Architectural design effect picture automatic generation method, system, medium and program product

Abstract

A kind of architectural design rendering automatic generation method, system, medium and program product, in the method, the three-dimensional geometric data of building group is acquired, and the spatial visual importance network of building group is established according to three-dimensional geometric data;Building group is divided into several visual correlation sub-regions based on spatial visual importance network, and the spatial entropy value of each visual correlation sub-region is calculated, and spatial entropy value is used to represent the spatial distribution complexity of building unit in visual correlation sub-region;According to spatial entropy value and the three-dimensional geometric data of building unit in visual correlation sub-region, the multi-level detail model corresponding to each visual correlation sub-region is generated;Based on multi-level detail model and connection weight, rendering pipeline is constructed;The output result of rendering pipeline is carried out visual consistency synthesis, and the architectural design rendering of building group is generated.The real-time and practicality of the architectural design rendering generation are improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of digital cultural product production, and particularly relates to a building design effect drawing automatic generation method, system, medium and program product. BACKGROUND

[0002] The building design effect drawing is an important link in the building design process, and plays an important role in showing the building design concept, expressing the design intention and effectively communicating with the customer. However, the traditional building effect drawing production mainly relies on the steps of manual modeling, adjusting material and light, setting rendering parameters and the like of the designer, and this process not only consumes time and effort, but also has high requirements for the professional skills of the designer, and meanwhile, problems such as non-uniform style and uneven quality of the effect drawing are prone to occur.

[0003] With the development of artificial intelligence technology, the building effect drawing generation technology based on deep learning has gradually emerged. This technology builds a generative adversarial network model, trains the model by using a large number of building effect drawing samples, so that the model can learn the style features and composition rules of the building effect drawing, thereby realizing the automatic generation of the building effect drawing. This method improves the efficiency of effect drawing generation to a certain extent, and reduces the dependence on the professional skills of the designer.

[0004] However, the existing building effect drawing generation technology based on deep learning has the problem of excessive consumption of computing resources when processing large-scale building groups. This is because the building group usually contains a large amount of geometric details and material information, and a large amount of feature data needs to be processed simultaneously in the generation process, which results in the need for a large amount of computing resources in the model training and inference stage, thereby reducing the real-time performance and practicality of the effect drawing generation. SUMMARY

[0005] The application provides a building design effect drawing automatic generation method, system, medium and program product, which is used for improving the real-time performance and practicality of building design effect drawing generation.

[0006] In a first aspect, the application provides a building design effect drawing automatic generation method, which obtains three-dimensional geometric data of a building group, and establishes a spatial visual importance network of the building group according to the three-dimensional geometric data, wherein the nodes of the spatial visual importance network represent building units, and the connection weights between the nodes are determined according to the visual occlusion relationship between the building units.

[0007] The building group is divided into a plurality of visual correlation sub-regions based on the spatial visual importance network, and the spatial entropy value of each visual correlation sub-region is calculated, wherein the spatial entropy value is used to represent the spatial distribution complexity of the building units in the visual correlation sub-region.

[0008] generate a multi-level detail model corresponding to each visual correlation sub-region according to the spatial entropy value and three-dimensional geometric data of the building units in the visual correlation sub-region;

[0009] construct a rendering pipeline based on the multi-level detail model and the connection weight;

[0010] perform visual consistency synthesis on the output result of the rendering pipeline to generate an architectural design rendering of the building group.

[0011] By adopting the technical solutions, the three-dimensional geometric data of the building group is obtained to establish a spatial visual importance network, which can accurately reflect the visual occlusion relationship between the building units, and then the building group is divided into visual correlation sub-regions and the spatial entropy value is calculated based on the network, which can quantitatively represent the spatial distribution characteristics of the building units in each sub-region. The multi-level detail model generated according to the spatial entropy value and the three-dimensional geometric data can reduce the model complexity while maintaining the visual effect, and the rendering pipeline constructed in combination with the connection weight can realize efficient rendering of the building group. Finally, the rendering result is subjected to visual consistency synthesis, which can reduce the visual discontinuity generated in the rendering process, improve the realism and coherence of the overall rendering, and improve the real-time performance and practicality of the architectural design rendering generation.

[0012] In combination with some embodiments of the first aspect, in some embodiments, generating a multi-level detail model corresponding to each visual correlation sub-region according to the spatial entropy value and three-dimensional geometric data of the building units in the visual correlation sub-region specifically includes:

[0013] constructing a geometric feature matrix for the building units in the visual correlation sub-region, each row of the geometric feature matrix representing a geometric feature vector of a building unit;

[0014] determining a dimension threshold for feature decomposition based on the spatial entropy value, and performing singular value decomposition on the geometric feature matrix to obtain a principal feature vector set;

[0015] sorting the principal feature vector set according to the singular value size, and performing feature reconstruction according to the dimension threshold to obtain the multi-level detail model.

[0016] By adopting the technical solutions, the geometric feature matrix is constructed for the building units in the visual correlation sub-region, and the geometric features of the building units are represented in the form of vectors, so that the feature extraction is more systematic and standardized. The dimension threshold for feature decomposition is determined based on the spatial entropy value, which can adaptively adjust the feature dimension according to the complexity of the spatial distribution. The principal feature vector set is obtained by performing singular value decomposition on the geometric feature matrix, and the feature reconstruction is performed after sorting according to the singular value size, which can reduce the data dimension while retaining the key geometric features, and the multi-level detail model generated can maintain the main geometric features of the building units while reducing the data amount, thereby improving the calculation efficiency of the subsequent rendering process.

[0017] In some embodiments in combination with the first aspect, in some embodiments, the rendering pipeline is constructed based on the multi-level detail model and the connection weights, specifically comprising:

[0018] a rendering task priority queue is constructed, and the priority of a task in the rendering task priority queue is determined according to the connection weights;

[0019] a detail level model is determined for each rendering task in the rendering task priority queue according to the distance from the preset viewpoint to the corresponding visually associated sub-region;

[0020] a rendering computation graph is established, and a node of the rendering computation graph represents a rendering task, and an edge between the nodes represents a task dependency relationship;

[0021] a multi-level cache structure is constructed according to the topological structure of the rendering computation graph;

[0022] the rendering task priority queue, the multi-level cache structure and the rendering computation graph are integrated into a rendering pipeline.

[0023] By adopting the above technical solution, the rendering task priority queue is constructed, and the priority of a task is determined according to the connection weights, so that the rendering order can be reasonably arranged, and the building units with greater visual impact are processed in priority. The detail level model is determined according to the distance from the preset viewpoint to the visually associated sub-region, so that high-detail models are used for buildings close to the viewpoint and low-detail models are used for buildings far from the viewpoint, which ensures the visual effect and improves the rendering efficiency. The rendering computation graph is established to represent the task dependency relationship and the multi-level cache structure is constructed, which can optimize the execution order of the rendering tasks and the data access efficiency. Efficient adaptive rendering of building groups is realized, which reduces the consumption of computing resources while ensuring the rendering quality.

[0024] In some embodiments in combination with the first aspect, before the three-dimensional geometric data of the building group is acquired, the method further comprises:

[0025] a hierarchical data structure of the building group is constructed, and the hierarchical data structure comprises the spatial topological relationship, the functional attribute and the material parameter of the building units;

[0026] a geometric constraint check is performed based on the hierarchical data structure, and a constraint violation metric index is generated, which is used to represent the degree of spatial conflict between the building units;

[0027] the building units with a constraint violation metric index less than a preset first threshold value are adjusted in position to a constraint violation metric index not less than the preset first threshold value, and three-dimensional geometric data satisfying the spatial constraints is obtained.

[0028] By adopting the technical solution, the hierarchical data structure containing the spatial topological relationship, the functional attribute and the material parameter is constructed, so that the spatial organization characteristics of the building group can be comprehensively described. The geometric constraint checking is performed based on the data structure, and the constraint violation metric index is generated, so that the spatial conflict degree between the building units can be quantitatively evaluated. The position of the building unit with the constraint violation metric index less than the preset first threshold is adjusted until the constraint violation metric index is not less than the preset first threshold, so that the unreasonable spatial relationship between the building units can be eliminated, and the three-dimensional geometric data obtained satisfies the spatial constraint requirement.

[0029] In combination with some embodiments of the first aspect, in some embodiments, the geometric constraint checking is performed based on the hierarchical data structure, and the constraint violation metric index is generated, specifically including:

[0030] The space index is constructed, and the bounding box information of the building unit is stored in the corresponding space division node;

[0031] The occlusion relationship between the building units is detected by using the space index, and an occlusion detection result is obtained;

[0032] The constraint graph is constructed based on the occlusion detection result, and the nodes of the constraint graph represent the building units, and the edges represent the spatial constraint relationship;

[0033] The connectivity analysis is performed on the constraint graph, and the constraint violation metric index is calculated.

[0034] By adopting the technical solution, the bounding box information of the building unit is stored in the space index, so that the building unit in the space can be quickly located and accessed, and the calculation complexity in the occlusion relationship detection is reduced. The occlusion detection by using the space index can accurately identify the spatial position relationship and the occlusion degree between the building units, and the redundant calculation caused by the one-by-one comparison of all the building units is reduced. The constraint graph constructed based on the occlusion detection result converts the spatial constraint relationship between the building units into a graph structure representation, which is convenient for quantitative analysis. The constraint violation metric index is calculated by performing the connectivity analysis on the constraint graph, so that the spatial conflict situation between the building units in the building group can be comprehensively evaluated, the efficiency and accuracy of the geometric constraint checking are improved, and the spatial layout of the building group is more reasonable.

[0035] In combination with some embodiments of the first aspect, in some embodiments, after the building design effect drawing of the building group is generated, the method further includes:

[0036] A plurality of sets of expert scoring samples are collected, and a scoring standard set is established, and the scoring standard set includes spatial layout rationality, visual coherence and material authenticity;

[0037] The building design effect drawing is evaluated in multiple dimensions based on the scoring standard set, and a quality score vector is obtained;

[0038] The rendering pipeline is adjusted according to the quality score vector until each dimension component of the quality score vector is greater than a preset second threshold.

[0039] By adopting the technical solution, the expert score sample is introduced to establish the score standard set, and the evaluation indexes of multiple dimensions such as spatial layout rationality, visual coherence and material authenticity are quantified, so that the quality evaluation of the effect picture has an objective basis. The quality score vector obtained through multi-dimensional quality evaluation can accurately reflect the performance level of the effect picture in different aspects. According to the quality score vector, the rendering pipeline is adjusted in a targeted manner, and through iterative optimization, each dimension component reaches the preset threshold requirement, which can continuously improve the overall quality of the effect picture, makes the generation process of the effect picture have clear quality control standards, ensures the professional level of the final effect picture, reduces the workload of repeated manual modification, and improves the automation degree and quality controllability of the effect picture generation.

[0040] In combination with some embodiments of the first aspect, in some embodiments, the building design effect picture is subjected to multi-dimensional quality evaluation based on the score standard set to obtain a quality score vector, specifically including:

[0041] Extracting a local image block of the building design effect picture and calculating a feature statistic of each local image block;

[0042] Performing cluster analysis on the local image block features to obtain a feature distribution map;

[0043] Calculating a spatial layout score, a visual coherence score and a material authenticity score according to the feature distribution map;

[0044] After normalization processing of the spatial layout score, the visual coherence score and the material authenticity score, a quality score vector is formed.

[0045] By adopting the technical solution, the local image block is extracted and the feature statistic is calculated, which can capture the detailed features and local visual effects in the effect picture. The feature distribution map obtained by performing cluster analysis on the local image block features reflects the spatial distribution law and organization mode of the visual elements in the effect picture. The spatial layout score, the visual coherence score and the material authenticity score are calculated based on the feature distribution map, realizing multi-angle quantitative evaluation of the quality of the effect picture. The quality score vector is formed by normalizing each score, so that the evaluation indexes of different dimensions have comparability, which can systematically measure the quality of the effect picture from the local to the whole, improving the explainability and reliability of the evaluation result. Through feature statistics and cluster analysis, quality problems in the effect picture can be accurately identified, providing specific improvement direction for optimization of the rendering pipeline.

[0046] In a second aspect, the embodiments of the present application provide an architectural design rendering automatic generation system, comprising: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is configured to store computer program codes, the computer program codes comprising computer instructions, and the one or more processors are configured to invoke the computer instructions to cause the system to perform the method described in the first aspect and any possible implementation manner of the first aspect.

[0047] In a third aspect, the embodiments of the present application provide a computer readable storage medium, comprising instructions, when the instructions are run on a system, the system performs the method described in the first aspect and any possible implementation manner of the first aspect.

[0048] In a fourth aspect, the embodiments of the present application provide a computer program product, characterized in that, when the computer program product is run on a system, the system performs the method described in any possible implementation manner of the first aspect.

[0049] The one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0050] 1. The present application provides an architectural design rendering automatic generation method, three-dimensional geometric data of a building group is obtained to establish a space visual importance network, which can accurately reflect the visual occlusion relationship between building units, and then the building group is divided into visual correlation sub-regions based on the network and the space entropy value is calculated, which can quantitatively represent the spatial distribution characteristics of building units in each sub-region. The multi-level detail model generated based on the space entropy value and the three-dimensional geometric data can reduce the model complexity on the premise of maintaining the visual effect, and the rendering pipeline constructed in combination with the connection weight can realize efficient rendering of the building group. Finally, the visual consistency of the rendering result is synthesized, which can reduce the visual discontinuity generated in the rendering process, improve the realism and coherence of the overall rendering, and improve the real-time performance and practicality of the architectural design rendering generation.

[0051] 2. The present application provides an architectural design rendering automatic generation method, a hierarchical data structure containing space topological relationship, functional attribute and material parameter is constructed, which can comprehensively describe the spatial organization characteristics of the building group. The geometric constraint checking is performed based on the data structure, and the constraint violation metric index is generated, which can quantitatively evaluate the spatial conflict degree between building units. The position of the building unit with the constraint violation metric index less than a preset first threshold is adjusted until the constraint violation metric index is not less than the preset first threshold, which can eliminate the unreasonable spatial relationship between building units, and the obtained three-dimensional geometric data meets the spatial constraint requirement.

[0052] 3, The application provides a kind of architectural design rendering automatic generation method, introduce expert score sample to establish score standard set, the evaluation index of multiple dimensions such as space layout rationality, visual coherence and material authenticity is quantified, so that the quality evaluation of rendering has objective basis.The quality score vector obtained by multidimensional quality evaluation can accurately reflect the performance level of rendering in different aspects.According to quality score vector, the rendering pipeline is adjusted specifically, and by iteration optimization, each dimension component reaches the preset threshold requirement, which can continuously improve the overall quality of rendering, make the generation process of rendering have clear quality control standard, ensure the professional level of final rendering, reduce the workload of artificial repeated modification, improve the degree of automation and quality controllability of rendering generation. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 It is a flowchart of one kind of architectural design rendering automatic generation method in the embodiment of the application.

[0054] Figure 2 It is a flowchart of one kind of quality evaluation and optimization method based on expert score in the embodiment of the application.

[0055] Figure 3 It is an entity device structure schematic diagram of one kind of architectural design rendering automatic generation system provided by the embodiment of the application. DETAILED DESCRIPTION

[0056] The terms used in the following embodiments of the application are only for the purpose of describing specific embodiments, and are not intended to be limiting to the application. As used in the specification and the appended claims of the application, the singular forms "a," "an," and "the" are intended to include both singular and plural forms, unless the context clearly indicates otherwise. It will be further understood that the terms "and / or" used in the application means any or all possible combinations of one or more of the listed items.

[0057] Hereinafter, the terms "first", "second" are only for the purpose of description, and cannot be understood as implying or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features, and in the description of the embodiments of the application, unless otherwise specified, the meaning of "multiple" is two or more.

[0058] In the following, an embodiment is used and combined with Figure 1 A kind of architectural design rendering automatic generation method in the embodiment of the application is described:

[0059] Please refer to Figure 1Fig. 1 is a flowchart of a method for automatically generating an architectural design effect drawing according to an embodiment of the present application.

[0060] S101, constructing a hierarchical data structure of the building group;

[0061] The system constructs a hierarchical data structure of the building group, which includes the spatial topological relationship, functional attribute, and material parameter of the building unit.

[0062] The system constructs a hierarchical data structure of the building group, which includes the spatial topological relationship, functional attribute, and material parameter of the building unit. The hierarchical data structure is a way of organizing data, which can clearly express the spatial position relationship between each building unit in the building group, the use function of the building unit, and the building material information. By constructing the hierarchical data structure, the necessary data basis is provided for subsequent steps such as geometric constraint checking and three-dimensional geometric data generation.

[0063] The system can use a tree structure to represent the hierarchical data structure of the building group. The root node of the tree structure represents the entire building group, the intermediate nodes represent different areas or building types, and the leaf nodes represent specific building units. The edges between nodes represent the spatial topological relationship, such as adjacency, containment, etc. Each node can also carry the functional attribute (such as residential, commercial, office, etc.) and material parameter (such as masonry, glass, metal, etc.) of the building unit. The system can automatically extract the spatial topological relationship, functional attribute, and material parameter of the building unit by analyzing architectural design drawings, BIM models, and other data sources, and construct the hierarchical data structure.

[0064] S102, performing geometric constraint checking based on the hierarchical data structure, and generating a constraint violation metric index;

[0065] The system performs geometric constraint checking based on the hierarchical data structure, and generates a constraint violation metric index, which is used to represent the degree of spatial conflict between building units. Specifically: constructing a spatial index, storing the bounding box information of the building unit in the corresponding spatial division node;

[0066] Using the spatial index to detect the occlusion relationship between building units, and obtaining an occlusion detection result;

[0067] Constructing a constraint graph based on the occlusion detection result, the nodes of the constraint graph representing the building units, and the edges representing the spatial constraint relationship;

[0068] Performing connectivity analysis on the constraint graph, and calculating the constraint violation metric index.

[0069] Based on the hierarchical data structure constructed in the previous step, the system performs geometric constraint checks and generates constraint violation metrics that characterize the degree of spatial conflict between building units. Geometric constraints refer to the spatial positional relationships that building units should satisfy, such as certain building units not obstructing each other, and certain building units needing to maintain a certain distance from each other. Through geometric constraint checks, spatial conflict problems existing in the building group can be discovered, providing a basis for subsequent adjustments to the positions of building units.

[0070] The system first constructs a spatial index, storing the bounding box (the smallest rectangle that can completely contain a building unit) information of each building unit in the corresponding spatial partitioning node. Common spatial indexes include octrees, kd-trees, and R-trees. Then, the system uses the spatial index to quickly detect whether there are occlusion relationships between building units, obtaining the occlusion detection results. Next, the system constructs a constraint graph based on the occlusion detection results. The constraint graph is a graph model where nodes represent building units and edges represent spatial constraints between building units (such as occlusion, being too close together, etc.). Finally, the system performs connectivity analysis on the constraint graph, counting the number of connected subgraphs. A higher number of connected subgraphs indicates a more severe spatial conflict, and a constraint violation metric is calculated accordingly.

[0071] When performing geometric constraint checks, if there are many types of constraints to be checked, the constraint graph may become too dense, leading to high computational complexity for connectivity analysis. To address this issue, the system can adopt an incremental constraint checking strategy. First, it checks more basic constraints (such as occlusion) to generate an initial constraint graph. Then, it gradually adds other types of constraints, updates the constraint graph, and performs incremental connectivity analysis simultaneously. Through incremental constraint checking, the system can reduce the computational complexity of the algorithm while meeting the requirements of geometric constraint checking.

[0072] S103. Adjust the position of building units whose constraint violation metric index is less than the preset first threshold until the constraint violation metric index is not less than the preset first threshold, and obtain three-dimensional geometric data that meets the spatial constraints.

[0073] The system adjusts the positions of building units whose constraint violation metric is less than a preset first threshold until the constraint violation metric is not less than the preset first threshold, thus obtaining three-dimensional geometric data that satisfies the spatial constraints. Following this, a spatial visual importance network for the building group is established based on the three-dimensional geometric data. Nodes in the spatial visual importance network represent building units, and the connection weights between nodes are determined based on the visual occlusion relationships between building units.

[0074] The system adjusts the positions of the building units with small constraint violation metrics (less than a preset threshold) until the constraint violation metrics are no longer less than the threshold, thereby obtaining the three-dimensional geometric data of the building group that satisfies the spatial constraints. The position adjustment eliminates the spatial conflicts between the building units by changing the spatial coordinates of the building units. After the position adjustment, the spatial layout of the building group can satisfy the design specifications and usage requirements.

[0075] The system can adopt a position adjustment algorithm based on physical simulation. Each building unit is regarded as a particle, and there is repulsion (from constraint violation) and attraction (from design intent) between the particles. The system determines the size of the repulsion according to the constraint violation metric and determines the size of the attraction according to the spatial relationship (such as symmetry, parallelism, etc.) between the building units. Then, the system simulates the motion trajectory of the particles using Newton's second law to obtain the new coordinates of the particles as the adjusted positions of the building units. In the iteration process, the system continuously detects the change of the constraint violation metric, and stops the iteration when the metric is no longer less than the threshold. In addition, the system can also perform collision detection on the building units after position adjustment to avoid introducing new spatial conflicts.

[0076] S104、based on the spatial visual importance network, the building group is divided into a plurality of visual correlation sub-regions, and the spatial entropy value of each visual correlation sub-region is calculated;

[0077] The system divides the building group into a plurality of visual correlation sub-regions based on the spatial visual importance network, and calculates the spatial entropy value of each visual correlation sub-region. The spatial entropy value is used to represent the complexity of the spatial distribution of the building units in the visual correlation sub-region.

[0078] The system first establishes the spatial visual importance network of the building group according to the three-dimensional geometric data obtained in the previous step. The network reflects the visibility relationship between the building units, and the nodes in the network represent the building units. The edge weight between the nodes is determined by the visibility of the building units. Then, the system performs community discovery based on the spatial visual importance network, and divides the building units with strong visibility into the same visual correlation sub-region. Finally, the system calculates the spatial entropy value of each sub-region, which measures the complexity of the spatial distribution of the building units and provides a basis for subsequent multi-level detail model generation.

[0079] The system can adopt a Graph Cut algorithm for visual association sub-region division. The spatial visual importance network is regarded as an undirected weighted graph, and the goal of the Graph Cut algorithm is to find a set of edges so that after deleting these edges, the graph is divided into several connected sub-graphs, while minimizing the sum of the weights of the cut edges. Intuitively, the sum of the weights of the cut edges reflects the strength of the visual association between different sub-regions. The visual association within a sub-region is strong, while the visual association between sub-regions is weak. After obtaining the visual association sub-regions, the system calculates the spatial entropy value of each sub-region. The calculation of the spatial entropy can be based on the spatial distribution histogram of the building units within the sub-region. The system uniformly divides the three-dimensional space and counts the number of building units falling into each grid, thereby obtaining the spatial distribution histogram. Then, the system calculates the Shannon entropy as the spatial entropy value according to the histogram.

[0080] S105, generating a multi-level detail model corresponding to each visual association sub-region according to the spatial entropy value and the three-dimensional geometric data of the building units within the visual association sub-region;

[0081] The system generates a corresponding multi-level detail model (LOD) for each visual association sub-region. Multi-level detail model refers to using different levels of geometric models to represent the same object at different viewing distances. The farther the viewing distance, the simpler the geometric model, and the closer the viewing distance, the more detailed the geometric model. The purpose of generating a multi-level detail model is to reduce the rendering calculation as much as possible while ensuring the visualization effect. The system determines the level of detail contained in the sub-region LOD model and the degree of simplification of each level according to the spatial entropy value of the sub-region and the three-dimensional geometric data of the building units.

[0082] For each visual association sub-region, the system first determines the LOD level that needs to be generated according to the spatial entropy value. Generally, the greater the spatial entropy value, the more complex the distribution of building units, and the more LOD levels that need to be generated. Then, for each LOD level, the system uses a geometric simplification algorithm to simplify the three-dimensional geometric data of the building units. Common geometric simplification algorithms include vertex clustering, edge collapse, and face merging. The vertex clustering algorithm reduces the number of model faces by merging vertices that are close in distance; the edge collapse algorithm merges two adjacent faces into one face by deleting edges in the model; and the face merging algorithm simplifies the model by merging faces that are coplanar or have a small angle. The system can combine vertex clustering, edge collapse, and face merging algorithms to obtain building unit LOD models with different degrees of simplification.

[0083] S106, constructing a rendering pipeline based on the multi-level detail model and the connection weight;

[0084] The system constructs a rendering pipeline for rendering the building group based on the multi-level detail model generated in the previous step and the connection weights of the spatial visual importance network. The rendering pipeline refers to a series of operation steps for converting a three-dimensional model into a two-dimensional image, usually including geometry processing, lighting calculation, texture mapping, shadow generation, post-processing, etc. The system determines the execution mode and order of each stage in the rendering pipeline according to the hierarchical structure of the LOD model and the connection weights between the visual association sub-regions, thereby achieving efficient and high-quality rendering of the building group.

[0085] The system can adopt an adaptive block rendering strategy to construct the rendering pipeline. The system divides the screen space into several rendering blocks, each corresponding to a visual association sub-region. For each rendering block, the system executes the rendering pipeline from coarse to fine according to the LOD level of the building unit it contains. In the geometry processing stage, the system loads the building unit geometry model of the corresponding LOD level, and transforms and projects the vertex coordinates. In the lighting calculation stage, the system calculates the vertex color according to the light source position and the material properties of the building unit. For different LOD levels, the system can choose lighting models of different complexity, such as Phong model, PBR model, etc. In the texture mapping stage, the system maps texture images to the surface of the building unit to provide surface details. Similarly, for different LOD levels, the system can use textures of different resolutions. In the shadow generation stage, the system calculates the shadow relationship between the building units. Considering the overhead of shadow calculation, the system can only generate dynamic shadows for high-LOD-level building units, while using pre-baked lighting maps for low-LOD-level building units. After completing each stage, the system obtains the rendering results of all building units in the sub-region. Finally, the system mixes the rendering results of the sub-regions based on the connection weights of the visual association sub-regions to obtain the final building object rendering image.

[0086] S107, visually consistent synthesis is performed on the output results of the rendering pipeline to generate the architectural design rendering of the building group.

[0087] The system performs post-processing on the initial rendering results output by the rendering pipeline to generate the final architectural design rendering. Since the rendering pipeline is executed in blocks, the initial rendering results may not be consistent between different visual association sub-regions, such as discontinuous lighting, abrupt shadow changes, etc. In order to eliminate these inconsistencies and improve rendering quality, the system needs to perform visually consistent synthesis on the initial rendering results. The goal of visually consistent synthesis is to adjust the visual properties such as color, brightness, contrast, etc. in the initial rendering results, so that adjacent sub-regions transition smoothly in vision, and the entire building group rendering image has global consistency.

[0088] The system can implement visual consistency synthesis by using an optimization-based method. The system constructs an energy function to measure the visual consistency of the rendered image. The energy function consists of two parts: a data term and a smoothing term. The data term measures the difference between the color of each pixel and the initial rendering result. The smaller the difference, the smaller the energy function value. The smoothing term measures the difference between the colors of adjacent pixels. The smaller the difference, the smaller the energy function value. The goal of the system is to find a set of pixel colors that minimizes the energy function value. This can be translated into a least squares optimization problem. Considering the high resolution of architectural design renderings, the scale of the optimization problem is large. Therefore, the system uses a multi-resolution optimization strategy. First, the system downsamples the initial rendering result to obtain a lower-resolution image; then, it performs optimization on the low-resolution image to obtain a result with better visual consistency; finally, the system upsamples the optimization result using bilinear interpolation and uses it as the initial value to perform optimization on the original resolution image to obtain the final visual consistency image. During the optimization process, the system can also adaptively adjust the weight of the smoothing term according to the connection weights of the visual correlation sub-regions. The greater the connection weight of adjacent regions, the higher the requirement for their smoothness, and the greater the weight of the smoothing term.

[0089] In the above embodiments, the three-dimensional geometric data of the building group is obtained to establish a spatial visual importance network, which can accurately reflect the visual occlusion relationship between building units, and then the building group is divided into visual correlation sub-regions based on the network and the spatial entropy value is calculated, which can quantitatively represent the spatial distribution characteristics of the building units in each sub-region. The multi-level detail model generated according to the spatial entropy value and the three-dimensional geometric data can reduce the model complexity while maintaining the visual effect, and the rendering pipeline constructed in combination with the connection weights can realize efficient rendering of the building group. Finally, the visual consistency synthesis of the rendering result can reduce the visual discontinuity generated in the rendering process, improve the realism and coherence of the overall rendering, and improve the real-time performance and practicality of the architectural design rendering generation.

[0090] On the basis of the first embodiment, in order to further improve the generation quality of the architectural design rendering, the embodiment of the present application also provides a quality evaluation and optimization method based on expert scoring. This method evaluates the quality of the generated rendering in multiple dimensions by establishing a standardized scoring system, and dynamically adjusts the rendering pipeline according to the evaluation results, thereby realizing quality closed-loop control of the rendering generation process. The following will be described in combination with Figure 2 The quality evaluation and optimization method based on expert scoring in the embodiment of the present application is described as follows:

[0091] Please refer to Figure 2 for a flowchart of the quality evaluation and optimization method based on expert scoring in the embodiment of the present application.

[0092] S201, collect a plurality of expert scoring samples to establish a scoring standard set;

[0093] The system collects a plurality of expert scoring samples to establish a scoring standard set, which includes spatial layout rationality, visual coherence, and material realism.

[0094] The system collects scoring data of different architectural design renderings from multiple experts as samples to establish a standardized scoring system. Expert scoring refers to experienced designers or expert reviewers in the field of architectural design subjectively evaluating and scoring various indicators of renderings. These indicators can include spatial layout rationality, visual element coherence, material map realism, and other aspects.

[0095] For different architectural design styles and types, the system can select multiple experts in the corresponding professional field to score, and through the collection and statistical analysis of a large number of samples, a set of general scoring rules and standards are summarized. For example, cluster analysis can be performed on sample data to extract statistical characteristics such as score interval, mean, and variance of each indicator, and the scoring weights of each indicator are summarized based on expert experience to form a standardized multi-dimensional scoring matrix.

[0096] In practical applications, the collection of expert scoring samples may encounter problems such as insufficient sample quantity, inconsistent scoring standards of individual experts, etc. Therefore, the system can expand the scale and scope of sample collection through crowdsourcing, and perform necessary preprocessing such as outlier removal and normalization on the original scoring data to ensure the representativeness and robustness of the scoring standard set. In addition, the system can periodically update and iterate the scoring standard set to adapt to the dynamic changes in architectural design aesthetics.

[0097] S202, extract local image blocks of the architectural design rendering and calculate the feature statistics of each local image block;

[0098] The system first divides the entire rendering into a plurality of local sub-regions, each sub-region being referred to as an image block. The division method can use fixed-size grid segmentation, adaptive block segmentation based on saliency, etc. Then, the system extracts multiple visual features for each image block and calculates the corresponding statistics. Common features can include color histograms, gradient histograms, texture features, etc., and the corresponding statistics include mean, variance, skewness, kurtosis, etc.

[0099] In feature extraction, the system can fully consider the high-level semantic characteristics of architectural design images. For example, a target detection and segmentation algorithm based on deep learning can be used to extract key structural elements such as building outlines, windows, and balconies, and then calculate their size ratios, alignment methods, and other layout features. For example, for interior decoration renderings, the style features of furniture, lighting, and ornaments can be extracted.

[0100] S203, cluster analysis of local image block features is performed to obtain a feature distribution map;

[0101] This step uses clustering analysis method to study the internal distribution pattern of local image block features, thereby depicting the visual characteristics of the overall rendering. Clustering analysis is an unsupervised learning method that can automatically classify similar data samples into a class and reveal the implicit structure of the data. By taking the feature vectors of all image blocks as input samples, the system can use classical algorithms such as K-means and spectral clustering to perform clustering and obtain a number of cluster centers, each center representing a typical feature pattern.

[0102] Further, the system can calculate the similarity between each image block feature vector and each cluster center to generate a feature distribution map. The position of the image block corresponds to the spatial layout of the rendering, the similarity corresponds to the degree of agreement between the image block feature and a certain cluster pattern, and the similarity value can be encoded with color or brightness to form an intuitive visual result. By observing the feature distribution map, the visual characteristics of the rendering in different areas can be determined, such as the harmony of color matching and the uniformity of texture arrangement.

[0103] S204, calculating the spatial layout score, visual coherence score, and material authenticity score according to the feature distribution map;

[0104] This step is based on the feature distribution map obtained above to quantitatively evaluate the architectural design rendering from the aspects of spatial layout, visual coherence, and material authenticity. The system first refers to the artificially annotated scoring samples to establish a mapping relationship between the feature distribution map and the scores. For example, regression analysis, support vector machines, and other machine learning methods can be used to take the statistical features of the feature distribution map (such as histograms, co-occurrence matrices, etc.) as input to train a prediction function model, thereby obtaining an automatic scoring mechanism for new renderings.

[0105] The spatial layout score reflects the rationality of the layout relationship of each structural element in the rendering, such as scale, proportion, and alignment. The system can extract geometric parameters such as area ratio and golden section ratio of key parts such as building contour, door and window, column and beam from the feature distribution map, and calculate the layout score combined with the building design specification. The visual coherence score measures the harmony and unity of the whole rendering in color matching, light transition, etc. The system can analyze the subject and object color, cold and warm contrast, light and shade level, etc. from the feature distribution map by using color aesthetics theory, and consider the composition rules such as shape repetition and rhythm to give a comprehensive score. The material reality score focuses on evaluating the degree of realism of the mapped material. The system can start from the texture features of the feature distribution map, such as roughness, glossiness, and transparency, and refer to the visual properties of real material samples to give a score.

[0106] It should be noted that due to the correlation and constraint relationship between different evaluation indicators, the system needs to balance the comprehensive influence of multiple indicators when scoring. For example, too dense spatial layout may affect the material reality, and too rich and jumping color matching may destroy the overall visual coherence. Therefore, the system can establish a dependency model between the scoring indicators, and through methods such as conditional random field and Bayesian network, the scores of each item are jointly inferred to obtain a more comprehensive and objective evaluation result. At the same time, for different design styles and application scenarios, the system can also flexibly adjust the weight proportion of each indicator to adapt to the individual preferences of users.

[0107] S205, normalize the spatial layout score, the visual coherence score and the material reality score to form a quality score vector;

[0108] This step integrates the three scoring indicators obtained in the previous step into a comprehensive quality score vector for subsequent overall optimization of the rendering pipeline. Since the value range and dimension of different scoring indicators are not the same, the system first needs to normalize them to map them to a unified interval (such as [0, 1]). Common normalization methods include maximum and minimum value normalization, zero mean and unit variance normalization, etc.

[0109] After normalization, the system combines the three scores into a three-dimensional column vector, and each component represents the score of the corresponding indicator. By introducing the concept of score vector, the overall quality of the rendering can be represented as a mathematical quantitative indicator, which is convenient for tracking the optimization process. At the same time, the score vector can also be used as a visualization tool to intuitively show the pros and cons of different design schemes in each indicator.

[0110] In constructing the score vector, the system can also dynamically adjust the weights of each component in combination with the user's interactive feedback. For example, an interactive interface can be designed to allow the user to manually drag to adjust the importance of each evaluation index, and the system accordingly adaptively generates a weighted score vector and updates the optimized focus direction accordingly. This interactive mechanism makes the evaluation process more flexible and variable, and fully respects the user's subjective preferences.

[0111] S206, adjust the rendering pipeline according to the quality score vector until each dimension component of the quality score vector is greater than a preset second threshold.

[0112] This step uses the quality score vector obtained in the foregoing to achieve closed-loop optimization of the rendering pipeline by setting a target threshold. The system first sets a quality threshold to be reached for each component of the score vector. When each component exceeds this threshold, it is considered that the current design effect meets the comprehensive quality requirements.

[0113] In the optimization process, the system adaptively adjusts the parameters and sub-modules of the rendering pipeline based on the score vector. Specifically, a gradient-based optimization algorithm such as SGD, Adam, etc. can be used to treat the score vector as a multi-objective function, and the gradient direction is solved to iteratively update the rendering parameters. Common adjustable parameters include the softness of global lighting, the sharpness of shadows, the highlight intensity of materials, the saturation contrast of colors, etc. At the same time, for some indicators with low scores, the system can selectively enhance the corresponding processing modules of the rendering pipeline, such as adding stylized filters, cartography optimization, and other additional post-processing effects, to focus on improving the performance of weak links.

[0114] The optimization process usually requires multiple iterations, and the system updates the score vector based on the rendered image of each iteration and dynamically adjusts the step size and direction of optimization until each component of the score vector exceeds the preset threshold. The entire closed-loop optimization mechanism realizes automatic feedback adjustment of rendering parameters and quality evaluation, improving the generation efficiency of design effects.

[0115] In the above embodiments, expert score samples are introduced to establish a score standard set, and multiple dimensions of evaluation indicators such as spatial layout rationality, visual coherence, and material authenticity are quantified, so that the quality evaluation of the rendering image has an objective basis. The quality score vector obtained through multi-dimensional quality evaluation can accurately reflect the performance level of the rendering image in different aspects. According to the quality score vector, the rendering pipeline is adjusted in a targeted manner, and through iterative optimization, each dimension component meets the preset threshold requirement, which can continuously improve the overall quality of the rendering image, make the generation process of the rendering image have clear quality control standards, ensure the professional level of the final rendering image, reduce the workload of repeated manual modification, and improve the automation degree and quality controllability of the rendering image generation.

[0116] The system in the embodiments of the present application will be described from the perspective of hardware processing. Please refer to Figure 3 FIG. 1 is a schematic diagram of an entity device structure of an architectural design effect picture automatic generation system provided by the embodiments of the present application.

[0117] It should be noted that Figure 3 The structure of the system shown is only an example, and should not bring any limitation to the functions and use range of the embodiments of the present application.

[0118] As Figure 3 shown, the system includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 302 or programs loaded from a storage portion 308 into a random access memory (RAM) 303, such as performing the method in the above embodiments. In the RAM 303, various programs and data required for system operation are also stored. The CPU 301, the ROM 302, and the RAM 303 are connected to each other through a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0119] The following components are connected to the I / O interface 305: an input portion 306 including a camera, an infrared sensor, and the like; an output portion 307 including a liquid crystal display (LCD), a speaker, and the like; a storage portion 308 including a hard disk and the like; and a communication portion 309 including a network interface card such as a LAN (Local Area Network) card, a modem, and the like. The communication portion 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, and the like is mounted on the drive 310 as needed, so that a computer program read therefrom is installed in the storage portion 308 as needed.

[0120] In particular, the processes described above with reference to the flowcharts can be implemented as a computer software program in accordance with embodiments of the present application. For example, embodiments of the present application include a computer program product comprising a computer program carried on a computer readable medium, the computer program comprising computer programs for executing the methods illustrated by the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via the communication section 309, and / or installed from the removable medium 311. When the computer program is executed by the central processing unit (CPU) 301, various functions defined in the present application are executed.

[0121] It should be noted that the computer readable medium shown in the embodiments of the present application can be a computer readable signal medium or a computer readable storage medium or any combination of the two. The computer readable storage medium may, for example, be but is not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination of the above. More specific examples of the computer readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, the computer readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus or device. In the present application, the computer readable signal medium can include a data signal carried in a baseband or as part of a carrier wave, in which the computer readable computer program is carried. Such a propagated data signal can take a variety of forms, including but not limited to an electromagnetic signal, an optical signal or any suitable combination of the above.

[0122] The computer program product of the present application can be a computer program embodied on a tangible medium or transmitted from a storage medium to a computer or a processor. The computer program product can be a plug-in, a cartridge or an external interface with the computer. The computer program product can also be loaded onto a computer or a processor to produce a special purpose machine, thereby causing the computer or the processor to perform the operations of the application.

[0123] As another aspect, the present application also provides a computer readable storage medium, which can be included in the system described in the above embodiments, or can exist separately without being assembled into the system. The storage medium carries one or more computer programs, which, when executed by a processor of a system, cause the system to implement the method provided in the above embodiments.

[0124] The above embodiments are only used to illustrate the technical solutions of the present application, but not limit the present application; even though the present application has been described in detail with reference to the foregoing embodiments, those ordinarily skilled in the art should understand: the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced equivalently; and the modification or replacement does not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0125] In the above embodiments, according to the context, the term "when" can be interpreted as meaning "if" or "after" or "in response to determining" or "in response to detecting". Similarly, according to the context, the phrase "on determining" or "if detecting (the stated condition or event)" can be interpreted as meaning "if determining" or "in response to determining" or "on detecting (the stated condition or event)" or "in response to detecting (the stated condition or event)".

[0126] In the above embodiments, all or part of the embodiments can be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present application are generated. The computer can be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The computer instructions can be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another computer readable storage medium, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired (such as coaxial cable, optical fiber, digital subscriber line) or wireless (such as infrared, wireless, microwave, etc.) manner. The computer readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. integrated with one or more available media. The available media can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid state disk) and the like.

[0127] Those of ordinary skill in the art understand that all or part of the processes in the above embodiments can be implemented by a computer program to instruct the relevant hardware, which can be stored in a computer readable storage medium. The program can include the processes of the above method embodiments when executed. The aforementioned storage medium includes ROM or random access memory (RAM), magnetic disk or optical disk, and various media that can store program codes.

Claims

1. A method for automatically generating architectural design renderings, characterized in that, include: Acquire three-dimensional geometric data of the building group, and establish a spatial visual importance network of the building group based on the three-dimensional geometric data. The nodes of the spatial visual importance network represent building units, and the connection weights between the nodes are determined according to the visual occlusion relationship between the building units. Based on the spatial visual importance network, the building group is divided into several visually related sub-regions, and the spatial entropy value of each visually related sub-region is calculated. The spatial entropy value is used to characterize the spatial distribution complexity of the building units within the visually related sub-region. Based on the spatial entropy value and the 3D geometric data of the building units within the visually associated sub-region, a multi-level detail model is generated for each visually associated sub-region, specifically including: A geometric feature matrix is ​​constructed for the building units within the visually associated sub-region, where each row of the geometric feature matrix represents a geometric feature vector of the building unit; The dimensional threshold for feature decomposition is determined based on the spatial entropy value, and singular value decomposition is performed on the geometric feature matrix to obtain the main feature vector set. The main feature vector set is sorted according to the size of the singular values, and the features are reconstructed according to the dimensional threshold to obtain a multi-level detail model; The rendering pipeline is constructed based on the multi-level detail model and the connection weights, specifically including: Construct a rendering task priority queue, wherein the task priority in the rendering task priority queue is determined according to the connection weight; The level of detail model is determined for each rendering task in the rendering task priority queue based on the distance from the preset viewpoint to the corresponding visually associated sub-region. Establish a rendering computation graph, where nodes represent rendering tasks and edges between nodes represent task dependencies. A multi-level cache structure is constructed based on the topology of the rendering computation graph; The rendering task priority queue, the multi-level cache structure, and the rendering computation graph are integrated into a rendering pipeline; The output of the rendering pipeline is visually consistent and composited to generate architectural design renderings of the building complex.

2. The method according to claim 1, characterized in that, Prior to acquiring the three-dimensional geometric data of the building complex, the method further includes: Construct a hierarchical data structure for the building complex, wherein the hierarchical data structure includes the spatial topological relationships, functional attributes, and material parameters of the building units; Geometric constraint checks are performed based on the hierarchical data structure to generate constraint violation metrics, which are used to characterize the degree of spatial conflict between the building units. For building units whose constraint violation metric is less than a preset first threshold, the positions are adjusted until the constraint violation metric is not less than the preset first threshold, thereby obtaining three-dimensional geometric data that satisfies spatial constraints.

3. The method according to claim 2, characterized in that, The step of performing geometric constraint checks based on the hierarchical data structure and generating constraint violation metrics specifically includes: Construct a spatial index and store the bounding box information of the building unit in the corresponding spatial partitioning node; The spatial index is used to detect the occlusion relationship between the building units to obtain the occlusion detection results; A constraint graph is constructed based on the occlusion detection results. The nodes of the constraint graph represent the building units, and the edges represent spatial constraint relationships. Connectivity analysis is performed on the constraint graph to calculate the constraint violation metric.

4. The method according to claim 1, characterized in that, After generating the architectural design renderings of the building complex, the method further includes: Several sets of expert rating samples were collected to establish a set of rating criteria, which included the rationality of spatial layout, visual coherence, and material authenticity. Based on the aforementioned scoring criteria set, a multi-dimensional quality assessment of the architectural design renderings is performed to obtain a quality score vector. The rendering pipeline is adjusted based on the quality score vector until all components of the quality score vector are greater than a preset second threshold.

5. The method according to claim 4, characterized in that, The multi-dimensional quality assessment of the architectural design renderings based on the scoring criteria set, to obtain a quality scoring vector, specifically includes: Extract local image patches from the architectural design renderings and calculate the feature statistics of each local image patch; Cluster analysis is performed on the features of the local image patches to obtain a feature distribution map; Calculate the spatial layout score, visual coherence score, and material realism score based on the feature distribution map; The spatial layout score, visual coherence score, and material realism score are normalized and then combined to form a quality score vector.

6. An automatic architectural design rendering generation system, characterized in that, The system includes: One or more processors and a memory; the memory is coupled to the one or more processors, the memory being used to store computer program code, the computer program code including computer instructions, the one or more processors invoking the computer instructions to cause the system to perform the method as described in any one of claims 1-5.

7. A computer-readable storage medium comprising instructions, characterized in that, When the instructions are executed on the system, the system performs the method as described in any one of claims 1-5.

8. A computer program product, characterized in that, When the computer program product is run on the system, it causes the system to perform the method as described in any one of claims 1-5.

Images