Urban historical style three-dimensional digitization translation method and system

Abstract

The present application provides a city historical style three-dimensional digitization translation method and system, adopts a complete closed loop of "collection-recovery-deconstruction-modeling-warehousing-application", combines architectural history, digital modeling, AI technology, CIM technology, realizes the whole chain digitization translation of historical style, and the specific steps are as follows: S1: multi-source heterogeneous data collaborative collection; S2: time and space reference unification and damaged area semantic recovery; S3: historical style five-level element semantic deconstruction and parameterization; S4: multi-LOD hierarchical three-dimensional modeling and style consistency constraint; S5: semantic annotation and historical style digital asset warehousing; S6: multi-scene compliance application output; The translation system is composed of the following modules: module one: multi-source data collection module; Module two: space reference fusion and AI recovery module; Module three: historical style semantic deconstruction and parameterization module; Module four: multi-LOD hierarchical modeling and style checking module; Module five: semantic annotation and digital asset library module.

Description

Technical Field

[0001] This invention relates to the field of digital technology for urban historical features, and more specifically to a method and system for three-dimensional digital translation of urban historical features. Background Technology

[0002] The historical features of a city are the core carriers of its cultural heritage, encompassing diverse elements such as the texture of historical blocks, traditional architectural styles, distinctive component details, and regional cultural symbols. They are a concentrated embodiment of the city's historical memory and regional characteristics. With the acceleration of urbanization, historical features are facing problems such as damage, aging, and inadequate protection. Digital translation has become an important means of protecting, inheriting, and revitalizing historical features. However, the existing methods and systems for digitally translating urban historical features still have the following shortcomings: Data collection is singular and fragmented: Existing technologies use drone oblique photography or ground laser scanning, which lack the dual-source collaborative collection of "spatial geometric data + historical context data". As a result, the digital results can only restore the geometric form of the building, but lack semantic information such as historical age, craftsmanship, and cultural connotation, and cannot achieve the translation of "historical authenticity". The restoration of damaged areas lacks historical constraints: For missing or damaged areas of historical buildings, existing technologies usually use ordinary AI to fill in the gaps or simple manual repairs, which cannot be constrained by the form and component style of the city's historical buildings, and are prone to problems such as "visual rationality but historical distortion". Fragmented deconstruction of historical features: Existing technologies only address the structure of historical features at the level of individual buildings, failing to cover the entire system of elements from macro-blocks to micro-cultural symbols. This makes it impossible to achieve parametric reuse and unified management of historical features, resulting in low modeling efficiency, poor consistency of features, and difficulty in adapting to the digital translation of large-scale historical blocks. Modeling and protection standards are disconnected: Existing hierarchical modeling only focuses on geometric accuracy and does not embed historical style protection standards and architectural form requirements into the modeling process. It lacks real-time style consistency verification and is prone to problems such as modeling results not matching the original historical appearance or failing to meet protection requirements. It cannot be directly used for protection planning and renovation design. Poor reusability of digital achievements: Existing digital achievements are single three-dimensional models, without the establishment of a digital asset library with associated historical semantics, and lack standardized connection with urban information model platforms. This makes it impossible to reuse historical landscape data in the entire chain of urban governance, planning approval, and cultural tourism revitalization, resulting in digital achievements being "built but not used". To address the shortcomings of existing technologies, a novel three-dimensional digital translation method and system for urban historical features is needed to resolve issues such as insufficient historical authenticity, poor consistency of features, and low reusability. Summary of the Invention

[0003] The technical solution adopted by this invention to achieve its technical objective is: a method and system for the three-dimensional digital translation of urban historical features. This method adopts a complete closed loop of "acquisition—restoration—deconstruction—modeling—database entry—application," combining architectural history, digital modeling, AI technology, and CIM technology to achieve the full-chain digital translation of historical features. The specific steps are as follows: S1: Collaborative acquisition of multi-source heterogeneous data; S2: Unification of spatiotemporal reference and semantic restoration of damaged areas; S3: Semantic deconstruction and parameterization of five levels of historical features; S4: Multi-LOD hierarchical 3D modeling and landscape consistency constraints; S5: Semantic annotation and historical digital asset storage; S6: Multi-scenario compliant application output; Based on the above digital translation methods, the corresponding translation system can provide hardware and software support for the implementation of the translation methods, forming a complete digital translation platform. This system consists of the following modules: Module 1: Multi-source data acquisition module; Module 2: Spatiotemporal Reference Fusion and AI Restoration Module; Module 3: Semantic Deconstruction and Parametricization of Historical Features; Module 4: Multi-LOD Layered Modeling and Landscape Verification Module; Module 5: Semantic Annotation and Digital Asset Repository Module; Module Six: Multi-Scenario Application Service Module.

[0004] As a further improvement of the present invention, S1: Multi-source heterogeneous data collaborative acquisition: For the historical blocks and urban areas to be protected, an integrated air-ground acquisition mode of "UAV oblique photography + ground three-dimensional laser scanning + historical document digitization" is adopted to construct a dual-source database of "spatial geometric data + historical context data" to ensure the comprehensiveness and accuracy of the data; Drone oblique photography: Using a multi-rotor drone equipped with a high-definition camera and GPS / IMU positioning module, aerial photography is carried out at a scale of 1:500 to obtain the overall real-world 3D point cloud (point cloud density ≥100 points / square centimeter) and building facade texture images of the historic district, covering the entire historic district area; Ground-based 3D laser scanning: Using a phase-type laser scanner (scanning accuracy ≤ ±2mm), millimeter-level scanning is performed on key historical buildings and distinctive components (such as brackets, brick carvings, and plaques) to obtain fine point cloud data, making up for the lack of accuracy of UAV aerial photography in microscopic details; Digitization of historical documents: Through scanning, OCR recognition, manual annotation and other methods, historical maps, old photos, building repair archives, local chronicles, architectural style historical materials, etc. are digitized to extract historical context data such as historical period, building technology, material color, cultural background and establish a structured document database.

[0005] As a further improvement of the present invention, S2: Spatiotemporal reference unification and semantic restoration of damaged areas: First, the multi-source data such as oblique photography point clouds, ground laser scanning point clouds, historical images, and topographic maps collected by the UAV in S1 are unified to the same CIM / GIS spatiotemporal coordinate system (such as the 2000 National Geodetic Coordinate System) through a coordinate registration algorithm (such as the ICP algorithm), and a timestamp is assigned to each type of data to construct a spatiotemporal base model containing spatial and temporal information; Secondly, for areas in the spatiotemporal base model that have missing building components, damaged texture, or faded color, semantic restoration is carried out using the method of "historical landscape map library retrieval + generative AI constraint restoration" to ensure that the restoration results are consistent with historical authenticity. Historical Landscape Atlas Retrieval: Search the historical landscape atlas database constructed for the target city to obtain form samples, component samples, and material and color samples that are from the same period, region, and building type as the damaged area; Generative AI-constrained restoration: Using retrieved historical samples as constraints, the geometric structure and texture information of the damaged area are generated by a generative AI model (such as a GAN model). During the generation process, a historical map discriminator performs real-time verification to ensure that the similarity between the generated result and the historical sample is ≥95%. The output restoration model conforms to historical authenticity and is integrated into the spatiotemporal basis model.

[0006] As a further improvement of the present invention, S3: semantic deconstruction and parameterization of five-level elements of historical features: based on architectural history theory and urban historical features protection standards, the urban historical features are deconstructed from top to bottom into five levels of parameterizable elements, so as to realize the standardization, reusability and editability of features elements; Level 1: Street fabric elements, parameters include road network pattern (road network density, street width), street scale (street spacing, cross-sectional dimensions), courtyard combination (courtyard layout, courtyard area), and skyline outline (building height, outline undulation). Level 2: Architectural form elements, parameters include structural type (timber structure, brick and stone structure, brick and wood mixed structure), bay width and depth (number of bays, depth dimensions), roof type (hipped roof, gable roof, overhanging roof), facade proportions (proportion of doors and windows to walls, eaves height). Level 3: Component details, including parameters such as door and window styles (lattice doors, decorative windows, door studs), bracket system (number of bracket layers, bracket spacing), column type (column diameter, column height, column base style), and carvings (carving patterns, dimensions, and locations). Level 4: Material and color elements, including parameters such as brick and stone material (blue bricks, gray tiles, stone types), pigment color spectrum (traditional pigment color values, fading coefficients), and aging process (degree of weathering, wear marks). Level 5: Cultural symbol elements, parameters include regional totems (regional characteristic patterns), plaques (font, content, size), couplets (text content, calligraphy style), and stone carving patterns (pattern style, carving technique). For each level of elements, a reusable, editable, and verifiable parametric component library is established. The component library is bidirectionally linked to the historical landscape map library, supporting parameter adjustment and dynamic updates.

[0007] As a further improvement of the present invention, S4: multi-LOD hierarchical 3D modeling and landscape consistency constraints: adopting a four-level hierarchical modeling method from LOD1 to LOD4, combined with a parametric component library and landscape consistency verification rules, to achieve hierarchical, high-precision, and highly consistent 3D modeling; LOD1 (street level): Based on the spatiotemporal base model, construct the overall outline and texture of the historical urban area / street, focusing on restoring the road network pattern, courtyard distribution and skyline outline to meet the needs of macro planning display. LOD2 (Architectural Level): Based on LOD1, it adds the external form of historical buildings and street space blocks, restores the overall form, width and depth of buildings, and roof form, and meets the display and analysis needs at the street scale. LOD3 (Component Level): Based on LOD2, it assembles components such as doors, windows, brackets, columns, and railings from the parametric component library to restore the detailed features of the building facade and meet the basic requirements of building renovation design. LOD4 (Cultural Gene Level): Based on LOD3, it adds micro-details such as interior space layout, painting, brick carving, and woodwork to restore the specific form and craftsmanship of cultural symbols and meet the needs of intangible cultural heritage inheritance and refined protection. During the modeling process, a style consistency verification rule is embedded to compare the modeling results with the parametric component library and the historical style map library in real time. The verification indicators include the form matching degree (≥95%), component proportion tolerance (≤±5%), material color difference range (ΔE≤3), and cultural symbol compliance. When the verification fails, an automatic prompt is given and a correction scheme that conforms to the historical map is recommended to ensure that the modeling results are "restored to their original state and rebuilt to their original state".

[0008] As a further improvement of the present invention, S5: Semantic annotation and historical appearance digital asset storage: For the three-dimensional models at LOD3 and LOD4 levels, multi-dimensional semantic tags are associated with each component. The tag content includes: age information (building construction year, renovation year), craft information (construction process, renovation process), material information (component material, pigment type), historical events (historical events related to the component / building), intangible cultural heritage information (associated intangible cultural heritage projects, inheritors), and protection level (building protection level, component protection requirements). Based on semantic annotation results, a searchable, parsable, traceable, and inheritable digital asset library of historical features is constructed. The asset library supports multi-dimensional retrieval by era, area, building type, component type, protection level, etc. At the same time, it opens standard interfaces such as CityGML, IFC, and BIMcollab to connect to the City Information Modeling (CIM) platform to achieve standardized reuse of data.

[0009] As a further improvement of the present invention, S6: Multi-scenario compliant application output: Based on the historical landscape digital asset library and 3D model, it supports multi-scenario compliant application output. All output results retain semantic traceability information and historical authenticity proof, and are adapted to different application needs. Historical Appearance Restoration and Display: Through a 3D visualization platform, the original appearance and restoration process of historical blocks / buildings are displayed, supporting zooming, panning, and detailed viewing; Repair design simulation: Based on the LOD4 model, simulate and verify the rationality and compliance of the repair plan. Urban renewal control plan comparison: Overlaying the 3D model with the urban renewal control plan scheme, analyzing the impact of the control plan scheme on the historical appearance, and assisting in planning approval; VR / AR immersive experience: Through VR / AR devices, immersive tours and virtual interactions of historical districts are realized, revitalizing historical culture; Digital Guide for Culture, Commerce and Tourism: Develop a digital guide system that combines semantic tags to provide tourists with services such as historical and cultural explanations and route navigation; Protection planning approval assistance: Provides precise digital support for the preparation and approval of historical and cultural preservation plans, improving approval efficiency and scientific rigor.

[0010] As a further improvement of the present invention, the spatiotemporal reference unification and AI image completion + historical map comparison in the semantic restoration of the damaged area in S2 include: Construct a historical landscape map database of the target city, including architectural styles, component styles, and color systems from different eras; For the damaged areas, generative AI is used to pre-generate the structure and texture, and then similarity retrieval and style calibration are performed with the historical map library to ensure that the restoration results are consistent with historical authenticity.

[0011] As a further improvement of the present invention, the parameterization of the five-level elements in the semantic deconstruction and parameterization of the historical features of S3 includes: Street fabric: road network pattern, street and alley scale, courtyard combination, skyline outline parameters; Architectural form: structural type, bay width and depth, roof type, eaves height, facade proportion parameters; Component details: door and window styles, bracket set type, column type, railings, and carving parameters; Material and color: Brick, tile and stone materials, pigment color spectrum, and antiquing process parameters; Cultural symbols: regional totems, plaques, couplets, and stone carving patterns.

[0012] As a further improvement of the present invention, module one: multi-source data acquisition module: includes a UAV aerial photography unit, a ground laser scanning unit, and a historical document digitization unit, which work together to acquire spatial geometric data and historical contextual data. UAV aerial photography unit: Composed of multi-rotor UAV, high-definition camera, GPS / IMU positioning module, it supports automatic aerial photography, flight path planning, and outputs overall 3D point cloud and texture image of historical blocks; Ground laser scanning unit: Composed of a phase laser scanner and a data transmission module, it supports millimeter-level scanning and outputs fine point cloud data of key buildings and distinctive components; The historical document digitization unit consists of a scanner, an OCR recognition module, and a manual annotation module. It supports the digitization and structured storage of historical documents and outputs a historical context database. Module Two: Spatiotemporal Reference Fusion and AI Restoration Module: Includes coordinate registration unit, point cloud denoising and refinement unit, historical image alignment unit, historical landscape map retrieval unit, and generative AI constraint restoration unit, realizing multi-source data fusion and damaged area restoration. Coordinate registration unit: Using the ICP algorithm, multi-source data are unified into the same CIM / GIS spatiotemporal coordinate system, timestamps are assigned, and a spatiotemporal base model is generated; Point cloud denoising is simplified and refined: Gaussian filtering algorithm is used to remove noise and redundant points in point cloud data, thereby improving the accuracy of point cloud; Historical image alignment unit: Aligns historical photos and maps with the current spatiotemporal base model to achieve spatiotemporal correlation; Historical Landscape Map Retrieval Unit: It uses a graph database to store historical landscape map samples, integrates a similarity retrieval algorithm, and supports multi-dimensional and accurate retrieval by era, region, building type, component type, material and color; Generative AI Constraint Restoration Unit: Integrates GAN generator and historical map discriminator. GAN generator is used to generate geometry and texture of damaged areas. Historical map discriminator uses historical map samples as ground values ​​to score and filter the historical authenticity of the generated results, and only outputs restoration models with similarity ≥95%. Module 3: Historical Feature Semantic Deconstruction and Parametricization Module: This module includes a five-level element classification unit, an architectural form library unit, a component parametricization unit, and a cultural symbol library unit, enabling the deconstruction and parametricization of historical feature elements. Five-level element classification unit: Based on architectural history and protection standards, the historical features are deconstructed into five levels of elements: street texture, building form, component details, material color, and cultural symbols, so as to achieve standardization of element classification; Architectural Form Library Unit: Stores the form parameters of historical buildings of different eras and types, is associated with the historical landscape map library, and supports dynamic updates; Parametric Component Units: Parametric modeling of various building components to build a reusable and editable parametric component library, supporting component drag-and-drop reuse and adaptive parameter adjustment; Cultural Symbols Database Unit: Stores cultural symbols samples such as regional totems, plaques, couplets, and stone carvings of the target city, associated with cultural connotation information, and supports retrieval and reuse; Module 4: Multi-LOD Layered Modeling and Feature Verification Module: Includes LOD1 to LOD4 modeling units, a feature consistency rule base unit, and a real-time verification and correction unit, realizing layered modeling and feature consistency control. LOD1~LOD4 modeling units: support four levels of layered modeling, corresponding to street level, building level, component level and cultural gene level modeling respectively, and integrate parametric component assembly function to improve modeling efficiency; Style Consistency Rule Base Unit: Stores verification rules for form matching degree, component proportion tolerance, material color difference range, cultural symbol compliance, etc., and supports custom adjustment of rules; Real-time verification and correction unit: During the modeling process, the verification rules are called in real time to verify the modeling results. If the verification fails, the unit will automatically prompt and recommend a correction plan. Module 5: Semantic Annotation and Digital Asset Library Module: Includes component semantic association unit, multi-dimensional tag library unit, CIM standard interface unit, and version management unit, realizing semantic annotation and digital asset management. Component semantic association unit: Supports associating multi-dimensional semantic tags with 3D model components, realizing the binding of components with historical information; Multi-dimensional tag library unit: stores tags such as era, craftsmanship, materials, historical events, intangible cultural heritage information, protection level, etc., and supports the addition of custom tags; CIM Standard Interface Unit: Supports international / domestic standards such as CityGML, IFC, and BIMcollab, enabling lossless data exchange with urban information modeling platforms, natural resource spatiotemporal platforms, and historical and cultural preservation platforms; Version Management Unit: Manages the versions of models, components, and tags in the digital asset library, supports version rollback, updates, and deletions, and ensures data traceability; Module Six: Multi-Scenario Application Service Module: Includes 3D visualization, restoration simulation, planning comparison, VR / AR interaction, and cultural, commercial, and tourism application units, enabling multi-scenario application output. 3D visualization unit: Supports real-time rendering, scaling, panning, and detailed viewing of 3D models, and outputs visualized display results; Repair Simulation Unit: Supports the simulation and modeling of building repair plans, analyzes the rationality of repair plans, and outputs simulation reports; Planning Comparison Unit: Supports the overlay and comparison of 3D models and control planning schemes, outputs comparison analysis reports, and assists in planning approval; VR / AR Interaction Unit: Provides VR / AR immersive experience interface, supports virtual roaming and interactive operation, and outputs VR / AR application results; Cultural, commercial and tourism application unit: Develop functions such as digital guides and cultural displays, output cultural, commercial and tourism application products, and revitalize historical features.

[0013] As a further improvement of the present invention, the AI ​​restoration unit integrates a generative adversarial network (GAN) and a historical landscape map retrieval engine to intelligently restore missing structures, blurred textures, and damaged components under the constraint of historical authenticity.

[0014] As a further improvement of the present invention, the parametric component assembly unit supports drag-and-drop component reuse, adaptive shape adjustment, and batch style consistency verification, which significantly improves modeling efficiency and ensures historical style uniformity.

[0015] As a further improvement of the present invention, the CIM data interface unit supports data interoperability with the city information model platform, the natural resource spatiotemporal platform, and the historical and cultural protection platform, enabling the reuse of historical landscape data throughout the entire chain of urban governance.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention can improve the authenticity of history: It adopts "historical landscape map retrieval + generative AI constraint restoration" to solve the pain point of "historical distortion" in traditional AI restoration. The restoration accuracy rate is ≥95%. It strictly follows the protection principle of "restoring the old as it was" to ensure the historical authenticity of digital translation, which is different from existing ordinary AI restoration technology.

[0017] 2. The present invention provides a more systematic approach to the management of landscape elements: It features a unique five-level semantic deconstruction system that covers the entire chain from macro-level street blocks to micro-level cultural genes. Combined with a parametric component library, it achieves standardization, reusability, and editability of landscape elements, improving modeling efficiency by ≥60% and component reuse rate by ≥80%, thus solving the problems of fragmented elements and low modeling efficiency in existing technologies.

[0018] 3. The present invention deeply integrates modeling and protection standards: the style consistency verification rules are embedded into the LOD layered modeling process to achieve "modeling is verification, output is compliance", ensuring that the modeling results comply with the historical style protection standards and can be directly used for repair design and planning approval, thus solving the problem of the disconnect between existing modeling and protection standards.

[0019] 4. The digital results of this invention have strong reusability: a digital asset library with associated multi-dimensional semantic tags is constructed, which supports multi-dimensional retrieval and standardized docking with the CIM platform, realizing the reuse of historical landscape data in the whole chain of urban governance, cultural tourism revitalization, and protection planning, and solving the pain point of existing results being "built but not used".

[0020] 5. This invention adopts a complete closed loop of "dual-source data acquisition + historical constraint-based AI restoration + five-level element parameterization + LOD hierarchical verification + CIM native integration", which is different from the simple superposition of existing single devices or single algorithms. Attached Figure Description

[0021] Figure 1 A flowchart illustrating the three-dimensional digital translation method for urban historical features.

[0022] Figure 2 A schematic diagram of the modules in a three-dimensional digital translation system for urban historical features.

[0023] Figure 3 A schematic diagram of the closed-loop process of a three-dimensional digital translation method for urban historical features. Detailed Implementation

[0024] The present invention will be further described below with reference to the accompanying drawings: Example: Figures 1 to 2 As shown: This invention provides a method and system for the three-dimensional digital translation of urban historical features. This method employs a complete closed loop of "acquisition—reconstruction—deconstruction—modeling—database entry—application," combining architectural history, digital modeling, AI technology, and CIM technology to achieve a full-chain digital translation of historical features. The specific steps are as follows: S1: Collaborative acquisition of multi-source heterogeneous data; S2: Unification of spatiotemporal reference and semantic restoration of damaged areas; S3: Semantic deconstruction and parameterization of five levels of historical features; S4: Multi-LOD hierarchical 3D modeling and landscape consistency constraints; S5: Semantic annotation and historical digital asset storage; S6: Multi-scenario compliant application output; Based on the above digital translation methods, the corresponding translation system can provide hardware and software support for the implementation of the translation methods, forming a complete digital translation platform. This system consists of the following modules: Module 1: Multi-source data acquisition module; Module 2: Spatiotemporal Reference Fusion and AI Restoration Module; Module 3: Semantic Deconstruction and Parametricization of Historical Features; Module 4: Multi-LOD Layered Modeling and Landscape Verification Module; Module 5: Semantic Annotation and Digital Asset Repository Module; Module Six: Multi-Scenario Application Service Module.

[0025] Among them, S1: Multi-source heterogeneous data collaborative acquisition: For the historical blocks and urban areas to be protected, an integrated air-ground acquisition mode of "UAV oblique photography + ground three-dimensional laser scanning + historical document digitization" is adopted to construct a dual-source database of "spatial geometric data + historical context data" to ensure the comprehensiveness and accuracy of the data; Drone oblique photography: Using a multi-rotor drone equipped with a high-definition camera and GPS / IMU positioning module, aerial photography is carried out at a scale of 1:500 to obtain the overall real-world 3D point cloud (point cloud density ≥100 points / square centimeter) and building facade texture images of the historic district, covering the entire historic district area; Ground-based 3D laser scanning: Using a phase-type laser scanner (scanning accuracy ≤ ±2mm), millimeter-level scanning is performed on key historical buildings and distinctive components (such as brackets, brick carvings, and plaques) to obtain fine point cloud data, making up for the lack of accuracy of UAV aerial photography in microscopic details; Digitization of historical documents: Through scanning, OCR recognition, manual annotation and other methods, historical maps, old photos, building repair archives, local chronicles, architectural style historical materials, etc. are digitized to extract historical context data such as historical period, building technology, material color, cultural background and establish a structured document database.

[0026] Among them, S2: Spatiotemporal benchmark unification and semantic restoration of damaged areas: First, the multi-source data such as oblique photogrammetric point cloud, ground laser scan point cloud, historical images, and topographic maps collected by the UAV in S1 are unified to the same CIM / GIS spatiotemporal coordinate system (such as the 2000 National Geodetic Coordinate System) through coordinate registration algorithms (such as the ICP algorithm), and a timestamp is assigned to each type of data to construct a spatiotemporal base model containing spatial and temporal information; Secondly, for areas in the spatiotemporal base model that have missing building components, damaged texture, or faded color, semantic restoration is carried out using the method of "historical landscape map library retrieval + generative AI constraint restoration" to ensure that the restoration results are consistent with historical authenticity. Historical Landscape Atlas Retrieval: Search the historical landscape atlas database constructed for the target city to obtain form samples, component samples, and material and color samples that are from the same period, region, and building type as the damaged area; Generative AI-constrained restoration: Using retrieved historical samples as constraints, the geometric structure and texture information of the damaged area are generated by a generative AI model (such as a GAN model). During the generation process, a historical map discriminator performs real-time verification to ensure that the similarity between the generated result and the historical sample is ≥95%. The output restoration model conforms to historical authenticity and is integrated into the spatiotemporal basis model.

[0027] Among them, S3: Semantic deconstruction and parameterization of five levels of historical features: Based on architectural history theory and urban historical features protection standards, the urban historical features are deconstructed from top to bottom into five levels of parameterizable elements, so as to realize the standardization, reusability and editability of features elements; Level 1: Street fabric elements, parameters include road network pattern (road network density, street width), street scale (street spacing, cross-sectional dimensions), courtyard combination (courtyard layout, courtyard area), and skyline outline (building height, outline undulation). Level 2: Architectural form elements, parameters include structural type (timber structure, brick and stone structure, brick and wood mixed structure), bay width and depth (number of bays, depth dimensions), roof type (hipped roof, gable roof, overhanging roof), facade proportions (proportion of doors and windows to walls, eaves height). Level 3: Component details, including parameters such as door and window styles (lattice doors, decorative windows, door studs), bracket system (number of bracket layers, bracket spacing), column type (column diameter, column height, column base style), and carvings (carving patterns, dimensions, and locations). Level 4: Material and color elements, including parameters such as brick and stone material (blue bricks, gray tiles, stone types), pigment color spectrum (traditional pigment color values, fading coefficients), and aging process (degree of weathering, wear marks). Level 5: Cultural symbol elements, parameters include regional totems (regional characteristic patterns), plaques (font, content, size), couplets (text content, calligraphy style), and stone carving patterns (pattern style, carving technique). For each level of elements, a reusable, editable, and verifiable parametric component library is established. The component library is bidirectionally linked to the historical landscape map library, supporting parameter adjustment and dynamic updates.

[0028] Among them, S4: Multi-LOD hierarchical 3D modeling and appearance consistency constraints: adopting a four-level hierarchical modeling method from LOD1 to LOD4, combined with a parametric component library and appearance consistency verification rules, to achieve hierarchical, high-precision, and highly consistent 3D modeling; LOD1 (street level): Based on the spatiotemporal base model, construct the overall outline and texture of the historical urban area / street, focusing on restoring the road network pattern, courtyard distribution and skyline outline to meet the needs of macro planning display. LOD2 (Architectural Level): Based on LOD1, it adds the external form of historical buildings and street space blocks, restores the overall form, width and depth of buildings, and roof form, and meets the display and analysis needs at the street scale. LOD3 (Component Level): Based on LOD2, it assembles components such as doors, windows, brackets, columns, and railings from the parametric component library to restore the detailed features of the building facade and meet the basic requirements of building renovation design. LOD4 (Cultural Gene Level): Based on LOD3, it adds micro-details such as interior space layout, painting, brick carving, and woodwork to restore the specific form and craftsmanship of cultural symbols and meet the needs of intangible cultural heritage inheritance and refined protection. During the modeling process, a style consistency verification rule is embedded to compare the modeling results with the parametric component library and the historical style map library in real time. The verification indicators include the form matching degree (≥95%), component proportion tolerance (≤±5%), material color difference range (ΔE≤3), and cultural symbol compliance. When the verification fails, an automatic prompt is given and a correction scheme that conforms to the historical map is recommended to ensure that the modeling results are "restored to their original state and rebuilt to their original state".

[0029] S5: Semantic annotation and digital asset storage of historical features: For the 3D models at LOD3 and LOD4 levels, multi-dimensional semantic tags are associated with each component. The tag content includes: age information (building construction year, renovation year), craft information (construction technique, renovation technique), material information (component material, pigment type), historical events (historical events related to the component / building), intangible cultural heritage information (related intangible cultural heritage projects, inheritors), and protection level (building protection level, component protection requirements). Based on semantic annotation results, a searchable, parsable, traceable, and inheritable digital asset library of historical features is constructed. The asset library supports multi-dimensional retrieval by era, area, building type, component type, protection level, etc. At the same time, it opens standard interfaces such as CityGML, IFC, and BIMcollab to connect to the City Information Modeling (CIM) platform to achieve standardized reuse of data.

[0030] Among them, S6: Multi-scenario compliant application output: Based on the historical landscape digital asset library and 3D model, it supports multi-scenario compliant application output. All output results retain semantic traceability information and historical authenticity proof, adapting to different application needs. Historical Appearance Restoration and Display: Through a 3D visualization platform, the original appearance and restoration process of historical blocks / buildings are displayed, supporting zooming, panning, and detailed viewing; Repair design simulation: Based on the LOD4 model, simulate and verify the rationality and compliance of the repair plan. Urban renewal control plan comparison: Overlaying the 3D model with the urban renewal control plan scheme, analyzing the impact of the control plan scheme on the historical appearance, and assisting in planning approval; VR / AR immersive experience: Through VR / AR devices, immersive tours and virtual interactions of historical districts are realized, revitalizing historical culture; Digital Guide for Culture, Commerce and Tourism: Develop a digital guide system that combines semantic tags to provide tourists with services such as historical and cultural explanations and route navigation; Protection planning approval assistance: Provides precise digital support for the preparation and approval of historical and cultural preservation plans, improving approval efficiency and scientific rigor.

[0031] Among them, the spatiotemporal benchmark unification of S2 and the AI ​​image completion + historical map comparison in the semantic restoration of the damaged area include: Construct a historical landscape map database of the target city, including architectural styles, component designs, and color schemes from different eras; For the damaged areas, generative AI is used to pre-generate the structure and texture, and then similarity retrieval and style calibration are performed with the historical map library to ensure that the restoration results are consistent with historical authenticity.

[0032] Among them, the five-level element parameterization in the semantic deconstruction and parameterization of the historical features of S3 includes: Street fabric: road network pattern, street and alley scale, courtyard combination, skyline outline parameters; Architectural form: structural type, bay width and depth, roof type, eaves height, facade proportion parameters; Component details: door and window styles, bracket set type, column type, railings, and carving parameters; Material and color: Brick, tile and stone materials, pigment color spectrum, and antiquing process parameters; Cultural symbols: regional totems, plaques, couplets, and stone carving patterns.

[0033] Module 1, the multi-source data acquisition module, comprises an UAV aerial photography unit, a ground laser scanning unit, and a historical document digitization unit. These three components work collaboratively to acquire spatial geometric data and historical contextual data. UAV aerial photography unit: Composed of multi-rotor UAV, high-definition camera, GPS / IMU positioning module, it supports automatic aerial photography, flight path planning, and outputs overall 3D point cloud and texture image of historical blocks; Ground laser scanning unit: Composed of a phase laser scanner and a data transmission module, it supports millimeter-level scanning and outputs fine point cloud data of key buildings and distinctive components; The historical document digitization unit consists of a scanner, an OCR recognition module, and a manual annotation module. It supports the digitization and structured storage of historical documents and outputs a historical context database. Module Two: Spatiotemporal Reference Fusion and AI Restoration Module: Includes coordinate registration unit, point cloud denoising and refinement unit, historical image alignment unit, historical landscape map retrieval unit, and generative AI constraint restoration unit, realizing multi-source data fusion and damaged area restoration. Coordinate registration unit: Using the ICP algorithm, multi-source data are unified into the same CIM / GIS spatiotemporal coordinate system, timestamps are assigned, and a spatiotemporal base model is generated; Point cloud denoising is simplified and refined: Gaussian filtering algorithm is used to remove noise and redundant points in point cloud data, thereby improving the accuracy of point cloud; Historical image alignment unit: Aligns historical photos and maps with the current spatiotemporal base model to achieve spatiotemporal correlation; Historical Landscape Map Retrieval Unit: It uses a graph database to store historical landscape map samples, integrates a similarity retrieval algorithm, and supports multi-dimensional and accurate retrieval by era, region, building type, component type, material and color; Generative AI Constraint Restoration Unit: Integrates GAN generator and historical map discriminator. GAN generator is used to generate geometry and texture of damaged areas. Historical map discriminator uses historical map samples as ground values ​​to score and filter the historical authenticity of the generated results, and only outputs restoration models with similarity ≥95%. Module 3: Historical Feature Semantic Deconstruction and Parametricization Module: This module includes a five-level element classification unit, an architectural form library unit, a component parametricization unit, and a cultural symbol library unit, enabling the deconstruction and parametricization of historical feature elements. Five-level element classification unit: Based on architectural history and protection standards, the historical features are deconstructed into five levels of elements: street texture, building form, component details, material color, and cultural symbols, so as to achieve standardization of element classification; Architectural Form Library Unit: Stores the form parameters of historical buildings of different eras and types, is associated with the historical landscape map library, and supports dynamic updates; Parametric Component Units: Parametric modeling of various building components to build a reusable and editable parametric component library, supporting component drag-and-drop reuse and adaptive parameter adjustment; Cultural Symbols Database Unit: Stores cultural symbols samples such as regional totems, plaques, couplets, and stone carvings of the target city, associated with cultural connotation information, and supports retrieval and reuse; Module 4: Multi-LOD Layered Modeling and Feature Verification Module: Includes LOD1 to LOD4 modeling units, a feature consistency rule base unit, and a real-time verification and correction unit, realizing layered modeling and feature consistency control. LOD1~LOD4 modeling units: support four levels of layered modeling, corresponding to street level, building level, component level and cultural gene level modeling respectively, and integrate parametric component assembly function to improve modeling efficiency; Style Consistency Rule Base Unit: Stores verification rules for form matching degree, component proportion tolerance, material color difference range, cultural symbol compliance, etc., and supports custom adjustment of rules; Real-time verification and correction unit: During the modeling process, the verification rules are called in real time to verify the modeling results. If the verification fails, the unit will automatically prompt and recommend a correction plan. Module 5: Semantic Annotation and Digital Asset Library Module: Includes component semantic association unit, multi-dimensional tag library unit, CIM standard interface unit, and version management unit, realizing semantic annotation and digital asset management. Component semantic association unit: Supports associating multi-dimensional semantic tags with 3D model components, realizing the binding of components with historical information; Multi-dimensional tag library unit: stores tags such as era, craftsmanship, materials, historical events, intangible cultural heritage information, protection level, etc., and supports the addition of custom tags; CIM Standard Interface Unit: Supports international / domestic standards such as CityGML, IFC, and BIMcollab, enabling lossless data exchange with urban information modeling platforms, natural resource spatiotemporal platforms, and historical and cultural preservation platforms; Version Management Unit: Manages the versions of models, components, and tags in the digital asset library, supports version rollback, updates, and deletions, and ensures data traceability; Module Six: Multi-Scenario Application Service Module: Includes 3D visualization, restoration simulation, planning comparison, VR / AR interaction, and cultural, commercial, and tourism application units, enabling multi-scenario application output. 3D visualization unit: Supports real-time rendering, scaling, panning, and detailed viewing of 3D models, and outputs visualized display results; Repair Simulation Unit: Supports the simulation and modeling of building repair plans, analyzes the rationality of repair plans, and outputs simulation reports; Planning Comparison Unit: Supports the overlay and comparison of 3D models and control planning schemes, outputs comparison analysis reports, and assists in planning approval; VR / AR Interaction Unit: Provides VR / AR immersive experience interface, supports virtual roaming and interactive operation, and outputs VR / AR application results; Cultural, commercial and tourism application unit: Develop functions such as digital guides and cultural displays, output cultural, commercial and tourism application products, and revitalize historical features.

[0034] The AI ​​restoration unit integrates a generative adversarial network (GAN) and a historical landscape map retrieval engine to intelligently restore missing structures, blurred textures, and damaged components under the constraint of historical authenticity.

[0035] The parametric component assembly unit supports drag-and-drop component reuse, adaptive shape adjustment, and batch style consistency verification, which significantly improves modeling efficiency and ensures historical style consistency.

[0036] The CIM data interface unit supports data exchange with urban information modeling platforms, natural resource spatiotemporal platforms, and historical and cultural protection platforms, enabling the reuse of historical landscape data throughout the entire urban governance chain.

[0037] The specific functions and operation procedures of this embodiment are as follows: The three-dimensional digital translation method of the present invention employs three-dimensional digital translation, and the specific steps are as follows: Collaborative acquisition of multi-source heterogeneous data: Drone oblique photography: A DJI Phantom 4 RTK drone, equipped with a 20-megapixel high-definition camera, was used to conduct aerial photography at a scale of 1:500 and a flight altitude of 100m, covering an area of ​​2.4 square kilometers in Pingjiang Historical District, and acquiring overall 3D point cloud (point cloud density of 120 points / square centimeter) and building facade texture images of the district. Ground 3D laser scanning: Using the Faro Focus S70 laser scanner, millimeter-level scanning was performed on 20 key buildings, including Couple Garden, Pingjiang Road Ancient Bridge, and key ancient residences, with a scanning accuracy of ±1.5mm, to obtain fine point cloud data of building components, brick carvings, brackets, etc. Digitization of historical documents: More than 300 documents were collected, including Suzhou Prefecture Annals, Pingjiang Historical District renovation archives, old photos from the Republic of China era, and historical materials on the form of traditional Suzhou buildings. Through scanning, OCR recognition, and manual annotation, information such as building age, construction techniques, materials and colors were extracted to establish a structured historical context database.

[0038] Spatiotemporal benchmark unification and semantic restoration of damaged areas: Unified spatiotemporal reference: Using the ICP algorithm, the UAV point cloud, laser point cloud, historical map, and current topographic map are unified to the 2000 National Geodetic Coordinate System, and a timestamp is assigned to each type of data (e.g., old photos from the Republic of China are labeled 1930, and current scanned data is labeled 2024), to construct a spatiotemporal base model of Pingjiang historical district; Damaged Area Restoration: For a damaged eaves corner in Couple's Garden and three missing brick carvings on Pingjiang Road, we searched the Suzhou Ancient City Historical Landscape Atlas Database (which contains more than 5,000 sets of samples of the form and components of Suzhou residences and gardens from the Ming and Qing Dynasties to the Republic of China period) to obtain eaves corner and brick carving samples of the same period and type. We generated the geometric structure and texture of the damaged area using a GAN generator, and verified it by a historical atlas discriminator to ensure that the similarity is ≥95%. We then integrated the restoration model into the spatiotemporal base model to achieve historical authenticity restoration.

[0039] Semantic deconstruction and parameterization of five levels of historical features: Based on Suzhou's traditional architectural history and ancient city protection standards, the appearance of Pingjiang historical district is deconstructed into five levels of elements, and a parametric component library is established: Urban fabric elements: Road network density 800m / km 2 The streets and alleys are 2-4m wide, the courtyards are laid out in the typical Jiangnan style, and the skyline height is controlled at 6-9m. Architectural elements: mainly brick-wood mixed structure, with 2-4 bays, 3-5m depth, and roof forms mainly gable roof and hip roof. The ratio of doors and windows to walls on the facade is 1:3. Component details: The doors and windows adopt the traditional Suzhou lattice doors and flower windows, the brackets are the typical doukou tiao shape of the Jiangnan region, the column diameter is 20-30cm, the column height is 3-4m, and the brick carving patterns are mainly flowers and auspicious patterns; Material and color elements: The walls are made of blue bricks (RGB values: 100, 100, 100), the roof is made of gray tiles (RGB values: 80, 80, 80), the pigments are traditional mineral pigments, and the aging process simulates natural weathering. Cultural symbol elements: the regional totem is Suzhou embroidery pattern, the font of the plaque is mainly regular script and clerical script, the couplets are mostly inscribed by literati in Jiangnan, and the stone carving patterns include auspicious patterns such as auspicious clouds and bats; A parametric component library containing over 1,000 components has been built, supporting component drag-and-drop reuse, adaptive parameter adjustment, and bidirectional association with the historical landscape map library.

[0040] Multi-LOD hierarchical 3D modeling and landscape consistency constraints: LOD1: Construct a white model of the overall outline and texture of the Pingjiang historical district, restore the road network pattern, courtyard distribution and skyline outline, and meet the needs of macro-planning and display of the ancient city. LOD2: Based on LOD1, more than 200 historical buildings and street space blocks are added to restore the overall shape, width and depth of the buildings and the form of the roof. LOD3: Assemble components such as doors, windows, brackets, columns, and brick carvings from the parametric component library to restore the details of the building facade and meet the basic requirements of renovation design; LOD4: Add interior space layout, Suzhou embroidery and painting, woodwork details, etc., to restore the specific form and craftsmanship of cultural symbols; During the modeling process, a style consistency verification rule is embedded, and the modeling results are compared with the component library and historical map library in real time. The modeling results are automatically corrected for 12 component proportion deviations and 8 color deviations to ensure that the form matching degree is ≥95%, the component proportion tolerance is ≤±5%, and the color difference ΔE is ≤3.

[0041] Semantic annotation and historical feature digital asset import: Associate semantic tags with each component of the LOD3 and LOD4 models. For example, the column of the main hall of the Couple Garden is labeled with "Year of construction: 1874, Material: Phoebe zhennan wood, Craftsmanship: Mortise and tenon structure, Protection level: Provincial, Related historical event: The founding of Couple Garden"; A digital asset database of historical features of Pingjiang historical districts has been constructed, supporting searches by era, building type, and protection level. The CityGML standard interface is open and connected to the Suzhou City Information Modeling (CIM) platform to achieve standardized data reuse.

[0042] Multi-scenario compliance application output: Historical Appearance Restoration and Display: Develop a 3D visualization platform to showcase the original appearance of Pingjiang historical blocks during the Ming and Qing dynasties and the restoration process, supporting online roaming and detailed viewing by the public; Repair Design Simulation: Based on the LOD4 model, the repair plan for the eaves of the Couple's Garden is simulated to verify the rationality of the plan and output a simulation report; Urban renewal control plan comparison: Overlay the 3D model with the Pingjiang historical district renewal control plan to analyze the impact of the control plan on the historical appearance and assist in planning approval; VR / AR Immersive Experience: VR equipment is deployed in the Pingjiang Road scenic area to enable virtual tours and historical scene restoration, thereby enhancing the cultural and tourism experience; Digital Guide for Culture, Commerce and Tourism: Develop a digital guide mini-program that combines semantic tags to provide tourists with architectural history explanations and route navigation services; The three-dimensional digital translation system of the present invention is based on the above translation method, and its configuration process is as follows: Multi-source data acquisition module: Drone aerial photography unit: It adopts DJI Phantom 4 RTK drone, equipped with a 20-megapixel high-definition camera and GPS / IMU positioning module, supports automatic aerial photography and flight path planning, and outputs 3D point cloud and texture images; Ground laser scanning unit: adopts Faro Focus S70 laser scanner, with scanning accuracy of ±1.5mm, data transmission rate of 100Mbps, and output of fine point cloud data; The historical document digitization unit uses an EPSON V850 scanner with an OCR recognition accuracy of ≥98%. The manual annotation module supports collaborative annotation by multiple people and outputs a structured historical context database.

[0043] Spatiotemporal reference fusion and AI restoration module: Coordinate registration unit: Employs ICP algorithm, with registration accuracy of ±3mm, and supports real-time registration of multi-source data; Point cloud denoising is refined and simple: Gaussian filtering algorithm is used, with denoising efficiency ≥90% and retention of ≥98% of effective point cloud; Historical image alignment unit: The SIFT feature matching algorithm is used to achieve an alignment accuracy of ±5mm, realizing the spatiotemporal association between historical images and the current model; Historical landscape map retrieval unit: It uses the Neo4j graph database to store historical landscape map samples, integrates the KNN similarity retrieval algorithm, has a retrieval response time of ≤1 second, and supports multi-dimensional accurate retrieval; Generative AI Constraint Restoration Unit: Integrates a GAN generator and a historical map discriminator. The GAN generator adopts the DCGAN architecture and generates a resolution of 1024×1024. The historical map discriminator uses historical samples as the ground truth and scores the generated results, only outputting restoration models with a score ≥95.

[0044] Historical features semantic deconstruction and parameterization module: Five-level element classification unit: Based on Suzhou's traditional architectural history and ancient city protection standards, the system achieves automatic classification of five-level elements with a classification accuracy of ≥96%. Architectural Form Library Unit: Stores over 500 sets of architectural form parameters for Suzhou from the Ming and Qing Dynasties to the Republic of China period, and supports dynamic updates; Parametric Component Units: Using Revit parametric modeling technology, a parametric component library containing more than 1,000 components is built, supporting component drag-and-drop reuse and adaptive parameter adjustment; Cultural Symbols Database Unit: Stores over 300 sets of Suzhou regional cultural symbols, associated with cultural connotation information, and supports retrieval and reuse.

[0045] Multi-LOD layered modeling and landscape verification module: LOD1~LOD4 Modeling Unit 41: Integrates 3ds Max and Revit modeling plugins, supports four-level layered modeling, and improves modeling efficiency by 65%; Style consistency rule library unit: Stores verification rules such as form matching degree and component proportion tolerance, and supports custom adjustment of rules; Real-time verification and correction unit: Real-time verification during modeling, with a verification response time of ≤0.5 seconds, automatically recommending correction schemes, and a correction accuracy of ≥98%.

[0046] Semantic annotation and digital asset library module: Component semantic association unit: Supports batch semantic annotation, improving annotation efficiency by 70% and association accuracy by ≥99%; Multi-dimensional tag library unit: contains 6 major categories and 30 subcategories of semantic tags, and supports custom tag addition; CIM Standard Interface Unit: Supports CityGML 2.0 and IFC4.0 standards, enabling lossless data exchange with the Suzhou CIM platform and the Natural Resources Spatiotemporal Platform; Version Management Unit: Supports version rollback, updates, and deletions, retains all version records, and ensures data traceability.

[0047] Multi-scenario application service module: 3D visualization unit: Utilizes WebGL technology to support real-time rendering of million-polygon models, ensuring smooth scaling and roaming without lag; Repair Simulation Unit: Integrates ANSYS simulation plugin, supports stress and deformation simulation of repair schemes, and outputs detailed simulation reports; Planning comparison unit: Supports overlay comparison of 3D models and control planning schemes, outputs comparison analysis reports, and assists in planning approval; VR / AR Interaction Unit: Supports VR devices such as HTC Vive and Quest, enabling immersive roaming and virtual interaction; Cultural, Commercial and Tourism Application Unit: Develop applications for multiple platforms such as WeChat mini-programs and APPs to realize functions such as digital tour guides and cultural displays.

[0048] In summary, this system operates stably, with high accuracy and efficiency in digital translation, achieving a restoration accuracy rate of ≥95%, a modeling efficiency improvement of ≥60%, a component reuse rate of ≥80%, and a semantic retrieval response time of ≤1 second. It can effectively realize the three-dimensional digital translation of urban historical features, and the results can be directly used for historical preservation, planning approval, and cultural tourism revitalization, demonstrating good practicality and industrial value.

[0049] Any technical solution that achieves the above-mentioned technical effects by utilizing the technical solutions described in this invention, or by designing similar technical solutions by those skilled in the art under the inspiration of the technical solutions described in this invention, falls within the protection scope of this invention.

Claims

1. A method and system for three-dimensional digital translation of urban historical features, characterized by: This method employs a complete closed loop of "collection—reconstruction—deconstruction—modeling—database entry—application," combining architectural history, digital modeling, AI technology, and CIM technology to achieve a full-chain digital translation of historical features. The specific steps are as follows: S1: Collaborative acquisition of multi-source heterogeneous data; S2: Unification of spatiotemporal reference and semantic restoration of damaged areas; S3: Semantic deconstruction and parameterization of five levels of historical features; S4: Multi-LOD hierarchical 3D modeling and landscape consistency constraints; S5: Semantic annotation and historical digital asset storage; S6: Multi-scenario compliant application output; Based on the above digital translation methods, the corresponding translation system can provide hardware and software support for the implementation of the translation methods, forming a complete digital translation platform. This system consists of the following modules: Module 1: Multi-source data acquisition module; Module 2: Spatiotemporal Reference Fusion and AI Restoration Module; Module 3: Semantic Deconstruction and Parametricization of Historical Features; Module 4: Multi-LOD Layered Modeling and Landscape Verification Module; Module 5: Semantic Annotation and Digital Asset Repository Module; Module Six: Multi-Scenario Application Service Module.

2. The method for three-dimensional digital translation of urban historical features according to claim 1, characterized in that: S1: Collaborative acquisition of multi-source heterogeneous data: For the historical blocks and urban areas to be protected, an integrated air-ground acquisition mode of "UAV oblique photography + ground 3D laser scanning + digitization of historical documents" is adopted to construct a dual-source database of "spatial geometric data + historical context data" to ensure the comprehensiveness and accuracy of the data; Drone oblique photography: Using a multi-rotor drone equipped with a high-definition camera and GPS / IMU positioning module, aerial photography is carried out at a scale of 1:500 to obtain the overall real-world 3D point cloud (point cloud density ≥100 points / square centimeter) and building facade texture images of the historic district, covering the entire historic district area; Ground-based 3D laser scanning: Using a phase-type laser scanner (scanning accuracy ≤ ±2mm), millimeter-level scanning is performed on key historical buildings and distinctive components (such as brackets, brick carvings, and plaques) to obtain fine point cloud data, making up for the lack of accuracy of UAV aerial photography in microscopic details; Digitization of historical documents: Through scanning, OCR recognition, manual annotation and other methods, historical maps, old photos, building repair archives, local chronicles, architectural style historical materials, etc. are digitized to extract historical context data such as historical period, building technology, material color, cultural background and establish a structured document database. The S2: Spatiotemporal benchmark unification and semantic restoration of damaged areas: First, the multi-source data such as oblique photogrammetric point clouds, ground laser scanning point clouds, historical images, and topographic maps collected by the S1 UAV are unified into the same CIM / GIS spatiotemporal coordinate system (such as the 2000 National Geodetic Coordinate System) through coordinate registration algorithms (such as the ICP algorithm), and a timestamp is assigned to each type of data to construct a spatiotemporal base model containing spatial and temporal information; Secondly, for areas in the spatiotemporal base model that have missing building components, damaged texture, or faded color, semantic restoration is carried out using the method of "historical landscape map library retrieval + generative AI constraint restoration" to ensure that the restoration results are consistent with historical authenticity. Historical Landscape Atlas Retrieval: Search the historical landscape atlas database constructed for the target city to obtain form samples, component samples, and material and color samples that are from the same period, region, and building type as the damaged area; Generative AI-constrained restoration: Using retrieved historical samples as constraints, the geometric structure and texture information of the damaged area are generated through a generative AI model (such as a GAN model). During the generation process, a historical map discriminator performs real-time verification to ensure that the similarity between the generated result and the historical sample is ≥95%. The output restoration model conforms to historical authenticity and is integrated into the spatiotemporal basis model. The S3: Semantic deconstruction and parameterization of five levels of historical features: Based on architectural history theory and urban historical features protection standards, the urban historical features are deconstructed from top to bottom into five levels of parameterizable elements, so as to realize the standardization, reusability and editability of features elements; Level 1: Street fabric elements, parameters include road network pattern (road network density, street width), street scale (street spacing, cross-sectional dimensions), courtyard combination (courtyard layout, courtyard area), and skyline outline (building height, outline undulation). Level 2: Architectural form elements, parameters include structural type (timber structure, brick and stone structure, brick and wood mixed structure), bay width and depth (number of bays, depth dimensions), roof type (hipped roof, gable roof, overhanging roof), facade proportions (proportion of doors and windows to walls, eaves height). Level 3: Component details, including parameters such as door and window styles (lattice doors, decorative windows, door studs), bracket system (number of bracket layers, bracket spacing), column type (column diameter, column height, column base style), and carvings (carving patterns, dimensions, and locations). Level 4: Material and color elements, including parameters such as brick and stone material (blue bricks, gray tiles, stone types), pigment color spectrum (traditional pigment color values, fading coefficients), and aging process (degree of weathering, wear marks). Level 5: Cultural symbol elements, parameters include regional totems (regional characteristic patterns), plaques (font, content, size), couplets (text content, calligraphy style), and stone carving patterns (pattern style, carving technique). For each level of elements, a reusable, editable, and verifiable parametric component library is established. The component library is bidirectionally linked to the historical landscape map library, supporting parameter adjustment and dynamic updates. S4: Multi-LOD hierarchical 3D modeling and landscape consistency constraints: adopting a four-level hierarchical modeling method from LOD1 to LOD4, combined with a parametric component library and landscape consistency verification rules, to achieve hierarchical, high-precision, and highly consistent 3D modeling; LOD1 (street level): Based on the spatiotemporal base model, construct the overall outline and texture of the historical urban area / street, focusing on restoring the road network pattern, courtyard distribution and skyline outline to meet the needs of macro planning display. LOD2 (Architectural Level): Based on LOD1, it adds the external form of historical buildings and street space blocks, restores the overall form, width and depth of buildings, and roof form, and meets the display and analysis needs at the street scale. LOD3 (Component Level): Based on LOD2, it assembles components such as doors, windows, brackets, columns, and railings from the parametric component library to restore the detailed features of the building facade and meet the basic requirements of building renovation design. LOD4 (Cultural Gene Level): Based on LOD3, it adds micro-details such as interior space layout, painting, brick carving, and woodwork to restore the specific form and craftsmanship of cultural symbols and meet the needs of intangible cultural heritage inheritance and refined protection. During the modeling process, a style consistency verification rule is embedded to compare the modeling results with the parametric component library and the historical style map library in real time. The verification indicators include form matching degree (≥95%), component proportion tolerance (≤±5%), material color difference range (ΔE≤3), and cultural symbol compliance. When the verification fails, an automatic prompt is given and a correction scheme that conforms to the historical map is recommended to ensure that the modeling results are "restored to the old as before and built to the old as before". S5: Semantic annotation and digital asset storage of historical features: For the 3D models at LOD3 and LOD4 levels, multi-dimensional semantic tags are associated with each component. The tag content includes: age information (building construction year, renovation year), craftsmanship information (construction technique, renovation technique), material information (component material, pigment type), historical events (historical events related to the component / building), intangible cultural heritage information (related intangible cultural heritage projects, inheritors), and protection level (building protection level, component protection requirements). Based on semantic annotation results, a searchable, parsable, traceable, and inheritable digital asset library of historical features is constructed. The asset library supports multi-dimensional retrieval by year, area, building type, component type, protection level, etc. At the same time, it opens standard interfaces such as CityGML, IFC, and BIMcollab to connect to the City Information Modeling (CIM) platform and realize the standardized reuse of data. The S6: Multi-scenario compliant application output: Based on the historical landscape digital asset library and 3D model, it supports multi-scenario compliant application output. All output results retain semantic traceability information and historical authenticity proof, adapting to different application needs. Historical Appearance Restoration and Display: Through a 3D visualization platform, the original appearance and restoration process of historical blocks / buildings are displayed, supporting zooming, panning, and detailed viewing; Repair design simulation: Based on the LOD4 model, simulate and verify the rationality and compliance of the repair plan. Urban renewal control plan comparison: Overlaying the 3D model with the urban renewal control plan scheme, analyzing the impact of the control plan scheme on the historical appearance, and assisting in planning approval; VR / AR immersive experience: Through VR / AR devices, immersive tours and virtual interactions of historical districts are realized, revitalizing historical culture; Digital Guide for Culture, Commerce and Tourism: Develop a digital guide system that combines semantic tags to provide tourists with services such as historical and cultural explanations and route navigation; Protection planning approval assistance: Provides precise digital support for the preparation and approval of historical and cultural preservation plans, improving approval efficiency and scientific rigor.

3. The method for three-dimensional digital translation of urban historical features according to claim 2, characterized in that: The spatiotemporal benchmark unification and semantic restoration of the damaged area in S2, which includes AI image completion and historical map comparison, includes: Construct a historical landscape map database of the target city, including architectural styles, component styles, and color systems from different eras; For the damaged areas, generative AI is used to pre-generate the structure and texture, and then similarity retrieval and style calibration are performed with the historical map library to ensure that the restoration results are consistent with historical authenticity.

4. The method for three-dimensional digital translation of urban historical features according to claim 2, characterized in that: The five-level element parameterization in the semantic deconstruction and parameterization of the historical features of S3 includes: Street fabric: road network pattern, street and alley scale, courtyard combination, skyline outline parameters; Architectural form: structural type, bay width and depth, roof type, eaves height, facade proportion parameters; Component details: door and window styles, bracket set type, column type, railings, and carving parameters; Material and color: Brick, tile and stone materials, pigment color spectrum, and antiquing process parameters; Cultural symbols: regional totems, plaques, couplets, and stone carving patterns.

5. The three-dimensional digital translation system for urban historical features according to claim 1, characterized in that: Module 1: Multi-source data acquisition module: This module includes an UAV aerial photography unit, a ground laser scanning unit, and a historical document digitization unit. These three components work together to acquire spatial geometric data and historical contextual data. UAV aerial photography unit: Composed of multi-rotor UAV, high-definition camera, GPS / IMU positioning module, it supports automatic aerial photography, flight path planning, and outputs overall 3D point cloud and texture image of historical blocks; Ground laser scanning unit: Composed of a phase laser scanner and a data transmission module, it supports millimeter-level scanning and outputs fine point cloud data of key buildings and distinctive components; The historical document digitization unit consists of a scanner, an OCR recognition module, and a manual annotation module. It supports the digitization and structured storage of historical documents and outputs a historical context database. Module Two: Spatiotemporal Reference Fusion and AI Restoration Module: Includes coordinate registration unit, point cloud denoising and refinement unit, historical image alignment unit, historical landscape map retrieval unit, and generative AI constraint restoration unit, realizing multi-source data fusion and damaged area restoration. Coordinate registration unit: Using the ICP algorithm, multi-source data are unified into the same CIM / GIS spatiotemporal coordinate system, timestamps are assigned, and a spatiotemporal base model is generated; Point cloud denoising is simplified and refined: Gaussian filtering algorithm is used to remove noise and redundant points in point cloud data, thereby improving the accuracy of point cloud; Historical image alignment unit: Aligns historical photos and maps with the current spatiotemporal base model to achieve spatiotemporal correlation; Historical Landscape Map Retrieval Unit: It uses a graph database to store historical landscape map samples, integrates a similarity retrieval algorithm, and supports multi-dimensional and accurate retrieval by era, region, building type, component type, material and color; Generative AI Constraint Restoration Unit: Integrates GAN generator and historical map discriminator. GAN generator is used to generate geometry and texture of damaged areas. Historical map discriminator uses historical map samples as ground values ​​to score and filter the historical authenticity of the generated results, and only outputs restoration models with similarity ≥95%. Module 3: Historical Feature Semantic Deconstruction and Parametricization Module: This module includes a five-level element classification unit, an architectural form library unit, a component parametricization unit, and a cultural symbol library unit, enabling the deconstruction and parametricization of historical feature elements. Five-level element classification unit: Based on architectural history and protection standards, the historical features are deconstructed into five levels of elements: street texture, building form, component details, material color, and cultural symbols, so as to achieve standardization of element classification; Architectural Form Library Unit: Stores the form parameters of historical buildings of different eras and types, is associated with the historical landscape map library, and supports dynamic updates; Parametric Component Units: Parametric modeling of various building components to build a reusable and editable parametric component library, supporting component drag-and-drop reuse and adaptive parameter adjustment; Cultural Symbols Database Unit: Stores cultural symbols samples such as regional totems, plaques, couplets, and stone carvings of the target city, associated with cultural connotation information, and supports retrieval and reuse; Module 4: Multi-LOD Layered Modeling and Feature Verification Module: Includes LOD1 to LOD4 modeling units, a feature consistency rule base unit, and a real-time verification and correction unit, realizing layered modeling and feature consistency control. LOD1~LOD4 modeling units: support four levels of layered modeling, corresponding to street level, building level, component level and cultural gene level modeling respectively, and integrate parametric component assembly function to improve modeling efficiency; Style Consistency Rule Base Unit: Stores verification rules for form matching degree, component proportion tolerance, material color difference range, cultural symbol compliance, etc., and supports custom adjustment of rules; Real-time verification and correction unit: During the modeling process, the verification rules are called in real time to verify the modeling results. If the verification fails, the unit will automatically prompt and recommend a correction plan. Module 5: Semantic Annotation and Digital Asset Library Module: Includes component semantic association unit, multi-dimensional tag library unit, CIM standard interface unit, and version management unit, realizing semantic annotation and digital asset management. Component semantic association unit: Supports associating multi-dimensional semantic tags with 3D model components, realizing the binding of components with historical information; Multi-dimensional tag library unit: stores tags such as era, craftsmanship, materials, historical events, intangible cultural heritage information, protection level, etc., and supports the addition of custom tags; CIM Standard Interface Unit: Supports international / domestic standards such as CityGML, IFC, and BIMcollab, enabling lossless data exchange with urban information modeling platforms, natural resource spatiotemporal platforms, and historical and cultural preservation platforms; Version Management Unit: Manages the versions of models, components, and tags in the digital asset library, supports version rollback, updates, and deletions, and ensures data traceability; Module Six: Multi-Scenario Application Service Module: Includes 3D visualization, restoration simulation, planning comparison, VR / AR interaction, and cultural, commercial, and tourism application units, enabling multi-scenario application output. 3D visualization unit: Supports real-time rendering, scaling, panning, and detailed viewing of 3D models, and outputs visualized display results; Repair Simulation Unit: Supports the simulation and modeling of building repair plans, analyzes the rationality of repair plans, and outputs simulation reports; Planning Comparison Unit: Supports the overlay and comparison of 3D models and control planning schemes, outputs comparison analysis reports, and assists in planning approval; VR / AR Interaction Unit: Provides VR / AR immersive experience interface, supports virtual roaming and interactive operation, and outputs VR / AR application results; Cultural, commercial and tourism application unit: Develop functions such as digital guides and cultural displays, output cultural, commercial and tourism application products, and revitalize historical features.

6. The three-dimensional digital translation system for urban historical features according to claim 5, characterized in that: The AI ​​restoration unit integrates a generative adversarial network (GAN) and a historical landscape map retrieval engine to intelligently restore missing structures, blurred textures, and damaged components under the constraint of historical authenticity.

7. The three-dimensional digital translation system for urban historical features according to claim 5, characterized in that: The parametric component assembly unit supports drag-and-drop component reuse, adaptive shape adjustment, and batch style consistency verification, which significantly improves modeling efficiency and ensures historical style consistency.

8. The three-dimensional digital translation system for urban historical features according to claim 5, characterized in that: The CIM data interface unit supports data exchange with urban information modeling platforms, natural resource spatiotemporal platforms, and historical and cultural protection platforms, enabling the reuse of historical landscape data throughout the entire urban governance chain.

Images