Engineering BIM model lightweight rendering method based on model semantic information

Abstract

The invention provides an engineering BIM model lightweight rendering method based on model semantic information, and the method comprises the steps: analyzing an engineering BIM model, and obtaining the semantic information of the engineering BIM model; performing octree space segmentation on the engineering BIM model, and establishing a model octree optimization structure of the engineering BIM model; semantic information of each tile of each level is obtained; and carrying out space elimination based on view cones and semantic elimination based on model semantics. By means of the method, the efficiency of octree index calculation and model rendering simplified calculation of the engineering BIM model can be improved, the data size of the BIM model rendered by a browser at the same time is reduced, the utilization rate of hardware such as a GPU is increased, the loading and rendering efficiency of the BIM model is improved, more BIM models can be loaded at the same time under the same hardware configuration condition, and the loading efficiency and the rendering efficiency of the BIM model are improved. And rendering application of massive BIM models is supported.

Description

Technical Field

[0001] The present invention belongs to the technical field of computer graphics, and in particular relates to a lightweight rendering method for an engineering BIM model based on model semantic information. Background Art

[0002] Ordinary three-dimensional models focus on expressing geometric shapes and appearance textures, and usually use triangular facets to restore geometric structures; BIM models (i.e., Building Information Models) represent buildings and related information in a digital way, and usually include indoor and outdoor multi-level geometric structures, multi-scale geometric structures from large areas to small local models, and geometric structures that conform to mathematical laws. At the same time, BIM models have a large amount of information such as structural grouping and design attributes.

[0003] At present, the technologies for implementing 3D rendering based on browsers mainly include WebGPU and WebGL, both of which can realize the rendering of 3D graphics using JavaScript in HTML5 Canvas elements. Among them, WebGL supports 3D graphics rendering capabilities in browser web pages and provides a high-level encapsulation API, which is easy to use and use, but lacks support for modern GPU features. At the same time, WebGL is mainly executed in a single thread and lacks support for multi-threading. Therefore, WebGL cannot adapt to complex 3D scenes and is more suitable for rapid development and deployment of conventional basic 3D graphics rendering applications. As an emerging technology, WebGPU natively supports a variety of modern graphics features, provides a low-level API that is closer to GPU hardware, supports multi-threaded execution of command buffers and data processing operations, and has more flexible and efficient resource management mechanisms. Through mechanisms such as low-level control, command buffers, and pipeline state objects, it can more efficiently utilize GPU resources. It is more suitable for 3D applications that require high performance and advanced graphics functions, and is suitable for rendering applications of BIM models with complex geometric characteristics.

[0004] BIM models have characteristics such as geometric regularity, complex file structure, and a large amount of structural attribute information. Currently, when BIM models are rendered through front-end browsers, there are the following shortcomings: at the current viewing angle, the amount of data loaded and rendered BIM models is large, which reduces the display rate of BIM models and reduces the user experience of viewing BIM models on the screen. Summary of the invention

[0005] In view of the defects of the prior art, the present invention provides a lightweight rendering method of an engineering BIM model based on model semantic information, which can effectively solve the above problems.

[0006] The technical solution adopted by the present invention is as follows:

[0007] The present invention provides a lightweight rendering method for an engineering BIM model based on model semantic information, comprising the following steps:

[0008] Step S1, the engineering BIM model includes a plurality of BIM components with a hierarchical relationship; the engineering BIM model is parsed to obtain semantic information of the engineering BIM model; wherein the semantic information of the engineering BIM model includes: the hierarchical inclusion relationship between BIM components, the texture material information of the BIM components, and the geometric structure of the BIM components;

[0009] Step S2, combining the semantic information of the engineering BIM model, performing octree space segmentation on the engineering BIM model, and establishing a model octree optimization structure of the engineering BIM model; the model octree optimization structure is a multi-level tile structure corresponding to the outline of the engineering BIM model;

[0010] Step S3, obtaining semantic information of each tile at each level, including relevant face information of the BIM components directly contained in the tile and BIM component inclusion relationship information;

[0011] The related face information of the BIM component refers to the face information of the BIM component located inside the tile; the BIM component inclusion relationship information refers to: for the BIM component A directly contained in the tile, if it does not further contain other BIM components inside, it is recorded that the BIM component A does not have an inclusion relationship; if it further contains BIM component B inside, the material of BIM component A is further determined according to the texture material information of the BIM component, and if there is a transparent material, it is recorded that BIM component A and BIM component B are in a visible inclusion relationship; otherwise, it is recorded that BIM component A and BIM component B are in an invisible inclusion relationship;

[0012] Step S4, concurrently requesting tiles of all levels of the engineering BIM model in the browser through WebGPU technology, and calculating the position of the 0th level tile in the three-dimensional scene as the initial position; storing the tiles of all levels with a hierarchical relationship and the initial position;

[0013] Step S5, spatial culling based on the viewing frustum:

[0014] When rendering is required, firstly, tiles of all levels with a hierarchical relationship are loaded, and displayed in the three-dimensional scene according to the initial position of the 0th level tile;

[0015] Obtain the view cone according to the camera position of the three-dimensional scene, and calculate the intersection boundary between the view cone and tiles at all levels; according to the intersection boundary, determine the tiles at all levels located inside the intersection boundary as the tiles at all levels located in the visible area, and remove the tiles of other levels located in the invisible area;

[0016] Step S6, semantic elimination based on model semantics:

[0017] According to the semantic information of tiles at all levels located in the visible area, the relevant facets of the BIM components directly contained in the tiles are rendered. At the same time, for the directly contained BIM components, if there is a BIM component B with an invisible relationship inside it, BIM component B is automatically filtered and not rendered. If there is a BIM component B with a visible relationship inside it, BIM component B is rendered;

[0018] This enables lightweight rendering of engineering BIM models based on model semantic information.

[0019] Preferably, the semantic information of the engineering BIM model also includes BIM component grouping information; the BIM component grouping information means: if multiple BIM components express the same feature, the multiple BIM components are regarded as an integral BIM component through semantic association, the BIM component outline is determined through the integral BIM component, and then its association relationship with the tile is determined.

[0020] Preferably, step S2 specifically comprises:

[0021] Step S2.1, determining the outline of the engineering BIM model according to the geometric structure of the engineering BIM model, and then establishing the minimum bounding volume of the engineering BIM model, which is the 0th level tile;

[0022] Step S2.2, dividing the 0th level tile into 8 sub-tiles of the same size. If a sub-tile does not contain any face of the BIM model, the sub-tile is removed; the remaining sub-tiles are sub-tiles of the 0th level tile and are also 1st level tiles;

[0023] Step S2.3, for each first-level tile, if the size of the BIM model it contains is smaller than the size threshold, or if the number of facets of the BIM model it contains is smaller than the facet number threshold, stop dividing it to the next level; otherwise, continue to divide it into 8 sub-tiles of the same size, and continue in this way to establish the model octree optimization structure of the engineering BIM model.

[0024] Preferably, it also includes:

[0025] Step S7, when the three-dimensional field of view is scaled or translated, the viewing cone is recalculated according to the new camera position, and steps S5 to S6 are repeated to recalculate the tiles and semantic information of the tiles in the visible area to achieve rendering of the visible BIM components in the visible area.

[0026] The lightweight rendering method of an engineering BIM model based on model semantic information provided by the present invention has the following advantages:

[0027] Through the method of the present invention, the efficiency of octree index calculation and model rendering simplification calculation of engineering BIM models can be improved, the amount of data of BIM models rendered simultaneously by the browser can be reduced, the utilization rate of hardware such as GPU can be improved, the loading and rendering efficiency of BIM models can be improved, more BIM models can be loaded simultaneously under the same hardware configuration conditions, and the rendering application of massive BIM models can be supported. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 An overall flow chart of a lightweight rendering method for an engineering BIM model based on model semantic information provided by the present invention;

[0029] Figure 2 A schematic diagram of the model octree optimization structure provided by the present invention;

[0030] Figure 3 A detailed flow chart of a lightweight rendering method for an engineering BIM model based on model semantic information provided by the present invention. DETAILED DESCRIPTION

[0031] In order to make the technical problems, technical solutions and beneficial effects solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0032] The present invention provides a lightweight rendering method for an engineering BIM model based on model semantic information, which relates to visual rendering of a three-dimensional model and an engineering BIM model, and in particular to a method for performing occlusion culling calculation based on BIM model semantic information to achieve lightweight rendering when rendering a three-dimensional BIM model in a browser. The method is mainly used for lightweight rendering and application of BIM three-dimensional models in industries such as engineering digital twins and smart cities.

[0033] See also Figure 1 The present invention provides a lightweight rendering method for an engineering BIM model based on model semantic information, comprising the following steps:

[0034] Step S1, the engineering BIM model includes a plurality of BIM components with a hierarchical relationship; the engineering BIM model is parsed to obtain the semantic information of the engineering BIM model; wherein the semantic information of the engineering BIM model includes: the hierarchical inclusion relationship between BIM components, the texture material information of the BIM components, and the geometric structure of the BIM components; of course, the semantic information of the engineering BIM model may further include model index, design parameters, management attributes, etc.;

[0035] To further improve rendering efficiency, the semantic information of the engineering BIM model also includes BIM component grouping information; the BIM component grouping information means: if multiple BIM components express the same feature, the multiple BIM components are regarded as a whole BIM component through semantic association, and the BIM component outline is determined through the whole BIM component, and then its association relationship with the tile is determined. In the subsequent description, if the BIM component has grouping information, the BIM component in the subsequent description refers to the whole BIM component.

[0036] Step S2, combining the semantic information of the engineering BIM model, performing octree space segmentation on the engineering BIM model, and establishing a model octree optimization structure of the engineering BIM model; the model octree optimization structure is a multi-level tile structure corresponding to the outline of the engineering BIM model;

[0037] See also Figure 2 , step S2 is specifically as follows:

[0038] Step S2.1, determining the outline of the engineering BIM model according to the geometric structure of the engineering BIM model, and then establishing the minimum bounding volume of the engineering BIM model, which is the 0th level tile;

[0039] Step S2.2, divide the 0th level tile into 8 sub-tiles of the same size. The specific division method is: perform spatial octree division in such a way that the length, width and height of the 0th level tile are divided into two, and obtain 8 sub-tiles of the same size.

[0040] According to the engineering BIM model and the outlines of BIM components at each level, if a sub-tile does not contain any face of the BIM model and is a blank sub-tile, the sub-tile is filtered out; the remaining sub-tiles are sub-tiles of the 0th level tile and are also 1st level tiles;

[0041] Of course, in actual applications, when filtering and removing each sub-tile, other solutions can also be selected, for example, presetting a minimum size; if the size of the BIM model contained in the sub-tile is smaller than the minimum size, it is also filtered and removed.

[0042] Step S2.3, for each first-level tile, if the size of the BIM model it contains is smaller than the size threshold, or if the number of facets of the BIM model it contains is smaller than the facet number threshold, stop dividing it to the next level; otherwise, continue to divide it into 8 sub-tiles of the same size, and continue in this way to establish the model octree optimization structure of the engineering BIM model.

[0043] Step S3, obtaining semantic information of each tile at each level, including relevant face information of the BIM components directly contained in the tile and BIM component inclusion relationship information;

[0044] The related face information of the BIM component refers to the face information of the BIM component located inside the tile; the BIM component inclusion relationship information refers to: for the BIM component A directly contained in the tile, if it does not further contain other BIM components inside, it is recorded that the BIM component A does not have an inclusion relationship; if it further contains BIM component B inside, the material of BIM component A is further determined according to the texture material information of the BIM component, and if there is a transparent material, it is recorded that BIM component A and BIM component B are in a visible inclusion relationship; otherwise, it is recorded that BIM component A and BIM component B are in an invisible inclusion relationship;

[0045] In practical applications, the various faces contained in each BIM component can be determined based on the semantic naming association, and the tile in which each face is located can be determined based on the coordinates of each face, so as to obtain the tile numbers corresponding to all faces of the same BIM component, and thus obtain the relevant face information of the BIM component directly contained in the tile.

[0046] Then, the inclusion relationship of multiple BIM components in each tile is parsed according to the semantic inclusion relationship of the BIM model. If BIM component X contains other BIM component Y, the material of BIM component X is further judged. If there is a transparent material, BIM component X and BIM component Y are recorded as a visible inclusion relationship. Otherwise, BIM component X and BIM component Y are recorded as an invisible inclusion relationship, and the relationship between BIM component X and other BIM components is recorded as no relationship, so as to obtain the semantic information of tiles at all levels; the semantic information of tiles at all levels and the model octree optimization structure are logically associated through the same tile naming.

[0047] Step S4, concurrently requesting tiles of all levels of the engineering BIM model in the browser through WebGPU technology, and calculating the position of the 0th level tile in the three-dimensional scene as the initial position; storing the tiles of all levels with a hierarchical relationship and the initial position;

[0048] For example, the correct spatial positioning of the engineering BIM model in the three-dimensional scene is obtained through the position and orientation of the 0th-level tile in the three-dimensional scene, which is recorded in the computer memory as the initial position of the model.

[0049] Step S5, spatial culling based on the viewing frustum:

[0050] When rendering is required, firstly, tiles of all levels with a hierarchical relationship are loaded, and displayed in the three-dimensional scene according to the initial position of the 0th level tile;

[0051] Obtain the view cone according to the camera position of the three-dimensional scene, and calculate the intersection boundary between the view cone and tiles at all levels; according to the intersection boundary, determine the tiles at all levels located inside the intersection boundary as the tiles at all levels located in the visible area, and remove the tiles of other levels located in the invisible area;

[0052] A specific implementation method is: obtain the viewing cone according to the camera position of the three-dimensional scene, perform intersection calculation with the tiles at all levels in the viewing cone and the model octree optimization structure, and use the directly intersecting tiles as visible node tiles; retain the visible node tiles and the first-level tiles inside each visible node tile, and discard tiles of other levels.

[0053] Step S6, semantic elimination based on model semantics:

[0054] According to the semantic information of tiles at all levels located in the visible area, the relevant facets of the BIM components directly contained in the tiles are rendered. At the same time, for the directly contained BIM components, if there is a BIM component B with an invisible relationship inside it, BIM component B is automatically filtered and not rendered. If there is a BIM component B with a visible relationship inside it, BIM component B is rendered;

[0055] This enables lightweight rendering of engineering BIM models based on model semantic information.

[0056] Step S7, when the three-dimensional field of view is scaled or translated, the viewing cone is recalculated according to the new camera position, and steps S5 to S6 are repeated to recalculate the tiles and semantic information of the tiles in the visible area to achieve rendering of the visible BIM components in the visible area.

[0057] Therefore, the visible nodes of the model are rendered based on WebGPU, and the amount of data for model rendering is reduced through two-level occlusion culling calculation to achieve lightweight rendering of the BIM model.

[0058] The present invention aims to provide a set of front-end lightweight rendering methods suitable for massive engineering BIM models. First, the semantic information (texture type, component size, inclusion relationship, etc.) of the engineering BIM model is extracted and stored; secondly, the octree structure is divided according to the minimum bounding volume of the BIM model, and the tiles at all levels are analyzed in combination with the semantic information, and the relevant semantic information (information such as inclusion relationship, semantic category, index number, etc.) is stored for each tile; then, when the front-end browser renders the BIM model, the network request of the model is reduced by the view frustum culling calculation, and the model rendering amount is reduced by the semantic culling calculation; finally, the lightweight rendering of the BIM model is realized in the browser by combining the view frustum culling and the semantic culling calculation. The method can improve the efficiency of the octree index calculation and the simplified calculation of the model rendering of the engineering BIM model, reduce the amount of data of the browser rendering the BIM model at the same time, improve the utilization rate of the hardware such as the GPU, and load more BIM models at the same time under the same hardware configuration conditions, and support the rendering application of massive BIM models.

[0059] An embodiment is described below:

[0060] Step 1: Analyze the semantic information of the engineering BIM model

[0061] a. Parse the name keywords in the semantics of the engineering BIM model, and determine the association or inclusion relationship of multiple BIM components through the semantic association of BIM component naming or grouping information. If multiple BIM components express the same feature, calculate the BIM component outline by treating the multiple BIM components as a whole BIM component through semantic association, and obtain the BIM component outline of each feature group.

[0062] b. Extract the model type, BIM component grouping information, BIM component inclusion relationship, model index, BIM component geometry, BIM component texture material information, BIM component design parameters, management attributes and other information of the BIM model to form an initial model group semantic library;

[0063] Step 2: BIM model tile space segmentation combined with engineering BIM model semantic information

[0064] a. Analyze the BIM model and calculate the minimum bounding volume outline, which is the 0th level tile;

[0065] The length, width and height of the minimum bounding volume outline are divided into two parts to form tiles at each level. When the BIM model is divided into spatial octrees, if the tile of an octree node does not contain any BIM model triangle, the node is filtered out. During the octree division process, the order of magnitude of each tile is judged. If the size of the BIM model in the tile is less than the size threshold or the total number of facets is less than the facet threshold, the octree division is stopped, otherwise the division continues, so that the model octree optimization structure R can be obtained. A ;

[0066] Step 3: During the octree partitioning process, the BIM components contained in each tile are parsed in real time. First, the multiple component patches of the same BIM component can be identified according to the semantic grouping information and semantic naming information of the BIM component. Then, the octree tile is calculated according to the patch coordinates, and the serial numbers of all tiles of the same BIM component are known. Thirdly, the semantic inclusion relationship of the BIM model is used to parse whether there is an inclusion relationship between multiple BIM components in the tile. If BIM component X contains other BIM components Y, and according to the semantic material information of the BIM model, if there is a transparent material, BIM component X and BIM component Y are recorded as a visible inclusion relationship. Otherwise, BIM component X and BIM component Y are recorded as an invisible inclusion relationship. Other model relationships are recorded as no relationship by default, thereby obtaining the semantic information R of tiles at all levels. DB ;

[0067] c. Model octree optimization structure R A and the semantic information of the tile R DB Use the same tile name but different suffixes for storage, thereby achieving a logical association between the two.

[0068] Step 4: Request the tile data of the BIM model through WebGPU technology in the browser and read the model octree optimization structure R A The 0th level tile is calculated, and the correct coordinates and azimuth of the 0th level tile in the three-dimensional scene are calculated. The initial position of the operation model is recorded in the computer memory;

[0069] Step 5: Space culling calculation based on the viewing cone

[0070] a. Calculate the viewing cone according to the camera position of the 3D scene. The viewing cone is surrounded by six planes: the near plane z1, the far plane z2, the top plane y1, the bottom plane y2, the left side x1 and the right side x2. At the same time, according to the definition of the 3D rendering camera, the closest distance between the rendering camera and the model is n, the farthest distance between the rendering camera and the model is f, and the horizontal field of view of the camera is FOV. x , Camera vertical field of view FOV y

[0071] Calculation of frustum:

[0072] Near plane z1: The plane established at the position n closest to the model;

[0073] Far plane z2: The plane established at the position farthest from the model by distance f;

[0074] The coordinate range of the top plane y1 is n / tan(FOV y / 2)~f / tan(FOV y / 2)

[0075] The coordinate range of the bottom plane y2 is -n / tan(FOV y / 2)~-f / tan(FOV y / 2)

[0076] The coordinate range of the left side x1 is -n / tan(FOV x / 2)~-f / tan(FOV x / 2)

[0077] The coordinate range of the right side x2 is n / tan(FOV x / 2)~F / tan(FOV x / 2)

[0078] b. Each level of the model octree tile corresponds to a different model rendering accuracy. The corresponding octree level Tree-n can be known according to the distance between the camera and the model. The coordinate range of the visible area can be known by intersecting the frustum with the 0th level tile. The model octree optimization structure R is traversed A The face patches of BIM components of all tiles under the Tree-n level, if the coordinates of the BIM component face patches are within the coordinate range of the visible area, the face patches of the BIM components of the corresponding node model number of the corresponding level are requested through the network to reduce the network request volume of the model;

[0079] Step 6: Semantic elimination calculation based on model semantics

[0080] When requesting a tile, we obtain the semantic information R of the tile. DB In the above example, the semantic information Linfo-n of the corresponding tile is requested. According to the semantic information Linfo-n, the BIM components that are not rendered are automatically filtered out during 3D rendering, and only other BIM components are rendered to reduce the amount of data for model rendering.

[0081] Step 7. When the 3D field of view is scaled or translated, the viewing cone is recalculated according to the new camera position, and steps 5 to 6 are repeated to recalculate the model octree node TreeN and semantic information LinfoN of the visible area. The node model is rendered based on WebGPU technology. The two-level culling calculation reduces the amount of data of the network transmission model and the amount of models rendered at the same time, thereby realizing lightweight rendering of the BIM model.

[0082] Compared with ordinary three-dimensional models, BIM models have more information such as grouping and management attributes, and the logical relationship of the model can be judged with the help of model semantic information. WebGPU technology has technical characteristics such as multi-threaded parallel computing, high-performance graphics rendering, and low communication overhead, which can better support the spatial operation of octree tiles. The method of the present invention innovatively involves the semantic information of the BIM model in the spatial occlusion calculation, and uses the efficient multi-threaded computing technology of WebGPU to realize the visual cone culling and semantic culling calculation of the BIM model, reducing the model network request volume and the amount of data required for rendering, improving the speed and accuracy of spatial operations, and realizing lightweight rendering of BIM models. Under the same hardware conditions, it can carry more three-dimensional model rendering, reduce the difficulty of using BIM models, and help promote the use of BIM models.

[0083] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be considered as the scope of protection of the present invention.

Claims

1. A lightweight rendering method for an engineering BIM model based on model semantic information, characterized in that: The following steps are involved: Step S1, the engineering BIM model includes a plurality of BIM components with a hierarchical relationship; Parsing the engineering BIM model to obtain semantic information of the engineering BIM model; wherein the semantic information of the engineering BIM model includes: hierarchical inclusion relationships between BIM components, texture material information of BIM components, and geometric structures of BIM components; Step S2, combining the semantic information of the engineering BIM model, performing octree space segmentation on the engineering BIM model, and establishing a model octree optimization structure of the engineering BIM model; the model octree optimization structure is a multi-level tile structure corresponding to the outline of the engineering BIM model; Step S3, obtaining semantic information of each tile at each level, including relevant face information of the BIM components directly contained in the tile and BIM component inclusion relationship information; The related face information of the BIM component refers to the face information of the BIM component located inside the tile; the BIM component inclusion relationship information refers to: for the BIM component A directly contained in the tile, if it does not further contain other BIM components inside, it is recorded that the BIM component A does not have an inclusion relationship; if it further contains BIM component B inside, the material of BIM component A is further determined according to the texture material information of the BIM component, and if there is a transparent material, it is recorded that BIM component A and BIM component B are in a visible inclusion relationship; otherwise, it is recorded that BIM component A and BIM component B are in an invisible inclusion relationship; Step S4, concurrently requesting tiles of all levels of the engineering BIM model in the browser through WebGPU technology, and calculating the position of the 0th level tile in the three-dimensional scene as the initial position; storing the tiles of all levels with a hierarchical relationship and the initial position; Step S5, spatial culling based on the viewing frustum: When rendering is required, firstly, tiles of all levels with a hierarchical relationship are loaded, and displayed in the three-dimensional scene according to the initial position of the 0th level tile; Obtain the view cone according to the camera position of the three-dimensional scene, and calculate the intersection boundary between the view cone and tiles at all levels; according to the intersection boundary, determine the tiles at all levels located inside the intersection boundary as the tiles at all levels located in the visible area, and remove the tiles of other levels located in the invisible area; Step S6, semantic elimination based on model semantics: According to the semantic information of tiles at all levels located in the visible area, the relevant facets of the BIM components directly contained in the tiles are rendered. At the same time, for the directly contained BIM components, if there is a BIM component B with an invisible relationship inside it, BIM component B is automatically filtered and not rendered. If there is a BIM component B with a visible relationship inside it, BIM component B is rendered; This enables lightweight rendering of engineering BIM models based on model semantic information.

2. The lightweight rendering method of an engineering BIM model based on model semantic information according to claim 1, characterized in that: The semantic information of the engineering BIM model also includes BIM component grouping information; the BIM component grouping information means: if multiple BIM components express the same feature, the multiple BIM components are regarded as an integral BIM component through semantic association, the BIM component outline is determined through the integral BIM component, and then its association relationship with the tile is determined.

3. The lightweight rendering method of an engineering BIM model based on model semantic information according to claim 1, characterized in that: Step S2 is specifically as follows: Step S2.1, determining the outline of the engineering BIM model according to the geometric structure of the engineering BIM model, and then establishing the minimum bounding volume of the engineering BIM model, which is the 0th level tile; Step S2.2, dividing the 0th level tile into 8 sub-tiles of the same size. If a sub-tile does not contain any face of the BIM model, the sub-tile is removed; the remaining sub-tiles are sub-tiles of the 0th level tile and are also 1st level tiles; Step S2.3, for each first-level tile, if the size of the BIM model it contains is smaller than the size threshold, or if the number of faces of the BIM model it contains is smaller than the face number threshold, stop dividing it into the next level; Otherwise, continue to divide it into 8 sub-tiles of the same size, and continue in this way, so as to establish the model octree optimization structure of the engineering BIM model.

4. The lightweight rendering method of an engineering BIM model based on model semantic information according to claim 1, characterized in that: Also includes: Step S7, when the three-dimensional field of view is scaled or translated, the viewing cone is recalculated according to the new camera position, and steps S5 to S6 are repeated to recalculate the tiles and semantic information of the tiles in the visible area to achieve rendering of the visible BIM components in the visible area.

Images