using the following steps: Using the visual field transparency coefficient as the first weighting factor and the visual focus weight value of each candidate embedding region as the second weighting factor, the first weighting factor and the second weighting factor are multiplied to obtain the initial adaptation index of each candidate embedding region. The initial adaptation index of each candidate embedding region is normalized, and the normalized initial adaptation index is used as the adaptation index of the corresponding candidate embedding region.

[0040] In specific implementation, firstly, the visual permeability coefficient of the current green landscape unit to be adapted, calculated in the aforementioned steps, and the visual focus weight values ​​determined for each candidate embedding region in the aforementioned steps are obtained. The visual permeability coefficient is a dimensionless value between 0 and 1, used to characterize the degree of visual obstruction by the green landscape unit. The visual focus weight value is the gray-scale mean extracted from the visual focus distribution map, used to quantify the visual significance of each candidate embedding region. The visual permeability coefficient is used as the first weighting factor, and the visual focus weight values ​​of each candidate embedding region are used as the second weighting factor. For each candidate embedding region, a product operation is performed. Combining the occlusion characteristics of the green landscape unit with the visual attention of its spatial location, green units with high visual transparency coefficients are preferentially matched with areas of high visual attention, while green units with low visual transparency coefficients can be matched with areas of relatively moderate visual attention. This achieves a precise match between occlusion characteristics and visual display requirements. Before the product operation, the visual focus weight values ​​of all candidate embedding regions within the same indoor space are globally normalized and mapped to the 0-1 range to eliminate the differences in visual focus weights caused by differences in image acquisition parameters and spatial layout between different indoor spaces. To address the issue of differing baseline values, the normalized visual focus weight value is multiplied by the visual field transparency coefficient to obtain the initial fit index for each candidate embedding region. This initial fit index reflects the degree of matching between the visual characteristics of the green landscape unit itself and the visual attention level of that region. The multiplication operation ensures that when the visual field transparency coefficient is high (i.e., the vegetation is denser), only regions with sufficiently high visual focus weight values ​​can obtain a high initial fit index. Conversely, when the vegetation is more transparent, regions with relatively low visual focus weight values ​​can also obtain an acceptable initial fit index, thus reflecting different transparency characteristics. The study identifies the differentiated needs of plants for spatial visual attention. Then, a minimum-maximum normalization process is performed on the initial adaptation indices of each candidate embedding region. This process involves iterating through all the initial adaptation indices of all candidate embedding regions, finding the minimum and maximum values, subtracting the minimum value from each initial adaptation index, and dividing by the range (the difference between the maximum and minimum values). After normalization, all adaptation indices are mapped to a value range of 0 to 1, eliminating the influence of differences in the absolute magnitude of visual focus weights in different batches of calculations. The normalized values ​​are then used as the adaptation indices for the corresponding candidate embedding regions.

[0041] It should be noted that the adaptation index in this application refers to the degree of adaptation of each candidate embedding area in terms of spatial enclosure and visual display effect. The closer the value is to 1, the more suitable the area is as the embedding location of the green landscape unit, providing a unified quantitative ranking basis for the subsequent output of the optimal embedding point.

[0042] In some embodiments, the process of outputting the candidate embedding region with the highest adaptation index as the optimal embedding point of the green landscape unit to be adapted is achieved through the following steps: Extract the maximum value from the adaptation index of each candidate embedding region, and determine the candidate embedding region corresponding to the maximum value as the target embedding region; Extract the set of pixel coordinates of the target embedding region in the spatial environment image. Based on the spatial mapping relationship between the spatial environment image and the target indoor space, convert the set of pixel coordinates into a set of physical coordinates in the target indoor space. Output the geometric center point of the set of physical coordinates as the optimal embedding point.

[0043] In specific implementation, firstly, the adaptation indices corresponding to each candidate embedding region calculated in the previous steps are constructed into a one-dimensional array. Each element in the array corresponds one-to-one with the label of the candidate embedding region. The maximum value is extracted from the array using a traversal comparison method. After traversal, the maximum value and the label of the candidate embedding region to which it belongs are obtained. The candidate embedding region corresponding to this label is determined as the target embedding region. The target embedding region refers to the continuous spatial range with the highest comprehensive adaptation degree among all candidate embedding regions. Then, the set of grid cell coordinates of the target embedding region in the two-dimensional top-view projection plane is obtained. Since the correspondence between the candidate embedding region and the set of grid cell coordinates has been established in the previous connected component analysis step, the coordinates of all grid cells contained in the target embedding region can be obtained by directly retrieving this correspondence. Through the pre-calibrated camera parameters and perspective projection transformation matrix, each coordinate point in the set of grid cell coordinates is transformed from the two-dimensional top-view projection plane coordinate system to the pixel coordinate system of the spatial environment image. The camera calibration adopts the Zhang Zhengyou calibration method. By taking multiple images of the checkerboard calibration board at different angles, the intrinsic parameter matrix and distortion coefficient of the camera, as well as the homography between the spatial environment image and the two-dimensional top-view projection plane, are obtained. The homography matrix is ​​a 3×3 transformation matrix used to describe the perspective mapping relationship between two planes. Multiplying the coordinates of each grid cell by the homography matrix yields its corresponding pixel coordinates in the spatial environment image. The set of all pixel coordinates is the set of pixel coordinates of the target embedding region in the spatial environment image. Then, based on the spatial mapping relationship between the spatial environment image and the target indoor space, the set of pixel coordinates is converted into the set of physical coordinates in the target indoor space. The spatial mapping relationship is achieved through the transformation relationship between the world coordinate system and the image coordinate system established during the aforementioned camera calibration process. That is, the pixel coordinates are back-projected to three-dimensional spatial points in the world coordinate system using the camera's intrinsic parameter matrix, distortion coefficients, and extrinsic parameter matrix. Since the target embedding region is located on the indoor ground or horizontal plane, the horizontal coordinates of the three-dimensional spatial points can be extracted to form the set of physical coordinates. Each coordinate in the set of physical coordinates is represented by the actual physical units of the indoor space. Then, the geometric center point is calculated for all coordinate points in the set of physical coordinates. The geometric center point is calculated by calculating the arithmetic mean of all x-coordinates and the arithmetic mean of all y-coordinates. The coordinate point formed by these two averages is taken as the optimal embedding point.

[0044] It should be noted that the optimal embedding point in this application is the specific coordinate of the location where the green landscape unit to be adapted is finally recommended to be placed within the target embedding area. This point is output as physical coordinates to the display device or construction guidance document to guide the actual placement of the green landscape unit.

[0045] In another aspect, in some embodiments, this application provides an indoor green landscape space adaptation determination system, with reference to... Figure 4 The figure is a schematic diagram of the structure of an indoor green landscape space adaptation determination system according to some embodiments of this application. The indoor green landscape space adaptation determination system includes: an acquisition module 201, a processing module 202, and an execution module 203, which are described below: The acquisition module 201 in this application is mainly used to acquire spatial environment images of the target indoor space and greening unit images of the greening landscape unit to be adapted. Processing module 202, in this application, is mainly used to extract visual enclosure features from the spatial environment image to obtain a spatial enclosure degree distribution map that characterizes the enclosure intensity of each area in the target indoor space. The processing module 202 is also used to perform plant canopy structure analysis on the greening unit image to obtain the visual permeability coefficient, which characterizes the degree of visual obstruction by the greening landscape unit to be adapted. In addition, the processing module 202 is also used to compare the visual field transparency coefficient with the preset enclosure threshold range, determine the target enclosure range that is compatible with the green landscape unit to be adapted, mark the continuous area in the spatial enclosure distribution map that falls into the target enclosure range as candidate embedding area, and determine the visual focus weight value of each candidate embedding area in the spatial environment image. The execution module 203 in this application is mainly used to construct the adaptation index of each candidate embedding region based on the visual field transparency coefficient and the visual focus weight value, and output the candidate embedding region with the highest adaptation index as the optimal embedding point of the green landscape unit to be adapted.

[0046] In addition, this application also provides a computer device, the computer device including a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the above-described indoor green landscape space adaptation determination method.

[0047] In some embodiments, reference Figure 5 The figure is a schematic diagram of the structure of a computer device implementing an indoor green landscape space adaptation determination method according to some embodiments of this application. The indoor green landscape space adaptation determination method in the above embodiments can be... Figure 5 The computer device shown is used to implement this, and the computer device includes at least one processor 301, a communication bus 302, a memory 303, and at least one communication interface 304.

[0048] The processor 301 can be a general-purpose central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more devices used to control the execution of the indoor green landscape space adaptation determination method in this application.

[0049] The communication bus 302 can be used to transmit information between the aforementioned components.

[0050] The memory 303 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CDROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disks or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 303 may exist independently and be connected to the processor 301 via the communication bus 302. The memory 303 may also be integrated with the processor 301.

[0051] The memory 303 stores program code for executing the solution of this application, and its execution is controlled by the processor 301. The processor 301 executes the program code stored in the memory 303. The program code may include one or more software modules. In the above embodiments, the determination of the indoor green landscape space adaptation method can be implemented by the processor 301 and one or more software modules in the program code in the memory 303.

[0052] Communication interface 304 uses any transceiver-like device for communicating with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.

[0053] In a specific implementation, as one example, a computer device may include multiple processors, each of which may be a single-core (single CPU) processor or a multi-core (multi CPU) processor. Here, a processor may refer to one or more devices, circuits, and / or processing cores used to process data (e.g., computer program instructions).

[0054] The aforementioned computer device can be a general-purpose computer device or a special-purpose computer device. In specific implementations, the computer device can be a desktop computer, a portable computer, a network server, a handheld digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. This application does not limit the type of computer device.

[0055] In addition, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for determining the adaptation of indoor green landscape space.

[0056] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0057] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for determining the suitability of indoor green landscape spaces, characterized in that, Includes the following steps: Acquire spatial environment images of the target indoor space and images of the green landscape units to be adapted; Visual enclosure features are extracted from the spatial environment image to obtain a spatial enclosure distribution map that characterizes the enclosure intensity of each area in the target indoor space; The plant canopy structure of the greening unit image is analyzed to obtain the visual permeability coefficient, which characterizes the degree of visual obstruction of the greening landscape unit to be adapted. The visual permeability coefficient is compared with the preset enclosure threshold range to determine the target enclosure range that is compatible with the green landscape unit to be adapted. The continuous areas in the spatial enclosure distribution map that fall into the target enclosure range are marked as candidate embedding areas, and the visual focus weight value of each candidate embedding area in the spatial environment image is determined. Based on the visual permeability coefficient and the visual focus weight value, an adaptation index is constructed for each candidate embedding region. The candidate embedding region with the highest adaptation index is output as the optimal embedding point of the green landscape unit to be adapted.

2. The method as described in claim 1, characterized in that, Extracting visual enclosure features from the spatial environment image to obtain a spatial enclosure distribution map representing the enclosure intensity of each area in the target indoor space specifically includes: Edge detection and plane segmentation are performed on the spatial environment image to extract vertical plane elements that are perpendicular to the horizontal plane in the spatial environment image, and spatial coordinate parameters are assigned to each vertical plane element. The spatial environment image is mapped onto a two-dimensional top-view projection plane based on the spatial coordinate parameters of each vertical plane element, and a sampling grid is established in the two-dimensional top-view projection plane with a preset step size; Calculate the horizontal distance between each sampling point in the sampling grid and each vertical plane element, and determine the enclosure strength value of each sampling point based on the corresponding horizontal distance and the view occlusion angle of each vertical plane element; The sampling grid is filled with the enclosure strength value of each sampling point to generate a spatial enclosure distribution map.

3. The method as described in claim 1, characterized in that, The visual permeability coefficient, which characterizes the degree of visual obstruction by the greening unit image, is obtained by performing plant canopy structure analysis on the greening unit image. Specifically, this includes: The image of the greening unit is segmented into plant subjects to extract the canopy outline region of the greening landscape unit to be adapted. Within the canopy outline region, the leaf and branch pixels and the gap pixels are distinguished, and the proportion of leaf and branch pixels within the canopy outline region is calculated to obtain the canopy leaf and branch coverage rate. Morphological skeleton extraction is performed on the canopy contour region to obtain the distribution of branches and main veins inside the canopy, and the structural void distribution index inside the canopy is determined based on the density of the distribution of branches and main veins. The canopy foliage coverage rate and the structural void distribution index are weighted and fused to generate the visual permeability coefficient of the green landscape unit to be adapted to the degree of visual obstruction.

4. The method as described in claim 1, characterized in that, The visual transparency coefficient is compared with a preset enclosure threshold range to determine the target enclosure range that is compatible with the green landscape unit to be adapted. Specifically, this includes: Retrieve a pre-stored green landscape adaptation mapping library, which stores the correlation between multiple enclosure degree threshold intervals and corresponding plant canopy structure features; Using the visual permeability coefficient as the retrieval key, query the enclosedness threshold range that matches the visual permeability coefficient in the green landscape adaptation mapping library; The range of enclosedness thresholds obtained from the query is determined as the target enclosedness range.

5. The method as described in claim 1, characterized in that, The continuous regions falling within the target enclosure range in the spatial enclosure distribution map are marked as candidate embedding regions, and the visual focus weight value of each candidate embedding region in the spatial environment image is determined, specifically including: Perform connected component analysis on the spatial enclosure degree distribution map, extract all continuous regions in the spatial enclosure degree distribution map whose enclosure intensity value falls into the target enclosure degree interval, and mark each extracted continuous region as a candidate embedding region. Visual saliency detection is performed on the spatial environment image to obtain a visual focus distribution map, where the pixel value of each pixel in the visual focus distribution map represents the visual saliency of that pixel. For each candidate embedding region, based on the mapping position of the candidate embedding region in the spatial environment image, the visual saliency feature value corresponding to the candidate embedding region is extracted from the visual focus distribution map, and the visual saliency feature value is used as the visual focus weight value of the corresponding candidate embedding region.

6. The method as described in claim 1, characterized in that, Specifically, the candidate embedding region with the highest adaptation index is output as the optimal embedding point of the green landscape unit to be adapted. Extract the maximum value from the adaptation index of each candidate embedding region, and determine the candidate embedding region corresponding to the maximum value as the target embedding region; Extract the set of pixel coordinates of the target embedding region in the spatial environment image. Based on the spatial mapping relationship between the spatial environment image and the target indoor space, convert the set of pixel coordinates into a set of physical coordinates in the target indoor space. Output the geometric center point of the set of physical coordinates as the optimal embedding point.

7. The method as described in claim 1, characterized in that, Image acquisition equipment with panoramic lenses is used to acquire spatial environment images of the target indoor space.

8. An indoor green landscape space adaptation determination system, characterized in that, The system includes: The acquisition module is used to acquire spatial environment images of the target indoor space and images of the greening units to be adapted to the green landscape units. The processing module is used to extract visual enclosure features from the spatial environment image to obtain a spatial enclosure distribution map that characterizes the enclosure intensity of each area in the target indoor space. The processing module is also used to perform plant canopy structure analysis on the greening unit image to obtain the visual permeability coefficient, which characterizes the degree of visual obstruction by the greening landscape unit to be adapted. The processing module is further configured to compare the visual field permeability coefficient with a preset enclosure threshold range, determine the target enclosure range that is compatible with the green landscape unit to be adapted, mark the continuous areas in the spatial enclosure distribution map that fall into the target enclosure range as candidate embedding areas, and determine the visual focus weight value of each candidate embedding area in the spatial environment image. The execution module is used to construct the adaptation index of each candidate embedding region based on the visual field transparency coefficient and the visual focus weight value, and output the candidate embedding region with the highest adaptation index as the optimal embedding point of the green landscape unit to be adapted.

9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the indoor green landscape space adaptation determination method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the indoor green landscape space adaptation determination method as described in any one of claims 1 to 7.

Images