Method for automatically generating a VR scene based on a parallel projection design drawing

Abstract

The application discloses a kind of VR scene automatic generation method based on parallel projection design drawing, comprising the following steps: S1, import two-dimensional CAD drawing containing architectural design information, parse the graphic primitive set in drawing;S2, component type identification is carried out to graphic primitive, and the mapping relationship between graphic primitive set and space semantics is established;S3, geometric reconstruction is carried out based on the mapping relationship between graphic primitive set and space semantics, and three-dimensional component model with unique name identification is generated;S4, multi-stage mapping model is established, and size conversion from drawing scale to virtual reality system space coordinate system is realized;S5, according to multi-stage mapping model, complete three-dimensional coordinate positioning and automatic assembly, and generate overall building or park-level three-dimensional structure layout;S6, the three-dimensional structure layout assembled is imported into virtual reality rendering engine, and virtual reality scene is generated.The application realizes the automatic, accurate and integrated conversion process from two-dimensional design drawing to three-dimensional virtual reality scene.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of computer-aided design analysis and modeling automation, and particularly relates to a VR scene automatic generation method based on parallel projection design drawings. BACKGROUND

[0002] In the design and visualization process of architecture, landscape, municipal and other fields, traditional design data mainly exists in the form of two-dimensional CAD drawings, such as plan and elevation drawings in DWG and DXF formats. These drawings usually contain the geometric contour, dimension label, layer structure and non-geometric information (such as line type, line weight, figure color and naming information) of the architectural components. In actual work, designers often use these two-dimensional drawings for scheme exchange, design verification and construction briefing. However, with the development of digital design tools and immersive visualization technology, traditional two-dimensional drawings have been difficult to meet the needs of modern architectural design for spatial expressiveness, interactivity and visual intuitiveness, especially in projects with complex building configurations, scale sensitivity or cross-departmental collaboration, where their limitations are more obvious.

[0003] The current three-dimensional modeling process widely used in the industry still requires manual modeling of components in two-dimensional CAD drawings one by one to form a three-dimensional model. This process is usually completed with modeling platforms such as SketchUp, 3dsMax, Revit, etc. Although it can finally output three-dimensional data that can be used for rendering or simulation, the modeling efficiency is extremely low, the operation is complex, and it is highly dependent on personnel skills. At the same time, in the absence of high-level information or elevation drawings, modelers often have to rely on experience to guess heights, opening positions and spatial relationships, which can easily lead to modeling errors, especially in large-scale scene or batch modeling tasks, the accumulation of errors will seriously affect the spatial authenticity of subsequent results. In addition, due to the differences in size units, drawing scales and figure expression habits used in various CAD drawings, without a unified unit conversion mechanism, it is easy to cause deviations between model size, scale and reality construction. Current modeling processes still rely on manual setting of unit scaling factors, which can cause model size distortion and visual scale imbalance due to insufficient experience or drawing understanding bias, seriously restricting spatial judgment and interaction effects in VR platforms.

[0004] In the process of converting two-dimensional drawings to three-dimensional VR scenes, there is also a problem of insufficient semantic recognition capability. Although two-dimensional CAD primitives contain attributes such as color, layer, and line type, they do not explicitly label component types or functional attributes. Existing technologies mostly rely on manual experience to determine the type of primitives. This approach lacks scalability and adaptability, making it difficult to efficiently migrate across drawings and projects, resulting in a lot of repetitive work. Meanwhile, in terms of facade information processing, existing systems usually only focus on geometric merging and vector cleaning of drawings, lack automated perspective classification and component height extraction mechanisms, and are particularly difficult to automatically adapt to non-standard perspectives such as inclined facades and inclined roofs, further increasing the technical threshold and implementation cost of modeling.

[0005] At the same time, there is no unified size mapping standard in VR integration. The size unit in CAD drawings is usually pixels or millimeters, while the unit system used in VR rendering engines such as Unity, Unreal, or self-developed engines may be "unit grid", "world unit", or custom proportion unit. In the existing modeling process, component sizes are often mapped using hard-coded proportions, lacking precise multi-level mapping and dynamic correction mechanisms, resulting in size proportion distortion, position offset, or scale incoordination between components in the three-dimensional scene generated in the VR environment. In addition, the existing VR scene construction process is usually separate from the modeling process. Even if a three-dimensional model has been completed, it still needs to be imported, adjusted, bound semantic information, and interactive logic, forming a fragmented engineering link, which is inefficient and prone to errors.

[0006] To overcome the above shortcomings, it is necessary to propose a new automated method that can achieve component type recognition, facade feature extraction, three-dimensional modeling, and multi-level size regression mapping based on two-dimensional parallel projection design drawings, and ultimately generate an immersive VR scene. This method should have high modeling intelligence, size conversion accuracy, and process closed-loop capability, truly realizing efficient, accurate, and intelligent conversion from two-dimensional drawings to three-dimensional VR scenes, meeting the future development needs of digital building design and spatial visualization. SUMMARY

[0007] One object of the present invention is to propose a VR scene automatic generation method based on parallel projection design drawings. The present invention makes full use of computer-aided drawing analysis, semantic recognition, three-dimensional modeling, and multi-level size regression mapping technologies, and describes in detail the entire process of automatically extracting component semantic information, facade features, and size parameters from two-dimensional CAD drawings, and generating an immersive virtual reality scene. This method has the advantages of high modeling efficiency, high size accuracy, accurate semantic recognition, and high automation of VR integration, effectively solving the problems of traditional manual modeling complexity, size mismatch, and model generation dependence on experience.

[0008] The VR scene automatic generation method based on parallel projection design drawing according to the embodiment of the application comprises the following steps:

[0009] S1, importing a two-dimensional CAD drawing containing architectural design information, and analyzing a grapheme set in the drawing;

[0010] S2, performing component type identification on the grapheme, and establishing a mapping relationship between the grapheme set and spatial semantics;

[0011] S3, performing geometric reconstruction based on the mapping relationship between the grapheme set and spatial semantics, and generating a three-dimensional component model with a unique name identifier;

[0012] S4, establishing a multi-stage mapping model between a drawing unit, a CAD pixel unit, an engine unit and actual metric dimensions, and realizing dimension conversion from a drawing scale to a virtual reality system spatial coordinate system;

[0013] S5, completing three-dimensional coordinate positioning and automatic assembly according to the multi-stage mapping model, and generating a three-dimensional structure layout of an overall building or a park;

[0014] S6, importing the assembled three-dimensional structure layout into a virtual reality rendering engine, and generating a virtual reality scene.

[0015] Optionally, the analyzing the grapheme set in the drawing comprises extracting a layer, a line type, a line weight, a color and a name attribute of each grapheme to form a grapheme attribute set.

[0016] Optionally, the component type identification comprises assisting in determining a component display priority and a sectioning attribute according to the grapheme attribute set, and simultaneously automatically classifying a direction and cutting an area of a multi-view elevation drawing.

[0017] Optionally, the S3 specifically comprises:

[0018] S31, constructing a two-dimensional topological graph by extracting line segment coordinates in the grapheme set bound with semantic information, and detecting whether the grapheme constitutes a closed structure by using a boundary tracking algorithm;

[0019] S32, establishing a mapping relationship between a closed graph and a component type according to the grapheme attribute set, identifying spatial entities such as a wall, a door, a window and a landscape component, and determining a component orientation in combination with direction information of the elevation drawing;

[0020] S33, performing a closed contour stretching operation to construct a three-dimensional block model for a wall type component, performing a nested cutting modeling operation for a door and window type component, and loading a corresponding three-dimensional simplified model in a preset component library for a special component;

[0021] S34, a unique component number is assigned to each generated three-dimensional construction model, which retains its original drawing number, layer name, semantic label and structure attribute;

[0022] S35, when there are abnormal conditions such as boundary unclosed or overlapping of graph elements in the graph element set, edge fitting and contour closure repair algorithm is executed, overlapping structures are automatically merged, and topologically consistent structure contours are output.

[0023] Optionally, the S4 specifically comprises:

[0024] S41, the scale information set in the drawing is extracted, and the conversion relationship between the drawing unit and the actual physical unit is established according to the unit scale marked on the drawing, so as to obtain the conversion ratio between the drawing unit and the metric unit;

[0025] S42, the pixel unit in the CAD drawing is converted into the internal space unit supported by the virtual reality engine, and the engine coordinate unit length corresponding to each CAD pixel is calculated according to the preset CAD resolution parameter and engine coordinate precision parameter;

[0026] S43, the coordinate mapping relationship between the engine space unit and the actual metric unit is established, and the size attribute and position attribute of all three-dimensional models are converted into actual values in meters based on the conversion relationship and the engine coordinate unit length;

[0027] S44, the conversion parameter set based on the conversion relationship and the coordinate mapping relationship is encapsulated to ensure that all component sizes and positions have consistent physical scale basis.

[0028] Optionally, the three-dimensional coordinate positioning and automatic assembly include that the system performs three-dimensional space positioning according to the center point coordinates and size information of each component model, and automatically completes assembly according to the space layout shown in the architectural structure drawing.

[0029] Optionally, the virtual reality rendering engine initializes material mapping according to the original color attribute of the graph element, loads light information, collision detection module and interaction logic to generate a virtual reality scene.

[0030] The VR scene automatic generation method based on parallel projection design drawing provided by the application has the advantages of high automation, closed loop structure, engineering adaptability, and significant beneficial effects.

[0031] Firstly, the present application can automatically identify the types and attributes of various components in two-dimensional CAD drawings by introducing a semantic rule-driven graphic element analysis mechanism, significantly improving the accuracy and consistency of component identification. Traditional methods usually rely on manual judgment and layer experience, while the present method uses attributes such as color, line type, line weight and name to construct component identification rules and complete semantic mapping, thereby avoiding human subjective bias and achieving automatic classification and organization at the component level, laying a semantic foundation for subsequent structural modeling.

[0032] Secondly, the present application introduces a view cutting and geometric parameter extraction mechanism in processing building facade information, which can automatically extract three-dimensional key parameters such as height and opening position from multiple facade drawings, realizing joint analysis of plan and facade drawings. This mechanism supports multi-view processing and cutting recognition, ensuring accurate construction of component models in the vertical dimension, effectively avoiding modeling distortion caused by height loss or incorrect inference in traditional methods, and enhancing the spatial restoration and structural consistency of models in the VR environment.

[0033] In addition, the present application constructs a multi-level regression size mapping model to model and correct the mapping relationship between CAD drawing units, virtual engine units and actual size units using a non-linear regression function. Compared with traditional linear scaling or fixed proportion conversion methods, this mapping mechanism is more flexible and accurate, and can adapt to different drawing scales and unit systems, ensuring that three-dimensional models have consistent size, proportion and spatial distribution with real buildings in the VR environment, and fundamentally solving the model distortion and scene misjudgment problems caused by unit mismatch.

[0034] Furthermore, the present application introduces an automatic component modeling process in the three-dimensional modeling stage, including stretching generation, Boolean subtraction and template calling modeling strategies, which not only improves the model construction efficiency, but also supports the automatic generation of various structural components such as walls, doors, windows and landscapes, reduces the dependence on traditional three-dimensional modeling tools, significantly reduces the modeling threshold, and improves the universality of engineering applications.

[0035] Finally, the present application proposes an integrated VR integration scheme, realizing the whole process automatic closed loop from drawing import to three-dimensional modeling and then to VR scene building. Component models can be directly imported into VR engines supporting real-time rendering and interaction after spatial positioning, with semantic labels and interactive control logic, forming an immersive digital scene that can be browsed, operated and scheduled, greatly improving the visualization level, delivery efficiency and user experience of architectural design.

[0036] The application breaks through the limitation of the traditional two-dimensional drawing modeling method, realizes technical innovation and process reengineering in component identification, facade understanding, size mapping, model generation and VR integration and multiple core links, has the remarkable advantages of reasonable structure, high calculation precision, strong adaptability and good generalization, and provides strong support for visual application of building information modeling (BIM), digital city, smart park and multiple scenes. BRIEF DESCRIPTION OF DRAWINGS

[0037] The accompanying drawings are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application, and do not limit the application. In the drawings:

[0038] Fig. 1 A flowchart of a VR scene automatic generation method based on parallel projection design drawings is provided for the application.

[0039] Fig. 2 A multi-stage size regression mapping relationship diagram of a VR scene automatic generation method based on parallel projection design drawings is provided for the application. DETAILED DESCRIPTION

[0040] The application will now be described in further detail with reference to the drawings. These drawings are simplified schematic diagrams, and only schematically show the basic structure of the application, and therefore only show the components related to the application.

[0041] REFERENCE Figs. 1-2 A VR scene automatic generation method based on parallel projection design drawings, comprising the following steps:

[0042] S1, importing a two-dimensional CAD drawing containing architectural design information, and analyzing the graph element set in the drawing;

[0043] S2, component type identification of the graph element, and establishing a mapping relationship between the graph element set and the spatial semantics;

[0044] S3, geometric reconstruction based on the mapping relationship between the graph element set and the spatial semantics, to generate a three-dimensional component model with a unique name identification;

[0045] S4, establishment of a multi-stage mapping model between the drawing unit, CAD pixel unit, engine unit and actual metric size, to realize size conversion from drawing scale to virtual reality system spatial coordinate system;

[0046] S5, according to the multi-stage mapping model, complete three-dimensional coordinate positioning and automatic assembly, to generate a whole building or park-level three-dimensional structure layout;

[0047] S6, importing the assembled three-dimensional structure layout into a virtual reality rendering engine, to generate a virtual reality scene.

[0048] In the embodiment, the parsing of the set of graphic elements in the drawing includes extracting the layer, line type, line weight, color and name attributes of each graphic element to form a set of graphic element attributes.

[0049] The application effectively improves the automation degree and geometric precision of three-dimensional scene construction. Through the reservation of component numbers, layer semantics and spatial semantic tags, bidirectional traceable binding between the drawing and the model data is realized, which facilitates subsequent model modification, dynamic interaction, material replacement and structure optimization operations. The generated virtual reality scene can be directly run in the rendering engine, with functions such as immersive roaming, component-level interaction, real-time lighting and collision detection, greatly improving the expressiveness and verification efficiency of architectural design results, and being suitable for architectural scheme demonstration, design collaborative review, planning optimization deduction, real estate sales display and intelligent park visual management and other scenes. The application integrates the fragmented links of two-dimensional expression, manual modeling and model import in the traditional design process to form an integrated, automated and highly consistent mapping method from drawings to VR space, significantly improving the applicability and engineering practical value of virtual reality scene generation in building, planning and urban information systems.

[0050] In the embodiment, the component type recognition includes auxiliary determination of component display priority and sectioning attributes according to the set of graphic element attributes, and automatic direction classification and area cutting of multi-view elevation drawings.

[0051] In the embodiment, the S3 specifically includes:

[0052] S31, constructing a two-dimensional topological graph by extracting the line segment coordinates in the set of graphic elements bound with semantic information, and detecting whether the graphic elements constitute a closed structure by using a boundary tracking algorithm;

[0053] S32, establishing a mapping relationship between the closed graphic and the component type according to the set of graphic element attributes, identifying spatial entities such as walls, doors, windows and landscape components, and determining the orientation of the components in combination with the direction information of the elevation drawing;

[0054] S33, for wall-type components, performing a closed contour stretching operation to construct a three-dimensional block model; for door and window-type components, performing a nested cutting modeling operation; for special components, loading a corresponding three-dimensional simplified model in a preset component library;

[0055] S34, giving each generated three-dimensional construction model a unique component number, and the three-dimensional construction model retains its original drawing number, layer name, semantic tag and structure attribute;

[0056] S35, when there are abnormal situations such as unclosed boundaries or overlapping graphic elements in the set of graphic elements, performing an edge fitting and contour closure repair algorithm to automatically merge overlapping structures and output a topologically consistent structure contour.

[0057] In this step, the system also gives the generated component model a unique component number and retains its original drawing structure attributes, including layer information, name, line weight, color, and spatial semantic identifier, etc., to provide a data basis for subsequent size conversion, scene assembly, and interactive recognition. At the same time, to improve the model integrity and robustness, this step also integrates contour repair and overlapping edge merging mechanism, which can repair the boundary breakage, line segment overlap, and non-closed graphics in the input graphics, and ensure the output model has spatial consistency and assembly capability. Through this step, the abstract component description in the two-dimensional drawing is upgraded to an object that exists in the three-dimensional space, laying a foundation for the component-level space and semantic basis for the final VR scene generation.

[0058] In the embodiment, the S4 specifically includes:

[0059] S41, extracting the scale information set in the drawing, establishing the proportion conversion relationship between the drawing unit and the actual physical unit according to the unit proportion marked in the drawing, and obtaining the conversion ratio between the drawing unit and the metric unit;

[0060] S42, converting the pixel unit in the CAD drawing into the internal space unit supported by the virtual reality engine, and calculating the engine coordinate unit length corresponding to each CAD pixel according to the preset CAD resolution parameter and engine coordinate precision parameter;

[0061] S43, establishing the coordinate mapping relationship between the engine space unit and the actual metric unit, and converting the size attribute and position attribute of all three-dimensional models into actual numerical values in meters based on the proportion conversion relationship and the engine coordinate unit length;

[0062] S44, encapsulating the proportion conversion relationship and the coordinate mapping relationship as a conversion parameter set to ensure that all component sizes and positions have a consistent physical scale basis.

[0063] The system analyzes the scale set in the drawing (such as 1:100, 1:200, etc.), constructs the scale conversion factor between the unit of the drawing and the actual size; secondly, according to the resolution parameter of the CAD drawing (such as how many pixels correspond to per meter), the pixel unit recorded in the drawing is mapped into an identifiable size value; further, according to the internal coordinate precision of the engine, the CAD unit is converted into the logical unit in the VR engine coordinate system; finally, according to the scaling factor between the engine unit and the actual meter unit, the size parameters and spatial coordinates of all components are finally converted into numerical values with physical scale significance. In addition, the system generates a unified conversion parameter set in this step, including the scale conversion coefficient, the coordinate scaling factor, the origin alignment rule, etc., which provides a unified reference for subsequent component positioning and scene assembly. The execution of this step ensures that the three-dimensional model can maintain the same design logic as the original drawing in terms of spatial scale, component position and scene structure after being imported into the rendering engine, so that the designer can obtain a real restored spatial perception and size reference in the VR environment.

[0064] In the embodiment, the three-dimensional coordinate positioning and automatic assembly includes that the system performs three-dimensional space positioning according to the center point coordinates and size information of each component model, and automatically completes assembly according to the spatial layout shown in the architectural structure drawing.

[0065] In the embodiment, the virtual reality rendering engine initializes the material mapping according to the original color attribute of the primitive, loads the light information, the collision detection module and the interaction logic to generate the virtual reality scene.

[0066] Example 1

[0067] In order to verify the feasibility and actual application effect of the present application in the implementation process, the "VR scene automatic generation method based on parallel projection design drawing" of the present application is applied to a provincial city green park planning project. The project goal is to build a large green landscape area with an area of about 36,000 square meters. The planning document is mainly provided in DWG format, including one general plan and four elevation drawings, which are the south main entrance elevation, the north landscape sketch elevation, the east fence elevation and the west platform structure elevation. The design team hopes to quickly convert the project drawing into an immersive VR scene in the preliminary review stage to assist in design decision-making, improve the reporting effect and reduce the model construction time.

[0068] In the actual application of the project, the design team first imports the two-dimensional drawing file in DWG format into the system, the system automatically identifies the graphic element types in the drawing and performs attribute analysis, and distinguishes wall, road, green belt, structure, tree pool, waterscape and other layers. Through the semantic rule engine, the graphic elements are quickly classified, and the type binding and attribute extraction of all graphic elements are completed in about 2 minutes. For the four elevation drawings in the drawing, the system calls the built-in direction classification and view cutting module, automatically identifies the view direction and cuts the components according to the layer label and boundary contour, and accurately extracts the height and opening data of 16 groups of wall components, 12 groups of doors and windows and 5 groups of landscape components.

[0069] In the three-dimensional modeling stage, the system automatically generates component blocks according to the extracted component geometric information. Taking the main entrance area as an example, the system identifies 4 groups of main wall components and models them by stretching, and the component size of each group of wall is between 6.0m*0.24m*3.2m and 9.5m*0.3m*3.8m. The corresponding door and window components automatically call the Boolean cutting logic to complete the hollowing processing, and then complete the model assembly. The modeling time of the entire main entrance area (about 420 square meters) three-dimensional model is only about 7 minutes, which is more than 12 times more efficient than the traditional 3dsMax modeling process (about 90 minutes).

[0070] In the size mapping process, the system constructs a multi-level size regression model according to the scale (1:100) and CAD unit (1 pixel = 0.005 meters) marked on the original drawing. The fitting error of the trained mapping function is kept within ±1.5 cm. Taking a group of green components as an example, its length on the drawing is 812 pixels, and the actual length of the three-dimensional component generated by the system after mapping is 4.059 meters. The actual length marked on the construction drawing is 4.04 meters, with an error of 0.019 meters, which meets the spatial accuracy requirements of the building pre-evaluation stage.

[0071] After the system completes the three-dimensional modeling of all components, the model data is automatically imported into the SSE graphics engine for VR integration, and the component semantic label, collision body information and interaction feedback logic are bound. After the user enters the scene through the VR terminal, he can freely roam the park, view the specific component parameters and space arrangement, and support path planning, space switching and component information query functions. The total time from the import of the two-dimensional drawing to the output of the VR scene is 42 minutes, while the average time of the traditional process (manual modeling + VR import + manual labeling) is 6.5 hours. The modeling and integration efficiency of the system is 1 / 9 of the traditional process, which greatly saves the labor cost and design time.

[0072] In order to further verify the performance of the system in multiple typical areas, the design team tested the system efficiency and accuracy of 5 representative areas, and the results are shown in the following table.

[0073] Table 1: Performance verification data table of the system of the application in the urban green space project (scene: VR design rapid modeling test of urban green space park of a certain province and city)

[0074]

[0075] It can be seen from the actual application and data comparison results of the embodiment that in the conversion process from two-dimensional drawings to three-dimensional VR scenes, the method significantly improves the recognition accuracy, size mapping accuracy and modeling integration efficiency, effectively solves the problems of low efficiency, mismatch and artificial dependence in traditional manual modeling, and provides an efficient, intelligent and strong digital solution for the field of municipal design, building scheme review, landscape planning and the like.

Claims

1. A method for automatically generating a VR scene based on parallel projection design drawings, characterized in that, The method comprises the following steps: S1, importing a two-dimensional CAD drawing containing architectural design information, and analyzing a set of drawing elements in the drawing; S2, identifying the types of the elements, and establishing a mapping relationship between the set of elements and spatial semantics; S3, performing geometric reconstruction based on the mapping relationship between the set of elements and spatial semantics, and generating a three-dimensional component model with a unique name identifier; S4, establishing a multi-stage mapping model between the drawing unit, the CAD pixel unit, the engine unit, and the actual metric size, and realizing size conversion from the drawing scale to the virtual reality system spatial coordinate system; S5, completing three-dimensional coordinate positioning and automatic assembly according to the multi-stage mapping model, and generating a three-dimensional structure layout of the whole building or park; S6, importing the assembled three-dimensional structure layout into a virtual reality rendering engine, and generating a virtual reality scene.

2. The method of claim 1, wherein, The analysis of the set of drawing elements includes extracting the layer, line type, line weight, color, and name attributes of each element to form an element attribute set.

3. The method of claim 1, wherein, The component type identification includes determining the display priority and sectioning attributes of the components with the assistance of the element attribute set, and automatically classifying the multi-view elevation drawing by direction and region.

4. The method of claim 1, wherein, S3 specifically comprises: S31, constructing a two-dimensional topological graph by extracting the line segment coordinates of the element set bound with semantic information, and detecting whether the elements constitute a closed structure by using a boundary tracking algorithm; S32, establishing a mapping relationship between the closed graph and the component type according to the element attribute set, identifying wall, door, window, and landscape component spatial entities, and determining the component orientation in combination with the direction information of the elevation drawing; S33, for wall-type components, performing a closed contour stretching operation to construct a three-dimensional block model; for door and window-type components, performing a nested cutting modeling operation; and for special components, loading a corresponding three-dimensional simplified model from a preset component library; S34, assigning a unique component number to each generated three-dimensional component model, and the three-dimensional component model retains its original drawing number, layer name, semantic label, and structure attribute; S35, when there are boundary unclosed or element overlapping abnormal conditions in the set of elements, performing an edge fitting and contour closure repair algorithm to automatically merge overlapping structures, and outputting a topologically consistent structure contour.

5. The method of claim 1, wherein, S4 specifically comprises: S41, extracting the scale information set in the drawing, establishing a scale conversion relationship between the drawing unit and the actual physical unit according to the unit scale marked in the drawing, and obtaining a conversion ratio between the drawing unit and the metric unit; S42, converting the pixel unit in the CAD drawing into an internal space unit supported by the virtual reality engine, calculating the engine coordinate unit length corresponding to each CAD pixel according to the preset CAD resolution parameter and engine coordinate precision parameter; S43, establishing a coordinate mapping relationship between the engine space unit and the actual metric unit, converting the size attribute and position attribute of all three-dimensional models into actual values in meters based on the scale conversion relationship and the engine coordinate unit length; S44, encapsulating the scale conversion relationship and the coordinate mapping relationship as a conversion parameter set to ensure that all component sizes and positions have a consistent physical scale basis.

6. The method of claim 1, wherein, The three-dimensional coordinate positioning and automatic assembly system comprises three-dimensional space positioning according to the center point coordinates and size information of each component model, and automatically completes assembly according to the space layout shown in the architectural structure drawing.

7. The method of claim 1, wherein the method further comprises: The virtual reality rendering engine initializes material mapping according to primitive color attributes, loads light information, a collision detection module, and interaction logic to generate a virtual reality scene.

Images