Modeling system integrating laser point cloud and 2D vector

Abstract

This invention relates to the field of 3D modeling technology and discloses a modeling system that integrates laser point clouds and 2D vectors. The system includes a data acquisition module, a data fusion module, a model generation module, a visualization module, an optimization and adjustment module, and a user feedback module. The data acquisition module acquires laser point cloud and 2D vector data and performs standardized preprocessing. Based on the data fusion module, the point cloud and vector data are registered, transformed, and weighted to generate high-precision fused data. The model generation module performs 3D reconstruction, detail enhancement, and quality verification, outputting the final model. The visualization module renders the scene and supports interactive operations. The optimization and adjustment module evaluates model efficiency and dynamically optimizes parameters. The user feedback module analyzes requirements and adaptively adjusts system settings. This invention improves the accuracy and reliability of 3D modeling, enhances the robustness and consistency of the modeling process, and improves the overall output quality and engineering applicability of the system.

Description

Technical Field

[0001] This invention relates to the field of 3D modeling technology, specifically a modeling system that integrates laser point clouds and 2D vectors. Background Technology

[0002] 3D modeling is a technique that uses computer software to construct object models in three-dimensional space, and it belongs to the field of computer graphics.

[0003] Currently, in the process of modeling by fusing laser point clouds and 2D vectors, point cloud data acquisition is easily affected by environmental occlusion and sensor noise interference, making it impossible to verify the coordinate consistency when the point cloud and vector data are aligned in real time. When the point cloud is missing or the registration deviation is large, it will cause geometric distortion of the 3D model and affect the modeling accuracy.

[0004] Therefore, a modeling system that integrates laser point clouds and two-dimensional vectors is proposed to solve the above problems. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a modeling system that integrates laser point clouds and two-dimensional vectors, solving the problem mentioned in the background that the coordinate consistency of point cloud and vector data cannot be verified in real time, as well as the problem affecting modeling accuracy.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a modeling system that integrates laser point clouds and two-dimensional vectors, comprising:

[0007] The data acquisition module uses the laser point cloud acquisition unit to generate three-dimensional point cloud data, acquires two-dimensional vector data through the two-dimensional vector import unit, and outputs standardized data through the data preprocessing unit.

[0008] The data fusion module receives the standardized data, aligns the data through a point cloud registration unit, converts two-dimensional vectors into three-dimensional references through a vector conversion unit, and outputs fused data through a fusion algorithm unit.

[0009] The model generation module receives the fused data, generates a basic model through the 3D reconstruction unit, adds features through the detail enhancement unit, and outputs the final model through the quality verification unit.

[0010] The visualization module receives the final model, generates a visualization scene through the rendering engine unit, allows users to operate it through the interactive control unit, and exports the model file through the output interface unit.

[0011] The optimization and adjustment module receives the final model, evaluates the model efficiency through the performance analysis unit, adjusts the modeling parameters through the parameter optimization unit, and outputs optimization instructions through the iterative update unit.

[0012] The user feedback module receives user input data, identifies user needs through the demand parsing unit, obtains operational feedback through the feedback collection unit, and adjusts system parameters through the adaptive learning unit.

[0013] Preferably, the process in the data acquisition module of generating three-dimensional point cloud data using a laser point cloud acquisition unit, acquiring two-dimensional vector data through a two-dimensional vector import unit, and outputting standardized data through a data preprocessing unit includes:

[0014] The laser scanning parameters, including scanning resolution, scanning angle, and scanning distance, are set through the laser point cloud acquisition unit.

[0015] Calculate the density and coverage of point cloud data to ensure data integrity;

[0016] When point cloud data contains missing data or noise, the data is repaired and interpolated and filtered.

[0017] The point cloud data is normalized in coordinates and standardized in format by the data preprocessing unit, and standardized point cloud data is output.

[0018] Meanwhile, the process of acquiring two-dimensional vector data by the two-dimensional vector import unit includes: importing vector file format, parsing vector elements, and performing geometric correction and scale alignment.

[0019] Preferably, the process of calculating the density and coverage of the point cloud data includes:

[0020] The point cloud density threshold is defined as the minimum number of points per cubic meter. When the actual density is lower than the threshold, a rescanning mechanism is triggered.

[0021] The coverage area is evaluated by calculating the intersection ratio between the point cloud bounding box and the target area. If the ratio is lower than the preset value, the scanning path is automatically adjusted.

[0022] The data repair operation uses an interpolation algorithm based on the weighted average of nearest neighbors to fill in missing data areas;

[0023] Noise processing involves statistical filtering to remove outliers and ensure the smoothness of point cloud data.

[0024] Preferably, the process of receiving the standardized data in the data fusion module, aligning the data through a point cloud registration unit, converting the two-dimensional vector into a three-dimensional reference through a vector conversion unit, and outputting the fused data through a fusion algorithm unit includes:

[0025] Extract feature points from the point cloud, including corner points, edge points, and planar points;

[0026] Registration is achieved by calculating the transformation matrix between point clouds using the iterative nearest-point algorithm, where the registration error is minimized using the following normalization formula:

[0027] ;

[0028] in, For registration error, Let be the transformation matrix. For points in the source point cloud, For the corresponding point in the target point cloud, Point-to-point quantity;

[0029] Set registration error threshold ,when If so, then re-perform the registration and adjust the transformation matrix parameters;

[0030] The process by which the vector transformation unit converts a two-dimensional vector into a three-dimensional reference includes: elevating the vector coordinates to three-dimensional space and mapping them based on the elevation information of the point cloud data;

[0031] The fusion algorithm unit adopts a weighted fusion strategy, which assigns weights based on the confidence levels of point clouds and vectors to generate fused data.

[0032] Preferably, the process of extracting point cloud feature points includes:

[0033] Feature points are identified using curvature calculation and normal vector analysis;

[0034] Feature point matching is achieved through descriptor alignment, including fast point feature histogram descriptors;

[0035] The transformation matrix calculation includes optimization of rotation and translation parameters, and the least squares method is used to minimize the registration error;

[0036] The weighting is based on the accuracy metrics of the data source, including the positioning accuracy of the laser point cloud and the geometric accuracy of the vector data.

[0037] Preferably, the process in the model generation module of receiving the fused data, generating a basic model through a 3D reconstruction unit, adding features through a detail enhancement unit, and outputting the final model through a quality verification unit includes:

[0038] Based on the fused data, a triangular mesh model is generated using the Poisson surface reconstruction algorithm;

[0039] The detail enhancement unit adds detail through multi-resolution analysis, including texture mapping and geometric subdivision;

[0040] The quality verification unit evaluates the model's integrity, geometric accuracy, and visual realism, and outputs a quality score.

[0041] When the quality score falls below the threshold, the model reconstruction process is triggered, and data fusion and reconstruction are performed again.

[0042] Preferably, the process by which the quality verification unit evaluates the geometric accuracy of the model includes:

[0043] The root mean square error between the model vertices and the original point cloud data is calculated as a geometric deviation index using the following normalization formula:

[0044] ;

[0045] in, The root mean square error, Let be the Euclidean distance between the i-th model vertex and its corresponding point cloud point. The number of vertices;

[0046] Set geometric error threshold ,when If the model's geometric accuracy is deemed unacceptable, the model reconstruction process will be triggered.

[0047] Visual realism is evaluated through rendering tests, including lighting consistency and shadow effects;

[0048] Integrity checks ensure the model can be used in subsequent applications by detecting voids and non-manifold structures in the model.

[0049] The quality score is calculated by combining geometric error, visual score, and integrity index, using a weighted average.

[0050] Preferably, the process of receiving the final model in the visualization module, generating a visualization scene through the rendering engine unit, allowing user operation through the interactive control unit, and exporting the model file through the output interface unit includes:

[0051] Load the final model data and apply physics to simulate lighting and materials through the rendering engine unit;

[0052] The interactive control unit supports user operations, including rotating, zooming, and panning views, as well as model editing functions;

[0053] The output interface unit supports exporting multiple file formats, including OBJ, STL, and point cloud formats;

[0054] Scene generation is performed in real time, accelerated by the graphics processor to ensure a smooth user experience.

[0055] Preferably, the optimization and adjustment module receives the final model, evaluates the model efficiency through a performance analysis unit, adjusts the modeling parameters through a parameter optimization unit, and outputs optimization instructions through an iterative update unit, including the following process:

[0056] The performance analysis unit evaluates model efficiency and monitors model generation time, memory usage, and computing resource utilization.

[0057] The parameter optimization unit dynamically adjusts modeling parameters, including reconstruction resolution and fusion weights, to balance accuracy and efficiency;

[0058] The iterative update unit automatically optimizes the system process based on performance feedback, reducing redundant calculations;

[0059] The optimization instructions include parameter adjustment suggestions and process refactoring commands to improve the overall system performance.

[0060] Preferably, the process of receiving user input data in the user feedback module, identifying user needs through the demand parsing unit, obtaining operational feedback through the feedback collection unit, and adjusting system parameters through the adaptive learning unit includes:

[0061] The requirement analysis unit identifies user requirements, parses the natural language and graphical instructions input by the user, and extracts key modeling requirements.

[0062] The feedback collection unit obtains feedback data through user interaction logs and satisfaction surveys;

[0063] The adaptive learning unit uses machine learning algorithms to analyze feedback patterns and optimize system default parameters.

[0064] System parameter adjustments include fusion strategy preferences, rendering quality settings, and output format priorities, to achieve a personalized modeling experience.

[0065] Compared with existing technologies, this invention provides a modeling system that integrates laser point clouds and two-dimensional vectors, which has the following advantages:

[0066] 1. In this invention, during the modeling process of fusing laser point clouds and two-dimensional vectors, the point cloud and vector data are verified and standardized in real time through a data acquisition module. This can solve the impact of environmental occlusion and sensor noise on data quality, avoid geometric distortion of the model caused by missing point clouds and registration deviations, and improve the accuracy and reliability of three-dimensional modeling.

[0067] 2. In this invention, during the data fusion stage, the point cloud registration unit and fusion algorithm unit in the data fusion module are used to achieve adaptive weight allocation and dynamic parameter optimization of point cloud and vector data. This can maintain the stability of the fusion results in complex scenarios and correct registration deviations in real time through the system feedback mechanism, thereby enhancing the robustness and consistency of the modeling process.

[0068] 3. In this invention, during the model generation and optimization stage, the 3D reconstruction unit and quality verification unit in the model generation module intelligently coordinate the conflicts between point cloud details and vector structures, avoid model holes and texture misalignment problems, ensure that the model meets practical requirements in terms of integrity and visual realism, and improve the overall output quality and engineering applicability of the system. Attached Figure Description

[0069] Figure 1 This is a schematic diagram of the architecture of the modeling system that integrates laser point cloud and two-dimensional vector in this invention. Detailed Implementation

[0070] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0071] Please see Figure 1 The specific implementation of the modeling system that integrates laser point clouds and two-dimensional vectors is as follows:

[0072] The data acquisition module uses the laser point cloud acquisition unit to generate three-dimensional point cloud data, acquires two-dimensional vector data through the two-dimensional vector import unit, and outputs standardized data through the data preprocessing unit.

[0073] The data fusion module receives standardized data, aligns the data through the point cloud registration unit, converts two-dimensional vectors into three-dimensional references through the vector conversion unit, and outputs fused data through the fusion algorithm unit.

[0074] The model generation module receives the fused data, generates a basic model through the 3D reconstruction unit, adds features through the detail enhancement unit, and outputs the final model through the quality verification unit.

[0075] The visualization module receives the final model, generates a visualization scene through the rendering engine unit, allows users to operate it through the interactive control unit, and exports the model file through the output interface unit.

[0076] The optimization and adjustment module receives the final model, evaluates the model efficiency through the performance analysis unit, adjusts the modeling parameters through the parameter optimization unit, and outputs optimization instructions through the iterative update unit.

[0077] The user feedback module receives user input data, identifies user needs through the demand parsing unit, obtains operational feedback through the feedback collection unit, and adjusts system parameters through the adaptive learning unit.

[0078] The data acquisition module utilizes a laser point cloud acquisition unit to generate 3D point cloud data, a 2D vector import unit to acquire 2D vector data, and a data preprocessing unit to output standardized data. The process includes:

[0079] The laser scanning parameters, including scanning resolution, scanning angle, and scanning distance, are set through the laser point cloud acquisition unit.

[0080] Calculate the density and coverage of the point cloud data to ensure data integrity. The point cloud density is calculated using the following normalization formula:

[0081] ;

[0082] in, Point cloud density, This represents the total number of points in the point cloud. The volume of the scanned region;

[0083] Specifically, volume The bounding box algorithm is used to calculate: first, the minimum bounding box of the point cloud is extracted, with its length, width, and height being respectively... ,but , points Obtained through point cloud index statistics;

[0084] Set density threshold ,when If so, a rescan mechanism will be triggered to improve data integrity;

[0085] When point cloud data contains missing data or noise, the data is repaired and interpolated and filtered.

[0086] The point cloud data is normalized in coordinates and standardized in format by the data preprocessing unit, and standardized point cloud data is output.

[0087] Meanwhile, the process of acquiring two-dimensional vector data by the two-dimensional vector import unit includes: importing vector file format, parsing vector elements, and performing geometric correction and scale alignment.

[0088] The process of calculating the density and coverage of point cloud data includes:

[0089] The point cloud density threshold is defined as the minimum number of points per cubic meter. When the actual density is lower than the threshold, a rescanning mechanism is triggered.

[0090] The coverage area is evaluated by calculating the intersection ratio between the point cloud bounding box and the target area. If the ratio is lower than the preset value, the scanning path is automatically adjusted.

[0091] The data repair operation uses an interpolation algorithm based on the nearest neighbor weighted average to fill in missing data regions, where the interpolated values ​​are calculated using the following normalization formula:

[0092] ;

[0093] in, These are the interpolated point coordinates. Let j be the weight of the j-th neighboring point. Let j be the coordinates of the j-th neighboring point. This represents the number of neighboring points;

[0094] The specific operations include: using the missing point as the center, the search radius... The initial setting is twice the average point distance of the point cloud, and the number of neighboring points. The default value is 5, and it will be increased gradually if it is insufficient. Find all neighboring points within the radius up to 5 times the average point distance. If the number of neighboring points is insufficient... This expands the search radius and enables higher-order interpolation;

[0095] Set interpolation error threshold When the actual interpolation error is greater than If so, then the backup repair strategy will be activated, including higher-order interpolation and data completion;

[0096] Noise processing involves statistical filtering to remove outliers and ensure the smoothness of point cloud data.

[0097] The data fusion module receives standardized data, aligns the data using a point cloud registration unit, converts two-dimensional vectors into three-dimensional references using a vector transformation unit, and outputs the fused data using a fusion algorithm unit. The process includes:

[0098] Extract feature points from the point cloud, including corner points, edge points, and planar points;

[0099] Registration is achieved by calculating the transformation matrix between point clouds using the iterative nearest-point algorithm, where the registration error is minimized using the following normalization formula:

[0100] ;

[0101] in, For registration error, Let be the transformation matrix. For points in the source point cloud, For the corresponding point in the target point cloud, Point-to-point quantity;

[0102] Set registration error threshold ,when If so, then re-perform the registration and adjust the transformation matrix parameters;

[0103] The process by which the vector transformation unit converts a two-dimensional vector into a three-dimensional reference includes: elevating the vector coordinates to three-dimensional space and mapping them based on the elevation information of the point cloud data;

[0104] The fusion algorithm unit adopts a weighted fusion strategy, which assigns weights based on the confidence levels of point clouds and vectors to generate fused data.

[0105] The process of extracting feature points from a point cloud includes:

[0106] Feature points are identified using curvature calculation and normal vector analysis;

[0107] Feature point matching is performed through descriptor alignment, including fast point feature histogram descriptors, where matching similarity is calculated using the following normalized formula:

[0108] ;

[0109] in, To score the similarity, and These are the descriptor vectors for two feature points;

[0110] Specifically, describing subvectors and The FPFH algorithm is used to generate the following: For each feature point, the histogram of differences in the normal vectors of its neighboring points is calculated, normalized to a 33-dimensional vector, and after similarity calculation, bidirectional matching verification is performed, i.e., simultaneous calculation... arrive and arrive The similarity is used as the average value as the final score;

[0111] Set similarity threshold ,when If the match fails, the feature points need to be extracted again and the matching parameters adjusted.

[0112] The transformation matrix calculation includes optimization of rotation and translation parameters, and the least squares method is used to minimize the registration error;

[0113] The weighting is based on the accuracy metrics of the data source, including the positioning accuracy of the laser point cloud and the geometric accuracy of the vector data.

[0114] The model generation module receives fused data, generates a basic model through 3D reconstruction units, adds features through detail enhancement units, and outputs the final model through quality verification units. The process includes:

[0115] Based on the fused data, a triangular mesh model is generated using the Poisson surface reconstruction algorithm;

[0116] The principle of the Poisson reconstruction algorithm is to solve the Poisson equation by constructing an implicit function of the point cloud. ,in The point cloud normal vector field is discretized and solved using the conjugate gradient method to generate an isosurface for extracting triangular meshes. The parameter settings include a depth of 8 and a sampling interval of 0.01.

[0117] The detail enhancement unit adds detail through multi-resolution analysis, including texture mapping and geometric subdivision;

[0118] Multi-resolution analysis uses wavelet transform: the model is decomposed into multiple levels, high-frequency components are used for detail enhancement, low-frequency components are used for basic shape, and texture mapping projects the 2D image onto the 3D model through UV mapping.

[0119] The smoothness of the reconstructed surface is evaluated by calculating the average curvature using the following normalization formula:

[0120] ;

[0121] in, For the mean curvature, Let be the curvature of the i-th vertex. The number of vertices;

[0122] The specific steps include: calculating the curvature of each vertex using the moving least squares method. That is, fitting the quadratic equation of the local surface, extracting the principal curvature and taking the absolute value. Before the curvature calculation, the model is preprocessed to remove isolated points and smooth noise.

[0123] Set curvature threshold ,when This triggers a detail smoothing process to optimize surface quality;

[0124] The quality verification unit evaluates the model's integrity, geometric accuracy, and visual realism, and outputs a quality score.

[0125] When the quality score falls below the threshold, the model reconstruction process is triggered, and data fusion and reconstruction are performed again.

[0126] The process of evaluating the geometric accuracy of a model using a quality verification unit includes:

[0127] The root mean square error between the model vertices and the original point cloud data is calculated as a geometric deviation index using the following normalization formula:

[0128] ;

[0129] in, The root mean square error, Let be the Euclidean distance between the i-th model vertex and its corresponding point cloud point. The number of vertices;

[0130] Set geometric error threshold ,when If the model's geometric accuracy is deemed unacceptable, the model reconstruction process will be triggered.

[0131] Visual realism is evaluated through rendering tests, including lighting consistency and shadow effects;

[0132] Specific testing methods: Phong lighting model was used to simulate diffuse reflection, specular reflection and ambient light, shadows were detected by depth mapping, and the consistency threshold was set to a brightness difference of less than 5%;

[0133] Integrity checks ensure the model can be used in subsequent applications by detecting voids and non-manifold structures in the model.

[0134] The quality score is calculated by combining geometric error, visual score, and integrity index, using a weighted average.

[0135] The visualization module receives the final model, generates a visualization scene through the rendering engine unit, allows user interaction through the interactive control unit, and exports the model file through the output interface unit. The process includes:

[0136] Load the final model data and apply physics to simulate lighting and materials through the rendering engine unit;

[0137] Rendering performance is evaluated by calculating the frame rate using the following normalization formula:

[0138] ;

[0139] in, For frame rate, For the number of frames rendered, For time intervals;

[0140] Specific operations include: using a high-performance timer at fixed time intervals. The system internally counts the number of rendered frames and monitors the GPU load. When the load exceeds 80%, it automatically enables level of detail technology to dynamically simplify the model.

[0141] Set frame rate threshold ,when If so, the rendering quality will be automatically reduced and the optimization mode will be enabled to ensure smoothness;

[0142] The interactive control unit supports user operations, including rotating, zooming, and panning views, as well as model editing functions;

[0143] The output interface unit supports exporting multiple file formats, including OBJ, STL, and point cloud formats;

[0144] Scene generation is performed in real time, accelerated by the graphics processor to ensure a smooth user experience.

[0145] The optimization and adjustment module receives the final model, evaluates its efficiency through the performance analysis unit, adjusts the modeling parameters through the parameter optimization unit, and outputs optimization instructions through the iterative update unit. The process includes:

[0146] The performance analysis unit evaluates model efficiency and monitors model generation time, memory usage, and computing resource utilization.

[0147] Dynamic adjustment algorithm: The gradient descent method is used to optimize the parameters, the learning rate is set to 0.01, and the objective function is the weighted sum of modeling efficiency and accuracy with a weight ratio of 1:1.

[0148] The parameter optimization unit dynamically adjusts modeling parameters, including reconstruction resolution and fusion weights, to balance accuracy and efficiency;

[0149] The iterative update unit automatically optimizes the system process based on performance feedback, reducing redundant calculations;

[0150] The optimization instructions include parameter adjustment suggestions and process refactoring commands to improve the overall system performance.

[0151] The user feedback module receives user input data, identifies user needs through the demand parsing unit, obtains operational feedback through the feedback collection unit, and adjusts system parameters through the adaptive learning unit. The process includes:

[0152] The requirement analysis unit identifies user requirements, parses the natural language and graphical instructions input by the user, and extracts key modeling requirements.

[0153] The feedback collection unit obtains feedback data through user interaction logs and satisfaction surveys;

[0154] The adaptive learning unit uses machine learning algorithms to analyze feedback patterns and optimize system default parameters.

[0155] System parameter adjustments include fusion strategy preferences, rendering quality settings, and output format priorities, to achieve a personalized modeling experience.

[0156] The operation steps of the modeling system that integrates laser point clouds and 2D vectors are as follows:

[0157] Step 1: Data Acquisition and Preprocessing

[0158] The system first starts working through the data acquisition module. The laser point cloud acquisition unit is responsible for generating 3D point cloud data, while the 2D vector import unit acquires 2D vector data. During the data acquisition process, the system sets key parameters, including scanning resolution and angle, and calculates point cloud density and coverage in real time to ensure data integrity. When there are missing or noisy points in the cloud data, the data repair operation uses an interpolation algorithm to repair them. The data preprocessing unit performs coordinate normalization and format standardization on the data and outputs standardized data. This step, through real-time verification and data purification, lays the foundation for subsequent fusion and avoids data deviation caused by environmental interference.

[0159] Step 2: Fusion of point cloud and vector data

[0160] After receiving standardized data, the data fusion module performs point cloud registration by feature point extraction and iterative nearest-point algorithm, calculates the transformation matrix to minimize registration error, and the vector conversion unit upscales the two-dimensional vector data to three-dimensional space and maps it based on point cloud elevation information to ensure coordinate consistency. The fusion algorithm unit adopts a weighted fusion strategy, dynamically allocating weights according to the confidence levels of point clouds and vectors to generate high-precision fused data. This step handles data conflicts in complex scenarios through adaptive weight adjustment and error threshold judgment, improving the stability and reliability of the fusion results.

[0161] Step 3: 3D Model Generation and Quality Verification

[0162] The model generation module initiates the 3D reconstruction process based on fused data. The 3D reconstruction unit uses the Poisson surface reconstruction algorithm to generate a triangular mesh base model. The detail enhancement unit adds texture and geometric details through multi-resolution analysis to enhance the realism of the model. The quality verification unit evaluates the model's integrity, geometric accuracy, and visual realism, and calculates indicators such as root mean square error. When the quality score is lower than the threshold, the model reconstruction process is triggered. This step intelligently resolves the contradictions between point clouds and vectors to prevent model holes and texture misalignment, ensuring that the output model meets engineering standards.

[0163] Step 4: Visualization and Interactive Control

[0164] The visualization module loads the final model data, the rendering engine unit applies physical-based rendering technology to simulate lighting and materials, and generates an immersive visualization scene. The interactive control unit supports user operations, including rotating, scaling, and panning the view, and exports files in various formats through the output interface unit. The system monitors rendering performance in real time, and automatically optimizes quality settings when performance falls below a threshold. This step improves user experience and model usability through rendering and interactive responses.

[0165] Step 5: System Optimization and User Feedback

[0166] The optimization and adjustment module continuously monitors model generation efficiency, the performance analysis unit evaluates resource usage, the parameter optimization unit dynamically adjusts reconstruction parameters to balance accuracy and speed, the user feedback module parses user input requirements, the feedback collection unit acquires operation data, and the adaptive learning unit optimizes system parameters through machine learning algorithms to achieve personalized settings. This step ensures the system's adaptive improvement through closed-loop feedback and iterative updates.

[0167] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0168] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A modeling system integrating laser point clouds and two-dimensional vectors, characterized in that, include: The data acquisition module uses the laser point cloud acquisition unit to generate three-dimensional point cloud data, acquires two-dimensional vector data through the two-dimensional vector import unit, and outputs standardized data through the data preprocessing unit. The data fusion module receives the standardized data, aligns the data through a point cloud registration unit, converts two-dimensional vectors into three-dimensional references through a vector conversion unit, and outputs fused data through a fusion algorithm unit. The model generation module receives the fused data, generates a basic model through the 3D reconstruction unit, adds features through the detail enhancement unit, and outputs the final model through the quality verification unit. The visualization module receives the final model, generates a visualization scene through the rendering engine unit, allows users to operate it through the interactive control unit, and exports the model file through the output interface unit. The optimization and adjustment module receives the final model, evaluates the model efficiency through the performance analysis unit, adjusts the modeling parameters through the parameter optimization unit, and outputs optimization instructions through the iterative update unit. The user feedback module receives user input data, identifies user needs through the demand parsing unit, obtains operational feedback through the feedback collection unit, and adjusts system parameters through the adaptive learning unit. The data fusion module receives the standardized data, aligns the data using a point cloud registration unit, converts the two-dimensional vector into a three-dimensional reference using a vector conversion unit, and outputs the fused data using a fusion algorithm unit. The process includes: Extract feature points from the point cloud, including corner points, edge points, and planar points; Registration is achieved by calculating the transformation matrix between point clouds using the iterative nearest-point algorithm, where the registration error is minimized using the following normalization formula: ； in, For registration error, The transformation matrix is... For points in the source point cloud, For the corresponding point in the target point cloud, Point-to-point quantity; Set registration error threshold ,when If so, then re-perform the registration and adjust the transformation matrix parameters; The process by which the vector transformation unit converts a two-dimensional vector into a three-dimensional reference includes: elevating the vector coordinates to three-dimensional space and mapping them based on the elevation information of the point cloud data; The fusion algorithm unit adopts a weighted fusion strategy, which assigns weights based on the confidence levels of point clouds and vectors to generate fused data; The process by which the model generation module receives the fused data, generates a basic model through a 3D reconstruction unit, adds features through a detail enhancement unit, and outputs the final model through a quality verification unit includes: Based on the fused data, a triangular mesh model is generated using the Poisson surface reconstruction algorithm; The detail enhancement unit adds detail through multi-resolution analysis, including texture mapping and geometric subdivision; The quality verification unit evaluates the model's integrity, geometric accuracy, and visual realism, and outputs a quality score. When the quality score falls below the threshold, the model reconstruction process is triggered, and data fusion and reconstruction are performed again.

2. The modeling system for fusing laser point clouds and two-dimensional vectors according to claim 1, characterized in that, The data acquisition module includes the following processes: generating 3D point cloud data using a laser point cloud acquisition unit, acquiring 2D vector data using a 2D vector import unit, and outputting standardized data using a data preprocessing unit. The laser scanning parameters, including scanning resolution, scanning angle, and scanning distance, are set through the laser point cloud acquisition unit. Calculate the density and coverage of point cloud data to ensure data integrity; When point cloud data contains missing data or noise, the data is repaired and interpolated and filtered. The point cloud data is normalized in coordinates and standardized in format by the data preprocessing unit, and standardized point cloud data is output. Meanwhile, the process of acquiring two-dimensional vector data by the two-dimensional vector import unit includes: importing vector file format, parsing vector elements, and performing geometric correction and scale alignment.

3. The modeling system for fusing laser point clouds and two-dimensional vectors according to claim 2, characterized in that, The process of calculating the density and coverage of point cloud data includes: The point cloud density threshold is defined as the minimum number of points per cubic meter. When the actual density is lower than the threshold, a rescanning mechanism is triggered. The coverage area is evaluated by calculating the intersection ratio between the point cloud bounding box and the target area. If the ratio is lower than the preset value, the scanning path is automatically adjusted. The data repair operation uses an interpolation algorithm based on the weighted average of nearest neighbors to fill in missing data areas; Noise processing involves statistical filtering to remove outliers and ensure the smoothness of point cloud data.

4. The modeling system for fusing laser point clouds and two-dimensional vectors according to claim 1, characterized in that, The process of extracting feature points from the point cloud includes: Feature points are identified using curvature calculation and normal vector analysis; Feature point matching is achieved through descriptor alignment, including fast point feature histogram descriptors; The transformation matrix calculation includes optimization of rotation and translation parameters, and the least squares method is used to minimize the registration error; The weighting is based on the accuracy metrics of the data source, including the positioning accuracy of the laser point cloud and the geometric accuracy of the vector data.

5. The modeling system for fusing laser point clouds and two-dimensional vectors according to claim 1, characterized in that, The process by which the quality verification unit evaluates the geometric accuracy of the model includes: The root mean square error between the model vertices and the original point cloud data is calculated as a geometric deviation index using the following normalization formula: ； in, The root mean square error, Let be the Euclidean distance between the i-th model vertex and its corresponding point cloud point. The number of vertices; Set geometric error threshold ,when If the model's geometric accuracy is deemed unacceptable, the model reconstruction process will be triggered. Visual realism is evaluated through rendering tests, including lighting consistency and shadow effects; Integrity checks ensure the model can be used in subsequent applications by detecting voids and non-manifold structures in the model. The quality score is calculated by combining geometric error, visual score, and integrity index, using a weighted average.

6. The modeling system for fusing laser point clouds and two-dimensional vectors according to claim 1, characterized in that, The visualization module receives the final model, generates a visualization scene through the rendering engine unit, allows user operation through the interactive control unit, and exports the model file through the output interface unit. Load the final model data and apply physics to simulate lighting and materials through the rendering engine unit; The interactive control unit supports user operations, including rotating, zooming, and panning views, as well as model editing functions; The output interface unit supports exporting multiple file formats, including OBJ, STL, and point cloud formats; Scene generation is performed in real time, accelerated by the graphics processor to ensure a smooth user experience.

7. The modeling system for fusing laser point clouds and two-dimensional vectors according to claim 1, characterized in that, The optimization and adjustment module receives the final model, evaluates the model efficiency through a performance analysis unit, adjusts the modeling parameters through a parameter optimization unit, and outputs optimization instructions through an iterative update unit. The process includes: The performance analysis unit evaluates model efficiency and monitors model generation time, memory usage, and computing resource utilization. The parameter optimization unit dynamically adjusts modeling parameters, including reconstruction resolution and fusion weights, to balance accuracy and efficiency; The iterative update unit automatically optimizes the system process based on performance feedback, reducing redundant calculations; The optimization instructions include parameter adjustment suggestions and process refactoring commands to improve the overall system performance.

8. The modeling system for fusing laser point clouds and two-dimensional vectors according to claim 1, characterized in that, The user feedback module receives user input data, identifies user needs through a demand parsing unit, obtains operational feedback through a feedback collection unit, and adjusts system parameters through an adaptive learning unit. The process includes: The requirement analysis unit identifies user needs, parses the natural language and graphical instructions input by the user, and extracts key modeling requirements. The feedback collection unit obtains feedback data through user interaction logs and satisfaction surveys; The adaptive learning unit uses machine learning algorithms to analyze feedback patterns and optimize system default parameters. System parameter adjustments include fusion strategy preferences, rendering quality settings, and output format priorities, to achieve a personalized modeling experience.

Images