A simplified method and system for urban rail engineering components

Abstract

This invention relates to the field of spatial modeling technology and discloses a method and system for simplifying urban rail engineering components. Existing technologies for global uniform polygon reduction lack the ability to distinguish the distribution characteristics of component face count and surface area. Manual parameter settings are difficult to achieve adaptive dynamic adjustment, and single-tool polygon reduction lacks a phased monitoring and feedback mechanism. This leads to insufficient reduction of redundancy in high-face-count components and excessive loss of key details in low-face-count components during the polygon reduction process, easily causing face count rebound or mesh topology distortion. This invention uses component face count priority ranking and surface area statistical indicators to drive dynamic polygon reduction control, combined with phased iterative termination judgment. It efficiently reduces the total face count of the model while adaptively protecting key details, effectively solving the problems of traditional methods that struggle to balance efficiency and quality, rely on manual parameter adjustment, and are prone to loss of control during the polygon reduction process.

Description

Technical Field

[0001] This invention relates to the field of spatial modeling technology, and in particular to a method and system for simplifying urban rail engineering components. Background Technology

[0002] In the field of 3D spatial modeling, especially in applications involving large-scale industrial components or building information models, such as urban rail engineering, the geometric complexity of the 3D model directly determines the rendering frame rate, storage overhead, and computational resource consumption. With the widespread adoption of laser scanning, BIM forward design, and high-precision parametric modeling technologies, the number of polygons in a single engineering model often reaches millions or even tens of millions. This places a heavy computational burden on subsequent stages such as real-time visualization, virtual assembly, collision detection, and 3D printing slicing. Therefore, reasonable geometric simplification of high-polygon 3D models, i.e., polygon reduction, has become a common key technology in computer graphics and digital engineering management.

[0003] Currently, mainstream model polygon reduction techniques are mainly divided into two categories. The first category is a global uniform polygon reduction strategy based on local geometric error measurement. For example, the quadratic error metric (QEM) algorithm is used to uniformly merge vertices or fold edges of all faces until the total number of faces is reduced to a certain preset threshold. The second category is a parametric polygon reduction method based on human experience. That is, the operator manually sets different simplification ratios for different components or regions based on visual observation, and then calls a general polygon reduction tool to perform batch processing. However, the existing technical solutions have the following significant shortcomings in practical applications: First, the global uniform polygon reduction strategy lacks the ability to distinguish the importance of components and the distribution characteristics of face area. It is very easy to lead to insufficient reduction of redundant data for high-face-count, large-size components, while the fine features of low-face-count, small-size components are oversimplified or even destroyed, resulting in a serious decrease in the structural integrity of the simplified model and the recognizability of key functional areas. Second, the manual parameter setting scheme relies heavily on the operator's personal experience. It is not only inefficient and highly subjective, but also, when the model contains hundreds or thousands of heterogeneous components, it is impossible for the operator to achieve adaptive adjustment of parameters, making it difficult to guarantee the consistency and reliability of the polygon reduction results. Third, most existing methods use a single face reduction tool to complete all simplification operations at once, lacking a phased monitoring and dynamic feedback mechanism for the face reduction process. When the model complexity exceeds the tool's processing capacity, secondary or multiple face reductions can easily lead to abnormal situations such as an increase in the number of faces instead of a decrease, or mesh topology distortion, causing the face reduction process to fail.

[0004] To address the aforementioned issues, this application abandons the traditional fixed threshold or globally uniform face reduction approach, instead providing a method and system for simplifying urban rail engineering components based on the distribution of component face counts and the statistical characteristics of face area. First, a priority processing sequence is established based on the number of facets of components to achieve precise allocation of computational resources. Then, by comparing the average and median face area of ​​each component in real time, the skewness of the area distribution is automatically determined, and this drives the dynamic reverse adjustment of the simplification value, ensuring that the face reduction intensity always matches the geometric characteristics of the component. Finally, a phased forced face reduction mechanism with termination condition judgment is introduced to effectively avoid invalid iterations. Therefore, this application significantly reduces the total number of facets in the model and improves rendering and computational efficiency while maximizing the preservation of key contours and detailed areas, effectively balancing the contradiction between face reduction efficiency and model quality, and providing reliable technical support for the lightweight processing of large-scale 3D data in urban rail engineering and other industrial fields. Summary of the Invention

[0005] Existing technologies for global uniform surface reduction lack the ability to distinguish the distribution characteristics of component face count and face area. Manual parameter settings are difficult to achieve adaptive dynamic adjustment, and single-tool surface reduction lacks a phased monitoring and feedback mechanism. This leads to a situation where insufficient reduction of redundancy in high-face-count components coexists with excessive loss of key details in low-face-count components, easily causing face count rebound or mesh topology distortion. This invention first proposes a method for simplifying components in urban rail transit engineering, specifically including the following steps: Step S1: Obtain the number of faces of each component in the original 3D model, and generate a component processing sequence in descending order of face count; Step S2: For the current component in the component processing sequence, statistically analyze the area distribution characteristics of all its faces, and obtain the average area and median area. Step S3: Based on the sign of the difference between the area mean and the area median, determine the initial adjustment direction of the area reduction intensity, use a variable step size strategy to iteratively adjust the simplified value, and update the area mean and area median of the current component after each adjustment until the size relationship between the area mean and the area median crosses in the opposite direction, then lock the current simplified value and complete the area reduction process for the current component. Step S4: Process the next component in the component processing sequence in sequence, repeat steps S2 to S3 until the initial reduction of all components is completed, and count the first total number of faces; Step S5: Determine whether the first total number of faces meets the preset threshold. If not, use the preset ratio of the current total number of faces as the target reduction amount, start the forced face reduction process, and monitor the second total number of faces after face reduction in real time. Step S6: If the second total number of faces does not show an effective reduction relative to the first total number of faces, then stop the face reduction and output the current total number of faces; if it shows an effective reduction, then repeat the iterative face reduction with the current total number of faces as the target reduction amount until the preset termination condition is met, and output the final total number of faces.

[0006] Furthermore, in step S3, the variable step size strategy dynamically adjusts the incremental step size of the simplified value based on the exponential decay function. The output value of the exponential decay function decreases as the number of iterations increases, and the initial magnitude of the simplified value is adaptively set according to the total number of components in the original 3D model. The larger the total number of components, the fewer the effective digits of the initial magnitude.

[0007] Furthermore, the exponential decay function is configured as follows: , where k is a fixed parameter characterizing the global curvature fluctuation of the model, and x is the simplified value corresponding to the current iteration round.

[0008] Furthermore, in step S3, the logic for determining a reverse crossover in the size relationship is as follows: If the median area is greater than the mean area in the initial state, continue to increase the simplified value until the current mean area is less than or equal to the current median area. When the median area is less than the mean area in the initial state, the simplified value is continuously reduced until the current mean area is greater than or equal to the current median area.

[0009] Furthermore, when performing the surface reduction process on the current component in step S4, the method also includes: identifying geometric feature regions in the current component where the curvature gradient exceeds a preset curvature threshold, freezing the surface reduction operation on the facets within the geometric feature regions, and setting a transition zone around the geometric feature regions, wherein the surface reduction intensity within the transition zone increases linearly from the inside to the outside.

[0010] Furthermore, the forced face reduction process in step S5 employs the following face reduction logic: firstly, adjacent triangular faces with a coplanar error less than the tolerance value are merged into quadrilateral faces, then isolated faces with an area less than the current global average area are deleted, and vertex resampling is performed on local regions where the normal vector dispersion is higher than the preset dispersion threshold.

[0011] Furthermore, the preset termination condition in step S6 is: the absolute value of the change in the total number of faces after two consecutive iterations of face reduction is less than or equal to the preset tolerance, or the ratio of the final total number of faces to the original number of faces of the original 3D model is less than or equal to the preset compression rate threshold.

[0012] Furthermore, before obtaining the number of faces of each component in the original 3D model in step S1, the method further includes: converting the original 3D model from a parametric format to an unstructured mesh format, including .obj, .stl, or .fbx formats, and dynamically adjusting the triangulation chord height tolerance according to the ratio of model volume to surface area during the conversion process to suppress redundant faces generated by excessive subdivision.

[0013] Furthermore, in step S3, the median area is calculated as follows: the area values ​​of all faces in the current component are sorted in ascending order. When the number of area values ​​is odd, the area value of the middle position is taken. When the number of area values ​​is even, the arithmetic mean of the area values ​​of the two middle positions is taken.

[0014] Secondly, the present invention also provides a simplified system for urban rail transit engineering components, applied to the aforementioned simplified method for urban rail transit engineering components, comprising: The preprocessing module is used to convert the original 3D model into a target format file, extract the number of faces of each component, and generate a component processing queue in descending order of the number of faces. The feature analysis module is used to traverse the component processing queue, count the area distribution of all faces of the current component one by one, and generate the average area and median area. The dynamic control module is used to determine the control direction of reducing surface intensity based on the sign of the difference between the area mean and the area median. It uses a nonlinear step size to iteratively adjust the simplified value until the relationship between the area mean and the area median is reversed, and outputs the optimized surface data of the current component. The sequence control module is used to call the feature analysis module and the dynamic control module in the order of the component processing queue until the entire queue is processed and the preliminary total number of faces is obtained. The forced reduction module is used to perform face merging and isolated face removal operations in a loop with a fixed proportion of the current total number of faces as the target when the initial total number of faces exceeds a preset threshold, and to monitor the actual number of faces reduced after each loop. The iterative decision module is used to determine whether to continue the loop based on the actual reduction in the number of faces. If the actual reduction in the number of faces is less than or equal to zero, it terminates immediately; otherwise, it continues to iterate with a fixed ratio as the target until the total number of faces converges, and the final simplified model is output.

[0015] The embodiments of the present invention have the following technical effects: Firstly, this solution establishes a priority processing sequence based on the number of facets of components and dynamically adjusts the simplified value in reverse based on the sign of the difference between the average and median area of ​​facets, thus achieving precise allocation of facet reduction resources. Components with higher facet counts are processed earlier in the sequence, ensuring that computing resources are prioritized for reducing major redundant data. This results in a rapid decrease in the total number of facets within the shortest number of iterations, significantly reducing memory usage and rendering load, and improving the computational efficiency of subsequent visualization, collision detection, and slicing processing. Secondly, this solution introduces an adaptive discrimination mechanism based on area statistics during the reduction process. This mechanism can automatically select to enhance or suppress the reduction intensity based on the skewness of the area distribution of the internal facets of each component, avoiding oversimplification caused by a globally uniform threshold. Simultaneously, by identifying key detail areas with curvature values ​​exceeding a set threshold and applying protection rules, combined with the coordinates of the protected areas marked by user interaction, high-curvature contours, functional mating surfaces, and decorative features retain their original geometric recognizability after reduction. This effectively maintains structural integrity and visual fidelity while compressing the model data volume. Thirdly, this scheme employs a first polygon reduction tool for dynamic coarse adjustment driven by statistical indicators, followed by a second polygon reduction tool with a phased forced polygon reduction strategy. It iterates using a preset percentage of the current total polygon count as the target reduction amount, monitoring the actual polygon reduction in real time after each iteration. Once the change in polygon count after two consecutive polygon reductions is less than or equal to a set tolerance, the process is considered converged and automatically terminated. This two-layer polygon reduction architecture, coupled with strict termination condition judgment, effectively avoids the polygon count bounce, mesh distortion, or invalid loop problems that easily occur when a single tool processes complex models, ensuring stable convergence of the polygon reduction process and controllability of the final output. Attached Figure Description

[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 A simplified method flow diagram for urban rail engineering components provided by an embodiment of the present invention; Figure 2 A simplified system structure diagram of urban rail engineering components provided in this embodiment of the invention; Figure 3 This is a schematic diagram of the iterative control logic of the second surface reduction tool provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the overall logical architecture of the simplified method for urban rail engineering components provided in an embodiment of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0019] Existing technologies for global uniform surface reduction lack the ability to distinguish the distribution characteristics of component face count and patch area. Manual parameter settings are also difficult to achieve adaptive dynamic adjustment, and single-tool surface reduction lacks a phased monitoring and feedback mechanism. This leads to a situation where insufficient reduction of redundancy in high-face-count components coexists with excessive loss of key details in low-face-count components, easily causing face count rebound or mesh topology distortion. This invention addresses these issues by using component face count priority ranking and patch area statistical indicators to drive dynamic surface reduction control, combined with phased iterative termination judgment. This efficiently reduces the total face count of the model while adaptively protecting key details, effectively solving the problems of traditional methods that struggle to balance efficiency and quality, rely on manual parameter adjustment, and are prone to loss of control during the surface reduction process.

[0020] Example 1 Figure 1 A simplified method flow diagram for urban rail transit engineering components provided by an embodiment of the present invention includes the following steps: Step S1: Obtain the face count of each component in the original 3D model and generate a component processing sequence in descending order of face count. Before obtaining the face count of each component in the original 3D model, the following steps are included: converting the original 3D model from a parametric format to an unstructured mesh format, including .obj, .stl, or .fbx formats. During the conversion process, the triangulation chord height tolerance is dynamically adjusted based on the ratio of model volume to surface area to suppress redundant faces caused by excessive subdivision.

[0021] The original 3D model is usually generated by parametric modeling software and contains non-geometric information such as feature trees and constraints. Its parsing requires additional processing of historical data, making it relatively prone to introducing errors. The choice of target file format directly affects the feasibility of face extraction and face reduction. Non-parametric mesh formats such as .obj and .stl strip away the parametric constraints of the original design software, retaining only geometric and topological information, making it easier to directly parse patch data. For example, in an .obj file, the "v" line defines vertex coordinates, and the "f" line defines patch vertex indices; the number of facets can be obtained by parsing the number of lines in the "f" line.

[0022] Another advantage of choosing the .obj or .stl format is its wide compatibility. Mainstream 3D software and rendering engines support these formats, ensuring that the reduced-polygon model can be reused across platforms. For example, when importing a reduced-polygon .stl model into 3D printing slicing software, the support structure can be generated directly without needing to repair the mesh. However, attention must be paid to accuracy loss during format conversion. Some software triangulates surfaces during export; improper subdivision parameters can lead to an abnormal increase in the number of faces. For example, when exporting a high-precision NURBS surface as .stl, the chord height tolerance must be set appropriately to avoid generating redundant faces due to excessive subdivision. After conversion, the integrity of the facet topology must be verified to ensure there are no cracks or overlapping faces, so as not to affect subsequent facet reduction results.

[0023] To facilitate direct parsing of patch topology data, it is exported as an unstructured mesh format file, such as .obj, .stl, or .fbx. These formats store geometric information in plaintext form, including vertex coordinates and patch indices. During the conversion, the triangulation chord height tolerance is dynamically adjusted based on the ratio of model volume to surface area to suppress redundant patches caused by excessive subdivision. After format conversion, all independent components in the model are traversed, and the total number of triangular or quadrilateral patches contained in each component is extracted as the face count of that component. By converting the original model to a general mesh format and extracting the face count of each component, a data foundation is provided for subsequent priority sorting by face count and targeted face reduction.

[0024] Step S2: For the current component in the component processing sequence, statistically analyze the area distribution characteristics of all its faces to obtain the average area and median area.

[0025] In practical applications, generating a component processing sequence requires accurate identification of independent components within the model. In mechanical assembly models, components typically correspond to physical assembly units, such as bolts and bearing seats. In architectural models, components can be defined as logical units such as floors and walls. Component independence is determined by mesh connectivity: if two faces share vertices and belong to the same closed body, they are classified as the same component. After face count extraction, the sorting algorithm needs to efficiently handle large-scale data. For example, for a model containing tens of thousands of components, a quicksort algorithm is used to sort them in descending order of face count, with a time complexity of O(nlogn), ensuring processing speed. The sorted sequence needs to be stored as a queue structure, popping high-face-count components one by one for processing. This design avoids repeated traversal of low-face-count components, improving computational efficiency.

[0026] The prioritization strategy for processing is based on the Pareto principle, which states that approximately 20% of high-polygon components consume 80% of the resources. For example, a car model might contain 50,000 polygons for the body, 10,000 for the tires, and 5,000 for the interior. Prioritizing the body quickly reduces the total polygon count, while preserving the details of the tires and interior. This strategy minimizes the impact on the model's visual quality while ensuring efficient polygon reduction.

[0027] Components are sorted in descending order of face count to generate a component processing sequence. After traversing all components in the model and extracting their face count values, these values ​​are sorted in descending order to generate a processing list ordered by face count from highest to lowest. A quicksort algorithm can be used to handle large-scale data, and the sorted sequence is stored as a queue structure. High-face-count components consume significantly more computing resources than low-face-count components; therefore, they are placed at the beginning of the processing sequence to ensure that face count reduction resources are prioritized for areas with the highest concentration of redundant data. This allows for a rapid decrease in the total face count in the early stages of face count reduction, maximizing resource release within a limited number of iterations.

[0028] Step S3: Based on the sign of the difference between the area mean and the area median, determine the initial adjustment direction of the area reduction intensity, use a variable step size strategy to iteratively adjust the simplified value, and update the area mean and area median of the current component after each adjustment, until the size relationship between the area mean and the area median crosses in the opposite direction, lock the current simplified value, and complete the area reduction process for the current component.

[0029] The variable step size strategy dynamically adjusts the incremental step size of the simplified value based on an exponential decay function. The output value of the exponential decay function decreases with the increase of the number of iterations, and the initial magnitude of the simplified value is adaptively set according to the total number of components in the original 3D model. The larger the total number of components, the fewer the significant digits of the initial magnitude. The median area is calculated as follows: the area values ​​of all faces in the current component are sorted in ascending order. When the number of area values ​​is odd, the middle area value is taken; when the number of area values ​​is even, the arithmetic mean of the two middle area values ​​is taken.

[0030] The exponential decay function is configured as follows: , where k is a fixed parameter characterizing the global curvature fluctuation of the model, and the value of k ranges from 0.01 to 1.0. The larger the value, the more sensitive the simplified value adjustment. x is the simplified value corresponding to the current iteration round.

[0031] Specifically, the logic for determining a reverse crossover in the size relationship is as follows: If the median area is greater than the mean area in the initial state, continue to increase the simplified value until the current mean area is less than or equal to the current median area. When the median area is initially less than the average area, the simplification value is continuously decreased until the current average area is greater than or equal to the current median area. Since increasing the simplification value will cause more small areas to be merged or deleted, thus lowering the overall average area, continuously increasing the simplification value will gradually bring the average area of ​​the current iteration closer to and eventually lower than the median area of ​​the current iteration. The areas of all faces in each component are sorted, and the average area V0 and median A0 are calculated. The median A0 is calculated to eliminate the influence of extreme values: when the number of area values ​​is odd, the middle value after sorting is taken; when the number of area values ​​is even, the arithmetic mean of the two middle values ​​after sorting is taken. The sorting of the area value set needs to consider computational efficiency and accuracy. For components containing tens of thousands of faces, a radix sort algorithm is used, with a time complexity of O(nk), suitable for large-scale data. After sorting, the calculation of the median needs to distinguish between odd and even cases to ensure accuracy. Taking a component containing 10,000 faces as an example, if the number of faces is even, the median is the average of the areas of the 5000th and 5001st faces; if the number is odd, the 5000th value is used directly. For instance, if a component's area distribution contains a few maxima, such as decorative reliefs, the mean V0 might be inflated, while the median A0 better reflects the actual area level of most faces. By comparing V0 and A0, the skewness of the area distribution can be accurately determined, allowing for the development of targeted face reduction strategies.

[0032] For the component to be processed, all its faces are traversed. The area value of each face is calculated based on its geometry. For triangular faces, the area is determined by half the modulus of the cross product of the two edge vectors. For quadrilateral faces, the area is calculated separately for each of the two triangular faces, and then summed. After collecting all face area values ​​into a numerical set, the set is sorted in ascending order, and its arithmetic mean is calculated as V0. For the median A0, if the number of area values ​​is odd, the value at the middle of the sorted position is used; if the number of area values ​​is even, the arithmetic mean of the two middle values ​​is used. The mean reflects the overall distribution level of the face area, while the median represents the area limit characteristic at the middle position. The comparison between the two accurately indicates whether the area distribution is right-skewed or left-skewed, providing a statistical basis for the subsequent selection of the face reduction direction.

[0033] Step S4: Process the next component in the component processing sequence in sequence, repeat steps S2 to S3 until the initial reduction of all components is completed, and obtain the first total number of faces.

[0034] In other words, when completing the surface reduction process of the current component, it also includes: identifying geometric feature regions in the current component where the curvature gradient exceeds a preset curvature threshold, freezing the surface reduction operation on the facets within the geometric feature regions, and setting a transition zone around the geometric feature regions, with the surface reduction intensity within the transition zone increasing linearly from the inside to the outside.

[0035] Based on the relationship between V0 and A0, the direction for adjusting the reduction parameters is selected, and the first reduction tool is used to sequentially reduce the surface area of ​​the components in the component processing sequence. The simplification value is dynamically adjusted until a preset condition is met, at which point the process stops. Specifically, when A0 > V0, it indicates that the area distribution is skewed to the right, meaning that there are many large redundant surface areas in the component. In this case, the simplification value needs to be gradually increased to enhance the reduction intensity until the current average V0 is reached. x Less than or equal to the current median A x When A0 < V0, it indicates that the area distribution is left-skewed, meaning that small-area detail patches dominate. In this case, the simplification value needs to be gradually reduced to suppress oversimplification until the current average V0 is reached. x Greater than or equal to the current median A x When A0=V0, it indicates that the area distribution is symmetrical, and the current simplified value remains unchanged.

[0036] The adjustment of simplified values ​​is based on the exponential function. The function is implemented where k is a fixed parameter related to model complexity, and x is a simplification value. The output of this function increases non-linearly with the simplification value, achieving a variable step size control effect of fast response in the early stage and fine approximation in the later stage. The setting of parameter k is related to the density of model patches: models with uniform patch distribution, such as mechanical parts, use a smaller k value to allow the simplification value to increase slowly; models with large differences in patch density, such as terrain meshes, use a larger k value to quickly respond to changes in area distribution.

[0037] The initial magnitude of the simplification value is set based on the total number of components in the model. The more components there are, the fewer digits the initial value needs to be precise, thus reducing the number of iterations required to process highly complex models. For example, when processing a building model with 200 components, the initial simplification value is set to 0.005, with an adjustment step size of 0.001 each time, ensuring a stable state is reached within 10 iterations. Furthermore, during the reduction of polygons using the first polygon reduction tool, it is necessary to identify key detail areas in the model, including areas with curvature values ​​greater than a set threshold and protected area coordinates annotated by the user through the interactive interface. Polygon reduction protection rules are set for these areas to prevent the destruction of key geometric features during polygon reduction. By independently adjusting the polygon reduction intensity for each component using statistically driven dynamic adjustments, the polygon reduction intensity is precisely matched to the individual geometric features of each component, overcoming the problem of detail loss or redundant residue caused by globally uniform parameters. Simultaneously, the key detail protection mechanism further ensures the usability of the model after polygon reduction.

[0038] like Figure 3The schematic diagram of the iteration control logic for the second face reduction tool is shown. Step S5: Determine whether the first total face count meets the preset threshold. If not, use the preset ratio of the current total face count as the target reduction amount, start the forced face reduction process, and monitor the second total face count after face reduction in real time. The face reduction logic adopted by the forced face reduction process is as follows: First, merge adjacent triangular faces with coplanarity error less than the tolerance value into quadrilateral faces, then delete isolated faces with an area less than the current global average area, and perform vertex resampling on local regions where the normal vector dispersion is higher than the preset dispersion threshold.

[0039] The total number of faces M after subtracting faces x If M x If the preset face count threshold is not met, the target face count is set to a preset proportion of the current total face count, and the second face count reduction tool is used to reduce the face count to obtain a new face count M. i After the first face reduction tool completes the initial face reduction for all components, the total number of faces M of the entire model is counted. x The face count is compared with a preset face count threshold. If the threshold is met, the result is output directly; otherwise, a forced face reduction phase is triggered. In this phase, the target face count is set to half of the current total face count, and a second face reduction tool is used to perform the face reduction operation. The face reduction logic of the second face reduction tool includes merging adjacent faces to reduce the face count and deleting faces with an area less than the current average V. x The model performs local mesh reconstruction on areas where the rate of curvature change exceeds a threshold. By introducing a second facet reduction tool for phased processing, forced simplification is further implemented on the basis of statistically driven coarse adjustment to ensure that the total number of faces in the model can approach the preset target.

[0040] Step S6: If the second total face count does not show an effective reduction relative to the first total face count, then terminate the face reduction and output the current total face count; if an effective reduction is shown, then repeat the iterative face reduction using a preset ratio of the current total face count as the target reduction amount until a preset termination condition is met, and output the final total face count. The preset termination condition is: the absolute value of the change in the total face count after two consecutive iterations of face reduction is less than or equal to a preset tolerance, or the ratio of the final total face count to the original face count of the original 3D model is less than or equal to a preset compression rate threshold.

[0041] If M i If the target number of faces is greater than or equal to the current total number of faces, stop reducing the number of faces; otherwise, update the target number of faces to the current M. i The preset ratio is used to repeat step S5 until the stopping condition is met, and the final total number of faces M is output. q After each second face reduction tool completes its operation, the newly obtained face count M is calculated. i Compare this to the total number of faces before the subtraction, if M iIf M is greater than or equal to the total number of faces before reduction, it indicates that further reduction can no longer generate positive returns, and the cycle should be terminated immediately; if M i If the target number of faces is less than the total number of faces before subtraction, then update the target number of faces to the current M. i If the number of faces is reduced to half, the second face reduction tool is called for the next iteration. Stopping conditions also include the change in the total number of faces after two consecutive face reductions being less than or equal to a set tolerance value, and the final total number of faces M. q The ratio of the number of faces to the original face count is less than a preset threshold. Through strict iteration termination condition checks, face count bounce and invalid loops are effectively avoided, ensuring that the face reduction process automatically terminates when it converges to the optimal solution, outputting a final model that meets the compression ratio requirements and has a complete structure.

[0042] like Figure 4 The schematic diagram of the overall logical architecture of the simplification method for urban rail transit engineering components shows that, through step-by-step progressive processing, the efficiency and quality of 3D model reduction are significantly improved. First, the original model is converted into a general mesh format, and the face count information of each component is directly extracted. A processing sequence is generated by sorting the face count from high to low, prioritizing the simplification of components with the highest face count to quickly reduce the overall computational burden. Next, by calculating the average and median area of ​​each component's facets, the system automatically judges the area distribution characteristics and dynamically adjusts the face reduction intensity: when the median is greater than the average, it indicates that the model has redundant large-area facets, and the system gradually increases the simplification intensity until the relationship between the two is reversed; conversely, the reduction rate is controlled to prevent the loss of small-area details. After initial optimization, a phased forced face reduction strategy is further adopted, first coarsely screening according to dynamic parameters, and then finely reducing according to the target ratio. During the process, the face count change is monitored in real time, and the process terminates immediately once a preset threshold is reached or the change stabilizes to avoid invalid operations. The entire method balances efficiency and accuracy, quickly reducing redundant data in high-face-count components while protecting key details through statistical indicators, ultimately ensuring structural integrity and usability while reducing model complexity.

[0043] This solution achieves precise allocation of face reduction resources by establishing a priority processing sequence for component face count and dynamically adjusting the simplification value based on the sign of the difference between the average and median face area. Components with higher face counts are processed earlier in the sequence, ensuring that computational resources are prioritized for reducing major redundant data. This results in a rapid decrease in the total face count within the shortest number of iterations, significantly reducing memory usage and rendering load, and improving the computational efficiency of subsequent visualization, collision detection, and slicing. An adaptive discrimination mechanism based on area statistics is introduced during face reduction. This mechanism automatically selects to enhance or suppress face reduction intensity based on the skewness of the face area distribution within each component, avoiding oversimplification caused by a globally uniform threshold. Simultaneously, by identifying key detail areas with curvature values ​​exceeding a set threshold and applying protection rules, combined with user-annotated protected area coordinates, high-curvature contours, functional surfaces, and decorative features retain their original geometric recognizability after face reduction, effectively maintaining structural integrity and visual fidelity while compressing model data.

[0044] Example 2 like Figure 2 As shown in the structural diagram of the simplified urban rail transit engineering component system, this invention also proposes a simplified urban rail transit engineering component system, using the simplified urban rail transit engineering component method as described in Example 1, specifically including the following: The preprocessing module converts the original 3D model into a target format file, extracts the face count of each component, and generates a component processing queue in descending order of face count. It also converts the original 3D model from a parametric format to an unstructured mesh target format file, which can be in .obj, .stl, or .fbx format. After conversion, the preprocessing module extracts the face count of each component in the model and sorts them in descending order of face count, generating a component processing queue. Simultaneously, during the format conversion process, the preprocessing module dynamically adjusts the triangulation chord height tolerance based on the ratio of model volume to surface area to suppress redundant facets caused by excessive subdivision, ensuring that subsequent processing is based on accurate and non-redundant geometric data.

[0045] The feature analysis module iterates through the component processing queue, statistically analyzing the area distribution of all faces of the current component, and generating the area mean and median. The feature analysis module also iterates through the component processing queue generated by the preprocessing module, statistically analyzing the area distribution of all faces in the current component. This module collects all face area values ​​of each component into a numerical set and sorts them in ascending order. When the number of area values ​​is odd, the middle area value is taken as the median A0; when the number of area values ​​is even, the arithmetic mean of the two middle area values ​​is taken as the median A0. Simultaneously, the arithmetic mean V0 of all area values ​​is calculated, and V0 and A0 are transmitted to the dynamic control module.

[0046] The dynamic control module determines the direction of reducing surface intensity based on the sign of the difference between the area mean and the area median. It iteratively adjusts the simplified value using a nonlinear step size until the relationship between the area mean and the area median reverses, then outputs the optimized surface data for the current component. After receiving V0 and A0 from the feature analysis module, the dynamic control module determines the direction of reducing surface intensity based on the sign of their difference. This module employs an exponential function-based approach. The simplified value is adjusted iteratively with a nonlinear step size, where k is a fixed parameter related to model complexity, and x is the simplified value. When A0 > V0, the dynamic adjustment module continuously increases the simplified value until the current average V0 is reached. x Less than or equal to the current median A x When A0 < V0, the dynamic control module continuously decreases the simplified value until the current average V. x Greater than or equal to the current median A x When A0=V0, the current simplified value remains unchanged. When the relationship between the area mean and the area median reverses, the dynamic control module locks the current simplified value and outputs the optimized surface data of the current component. During the surface reduction process, the dynamic control module also identifies regions with curvature values ​​greater than a set threshold and receives the coordinates of protected areas annotated by the user through the interactive interface. It sets surface reduction protection rules for these key detail regions to maintain the structural integrity and visual fidelity of the model after surface reduction.

[0047] The sequence control module sequentially calls the feature analysis module and the dynamic control module according to the component processing queue, until the entire queue is processed and a preliminary total face count is obtained. The sequence control module calls the feature analysis module and the dynamic control module sequentially according to the component processing queue, ensuring that each component completes independent adaptive face reduction processing based on its own area statistical characteristics. After all components have been processed, the sequence control module accumulates the face count of each component after face reduction, obtains a preliminary total face count, and transmits the preliminary total face count to the forced reduction module.

[0048] The forced reduction module is used to repeatedly perform face merging and isolated face removal operations with a target of a fixed proportion of the current total face count when the initial total face count exceeds a preset threshold, and monitors the actual face reduction after each loop. Upon receiving the initial total face count, the forced reduction module determines whether it exceeds the preset threshold. If it does, the forced reduction module uses half of the current total face count as the target face count and initiates a forced face reduction process. The face reduction logic executed by this module includes merging adjacent faces to reduce the face count, deleting isolated faces with an area smaller than the current average area, and performing local mesh reconstruction for regions with a curvature change rate higher than the curvature change rate threshold. After each forced face reduction operation, the forced reduction module generates a new face count and transmits this new face count to the iterative decision module.

[0049] The iterative decision module determines whether to continue the iteration based on the actual reduction in the number of faces. If the actual reduction is less than or equal to zero, the iteration terminates immediately; otherwise, it continues iterating at a fixed ratio until the total number of faces converges, outputting the final simplified model. After receiving the new face count from the forced simplification module, the iterative decision module compares this new face count with the total number of faces before the current reduction. If the new face count is greater than or equal to the total number of faces before the reduction, the iterative decision module immediately terminates the reduction process and outputs the current total number of faces. If the new face count is less than the total number of faces before the reduction, the iterative decision module updates the target face count with half of the current new face count and calls the forced simplification module again for the next iteration. The iterative decision module also monitors the change in the total number of faces after two consecutive reductions. When the change is less than or equal to a set tolerance value, the reduction process is considered converged, and the loop terminates. Furthermore, the iteration also terminates when the ratio of the final total number of faces to the original number of faces is less than a preset threshold. Finally, the iterative decision module outputs the final total number of faces M that meets the compression ratio requirement and convergence condition. q This completes the entire model simplification process. The above modules are interconnected and drive each other step by step, together forming a complete adaptive polygon reduction system that achieves the optimal balance between polygon reduction efficiency and model quality without manual intervention.

[0050] This invention provides a simplified system for urban rail transit components. After employing a first surface reduction tool for dynamic coarse adjustment driven by statistical indicators, it further introduces a second surface reduction tool with a phased forced surface reduction strategy. The system iteratively reduces the surface count by a preset percentage of the current total surface count, monitoring the actual surface reduction in real time after each iteration. If the surface count change is less than or equal to a set tolerance after two consecutive surface reductions, the process is considered converged and automatically terminated. This two-layer surface reduction architecture, coupled with strict termination condition judgment, effectively avoids surface count bounce, mesh distortion, or invalid loops that easily occur when using a single tool to process complex models, ensuring stable convergence of the surface reduction process and controllability of the final output.

[0051] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the technical solutions of the embodiments of the present invention.

Claims

1. A method of simplifying a metro engineering component, characterized by, Includes the following steps: Step S1: Obtain the number of faces of each component in the original 3D model, and generate a component processing sequence in descending order of the number of faces; Step S2: For the current component in the component processing sequence, statistically analyze the area distribution characteristics of all its faces to obtain the average area and the median area. Step S3: Based on the sign of the difference between the area mean and the area median, determine the initial adjustment direction of the area reduction intensity, use a variable step size strategy to iteratively adjust the simplified value, and update the area mean and area median of the current component after each adjustment until the size relationship between the area mean and the area median crosses in the opposite direction, then lock the current simplified value and complete the area reduction process for the current component. Step S4: Process the next component in the component processing sequence in sequence, and repeat steps S2 to S3 until the initial reduction of all components is completed, and the first total number of faces is obtained. Step S5: Determine whether the first total number of faces meets the preset threshold. If not, use the preset ratio of the current total number of faces as the target reduction amount, start the forced face reduction process, and monitor the second total number of faces after face reduction in real time. Step S6: If the second total number of faces does not show an effective reduction relative to the first total number of faces, then stop the face reduction and output the current total number of faces; if it shows an effective reduction, then repeat the iterative face reduction with the current total number of faces as the target reduction amount until the preset termination condition is met, and output the final total number of faces.

2. The method according to claim 1, characterized in that, In step S3, the variable step size strategy dynamically adjusts the incremental step size of the simplified value based on the exponential decay function. The output value of the exponential decay function decreases as the number of iterations increases, and the initial magnitude of the simplified value is adaptively set according to the total number of components in the original 3D model. The larger the total number of components, the fewer the effective digits of the initial magnitude.

3. The method according to claim 2, wherein The exponential decay function is configured as follows: ； Where k is a fixed parameter characterizing the global curvature fluctuation of the model, and x is the simplified value corresponding to the current iteration round.

4. The urban rail engineering component simplification method according to claim 1, characterized in that, In step S3, the logic for determining the reverse crossover of the size relationship is as follows: When the median area is greater than the mean area in the initial state, the simplified value is continuously increased until the current mean area is less than or equal to the current median area. When the median area is less than the mean area in the initial state, the simplified value is continuously reduced until the current mean area is greater than or equal to the current median area.

5. The urban rail engineering component simplification method according to claim 1, characterized in that, When performing the surface reduction process on the current component in step S4, the method further includes: identifying geometric feature regions in the current component where the curvature gradient exceeds a preset curvature threshold, freezing the surface reduction operation on the patches within the geometric feature regions, and setting a transition zone around the geometric feature regions, wherein the surface reduction intensity within the transition zone increases linearly from the inside to the outside.

6. The method according to claim 5, wherein The forced face reduction process in step S5 uses the following face reduction logic: firstly, adjacent triangular faces with coplanar errors less than the tolerance value are merged into quadrilateral faces, then isolated faces with areas smaller than the current global average area are deleted, and vertex resampling is performed on local regions where the normal vector dispersion is higher than the preset dispersion threshold.

7. The urban rail engineering component simplification method according to claim 1, characterized in that, The preset termination condition in step S6 is: the absolute value of the change in the total number of faces after two consecutive iterations of face reduction is less than or equal to the preset tolerance, or the ratio of the final total number of faces to the original number of faces of the original 3D model is less than or equal to the preset compression rate threshold.

8. The urban rail engineering component simplification method according to claim 1, characterized in that, Before obtaining the number of faces of each component in the original 3D model in step S1, the method further includes: converting the original 3D model from a parametric format to an unstructured mesh format, wherein the unstructured mesh format includes .obj, .stl or .fbx format, and dynamically adjusting the triangulation chord height tolerance according to the ratio of model volume to surface area during the conversion process to suppress redundant faces generated by excessive subdivision.

9. The urban rail engineering component simplification method according to claim 1, characterized in that, In step S3, the median area is calculated as follows: the area values ​​of all facets in the current component are sorted in ascending order. When the number of area values ​​is odd, the area value of the middle position is taken. When the number of area values ​​is even, the arithmetic mean of the area values ​​of the two middle positions is taken.

10. A system for simplifying a metro engineering component, for performing the method for simplifying a metro engineering component according to any one of claims 1 to 9, characterized in that, include: The preprocessing module is used to convert the original 3D model into a target format file, extract the number of faces of each component, and generate a component processing queue in descending order of the number of faces. The feature analysis module is used to traverse the component processing queue, count the area distribution of all faces of the current component one by one, and generate the average area and the median area. The dynamic control module is used to determine the control direction of reducing surface intensity based on the sign of the difference between the area mean and the area median, and to iteratively adjust the simplified value using a nonlinear step size until the relationship between the area mean and the area median is reversed, and then output the optimized surface data of the current component. The sequence control module is used to sequentially call the feature analysis module and the dynamic control module according to the order of the component processing queue until the entire queue is processed and the preliminary total number of faces is accumulated. The forced reduction module is used to perform face merging and isolated face removal operations cyclically with a fixed proportion of the current total number of faces as the target when the initial total number of faces exceeds a preset threshold, and to monitor the actual face reduction amount after each cycle. The iterative decision module is used to determine whether to continue the loop based on the actual reduction in the number of faces. If the actual reduction in the number of faces is less than or equal to zero, the loop terminates immediately; otherwise, it continues to iterate with the fixed ratio as the target until the total number of faces converges, and the final simplified model is output.

Images