Method and system for virtually repairing cultural building based on three-dimensional scanning

Abstract

The invention relates to the technical field of building virtual restoration, in particular to a cultural building virtual restoration method and system based on three-dimensional scanning, and the method comprises the steps: carrying out the multi-mode digital recording of a target building, obtaining the geometric morphology information, surface texture information and material spectrum information of the target building, and constructing a reference digital model; synchronously acquiring environment pedigree data and humanistic inheritance pedigree data of a region where the target building is located; performing damage exploration on the reference digital model to generate a building damage map; the building damage map is used for representing component deformation characteristics, material deterioration characteristics and accessory decoration information loss characteristics; and performing multi-source coupling analysis of the damage mechanism on the building damage map, the environment pedigree data and the humanistic inheritance pedigree data. Through the method, multi-source information can be fused, a damage mechanism can be analyzed, restoration rehearsal can be carried out in a virtual space, and the sustainability of historical and cultural building protection is improved.

Description

Technical Field

[0001] This invention relates to the technical field of virtual building restoration, and in particular to a method and system for virtual restoration of cultural buildings based on 3D scanning. Background Technology

[0002] Digital technology has become an indispensable tool in the restoration and maintenance of traditional villages and historical buildings. Currently, the industry widely employs technologies such as 3D laser scanning and oblique photography to collect data on the building itself, creating digital 3D models to record its current state and assist in surveying and display. For example, related research and practice have enabled the precise measurement of building geometric dimensions and the mapping and restoration of surface textures through point cloud data processing, forming static digital archives.

[0003] However, the formulation of existing restoration strategies relies on static diagnosis of the current visible damage and expert experience, lacking a systematic analysis of the root causes of the damage and a quantitative assessment of the long-term effects of different restoration measures. Specifically, although current methods can record surface phenomena such as cracks, weathering, and missing parts in detail, it is difficult to scientifically answer which factor among environmental stresses such as freeze-thaw cycles and wet-dry changes is dominant, or whether it is exacerbated by human activities such as improper historical modifications and changes in usage functions. As a result, restoration plans often address the surface symptoms without fundamentally suppressing the driving factors that cause the damage, which may lead to repairs that result in further damage. At the same time, it is difficult to accurately assess which strategy can achieve the longest-lasting protection effect with minimal intervention and optimal cost before intervention.

[0004] Therefore, there is an urgent need for a virtual restoration method for cultural buildings based on 3D scanning that can integrate multi-source information, analyze damage mechanisms, and perform virtual simulations in virtual space. Summary of the Invention

[0005] This invention provides a method and system for virtual restoration of cultural buildings based on 3D scanning, which can effectively solve the problems in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: Virtual restoration methods for cultural buildings based on 3D scanning include: Multimodal digital recording of the target building is performed to obtain its geometric shape, surface texture, and material spectral information, and a baseline digital model is constructed based on this information. Simultaneously, environmental and cultural heritage data of the region where the target building is located are also acquired. The baseline digital model is subjected to damage investigation to generate a building damage map; the building damage map is used to characterize the deformation characteristics of components, the material deterioration characteristics, and the loss characteristics of attached decoration information. Multi-source coupled analysis of the damage mechanism is performed on the building damage map, the environmental spectrum data, and the cultural heritage spectrum data to identify the dominant environmental stress factors and the influence paths of human activities that lead to the damage; Based on the dominant environmental stress factor and the influence path of human activities, a virtual intervention experiment under multiple scenarios is executed in a preset virtual space to obtain the virtual experiment results under each scenario. The virtual intervention experiment includes applying simulated intervention measures to the benchmark digital model and calculating the morphological stability response characteristics of the benchmark digital model within a set time span after the intervention. Based on the results of virtual experiments in various scenarios, intervention and repair strategies that meet the preset morphological stability threshold and intervention ethical constraints are determined and implemented.

[0007] Furthermore, the multimodal digital recording includes: Point clouds of the building's main structure were obtained using a terrestrial 3D laser scanner. Oblique photography using drones was employed to acquire images of the roof and its macroscopic shape. Hyperspectral imaging equipment was used to collect spectral data on the composition and weathering characteristics of the surface material.

[0008] Furthermore, the baseline digital model is subjected to damage investigation to generate a building damage map, including: The benchmark digital model is deconstructed semantically according to traditional building regulations; For each type of component, geometric distortion, texture damage, and material defects are detected by using curvature analysis of its point cloud data, spectral analysis of its texture image, and feature offset analysis of its spectral data. The various test results are integrated, labeled, and classified according to spatial location to form the building damage map.

[0009] Furthermore, the environmental spectrum data includes historical climate sequences, microenvironmental temperature and humidity field distribution, and atmospheric pollutant deposition flux data; The data on the cultural heritage includes documents and oral records related to traditional construction techniques, maintenance customs, and functional adaptations.

[0010] Furthermore, a multi-source coupling analysis of the damage mechanism is conducted, and a causal inference model based on graph neural networks is adopted. The nodes of the causal inference model represent building components, environmental factors and human activities, and the edges represent the interaction relationships and intensity among them. The dominant environmental stress factors and the influence paths of human activities are inferred and output through the causal inference model.

[0011] Furthermore, virtual intervention trials in multiple scenarios include: Define multiple intervention measures; In the simulation environment, each intervention is mapped to a modification of a specific parameter of the baseline digital model; The building physics performance simulation engine is driven to calculate the deformation accumulation, stress distribution and damage evolution of the benchmark digital model after parameter modification within a set number of years after loading a regional environmental load sequence, thereby obtaining the morphological stability response characteristics.

[0012] Furthermore, the intervention ethical constraints include the principles of minimal intervention, reversibility, material and process compatibility, and architectural integrity. The morphological stability threshold is a pre-set allowable deformation and stress standard based on the cultural relic protection level, structural type, and current degree of damage of the target building.

[0013] Furthermore, the specific parameters include material strength, surface permeability, and connection stiffness.

[0014] Furthermore, the intervention measures include any one or a combination of two or more of cleaning and reinforcement, structural support, material replacement, and surface sealing.

[0015] On the other hand, the present invention also provides a virtual restoration system for cultural buildings based on 3D scanning, comprising: The architectural recording module is used to acquire the geometric shape information, surface texture information and material spectral information of the target building, and to construct a benchmark digital model based on these; and simultaneously acquires the environmental spectrum data and cultural heritage spectrum data of the region where the target building is located. The damage detection module is used to detect damage on the benchmark digital model and generate a building damage map that characterizes the deformation features of components, the deterioration features of materials, and the loss features of attached decorative information. The analysis module is used to perform multi-source coupled analysis of the damage mechanism of the building damage map, the environmental spectrum data and the cultural heritage spectrum data, and to identify the dominant environmental stress factors and the influence paths of human activities that lead to the damage. The virtual experiment module is used to perform multi-scenario intervention virtual experiments in a preset virtual space based on the dominant environmental stress factor and the influence path of human activities, and to obtain the virtual experiment results under each scenario. The intervention virtual experiment includes applying simulated intervention measures to the benchmark digital model and calculating the morphological stability response characteristics of the benchmark digital model after intervention within a set time span. The strategy determination module is used to determine and execute intervention and repair strategies that meet the preset morphological stability threshold and intervention ethical constraints based on the results of virtual experiments in various scenarios.

[0016] The technical solution of this invention can achieve the following technical effects: By coupling multimodal digital models with environmental and cultural data, a dynamic simulation method integrating architecture, environment, and culture can be constructed. This method can record the current state of a building and simulate its long-term evolution under different restoration strategies, enabling a virtual projection of the life cycle of historical buildings. Through multi-source coupling analysis of damage mechanisms, a quantitative correlation can be established between building damage phenomena and environmental stress factors and the influence paths of human activities. By constructing the coupling relationship between damage characteristics and environmental factors and human activities, the key driving factors that dominate damage can be identified. By simulating various intervention scenarios in virtual space, the comprehensive performance of different strategies in terms of morphological stability and intervention ethics can be evaluated, achieving global optimization of restoration schemes. This can meet the morphological stability threshold requirements and comply with the ethical constraints of minimum intervention and maximum protection, avoiding the vicious cycle of repairing and then destroying in traditional restoration.

[0017] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the three-dimensional scanning virtual restoration method for cultural buildings according to the present invention. Figure 2 This is a structural block diagram of the three-dimensional scanning virtual restoration system for cultural buildings according to the present invention. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0022] like Figure 1 As shown, the virtual restoration method for cultural buildings based on 3D scanning of the present invention specifically includes the following steps: Step S1: Perform multimodal digital recording of the target building to obtain its geometric shape information, surface texture information, and material spectral information, and construct a benchmark digital model based on these information; and simultaneously obtain environmental spectrum data and cultural heritage spectrum data of the region where the target building is located. Step S2: Conduct damage investigation on the benchmark digital model to generate a building damage map; the building damage map is used to characterize the deformation characteristics of components, the material deterioration characteristics, and the loss characteristics of auxiliary decoration information; Step S3: Perform multi-source coupling analysis of the damage mechanism on the building damage map, the environmental spectrum data, and the cultural heritage spectrum data to identify the dominant environmental stress factors and the influence paths of human activities that lead to the damage; Step S4: Based on the dominant environmental stress factor and the influence path of human activities, execute virtual intervention experiments under multiple scenarios in a preset virtual space to obtain the results of virtual experiments under each scenario; the virtual intervention experiment includes applying simulated intervention measures to the benchmark digital model and calculating the morphological stability response characteristics of the benchmark digital model within a set time span after intervention. Step S5: Based on the results of the virtual experiments in each scenario, determine the intervention and repair strategies that meet the preset morphological stability threshold and intervention ethical constraints, and then execute them.

[0023] In this embodiment, by coupling a multimodal digital model with environmental and cultural data, a dynamic simulation method integrating architecture, environment, and culture is constructed. This method can record the current state of a building and simulate its long-term evolution under different restoration strategies, enabling a virtual projection of the life cycle of historical buildings. Through multi-source coupling analysis of damage mechanisms, a quantitative correlation is established between building damage phenomena and environmental stress factors and the influence paths of human activities. By constructing the coupling relationship between damage characteristics and environmental factors and human activities, the key driving factors that dominate damage are identified. By simulating various intervention scenarios in virtual space, the comprehensive performance of different strategies in terms of morphological stability and intervention ethics can be evaluated, achieving global optimization of restoration plans. This approach can meet the morphological stability threshold requirements while adhering to the ethical constraints of minimal intervention and maximum protection, avoiding the vicious cycle of repairing and then re-damaging in traditional restoration methods.

[0024] In a specific implementation, as one example, given that building damage is the result of the coupling effect of environmental stress and human activities, it is necessary to simultaneously acquire multi-dimensional features of the building itself and data on external influencing factors to construct a spatiotemporally correlated benchmark digital model. This example achieves multimodal digital recording of the target building by constructing a multi-source data collaborative acquisition system, as detailed below: Step S11: Deploy multiple automatic temperature and humidity recorders and one mini weather station within the target building's courtyard and along its main facade. The mini weather station continuously records temperature, relative humidity, precipitation, wind speed, and wind direction data at 10-minute intervals. All recording devices have built-in clocks that synchronize with the BeiDou satellite time signal via the NTP protocol, with a time synchronization error of less than 1 second. This monitoring system provides continuous and accurate time-series and microenvironmental field data for linking environmental data with building status.

[0025] Step S12: Based on the spatiotemporal reference established in Step S11, plan and execute a collaborative acquisition operation of ground laser scanning, UAV oblique photography, and hyperspectral imaging; a phase-detection scanner is used for the ground 3D laser scanning operation; 12 scanning stations are planned inside and outside the building to ensure that all facades, interior spaces, and main structural components are unobstructed; before scanning at each station, at least 3 targets with known coordinates are placed in the field of view; the scanning obtains a high-density point cloud containing coordinate and reflection intensity information, and the scanning density at a single station is better than 2 mm at a distance of 10 meters; point cloud registration is performed using targets to generate an overall point cloud model, and the coordinate residuals at the control points are verified to be less than ±5 mm; The oblique photography operation using a multi-rotor drone equipped with a five-lens oblique photography camera; the flight path design ensures that the image coverage of building roofs, complex ridges and eaves has an overlap rate of more than 85%; the drone flies at an altitude of about 30 meters above the highest point of the building, and the ground sampling distance is better than 3 centimeters; during the flight, the drone's GNSS module records the camera station coordinates of each image in real time; through aerial triangulation software, combined with ground control points, the precise exterior orientation elements of the image are calculated to generate an orthophoto map with real geographic coordinates and a real-scene 3D mesh model; Hyperspectral imaging acquisition employed a ground-based push-broom hyperspectral imager to collect data on the two main weathered facades of the building, facing south and west. The imager's spectral range covered 400-1000 nanometers, with a spectral resolution better than 5 nanometers. During acquisition, a standard reflectance whiteboard was placed within the field of view for subsequent radiometric calibration. The hyperspectral data cube generated by the push-broom technique contained a continuous spectral reflectance curve for each pixel. Through spectral feature analysis, diagnostic absorption characteristics of different materials such as bricks, clay tiles, wood, and mineral pigments could be identified, and the degree of surface weathering could be quantified based on specific spectral indices, such as the characteristic absorption depth ratio of weathered clay minerals.

[0026] Step S13: Import the collected laser point cloud, real-scene 3D mesh model, and hyperspectral data into the same data processing platform. First, using the high-precision laser point cloud as a geometric reference, the real-scene 3D mesh model is precisely registered using an iterative nearest-point algorithm. Then, using the geometric information inherent in the hyperspectral data, it is spatially aligned with the registered real-scene model through feature matching. After registration, each triangular facet or vertex of the 3D mesh model, in addition to having geometric coordinates and RGB color attributes, is also assigned a spectral feature vector or derived material classification label from the corresponding hyperspectral pixel. Thus, a reference digital model integrating millimeter-level geometric accuracy, centimeter-level texture realism, and material spectral attributes is generated. The reference digital model is stored in an open format, and its geometric, texture, and attribute information can be independently accessed and called by subsequent analysis modules.

[0027] Step S14: Environmental spectrum data acquisition is carried out simultaneously with on-site data collection; daily meteorological observation data of the county / city where the target building is located over the past 50 years are obtained, including maximum / minimum temperature, precipitation, and average wind speed; statistical analysis is performed on the monitoring data of the micro-meteorological station deployed on-site for 30 consecutive days, and combined with computational fluid dynamics simulation, interpolation is used to generate typical daily temperature and humidity field distribution maps with a resolution of 1m×1m within 50 meters around the target building in winter and summer; by reviewing local environmental quality annual reports and relevant research literature, the annual average concentration data of sulfur dioxide and nitrogen oxides in the region over the past ten years are obtained, and the approximate value of its deposition flux under atmospheric stability is estimated based on the general dry deposition model.

[0028] Step S15: The data on the lineage of cultural heritage was obtained through literature research and field interviews; the county annals and village archives of the county where the village is located, as well as the records of the various surveys of the residence, were systematically reviewed; the records of the building’s initial construction date and major renovations were selected and digitized.

[0029] In this embodiment, by establishing a precise spatiotemporal benchmark, collaboratively collecting multimodal data, and structurally fusing and associating multi-source information, a complete data foundation is formed to support the shift from static recording to dynamic mechanism analysis and predictive decision-making. The core difference between this embodiment and existing isolated, single-dimensional data collection lies in the fact that by constructing machine-readable explicit relationships between data of various dimensions based on a unified spatiotemporal framework, the building status, environmental effects, and human history can be jointly examined in a computable analytical model.

[0030] In some embodiments of the present invention, existing damage analysis of digital building models typically employs general point cloud or image processing algorithms. For example, geometric anomaly regions are found by calculating the curvature change of the overall point cloud, or the overall texture image is segmented to identify color differences and patches. The above methods treat the building as a homogeneous entity, and its analysis unit is disconnected from the structural and decorative components in the architectural sense. The resulting anomaly detection results are a series of discrete geometric clusters or pixel blocks without clear engineering semantics. Protection experts need to invest a lot of effort in manual interpretation, classification, and association to map them to specific beams, columns, walls, or sculpted components and determine their damage type. This process is inefficient, highly subjective, and makes it difficult to systematically establish the correlation between geometric deformation, surface texture damage, and deep material degradation. Based on the above issues, the analytical units for diagnosing damage to cultural heritage buildings must be consistent with the structural logic and cultural characteristics of the building itself. Therefore, it is necessary to first deconstruct the digital model into semantic components with clear functions and material properties according to traditional building regulations. Then, for each type of component, its geometric, visual, and spectral data characteristics are customized and interpreted to achieve targeted identification and correlation from data anomalies to engineering defects. Specifically, the following operations are performed: Step S21: Invoke the pre-built component knowledge base for the target architectural culture type; this knowledge base is established based on local traditional construction rules, craftsman interview records and architectural survey data, and is used to define a component classification system from broad categories to specific items; for example, for Manchu dwellings, the system can include four major categories: timber frame, masonry walls, roof tiles, and decoration; among them, the timber frame category can be further subdivided into columns, beams, purlins, rafters, etc.; the masonry wall category can be divided into gable walls, eaves walls, etc. according to location; The knowledge base is used to semi-automatically deconstruct the benchmark digital model. First, based on the spatial distribution of point clouds and geometric features such as cylindricity and flatness, a region growing and rule classification algorithm is used to initially segment candidate component instances. Then, through a human-computer interaction interface, engineers verify, correct, and finally semantically annotate the automatic segmentation results according to the knowledge base. Each confirmed component instance is given a unique identifier and associated with its type, theoretical material, and structural role attributes.

[0031] Step S22: For each semantically annotated component instance, perform the following quantitative analysis in parallel: a) Geometric distortion detection based on point cloud curvature: Extract the point cloud subset corresponding to the component and calculate its local surface curvature; compare the calculated curvature distribution with the baseline geometric model that the component should have in its intact state; for planar components such as walls and floors, detect their deviation from the best-fit plane and local depressions; for linear components such as columns and purlins, detect their straightness or standard cylindricity deviations; the output results are the geometric deviation of each point on the surface of the component and its spatial distribution map. b. Texture damage detection based on texture spectrum; obtain high-resolution texture image blocks mapped to the surface of the component; divide the image into blocks and perform a two-dimensional Fourier transform on each sub-block to convert it to the frequency domain; analyze its spectral characteristics: intact and homogeneous texture regions have spectral energy concentrated in the low-frequency part; while cracks, peeling, holes and other damage will introduce high-frequency components or cause spectral energy to concentrate in a specific direction; by training a support vector machine classifier, distinguish the spectral patterns representing damage and integrity, thereby locating the damaged area on the component texture and initially classifying it as linear cracks or planar defects; c. Material lesion detection based on spectral shift; locate the hyperspectral reflectance curve associated with each pixel on the surface of the component; retrieve the reference spectrum matching the material labeled on the component from the pre-stored standard healthy material spectral database; calculate the matching degree or deviation index between the spectrum of each pixel and the reference spectrum in the characteristic absorption band; for example, for blue bricks, focus on the depth and morphological changes of the absorption characteristics of its clay minerals near 2200 nm; for wood, focus on the changes in the characteristic absorption peak of cellulose; areas with deviation exceeding the set threshold are identified as material lesions, such as weathering, salting out, and mold growth, and are classified according to the degree of deviation.

[0032] Step S23: Establish a component-based fusion analysis process; overlay and register the geometric deviation distribution map, texture damage classification map, and material lesion grade map obtained on the same component according to their common three-dimensional spatial coordinate system; the rules for implementing fusion decisions are as follows: When a certain spatial location is simultaneously detected to have geometric deviation, clear crack texture, and obvious material spectral shift, it is determined that there is serious structural or complex damage, and its detection confidence level is set to the highest. When a region only shows a material spectral shift without obvious geometric or textural anomalies, it is recorded as an early or potential lesion; Referring to industry standards such as "Classification and Illustration of Diseases of Stone Cultural Relics", and combining geometric deformation, the proportion of damaged area and the degree of material disease, a comprehensive classification of each damaged area on the component is carried out; The results of the above fusion analysis are integrated into the baseline digital model in the form of vector graphics layers and attribute database records to form a building damage atlas. In the building damage atlas, each damage record is associated with its component identifier, three-dimensional spatial location, damage type, severity level, and the values ​​of various detection indicators on which it is based.

[0033] In this embodiment, the resulting damage map directly organizes information using building components as carriers, clearly indicating which components have specific damage types and degrees, eliminating the complex manual translation process required for general algorithm results. This allows the diagnostic conclusions to be directly used by structural engineers and conservationists for safety assessments and restoration planning, improving information utilization efficiency. By mandating the comparison and fusion of geometric, textural, and spectral evidence at the same spatial location, it is possible to distinguish between single surface phenomena and deep-seated damage caused by multiple coupled factors. For example, it can distinguish between areas with only surface contamination and areas with both internal hollowing and surface cracking.

[0034] In practical implementation, as an example, existing methods for analyzing the causes of building damage mostly remain at the level of qualitative inference or simple statistical analysis. Qualitative inference relies on the personal experience of experts, and the conclusions are difficult to quantify, verify, and reproduce. Traditional statistical analysis methods directly correlate component damage indicators with environmental monitoring data, failing to express the structural or functional relationships between components. They are also unable to handle discrete and temporally sequential influencing factors such as historical and cultural events. Statistical correlation cannot effectively distinguish causal relationships, nor can it reveal the specific action path from cause to effect. This embodiment constructs a heterogeneous graph that integrates building components, environmental factors, and cultural event nodes, and uses graph neural networks to learn the causal relationships in the graph to achieve a structured analysis of the damage mechanism. The specific implementation steps are as follows: Step S31: Construct a heterogeneous graph of damage mechanisms. The heterogeneous graph of damage mechanisms includes three types of nodes and four types of edges. The node types include building component nodes, environmental stress factor nodes, and human activity event nodes. Building component nodes directly correspond to each independent component after semantic deconstruction in step S21. Their node attributes are derived from the building damage map generated in step S2, including the component's identifier, type, material, spatial coordinates, and quantified damage feature vector. Environmental stress factor nodes are created based on the environmental spectrum data obtained in step S14, such as macro-climate sequence indicators, micro-environmental field grid cell attributes, and pollutant deposition flux data. Human activity event nodes are created based on the structured records obtained in step S15. Each record is a node, and its attributes include event type, occurrence time, and text description of the components involved. Edges are used to define the relationships between nodes; spatial adjacency edges connect building component nodes that are physically in contact or adjacent, reflecting possible force transmission or disease spread paths; environmental exposure edges connect environmental factor nodes with building component nodes that may be directly affected, and the connection relationship is determined based on spatial location and physical exposure logic; historical intervention edges point from human activity event nodes to building component nodes that are clearly involved in their records; factor association edges connect environmental factor nodes that have known physical or statistical associations, or human event nodes that have logical temporal relationships.

[0035] Step S32: To ensure the graph model has a reasonable causal framework, the graph structure is initialized and constrained based on domain knowledge before training; all environmental exposure edges are uniformly set to point from environmental nodes to component nodes; all historical intervention edges are uniformly set to point from event nodes to component nodes; in addition, based on the occurrence time of human events and the possible start time of component damage, temporal logic is applied to filter out obviously unreasonable backward connections; the initial weights of some edges can be pre-assigned based on industry standards, empirical formulas or physical models, for example, the edge weights between the freeze-thaw factor and wall components are initialized based on the material freeze-thaw resistance index.

[0036] Step S33: Model training and causal inference based on graph neural networks; a graph neural network model capable of handling heterogeneous graph relationships, such as a relational graph convolutional network, is adopted; the training objective of the model is to predict the damage state vector of each building component node with the structure of the entire heterogeneous graph and the attributes of all nodes as input; the training data needs to come from multiple building cases with complete data chains; by optimizing the difference between the predicted damage state and the actual damage state, the model learns the damage evolution law jointly encoded by the node attributes and edge relationships in the graph; After the model is trained, causal inference is performed on specific severely damaged components in the building to be analyzed. Using graph interpretation techniques, the contribution of all environmental and human nodes in the graph to the damage state of the target component is calculated. Nodes with contribution values ​​exceeding a set threshold are identified as dominant environmental stress factors and key human activities. By tracing the reverse path from the target component node along the high-contribution incoming edge, one or more influence paths from the driving factors to the damage result can be extracted. The node sequence and edges in the path constitute a specific explanation of the damage mechanism.

[0037] In this embodiment, by elevating the attribution of damage from a list of factors to a path description, the output contains specific causal chains, enhancing the depth and logic of the mechanistic explanation; by calculating the contribution of nodes, the degree of influence of multiple factors is quantitatively ranked; by modeling the damaged system as a structured network that integrates multiple types of entities and multiple relationships, and using graph neural networks to learn and invert the internal mechanisms of this network from the data, the output is a causal hypothesis with clear path support based on the simulation of the interaction of the entire network, rather than a simple statistical correlation index.

[0038] In practical implementation, as an example, in cultural heritage protection projects, traditional methods for selecting intervention measures rely on experience, field experiments, and static calculations. These methods struggle to quantitatively assess the differences in long-term performance of different measures under the combined influence of the building's actual environment and historical issues before implementation. Existing digital technologies primarily focus on the static visual effects after intervention or instantaneous structural verification under specific load conditions, lacking the ability to continuously and dynamically simulate the coupled effects of measures, environment, and time, and thus failing to provide predictive data on the long-term evolution trend of morphological stability. Therefore, this embodiment constructs a parameterized, time-series virtual experimental environment. This virtual environment quantifies intervention measures as modifications to the physical parameters of the building's digital model and drives a physical simulation engine to simulate and calculate the continuous mechanical and physical responses of the modified model over a long period under time-varying load sequences constructed based on real environmental data, thereby predicting its morphological stability characteristics. The specific implementation steps are as follows: Step S41: Establish a mapping library between intervention measures and model parameters; define a digital measure library containing various typical protective intervention measures, such as surface cleaning and reinforcement, structural support, material replacement, surface sealing, and their combinations; for each measure in the measure library, clarify the modification rules for its application to the baseline digital model; the modification rules specifically specify which physical parameters of which components in the model will be changed by the measure; these parameters include material strength, elastic modulus, density, surface permeability coefficient, stiffness and damping characteristics of connection nodes; for example, a measure of "brick surface ethyl silicate permeation reinforcement" is defined as increasing the elastic modulus parameter of the target brick masonry surface to a specific multiple of the original value and reducing its surface capillary water absorption coefficient to a specific percentage; Step S42: Integrate and construct a multiphysics simulation model and environmental load sequence; based on the geometric and material property data of the benchmark digital model, generate a computational mesh model suitable for finite element or discrete element simulation software; this computational mesh model can distinguish different material regions and characterize the connection relationship between components; simultaneously, using the environmental spectrum data obtained in step S1, construct a regional environmental time-varying load sequence representing a future set number of years; the compilation of this sequence is based on the dominant environmental stress factors identified in step S3 for design and reinforcement; for example, if the dominant factors are freeze-thaw cycles and wind-driven rain, the load sequence should include periodic temperature curves and corresponding wind speed and wind direction rainfall intensity curves generated based on historical statistics; this sequence is input in the form of time steps as the boundary conditions of the simulation model; Step S43: Perform multi-scenario simulation calculations; set up multiple test scenarios, each corresponding to one or a set of intervention measures selected from the measure library; for each scenario, modify the corresponding physical parameters of the benchmark calculation grid model according to the mapping rules in the first step to generate a dedicated "post-intervention simulation model" for that scenario; subsequently, load the time-varying environmental load sequence constructed in the second step into the model; start the building physical performance simulation engine that integrates structural mechanics and heat and moisture transfer coupling analysis modules; the engine performs time history analysis to calculate the dynamic response of the "post-intervention simulation model" over the entire set period under continuous environmental loads; extract key response features characterizing morphological stability from the complete simulation results, mainly including: displacement time history curves of key monitoring points and their cumulative deformation at the end of the period; stress time history curves of key stressed components and their peak distribution; damage area evolution diagram and damage factor development curve based on material damage criteria; Step S44: Generate structured virtual test results; For each scenario, the intervention measures, parameter modification details, and deformation accumulation, stress distribution, and damage evolution data output by simulation calculation are structured, correlated, and encapsulated to form the virtual test results for that scenario; The results of all scenarios together constitute a set of virtual test results for strategy comparison.

[0039] In this implementation, by parameterizing the intervention measures onto the physical properties of the benchmark digital model and driving the coupled simulation engine to perform long-term time history analysis under a time-varying load sequence constructed based on real environmental data, it is possible to output quantitative prediction data on deformation accumulation, stress distribution and damage evolution of different intervention schemes within a set number of years in the future. This allows for an objective quantitative comparison of the long-term morphological stability impact of different measures or combinations of measures based on a unified physical simulation benchmark before the intervention is implemented. This enables the selection of the optimal repair strategy that can most effectively resist the dominant environmental stress and avoid triggering secondary long-term risks.

[0040] In a specific implementation, as one example, candidate solutions are first screened based on preset technical standards. Then, a multi-criteria decision analysis model is used to comprehensively calculate and rank the technical performance data of the screened solutions with ethical principles that have been converted into quantifiable indicators, thereby outputting an optimization strategy and generating an executable guidance document. The specific implementation steps are as follows: Step S51: Construct a quantitative evaluation index system, which includes technical performance thresholds and ethical constraint indicators. Technical performance thresholds are set based on the target building's cultural relic protection level, structural type, and current degree of damage, referring to industry standards such as the "Technical Standard for Maintenance and Reinforcement of Ancient Wooden Structures." Specifically, this is manifested as a series of permissible upper limits, such as the upper limit of cumulative deformation at key points at the end of the simulation period, and the upper limit of the ratio of the maximum equivalent stress of components to the allowable stress of materials. Ethical constraints are achieved by defining quantifiable or gradeable evaluation indicators: quantifying the principle of minimum intervention by calculating the intervention impact index based on the volume of material replacement and the scale of new components involved in the intervention; scoring the reversibility level based on the degree of removability or restoration of the measures; evaluating material compatibility by comparing the differences in key physical parameters between new and old materials; and evaluating the integrity of the architectural style by calculating the deviation between the digital model after intervention and the original baseline model at key geometric feature points and the similarity of the visual texture of the main facade. Step S52: Traverse all scenarios in the virtual test result set; for each scenario, extract the key performance data of its simulation output, including the final deformation, peak stress, and newly added damage area; compare the above data with the various technical performance thresholds set in the first step; any scenario whose performance data exceeds its corresponding threshold upper limit is judged as a technically infeasible solution and excluded; scenarios in which all performance data meet the threshold requirements are marked as technically feasible scenarios and enter the comprehensive evaluation stage. Step S53: Construct a decision matrix for all technically feasible scenarios. Rows in the matrix represent each feasible scenario, and columns represent various evaluation criteria, including performance data such as deformation and stress values ​​extracted from virtual experiment results, and ethical quantitative indicators such as the intervention impact index and reversibility score defined in the first step. Standardize the data under each criterion to eliminate the influence of dimensions. Then, use multi-criteria decision analysis methods, such as weighted sum models, to calculate the comprehensive score for each scenario. The weights of each evaluation criterion are predetermined through expert consultation. The allocation of weights should reflect the specific conservation value focus of the building; for example, for national-level protected buildings, the weights of the principles of architectural integrity and reversibility will be correspondingly increased. All technically feasible scenarios are ranked in descending order based on the comprehensive score, and the top-ranked scheme is determined as the recommended optimized intervention and restoration strategy. Step S54: Integrate the recommended optimization strategies and their corresponding detailed measures, virtual test performance prediction data, and scores and ranking results on each evaluation criterion to generate a structured repair strategy guidance document; the repair strategy guidance document also clarifies the specific technical requirements that must be followed in construction to meet relevant ethical constraints, such as specified traditional process procedures, recommended material lists, and special structural details designed to achieve reversibility; Step S55: Conduct on-site construction according to the repair strategy guidance document; during the construction process, use mobile terminals to locate the site and verify the process by referring to the benchmark digital model and the damage map; at the same time, the system records key data in the construction process to form a construction process log, which is used to update the building digital archive.

[0041] In this embodiment, by establishing a quantitative technical threshold and ethical index evaluation system, and using a multi-criteria decision-making model to comprehensively rank the schemes that pass the threshold screening, this method can output a repair strategy from a unified and transparent computational framework that meets the requirements of long-term morphological stability in terms of technical performance, while achieving the optimal balance in core protection ethics such as minimal intervention, reversibility, material compatibility, and morphological integrity. This ensures the reliability of the strategy, improves the efficiency of multi-disciplinary collaborative decision-making and the traceability of decision-making basis, and directly transforms the optimized scheme into a clear technical document to guide construction.

[0042] Based on the same inventive concept as the three-dimensional scanning virtual restoration method for cultural buildings described in the foregoing embodiments, this invention also provides a three-dimensional scanning virtual restoration system for cultural buildings, such as... Figure 2 As shown, the system includes: The architectural recording module is used to acquire the geometric shape information, surface texture information and material spectral information of the target building, and to construct a benchmark digital model based on these; and simultaneously acquires the environmental spectrum data and cultural heritage spectrum data of the region where the target building is located. The damage detection module is used to detect damage on the benchmark digital model and generate a building damage map that characterizes the deformation features of components, the deterioration features of materials, and the loss features of attached decorative information. The analysis module is used to perform multi-source coupled analysis of the damage mechanism of the building damage map, the environmental spectrum data and the cultural heritage spectrum data, and to identify the dominant environmental stress factors and the influence paths of human activities that lead to the damage. The virtual experiment module is used to perform multi-scenario intervention virtual experiments in a preset virtual space based on the dominant environmental stress factor and the influence path of human activities, and to obtain the virtual experiment results under each scenario. The intervention virtual experiment includes applying simulated intervention measures to the benchmark digital model and calculating the morphological stability response characteristics of the benchmark digital model after intervention within a set time span. The strategy determination module is used to determine and execute intervention and repair strategies that meet the preset morphological stability threshold and intervention ethical constraints based on the results of virtual experiments in various scenarios.

[0043] The system described above in this invention can effectively realize the virtual restoration method for three-dimensional scanned cultural buildings, and the technical effects it can achieve are as described in the above embodiments, which will not be repeated here.

[0044] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Thus, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A virtual restoration method for cultural buildings based on 3D scanning, characterized in that, include: Multimodal digital recording of the target building is performed to obtain its geometric shape, surface texture, and material spectral information, and a benchmark digital model is constructed based on this information. Simultaneously, environmental and cultural heritage data of the region where the target building is located are acquired. The baseline digital model is subjected to damage investigation to generate a building damage map; The architectural damage map is used to characterize the deformation features of components, the deterioration features of materials, and the loss of information on attached decorations. Multi-source coupled analysis of the damage mechanism is performed on the building damage map, the environmental spectrum data, and the cultural heritage spectrum data to identify the dominant environmental stress factors and the influence paths of human activities that lead to the damage; Based on the dominant environmental stress factor and the influence path of human activities, multi-scenario virtual intervention experiments are conducted in a preset virtual space to obtain the virtual experiment results under each scenario. The virtual intervention experiment includes applying simulated intervention measures to the baseline digital model and calculating the morphological stability response characteristics of the baseline digital model within a set time span after the intervention. Based on the results of virtual experiments in various scenarios, intervention and repair strategies that meet the preset morphological stability threshold and intervention ethical constraints are determined and implemented.

2. The virtual restoration method for cultural buildings based on three-dimensional scanning according to claim 1, characterized in that, The multimodal digital recording includes: Point clouds of the building's main structure were obtained using a terrestrial 3D laser scanner. Oblique photography using drones was employed to acquire images of the roof and its macroscopic shape. Hyperspectral imaging equipment was used to collect spectral data on the composition and weathering characteristics of the surface material.

3. The virtual restoration method for cultural buildings based on three-dimensional scanning according to claim 1, characterized in that, The baseline digital model is subjected to damage investigation to generate a building damage map, including: The benchmark digital model is deconstructed semantically according to traditional building regulations; For each type of component, geometric distortion, texture damage, and material defects are detected by using curvature analysis of its point cloud data, spectral analysis of its texture image, and feature offset analysis of its spectral data. The various test results are integrated, labeled, and classified according to spatial location to form the building damage map.

4. The virtual restoration method for cultural buildings based on three-dimensional scanning according to claim 1, characterized in that, The environmental spectrum data includes historical climate sequences, microenvironmental temperature and humidity field distribution, and atmospheric pollutant deposition flux data. The data on the cultural heritage includes documents and oral records related to traditional construction techniques, maintenance customs, and functional adaptations.

5. The virtual restoration method for cultural buildings based on three-dimensional scanning according to claim 1, characterized in that, A multi-source coupling analysis of the damage mechanism is performed, and a causal inference model based on graph neural network is adopted. The nodes of the causal inference model represent building components, environmental factors and human activities, and the edges represent the interaction relationship and intensity between them. The dominant environmental stress factor and the influence path of human activities are inferred and output through the causal inference model.

6. The virtual restoration method for cultural buildings based on three-dimensional scanning according to claim 1, characterized in that, Virtual intervention trials in multiple scenarios, including: Define multiple intervention measures; In the simulation environment, each intervention is mapped to a modification of a specific parameter of the baseline digital model; The building physics performance simulation engine is driven to calculate the deformation accumulation, stress distribution and damage evolution of the benchmark digital model after parameter modification within a set number of years after loading a regional environmental load sequence, thereby obtaining the morphological stability response characteristics.

7. The virtual restoration method for cultural buildings based on three-dimensional scanning according to claim 1, characterized in that, The ethical constraints on intervention include the principles of minimal intervention, reversibility, material and process compatibility, and architectural integrity. The morphological stability threshold is a pre-set allowable deformation and stress standard based on the cultural relic protection level, structural type, and current degree of damage of the target building.

8. The virtual restoration method for cultural buildings based on three-dimensional scanning according to claim 6, characterized in that, The specific parameters include material strength, surface permeability, and connection stiffness.

9. The virtual restoration method for cultural buildings based on three-dimensional scanning according to claim 6, characterized in that, The intervention measures include any one or a combination of two or more of the following: cleaning and reinforcement, structural support, material replacement, and surface sealing.

10. A virtual restoration system for cultural buildings based on 3D scanning, characterized in that, include: The architectural recording module is used to acquire the geometric shape, surface texture, and material spectral information of the target building, and to construct a baseline digital model based on these information. Simultaneously, environmental and cultural heritage data of the region where the target building is located are acquired. The damage detection module is used to detect damage on the benchmark digital model and generate a building damage map that characterizes the deformation features of components, the deterioration features of materials, and the loss features of attached decorative information. The analysis module is used to perform multi-source coupled analysis of the damage mechanism of the building damage map, the environmental spectrum data and the cultural heritage spectrum data, and to identify the dominant environmental stress factors and the influence paths of human activities that lead to the damage. The virtual experiment module is used to perform multi-scenario intervention virtual experiments in a preset virtual space based on the dominant environmental stress factor and the influence path of human activities, and to obtain the virtual experiment results under each scenario. The intervention virtual experiment includes applying simulated intervention measures to the benchmark digital model and calculating the morphological stability response characteristics of the benchmark digital model after intervention within a set time span. The strategy determination module is used to determine and execute intervention and repair strategies that meet the preset morphological stability threshold and intervention ethical constraints based on the results of virtual experiments in various scenarios.

Images