A lightweight method and system for a highway engineering three-dimensional real scene model and a medium

Abstract

The application discloses a lightweight method and system of a highway engineering three-dimensional real scene model and a medium, and relates to the technical field of three-dimensional real scene modeling. The method first performs engineering semantic segmentation on an input three-dimensional grid model and extracts design parameters in BIM / GIS. Then, a multi-level constraint model is constructed, which integrates semantic constraints, engineering precision constraints and design parameter constraints. Finally, an improved secondary error measurement algorithm is used to iteratively simplify under the guidance and restriction of the multi-level constraint model. The application solves the problems of semantic information loss, local precision out of control and disconnection with design intent caused by the application of traditional pure geometric simplification methods in highway engineering, and can intelligently retain key engineering features, control geometric errors and conform to design forms while greatly reducing model data, thereby generating a high-quality lightweight model suitable for professional analysis and application.

Description

Technical Field

[0001] This invention relates to the field of 3D reality modeling technology, specifically to a lightweight method, system, and medium for creating 3D reality models of highway engineering. Background Technology

[0002] With the increasing maturity of 3D reality reconstruction technologies such as UAV oblique photogrammetry and LiDAR scanning, the highway engineering industry can quickly and cost-effectively acquire large-scale, high-precision 3D reality models of the ground surface and structures.

[0003] However, high-precision 3D reality models are usually composed of a massive number of triangular meshes. A medium-sized road segment model may contain hundreds of millions or even billions of triangular meshes, causing the model to load slowly, render lag, or even fail to open on personal computers, web browsers, and mobile devices, which seriously restricts its popularization and efficient collaboration in daily engineering applications.

[0004] To address the issue of excessively large model data volumes, various mesh simplification algorithms have been proposed in the field of computer graphics. Among them, the Quadratic Error Metric (QEM) algorithm is one of the most classic and widely used. The core idea of ​​the QEM algorithm is to remove vertices and faces by iteratively folding edges in the mesh. Each fold prioritizes selecting the edge that has the least impact on the overall geometry of the model, i.e., minimizing the sum of squared geometric errors introduced after folding. This algorithm is a purely geometry-driven process, and the optimal solution is to minimize the change in the overall geometry of the model.

[0005] However, when the traditional QEM algorithm is directly applied to highly specialized highway engineering scenarios, the following serious defects are exposed.

[0006] (1) Loss of semantic information: The QEM algorithm treats all vertices equally and cannot understand the engineering semantics of the model. Therefore, it may easily collapse the vertices representing the contours of key elements such as "lane lines" and "curb stones", causing the boundaries of these elements to become blurred, jagged or even disappear completely, resulting in serious distortion of the semantic information of the model. This makes the simplified model unusable in applications that require clear identification of these elements (such as traffic facility inspection and design review).

[0007] (2) Loss of control over engineering accuracy: The QEM algorithm aims to minimize the overall geometric error, but it cannot guarantee the accuracy of key local areas. For example, in earthwork calculation, small changes in road surface elevation can accumulate into huge engineering errors. Traditional QEM simplification may lead to elevation or planar displacement in some areas far exceeding the allowable error range of the project, causing the lightweight model to lose its value for quantitative engineering analysis.

[0008] (3) Disconnect from design intent: The real-world model is a capture of the current situation and may include irrelevant details such as noise and vegetation, while also reflecting the design form. Traditional algorithms cannot distinguish which design features should be retained (standard cross-sections of roads, arch ribs of bridges, etc.) and which are redundant noises that can be removed. The simplified model may have uncontrollable deviations from the original BIM design model in key lines, making it impossible to conduct an effective "design-real-world" comparison.

[0009] Therefore, there is an urgent need for a new lightweight technology that can understand the engineering scenario and meet professional requirements for 3D reality models of highway engineering. Summary of the Invention

[0010] This invention provides a lightweight method, system, and medium for creating a 3D reality model of highway engineering, in order to solve the technical problems that arise when applying existing mesh simplification algorithms to 3D reality models of highway engineering, such as loss of semantic information, inability to guarantee engineering accuracy in key local areas, and serious discrepancies between the simplified results and the original design intent.

[0011] This invention is achieved through the following technical solution:

[0012] A first aspect of the present invention provides a lightweight method for creating a three-dimensional reality model of a highway engineering project, comprising:

[0013] Obtain the highway engineering triangular mesh model, as well as the corresponding BIM design model and GIS database. The highway engineering triangular mesh model is derived from real-scene 3D reconstruction or a fine BIM model.

[0014] Semantic segmentation processing is performed on the highway engineering triangular mesh model, and an engineering semantic label is assigned to each mesh cell.

[0015] Key design parameters are extracted from the BIM design model and GIS database to obtain a set of key design parameters;

[0016] Based on the engineering semantic tags and the key design parameter set, a multi-level constraint model is constructed, which includes a semantic constraint layer, an engineering accuracy constraint layer, and a design parameter constraint layer.

[0017] An improved quadratic error metric algorithm is used to perform iterative simplification on the triangular mesh model of the highway engineering. The improved quadratic error metric algorithm is as follows: in each decision of the edge folding candidate operation, the edge folding candidate operation is evaluated according to the multi-level constraint model. If the multi-level constraint conditions are met, the edge folding candidate operation is executed.

[0018] When the preset simplification termination condition is met, the iteration stops and the simplified lightweight 3D reality model is output.

[0019] Furthermore, the step of performing semantic segmentation processing on the highway engineering triangular mesh model includes:

[0020] The highway engineering triangular mesh model is automatically segmented using a deep learning-based 3D semantic segmentation network, and each mesh cell is labeled with an engineering semantic label based on the automatic segmentation results; or...

[0021] Based on the rule-based method of geometric features and spatial relationships, the grid cells in the triangular mesh model are divided into at least one engineering semantic category, and each grid cell is labeled with an engineering semantic label according to the division result.

[0022] The engineering semantic categories include asphalt pavement, roadbed, lane lines, curbstone, guardrail, bridge bearing, slope, and greening.

[0023] Furthermore, the key design parameters include: road centerline equation, design longitudinal section, standard cross section, design coordinates of key control points, and LOD level of components.

[0024] Furthermore, the semantic constraint layer is configured to evaluate at least one of the following constraints on the edge folding candidate operation:

[0025] Based on a predefined semantic protection list, a maximum value is appended to the folding cost of edges belonging to engineering semantic tags in the semantic protection list to prevent folding operations from being performed on them.

[0026] The simplification level of each edge is adjusted based on the simplification weight coefficient assigned to the semantic labels of each project; where a larger simplification weight coefficient indicates a higher degree of simplification allowed.

[0027] Furthermore, the engineering accuracy constraint layer is configured to evaluate at least one of the following constraints for the edge folding candidate operation:

[0028] Calculate the local area elevation change caused by performing the edge folding candidate operation. If the local area elevation change exceeds the first engineering tolerance threshold, then prohibit the edge folding candidate operation.

[0029] Calculate the planar displacement caused by performing the edge folding candidate operation. If the planar displacement exceeds the second engineering tolerance threshold, then prohibit the edge folding candidate operation.

[0030] Furthermore, the engineering accuracy constraint layer is also configured to evaluate the following constraints for the edge folding candidate operation:

[0031] When calculating the initial geometric error of an edge, a local curvature factor is introduced for the region where the edge is located, which amplifies the folding cost of edges in high curvature regions.

[0032] Furthermore, the design parameter constraint layer is configured to evaluate at least one of the following constraints for the edge folding candidate operation:

[0033] Calculate the minimum distance between the new vertex formed after performing the edge folding candidate operation and the corresponding geometric element in the design model to obtain the design fit distance, and prohibit edge folding candidate operations that cause the design fit distance to exceed the preset maximum deviation distance;

[0034] Calculate the minimum distance between the new vertex formed after performing the edge folding candidate operation and the corresponding geometric element in the design model to obtain the design fit distance, and add the design fit distance as a penalty term to the edge folding cost;

[0035] Based on the LOD level of the components, different maximum simplification rate thresholds are set for mesh regions of different LOD levels, and the maximum simplification rate thresholds are ensured not to be exceeded during the simplification process.

[0036] Furthermore, the improved quadratic error metric simplification algorithm calculates the edge folding cost using the following formula:

[0037] ;

[0038] in, To represent the geometric error of the simplified algorithm using a quadratic error metric, Represents the simplified weight coefficients of the edges. This represents the local curvature factor of the region containing the edge. Indicates the design fit distance. Indicates the weight of design fit. Indicates a penalty item. Represents the cost of edge folding;

[0039] If the edge folding candidate operation violates any one of the semantic constraint layer, engineering accuracy constraint layer, and design parameter constraint layer, then its folding cost is set to infinity.

[0040] During the iterative simplification of the highway engineering triangular mesh model, each iteration selects the folding cost. Perform a fold operation on the edge with the smallest value and update the fold cost of the affected edge until the simplified termination condition is met.

[0041] A second aspect of the present invention provides a lightweight system for three-dimensional reality models of highway engineering, comprising:

[0042] The data input module is used to acquire the highway engineering triangular mesh model and the corresponding BIM design model and GIS database. The highway engineering triangular mesh model is derived from real-scene 3D reconstruction or fine BIM model.

[0043] The semantic analysis module is used to perform semantic segmentation processing on the highway engineering triangular mesh model and assign engineering semantic labels to each mesh cell.

[0044] The parameter extraction module is used to extract key design parameters from the BIM design model and / or GIS database to obtain a set of key design parameters;

[0045] The constraint fusion module is used to construct a multi-level constraint model containing a semantic constraint layer, an engineering accuracy constraint layer, and a design parameter constraint layer based on the engineering semantic tags and the key design parameter set.

[0046] The model simplification module is used to perform iterative simplification on the highway engineering triangular mesh model using an improved quadratic error metric algorithm. The improved quadratic error metric algorithm is as follows: in each decision of the edge folding candidate operation, the edge folding candidate operation is evaluated according to the multi-level constraint model. If the multi-level constraint conditions are met, the edge folding candidate operation is executed.

[0047] The results output module is used to stop the iteration when the preset simplification termination condition is met, and output the simplified lightweight 3D reality model.

[0048] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the lightweight method for three-dimensional real-scene models of highway engineering as described in any one of the first aspects of the present invention.

[0049] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0050] This invention is an improvement on the classic QEM algorithm. By combining a multi-level constraint model that incorporates semantics, engineering accuracy, and design parameter constraints, it can significantly reduce the amount of model data while intelligently preserving key engineering semantic features, strictly controlling geometric deformation within the allowable error range, and ensuring a high degree of consistency between the simplified model and the design model.

[0051] By introducing a semantic constraint layer into the simplified decision-making process, this invention can automatically identify and prioritize the protection of the geometric contours of key engineering elements such as lane lines, curbs, and guardrails, effectively avoiding the semantic information loss problem of the QEM algorithm, and ensuring that the simplified model maintains engineering recognizability both visually and structurally.

[0052] By introducing an engineering accuracy constraint layer, the model scale deviation during the simplification process is strictly limited by hard constraints, ensuring the geometric accuracy of the lightweight model in key areas, meeting the requirements of engineering measurement and deformation analysis, and upgrading the simplified model into analyzable engineering data.

[0053] By introducing a design parameter constraint layer and using BIM design parameters to guide the simplification process, noise in the real-world model can be effectively removed and features that conform to the design form can be strengthened. This allows the simplified real-world model to be compared and analyzed with the BIM design model with high precision, providing a high-quality data foundation for digital twin applications.

[0054] By using multi-level constraints, a large number of infeasible folding operations are eliminated in the early stages of simplification, reducing invalid calculations and improving the overall efficiency of the simplification process while ensuring quality. Attached Figure Description

[0055] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0056] Figure 1 This is a flowchart of a method for lightweighting a three-dimensional real-scene model of a highway engineering project according to an embodiment of the present invention;

[0057] Figure 2 This is a schematic diagram illustrating an iterative simplification based on an improved QEM algorithm according to an embodiment of the present invention;

[0058] Figure 3 This is a block diagram of a lightweight system for a three-dimensional real-scene model of a highway engineering project according to an embodiment of the present invention;

[0059] Figure 4 This is a schematic diagram illustrating the application of a lightweight method according to an embodiment of the present invention. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of this invention are only for explaining this invention and are not intended to limit this invention.

[0061] It should be noted that the terms "comprising" and "having" and any variations thereof in the specification, claims, and accompanying drawings of this invention are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to other steps or units inherent in the device.

[0062] The terminology used in the various embodiments of the invention is for the purpose of describing particular embodiments only and is not intended to limit the various embodiments of the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of the invention pertain. The terms (such as those defined in commonly used dictionaries) are to be interpreted as having the same meaning as in the context of the relevant technical field and are not to be interpreted as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of the invention.

[0063] This invention aims to address the shortcomings of existing quadratic error measurement algorithms by proposing an improved quadratic error measurement algorithm that combines semantic and multi-constraint fusion, applied to the lightweighting of 3D reality models for highway engineering. The core objective of this invention is to intelligently preserve the engineering semantic features of the model, strictly control the geometric accuracy of key locations, and ensure that the simplified result remains consistent with the original design intent while significantly reducing the model's data volume. This achieves an intelligent lightweighting solution that moves from simple geometric simplification to semantic and engineering fidelity, generating a simplified 3D solid model that balances lightweight design and accuracy, suitable for professional analysis.

[0064] Please see Figure 1 The present invention proposes a lightweight method for a three-dimensional real-scene model of highway engineering, which includes the following steps S100~S500.

[0065] S100: Obtain the triangular mesh model of the highway project, as well as the corresponding BIM design model and GIS database. The triangular mesh model of the highway project is derived from real-scene 3D reconstruction or a detailed BIM model.

[0066] Obtain the triangular mesh model M of the highway engineering project to be simplified. This model M is usually derived from real-scene 3D reconstruction technologies such as UAV oblique photogrammetry and ground-based LiDAR scanning, or it can be converted from a high-level-of-detail (LOD) fine BIM model. At the same time, obtain the BIM design model corresponding to this road section or structure, as well as the GIS database.

[0067] The BIM design model contains design information such as road centerline, design longitudinal section, standard cross section, and bridge bearing coordinates, while the GIS database contains data such as road section administrative divisions and topographic contour lines.

[0068] Among them, the BIM design model, GIS data and triangular mesh model M have a coordinate correspondence in space. Spatial alignment can be achieved through coordinate transformation or feature point matching to ensure the spatial consistency of subsequent constraint calculations. In subsequent processing, the design data and the real scene model can be integrated.

[0069] This step provides the data foundation for subsequent model simplification processing.

[0070] S200 performs semantic segmentation on the triangular mesh model of highway engineering and assigns an engineering semantic label to each mesh cell; it extracts key design parameters from the BIM design model and / or GIS database to obtain a set of key design parameters.

[0071] Among them, engineering semantic tags include, but are not limited to, asphalt pavement, roadbed, lane lines, curb stones, guardrails, bridge bearings, slopes, and greening.

[0072] In this step, the segmentation method can employ deep learning-based segmentation techniques, such as pre-trained 3D semantic segmentation networks like PointNet++ and RandLA-Net. The triangular mesh model M is automatically segmented using the pre-trained 3D semantic segmentation network. Based on the automatic segmentation results, the component type of each mesh unit (triangular facets and vertices) is obtained, and engineering semantic labels are assigned to each triangular facet and vertex.

[0073] The segmentation method can also be based on a rule-based segmentation engine, which divides the mesh according to the geometric features (such as normal vectors and elevations) and spatial relationships of the triangular mesh model M, divides each mesh unit into different semantic regions, and finally selects the engineering semantic label for each mesh unit based on the segmentation results.

[0074] Key design parameters are obtained by parsing BIM design models or GIS databases. These parameters include, but are not limited to:

[0075] Road centerline equations: such as straight line segments, circular curve segments, and transition curve segments;

[0076] Design longitudinal profile: includes the design elevation at each station;

[0077] Standard cross section: Standard cross section template and its unfolded form at a specific station, including the road camber curve equation;

[0078] Design coordinates of key control points: such as the center of bridge piers, road boundary points, and slope toe lines;

[0079] LOD level of a component: The level of detail information defined in the design phase for each major component (such as road surface, guardrail, bridge). For example, LOD 300 indicates that the component has precise geometry and dimensions, while LOD 200 indicates that it is a general system or component with approximate shape and size.

[0080] The extracted key parameters are combined to form the key design parameter set P.

[0081] S300, based on engineering semantic tags and key design parameter set P, constructs a multi-level constraint model including semantic constraint layer, engineering accuracy constraint layer and design parameter constraint layer.

[0082] The core improvement of this invention lies in constructing a three-layer filtering constraint mechanism. During the execution of the QEM algorithm, the constraints of multi-level constraint models are integrated to establish a triple filtering mechanism for each edge folding candidate operation in the simplification process. This includes edge folding constraints based on engineering semantic labels, edge folding constraints based on engineering accuracy, and edge folding constraints based on design parameters. This achieves a significant reduction in model data volume while intelligently retaining key engineering semantic features, strictly controlling geometric deformation within the allowable error range of engineering, and ensuring a high degree of consistency between the simplified model and the design model.

[0083] Specifically, by defining a semantic protection list, such as a list of key protected elements {white lane lines, yellow lane lines, curb edges}, any edge is prohibited from being collapsed if the semantic label of either of its two endpoints belongs to this list.

[0084] In a preferred embodiment, the semantic constraint layer is configured to append a maximum value to the folding cost of edges whose engineering semantic labels belong to the semantic protection list. That is, during each iteration simplification of the QEM algorithm, when calculating the folding cost of the current edge folding candidate operation, if the engineering semantic label of any endpoint of the edge is included in the semantic protection list, a maximum semantic penalty term is appended to its folding cost, such as... The cost of folding becomes infinite, thus preventing it from being selected and prohibiting the folding operation on that side.

[0085] In a preferred embodiment, a semantically simplified weight matrix is ​​defined. Assign a simplified weight coefficient with a positive real number to each semantic label. For example, set... (Slope) = 2.0, (Asphalt pavement) = 1.0, (Bridge bearing) = 0.1. The larger the simplification weight coefficient, the lower the sensitivity of the semantic region to geometric changes, allowing for a greater degree of simplification.

[0086] Furthermore, the semantic constraint layer is configured to adjust the simplification level of each edge based on the simplification weight coefficient assigned to each project semantic label. By assigning corresponding simplification weight coefficients to different project semantic labels, the folding cost of the edge after adjustment by the simplification weight coefficient is calculated when calculating the folding cost. The larger the simplification weight coefficient, the smaller the folding cost and the higher the allowed degree of simplification.

[0087] Preferably, in the semantic constraint layer, semantic protection list constraints and simplified weight coefficient constraints can be performed simultaneously. That is, the semantic constraint layer is configured to execute the following two constraints simultaneously during the cost evaluation of edge folding candidate operations:

[0088] (1) Based on the predefined semantic protection list, a maximum value is added to the folding cost of the edges of the engineering semantic tags belonging to the semantic protection list;

[0089] (2) Adjust the simplification degree of each side based on the simplification weight coefficient assigned to the semantic labels of each project.

[0090] In the first constraint, a semantic penalty term is added to the calculation of the folding cost of the QEM algorithm. The semantic penalty term is zero, but when the edge to be evaluated triggers the first constraint, it is assigned an infinite value.

[0091] Engineering accuracy constraints include elevation constraints and planar constraints. Specifically, the engineering accuracy constraint layer is configured to enforce at least one of the following constraints when performing cost evaluation on edge folding candidate operations:

[0092] (1) Calculate the local elevation change caused by the execution of the edge folding candidate operation. If the local elevation change exceeds the first engineering tolerance threshold, the edge folding candidate operation is prohibited.

[0093] (2) Calculate the planar displacement caused by the execution of the edge folding candidate operation. If the planar displacement exceeds the second engineering tolerance threshold, the edge folding candidate operation is prohibited.

[0094] For an edge e to be folded, its target folding point is: The algorithm predicts that this fold will occur in... The maximum possible elevation change that may be caused within and around the area is ,like If the preset elevation tolerance threshold is 2cm, then this folding is prohibited.

[0095] Similarly, calculate the maximum displacement in the XY plane that the folding may cause. ,like If the preset plane tolerance threshold is met, then this folding is prohibited.

[0096] Furthermore, the engineering accuracy constraints also include curvature constraints, and the engineering accuracy constraint layer is configured as follows:

[0097] (3) When calculating the initial geometric error of the edge, the local curvature factor of the region where the edge is located is introduced, so that the folding cost of the edge in the high curvature region is amplified.

[0098] For example, the principal curvature of the region containing the edge Let the local curvature factor of the region containing the edge be... (α is the magnification factor). The greater the curvature, the more the folding cost is magnified, thus protecting features such as curves and edges.

[0099] The design parameter constraints guide the design fit and calculate the new vertices after folding. Minimum distance to the corresponding geometric element in the BIM design model We take this minimum distance as the optimization objective, that is, the new vertex after folding. The closer the folded model is to the corresponding geometric element in the BIM design model, the better it fits the design model.

[0100] There are two strategies for design parameter constraints:

[0101] (1) Hard constraint: Set a maximum deviation distance ,like Folding is prohibited.

[0102] (2) Soft constraints: As a penalty term, it is added to the function for calculating the folding cost, which encourages the algorithm to prioritize simplifying points that deviate from the design model (which may be noise or irrelevant details), thereby protecting points that fit the design shape.

[0103] Accordingly, the design parameter constraint layer is configured to evaluate at least one of the following constraints for edge folding candidate operations:

[0104] (1) Calculate the minimum distance between the new vertex formed after performing the edge folding candidate operation and the corresponding geometric element in the design model to obtain the design fit distance. Edge folding candidate operations that cause the design fit distance to exceed the preset maximum deviation distance are prohibited.

[0105] (2) Calculate the minimum distance between the new vertex formed after performing the edge folding candidate operation and the corresponding geometric element in the BIM design model to obtain the design fit distance. Add the design fit distance as a penalty term to the edge folding cost.

[0106] Furthermore, the design parameter constraints also include BIM-LOD level mapping constraints, which set a maximum simplification rate threshold for the corresponding grid area based on the LOD level of the BIM component in the design. For example, the component simplification rate of LOD300 cannot exceed 50%, while the component simplification rate of LOD200 can be up to 80%.

[0107] The design parameter constraint layer is also configured to evaluate the following constraints for edge-folding candidate operations:

[0108] (3) Based on the LOD level of the component, set a different maximum simplification rate threshold for the grid area of ​​different LOD levels, and ensure that the maximum simplification rate threshold is not exceeded during the simplification process.

[0109] S400 employs an improved quadratic error metric (QEM) algorithm to perform iterative simplification on the triangular mesh model of highway engineering. The improved quadratic error metric algorithm is as follows: in each decision of edge folding candidate operation, the edge folding candidate operation is evaluated based on the multi-level constraint model. If the multi-level constraint conditions are met, the edge folding candidate operation is executed.

[0110] like Figure 2 As shown, the iterative simplification using the improved QEM algorithm includes the following steps.

[0111] S401 initializes a priority queue (heap) for each edge in the 3D reality model M, with the initial cost being the geometric error calculated by the classic QEM algorithm. .

[0112] S402, for each candidate edge to be folded, its folding cost is calculated by the following formula:

[0113]

[0114] In the formula, This represents the cost of folding an edge. Represents the simplified weight coefficients of the edges. This represents the local curvature factor of the region containing the edge. Indicates the design fit distance. Indicates the weight of design fit. This is a penalty item.

[0115] in, The term is a conditional function; if the folding operation of the edge violates any of the following: semantic protection list, hard constraints on engineering precision, or the maximum simplification rate threshold for the LOD level, then... (infinity), otherwise .

[0116] S403, retrieve the current folding cost from the priority queue. Find the smallest edge and determine if it is infinite.

[0117] S404, if If the value is infinity, it means that this edge is protected by constraints, so skip it and take the next edge.

[0118] If not, then perform an edge collapse operation.

[0119] S405, After collapsing, update the model topology and recalculate the affected edges according to step S402. .

[0120] S406, Repeat steps S403~S405 until the simplified termination condition is met.

[0121] S500 stops iterating when the preset simplification termination condition is met, and outputs a simplified lightweight 3D reality model.

[0122] A simplified termination condition can be set to reduce the total number of triangle faces in the model to a preset target value, or to reduce the number of edges in the priority queue to a certain threshold. All Or, until the preset maximum number of iterations is reached.

[0123] After the iteration stops, the final lightweight triangular mesh model is output. This model has a significantly reduced data size while strictly adhering to preset semantic, accuracy, and design constraints. This lightweight mesh model can be further re-associated with texture and attribute information to generate a lightweight 3D reality model that can be directly used for web publishing, mobile display, or loading into professional analysis software.

[0124] Based on the same inventive concept, this invention proposes a lightweight system for three-dimensional real-scene models of highway engineering, such as... Figure 3 As shown, the system includes:

[0125] The data input module 301 is used to acquire the highway engineering triangular mesh model and the corresponding BIM design model and GIS database. The highway engineering triangular mesh model is derived from real-scene 3D reconstruction or fine BIM model.

[0126] The semantic analysis module 302 is used to perform semantic segmentation processing on the triangular mesh model of highway engineering and assign engineering semantic labels to each mesh cell.

[0127] The parameter extraction module 303 is used to extract key design parameters from the BIM design model and GIS database to obtain a set of key design parameters.

[0128] The constraint fusion module 304 is used to construct a multi-level constraint model based on engineering semantic tags and key design parameter sets, which includes a semantic constraint layer, an engineering accuracy constraint layer, and a design parameter constraint layer.

[0129] Model simplification module 305 is used to perform iterative simplification of the highway engineering triangular mesh model using an improved quadratic error metric algorithm; wherein, the improved quadratic error metric algorithm is:

[0130] In each decision on an edge folding candidate operation, the edge folding candidate operation is evaluated based on a multi-level constraint model. If the multi-level constraint conditions are met, the edge folding candidate operation is executed.

[0131] The result output module 306 is used to stop the iteration and output a simplified lightweight 3D reality model when the preset simplification termination condition is reached.

[0132] Furthermore, the data input module 301 includes a first input submodule and a second input submodule, which respectively receive triangular mesh model, BIM design model and GIS data.

[0133] Furthermore, the semantic analysis module 302 integrates a deep learning model or rule engine for performing semantic segmentation.

[0134] Furthermore, the constraint fusion module 304 provides a user interface or configuration file for users to set the parameters and rules of constraints at each layer, and builds the constraint fusion model in memory.

[0135] Furthermore, the folding cost of the edge is calculated using the following formula:

[0136]

[0137] in, To represent the geometric error of the simplified algorithm using a quadratic error metric, Represents the simplified weight coefficients of the edges. This represents the local curvature factor of the region containing the edge. Indicates the design fit distance. Indicates the weight of design fit. Indicates a penalty item. This represents the cost of folding an edge.

[0138] A simplified process for an application example is as follows: Figure 4 As shown, the lightweight system of this invention was used to perform lightweight processing on a section of highway real-scene 3D model, and the process is as follows.

[0139] S1, Input: A 5-kilometer-long oblique photogrammetry model of a highway (200 million triangular facets) and the BIM design model of that section of the highway;

[0140] S2, Semantic Segmentation: The pre-trained RandLA-Net network was used to segment the grid, successfully dividing the model into parts such as "asphalt pavement", "white lane lines", "green guardrail", and "slope".

[0141] S3, Extract Design Parameters: Extract the road centerline coordinates, design longitudinal slope, and standard cross-section (including road camber curve) from the BIM model.

[0142] S4, Configuration Constraints:

[0143] Semantic constraints: Add "white lane lines" and "green guardrail edges" to the semantic protection list. Set semantic weights: (Slope) = 1.8 (Road surface) = 1.0, (Guardrail) = 0.5.

[0144] Accuracy constraints: Set elevation tolerance m, plane tolerance m.

[0145] Design constraints: Soft constraint pattern is adopted, and the following constraints are set... And set the maximum simplification rate of all BIM components to 70% according to their LOD level.

[0146] S4, Execute the simplification algorithm: Run the improved QEM algorithm, aiming to simplify to 20 million faces (simplification rate of 90%).

[0147] S5, Output: Lightweight 3D reality model, where:

[0148] The lane lines are clear and continuous, and the guardrail outlines are distinct (semantic constraints are in effect).

[0149] The elevation error of the entire road section is less than 3cm, which meets the requirements for earthwork calculation (accuracy constraints are in effect).

[0150] The simplified road cross slope matches the road crown curve designed in BIM with a very high degree of consistency (design constraints are in effect).

[0151] Secondary areas such as slopes have been simplified to the greatest extent possible, effectively reducing the amount of data.

[0152] Embodiments of the present invention also provide an electronic device, which includes a processor and a memory, wherein the number of processors may be one or more. The memory, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules. The processor executes various functional applications and data processing of the electronic device by running the software programs, instructions, and modules stored in the memory, thereby realizing the lightweight method for three-dimensional real-scene models of highway engineering according to any of the above embodiments of the present invention.

[0153] The memory may primarily comprise a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a given function; the data storage area may store data created based on terminal usage. Furthermore, the memory may include high-speed random access memory (RAM) and non-volatile memory, such as at least one disk storage device, flash memory, or other non-volatile solid-state storage device. In some instances, the memory may further include memory remotely located relative to the processor, which can be connected to the electronic device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks (LANs), mobile communication networks, and combinations thereof.

[0154] Embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the lightweight method for three-dimensional real-scene models of highway engineering according to any embodiment of the present invention.

[0155] The computer storage medium of this invention can be any combination of one or more computer-readable media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0156] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0157] Embodiments of the present invention also provide a computer program product that, when run on a computer, causes the computer to execute the lightweight method for three-dimensional real-scene models of highway engineering according to any of the above embodiments of the present invention.

[0158] The above embodiments are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the appended claims.

Claims

1. A lightweight method for creating a 3D reality model of a highway engineering project, characterized in that, include: Obtain the highway engineering triangular mesh model, as well as the corresponding BIM design model and GIS database. The highway engineering triangular mesh model is derived from real-scene 3D reconstruction or a fine BIM model. Semantic segmentation processing is performed on the highway engineering triangular mesh model, and an engineering semantic label is assigned to each mesh cell. Key design parameters are extracted from the BIM design model and GIS database to obtain a set of key design parameters; Based on the engineering semantic tags and the key design parameter set, a multi-level constraint model is constructed, which includes a semantic constraint layer, an engineering accuracy constraint layer, and a design parameter constraint layer. An improved quadratic error metric algorithm is used to perform iterative simplification on the triangular mesh model of the highway engineering. The improved quadratic error metric algorithm is as follows: in each decision of the edge folding candidate operation, the edge folding candidate operation is evaluated according to the multi-level constraint model. If the multi-level constraint conditions are met, the edge folding candidate operation is executed. When the preset simplification termination condition is met, the iteration stops and the simplified lightweight 3D reality model is output. The improved quadratic error metric simplification algorithm calculates the edge folding cost using the following formula: ； in, To represent the geometric error of the simplified algorithm using a quadratic error metric, Represents the simplified weight coefficients of the edges. This represents the local curvature factor of the region containing the edge. Indicates the design fit distance. Indicates the weight of design fit. Indicates a penalty item. Represents the cost of edge folding; If the edge folding candidate operation violates any one of the semantic constraint layer, engineering accuracy constraint layer, and design parameter constraint layer, then its folding cost is set to infinity. During the iterative simplification of the highway engineering triangular mesh model, each iteration selects the folding cost. Perform a fold operation on the edge with the smallest value and update the fold cost of the affected edge until the simplified termination condition is met.

2. The lightweight method for a three-dimensional real-scene model of highway engineering according to claim 1, characterized in that, The steps for performing semantic segmentation processing on the highway engineering triangular mesh model include: The highway engineering triangular mesh model is automatically segmented using a deep learning-based 3D semantic segmentation network, and each mesh cell is labeled with an engineering semantic label based on the automatic segmentation results; or... Based on the rule-based method of geometric features and spatial relationships, the grid cells in the triangular mesh model are divided into at least one engineering semantic category, and each grid cell is labeled with an engineering semantic label according to the division result. The engineering semantic categories include asphalt pavement, roadbed, lane lines, curbstone, guardrail, bridge bearing, slope, and greening.

3. The lightweight method for a three-dimensional real-scene model of highway engineering according to claim 1, characterized in that, The key design parameters include: road centerline equation, design longitudinal section, standard cross section, design coordinates of key control points, and LOD level of components.

4. The lightweight method for a three-dimensional real-scene model of highway engineering according to claim 1, characterized in that, The semantic constraint layer is configured to evaluate at least one of the following constraints on the edge folding candidate operation: Based on a predefined semantic protection list, a maximum value is appended to the folding cost of edges belonging to engineering semantic tags in the semantic protection list to prevent folding operations from being performed on them. The simplification level of each edge is adjusted based on the simplification weight coefficient assigned to the semantic labels of each project; where a larger simplification weight coefficient indicates a higher degree of simplification allowed.

5. The lightweight method for a three-dimensional real-scene model of highway engineering according to claim 4, characterized in that, The engineering accuracy constraint layer is configured to evaluate at least one of the following constraints for the edge folding candidate operation: Calculate the local area elevation change caused by performing the edge folding candidate operation. If the local area elevation change exceeds the first engineering tolerance threshold, then prohibit the edge folding candidate operation. Calculate the planar displacement caused by performing the edge folding candidate operation. If the planar displacement exceeds the second engineering tolerance threshold, then prohibit the edge folding candidate operation.

6. The lightweight method for a three-dimensional real-scene model of highway engineering according to claim 5, characterized in that, The engineering accuracy constraint layer is also configured to evaluate the following constraints for the edge folding candidate operations: When calculating the initial geometric error of an edge, a local curvature factor is introduced for the region where the edge is located, which amplifies the folding cost of edges in high curvature regions.

7. The lightweight method for a three-dimensional real-scene model of highway engineering according to claim 1, characterized in that, The design parameter constraint layer is configured to evaluate at least one of the following constraints for the edge folding candidate operation: Calculate the minimum distance between the new vertex formed after performing the edge folding candidate operation and the corresponding geometric element in the design model to obtain the design fit distance, and prohibit edge folding candidate operations that cause the design fit distance to exceed the preset maximum deviation distance; Calculate the minimum distance between the new vertex formed after performing the edge folding candidate operation and the corresponding geometric element in the BIM design model to obtain the design fit distance, and add the design fit distance as a penalty term to the edge folding cost; Based on the LOD level of the components, different maximum simplification rate thresholds are set for mesh regions of different LOD levels, and the maximum simplification rate thresholds are ensured not to be exceeded during the simplification process.

8. A lightweight system for a three-dimensional real-scene model of a highway engineering project, characterized in that, include: The data input module is used to acquire the highway engineering triangular mesh model and the corresponding BIM design model and GIS database. The highway engineering triangular mesh model is derived from real-scene 3D reconstruction or fine BIM model. The semantic analysis module is used to perform semantic segmentation processing on the highway engineering triangular mesh model and assign engineering semantic labels to each mesh cell. The parameter extraction module is used to extract key design parameters from the BIM design model and GIS database to obtain a set of key design parameters; The constraint fusion module is used to construct a multi-level constraint model containing a semantic constraint layer, an engineering accuracy constraint layer, and a design parameter constraint layer based on the engineering semantic tags and the key design parameter set. The model simplification module is used to perform iterative simplification on the highway engineering triangular mesh model using an improved quadratic error metric algorithm. The improved quadratic error metric algorithm is as follows: in each decision of the edge folding candidate operation, the edge folding candidate operation is evaluated according to the multi-level constraint model. If the multi-level constraint conditions are met, the edge folding candidate operation is executed. The results output module is used to stop the iteration when the preset simplification termination condition is met, and output the simplified lightweight 3D reality model. The improved quadratic error metric simplification algorithm calculates the edge folding cost using the following formula: ； in, To represent the geometric error of the simplified algorithm using a quadratic error metric, Represents the simplified weight coefficients of the edges. This represents the local curvature factor of the region containing the edge. Indicates the design fit distance. Indicates the weight of design fit. Indicates a penalty item. Represents the cost of edge folding; If the edge folding candidate operation violates any one of the semantic constraint layer, engineering accuracy constraint layer, and design parameter constraint layer, then its folding cost is set to infinity. During the iterative simplification of the highway engineering triangular mesh model, each iteration selects the folding cost. Perform a fold operation on the edge with the smallest value and update the fold cost of the affected edge until the simplified termination condition is met.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the lightweight method for the three-dimensional reality model of highway engineering as described in any one of claims 1-7.

Images