Method for extracting and storing along-line building space data based on traffic digital twinning

Abstract

The application discloses a method for extracting and storing along-line building space data based on traffic digital twinning, which comprises the following steps: obtaining along-line engineering CAD data and traffic digital twinning base map data, and generating a joint point set input structure; calling a coherent point drift algorithm to establish a continuous deformation mapping relationship, and generating an initial deformation field; introducing a Bayesian parameter estimation mechanism to generate a Bayesian deformation field; performing segmented deformation inference and cross-segment continuous consistency processing on the joint point set input structure to generate an along-line consistent deformation field and a segmented uncertainty distribution; performing discretization inversion processing and substantial re-projection processing to generate corrected building geometric data; generating building attribute data based on the along-line consistent deformation field; and completing data storage based on the corrected building geometric data and the building attribute data. The application realizes high-precision alignment, stable correction and automatic storage of along-line building space data.

Description

Technical Field

[0001] This invention relates to the field of GIS spatial data processing and spatial database storage technology, and in particular to a method for extracting and storing spatial data of buildings along transportation routes based on traffic digital twins. Background Technology

[0002] With the advancement of digital twin city and transportation infrastructure digitalization, a large amount of historical engineering CAD data needs to be converted into building spatial data that can be directly used in GIS systems and spatial databases to support real estate management, infrastructure surveys, and transportation digital twin applications. However, existing engineering CAD data is usually expressed using an independent engineering coordinate system, and its spatial reference method differs significantly from that of GIS systems. Direct conversion can easily lead to spatial offset problems.

[0003] In existing technologies, building outline features and building annotation information are usually stored separately on different layers. Attribute extraction often relies on manual or simple spatial connection operations, making it difficult to achieve automated and batch processing while ensuring spatial consistency. Especially in roadside engineering scenarios, building features are linearly distributed along the road, and local deformation, coordinate drift, and segmentation errors are easily amplified, leading to unstable spatial matching.

[0004] On the other hand, existing coordinate processing technologies generally rely on fixed projection parameters or static projection files, making it difficult to adapt to non-standard central meridians and engineering projection parameter configurations present in engineering projects, resulting in insufficient accuracy in the conversion of engineering coordinates to network map coordinates. Simultaneously, traditional data entry processes focus on geometric transformations, lacking systematic processing of building attribute data and spatial uncertainties, which affects the reliability of spatial analysis in subsequent digital twin scenarios.

[0005] Therefore, how to provide a method for extracting and storing spatial data of buildings along the route based on traffic digital twins is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] One objective of this invention is to propose a method for extracting and storing spatial data of buildings along transportation routes based on traffic digital twins. This invention uses CAD data of engineering projects along the route and traffic digital twin base map data as its data foundation. By constructing a linear reference structure and a joint point set input structure along the route, it introduces a coherent point drift algorithm and a Bayesian parameter estimation mechanism to progressively complete the continuous deformation mapping from the engineering coordinate domain to the traffic digital twin base map coordinate domain, segmented consistency inference, and uncertainty modeling. Furthermore, it combines the segmented uncertainty distribution to perform discretized inversion processing of engineering projection parameters, achieving stable reprojection and accurate alignment of building outline elements and building annotation elements. Based on this, this invention automatically generates building attribute data through annotation correspondence and annotation association confidence, and uniformly writes the corrected building geometric data and building attribute data into the traffic digital twin spatial database. This invention effectively improves the stability, consistency, and automation of spatial data processing for buildings along the route, and is suitable for large-scale building spatial data construction and management in traffic digital twin scenarios.

[0007] The method for extracting and storing spatial data of buildings along a transportation route based on traffic digital twins according to embodiments of the present invention includes the following steps:

[0008] Acquire CAD data of engineering projects along the route and digital twin base map data of traffic, construct a linear reference structure along the route, and generate a joint point set input structure;

[0009] The coherent point drift algorithm is invoked, and the joint point set input structure is used as the multi-source point set input to establish a continuous deformation mapping relationship from the engineering coordinate domain to the traffic digital twin base map coordinate domain, thereby generating the initial deformation field.

[0010] In the coherent point drift algorithm, a Bayesian parameter estimation mechanism is introduced. The initial deformation field is used as the posterior inference initialization input, and the posterior inference processing is performed on the joint point set input structure to obtain the Bayesian deformation field.

[0011] Based on the linear reference structure along the line, the joint point set input structure is divided into a set of segmented subsets along the line according to the fixed mileage segment length. A conditional coherent point drift sub-model is constructed, and segmented deformation inference and cross-segment continuous consistency processing are performed to obtain the consistent deformation field along the line and the segmented uncertainty distribution.

[0012] Based on the consistent deformation field and piecewise uncertainty distribution along the line, discretization inversion processing and substantial reprojection processing are performed to generate corrected building geometry data and corrected annotation data.

[0013] Based on the consistent deformation field along the line, coordinate mapping processing is performed on the correction annotation data to determine the annotation correspondence between the correction annotation data and the correction building geometry data, and building attribute data is calculated and generated.

[0014] Geometric standardization and topological consistency correction are performed on the corrected building geometry data, and the data is written into the traffic digital twin spatial database along with the building attribute data.

[0015] Optionally, the generation of the joint point set input structure includes:

[0016] Acquire CAD data of the project along the route, perform layer parsing processing, and identify and extract building outline elements representing the outer boundary of the building and building annotation elements representing building text labels;

[0017] Geometric discretization and equidistant sampling are performed on the building outline elements to determine the set of building outline control points.

[0018] Perform geometric representation unification processing on architectural annotation elements, mapping architectural annotation elements to annotation point sets;

[0019] Obtain traffic digital twin base map data, extract road centerline elements representing the direction of the road center, perform geometric discretization processing, and convert them into a set of geometric points of the road centerline;

[0020] Select a set of reference points from the digital twin base map data of the transportation system;

[0021] Based on the geometric point set of the road centerline, a linear reference structure is constructed along the line as a unified spatial index framework to complete the position calibration of the building outline control point set and annotation point set along the line.

[0022] The control point set and annotation point set of the building outline along the completed location, as well as the geometric point set of the road centerline and the base map reference point set, are organized in a unified manner to generate a joint point set input structure.

[0023] Optionally, the generation of the initial deformation field includes:

[0024] The joint point set input structure is processed by point set role classification, and the building outline control point set, the road centerline geometric point set, and the base map reference point set are marked as input point sets participating in deformation estimation, respectively.

[0025] Based on the engineering coordinate domain and the traffic digital twin base map coordinate domain, the source point set and target point set of the coherent point drift algorithm are determined. The source point set is composed of the point set representation of the building outline control point set and the road centerline geometric point set in the engineering coordinate domain, and the target point set is composed of the point set representation of the base map reference point set in the traffic digital twin base map coordinate domain.

[0026] In the coherent point drift algorithm, a point set probability correspondence is established between the source point set and the target point set. The source point set is regarded as a probability distribution sample generated from the target point set. Continuous non-rigid deformation estimation is performed to obtain a continuous deformation mapping relationship.

[0027] Based on the continuous deformation mapping relationship, the overall deformation calculation is performed on the control point set of the building outline and the geometric point set of the road centerline to generate the initial deformation field.

[0028] Optionally, the generation of the Bayesian deformation field includes:

[0029] In the continuous deformation mapping framework of the coherent point drift algorithm, a Bayesian parameter estimation mechanism is introduced to model the internal representation of the initial deformation field as random variables and to model the matching uncertainty formed under the action of continuous deformation mapping as random variables, thereby generating random variable representations.

[0030] Based on the joint point set input structure, a probabilistic observation model for point set matching is constructed, and prior constraints are introduced to constrain the smoothness of the continuous deformation space.

[0031] Using the initial deformation field as the initial input for posterior inference, Bayesian posterior inference processing is performed on the joint point set input structure under the combined action of the probabilistic observation model and prior constraints. This updates the matching probability distribution of the joint point set input structure and simultaneously updates the continuous deformation mapping relationship, resulting in a converged posterior probability distribution.

[0032] After completing the Bayesian posterior inference process, statistical features of the continuous deformation mapping relationship are extracted based on the convergent posterior probability distribution to generate the Bayesian deformation field.

[0033] Optionally, the generation of the consistent deformation field and piecewise uncertainty distribution along the line includes:

[0034] Based on the linear reference structure along the route, the input structure of the joint point set is divided into segments along the route according to the preset fixed mileage segment length to obtain a set of segments along the route.

[0035] For each segment subset along the line, a corresponding conditional coherence point drift sub-model is constructed by combining the coherence point drift algorithm. The Bayesian deformation field is used as the deformation initialization input to perform local deformation inference processing on the segment subset along the line and generate the initial segment deformation result.

[0036] In the conditional coherent point drift sub-model, the tangential direction information corresponding to the direction along the road in the geometric point set of the road centerline is introduced as the direction consistency constraint input, and the direction consistency modulation processing is performed on the initial segmented deformation result to obtain the segmented deformation result;

[0037] Based on the piecewise deformation results and the Bayesian deformation field, cross-segment continuous consistency processing is performed on the piecewise deformation results between adjacent segment subsets along the line to generate a consistent deformation field along the line.

[0038] After completing the cross-segment continuous consistency processing, the deformation uncertainty of each segment subset along the line during the consistency process is aggregated to form a segment uncertainty distribution.

[0039] Optionally, the generation of the corrected architectural geometry data and corrected annotation data includes:

[0040] Based on the consistent deformation field and piecewise uncertainty distribution along the line, the discretized inversion search space of the central meridian parameters and the east pseudo offset parameters in the engineering projection coordinate system is determined.

[0041] In the discretized inversion search space, a set of parameter combinations is constructed. For each parameter combination, a tentative reprojection process is performed on the building outline elements and building annotation elements in combination with the consistent deformation field along the line, and the spatial deviation index is calculated.

[0042] Based on the piecewise uncertainty distribution, uncertainty-perceived weighted processing is performed on the spatial deviation index to generate a comprehensive evaluation result.

[0043] Based on the comprehensive evaluation results, the optimal values ​​for the central meridian parameter and the east pseudo-offset parameter are determined from the parameter combination set, forming the central meridian parameter value set and the east pseudo-offset parameter value set.

[0044] Based on the set of parameters of the central meridian and the set of parameters of the east pseudo offset, a substantial reprojection process is performed on the building outline elements and building annotation elements to generate corrected building geometric data and corrected annotation data.

[0045] Optionally, the generation of the building attribute data includes:

[0046] Based on the consistent deformation field along the line, coordinate mapping processing is performed on the correction annotation data to map it to a unified coordinate domain consistent with the correction building geometry data, thus obtaining the annotation point set representation;

[0047] Within a unified coordinate domain, based on the spatial consistency results of the annotation point set representation and the building outline control point set in the consistent deformation field along the line, the spatial matching relationship is calculated to determine the one-to-one annotation correspondence between the corrected annotation data and the corrected building geometry data;

[0048] Based on the annotation correspondence, the spatial consistency score between the corrected annotation data and the corresponding building outline elements is calculated, and uncertainty-aware modulation processing is performed in combination with the deformation uncertainty distribution to generate annotation association confidence.

[0049] If the confidence level of the annotation association meets the preset confidence threshold, the building identification information carried in the corrected annotation data will be attached to the corresponding building outline elements.

[0050] Based on the confidence level of the association between building identification information and corresponding annotations, attribute calculations are performed on building outline elements to generate building attribute data.

[0051] Optionally, the data import process includes:

[0052] Based on the preset geometric data storage specifications in the transportation digital twin spatial database, geometric standardization processing is performed on the corrected building geometric data;

[0053] Based on the topological constraint rules of the traffic digital twin spatial database, topological consistency correction is performed on the corrected building geometric data after geometric standardization to generate corrected building geometric data.

[0054] The corrected building geometry data that meets the constraints of geometric standardization and topological consistency are structured and organized with the corresponding generated building attribute data to generate building space data records.

[0055] According to the spatial indexing rules and attribute field constraints of the transportation digital twin spatial database, the building spatial data records are written into the transportation digital twin spatial database.

[0056] The beneficial effects of this invention are:

[0057] First, this invention constructs a linear reference structure along the route and a joint point set input structure, and introduces a Bayesian parameter estimation mechanism based on the coherent point drift algorithm to achieve continuous deformation mapping and uncertainty modeling from the engineering coordinate domain to the traffic digital twin base map coordinate domain. This ensures that the building outline control point set, annotation point set, and road centerline geometric point set maintain spatial consistency at both the overall and local scales, effectively reducing alignment errors caused by engineering coordinate differences, data noise, and local distortions, and significantly improving the stability and reliability of the extracted building spatial data along the route.

[0058] Secondly, this invention constructs a conditional coherent point drift sub-model by segmenting subsets along the road, and combines the tangential direction constraint of the geometric point set of the road centerline with the continuous consistency processing across segments to generate a consistent deformation field and segmented uncertainty distribution along the road. This enables the continuous deformation mapping to have clear structural constraints along the road, avoiding the breakage and splicing errors of the segmented deformation results, thereby significantly improving the continuity and overall consistency of building spatial data processing in long-distance road scenarios.

[0059] Furthermore, this invention utilizes the consistent deformation field and segmented uncertainty distribution along the route to perform discretization inversion processing on the central meridian parameters and the eastern pseudo-offset parameters. It also automatically generates building attribute data by combining the annotation correspondence and annotation association confidence, thereby achieving high-quality storage of corrected building geometric data and building attribute data. This reduces manual intervention and reliance on experience, and improves the automated construction efficiency and application reliability of building spatial data along the route in the traffic digital twin spatial database. Attached Figure Description

[0060] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0061] Figure 1 This is an overall flowchart of a method for extracting and storing spatial data of buildings along a transportation route based on traffic digital twins, as proposed in this invention.

[0062] Figure 2 This is a schematic diagram of continuous deformation inference based on the coherence point drift algorithm and the introduction of a Bayesian parameter estimation mechanism in this invention.

[0063] Figure 3 This is a schematic diagram of the segmented deformation inference and cross-segment continuity consistency processing based on the linear reference structure along the line in this invention. Detailed Implementation

[0064] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0065] refer to Figure 1-3 The method for extracting and storing spatial data of buildings along the transportation route based on digital twins includes the following steps:

[0066] Acquire CAD data of the engineering projects along the route and digital twin base map data of the traffic system. Extract building outline elements and building annotation elements from the CAD data of the engineering projects along the route. Extract road centerline elements and base map reference point set from the digital twin base map data of the traffic system. Discretize the building outline elements into a set of building outline control points, the building annotation elements into a set of annotation points, and the road centerline elements into a set of road centerline geometric points. Construct a linear reference structure along the route based on the road centerline geometric point set. Map the building outline control point set and the annotation point set to the linear reference structure along the route to generate a joint point set input structure. The joint point set input structure includes the building outline control point set, the annotation point set, the road centerline geometric point set, and the base map reference point set.

[0067] Based on the joint point set input structure, the coherent point drift algorithm is called to take the building outline control point set, road centerline geometric point set, and base map reference point set in the joint point set input structure as multi-source point set input. In the coherent point drift algorithm, a continuous deformation mapping relationship from the engineering coordinate domain to the traffic digital twin base map coordinate domain is established to generate the initial deformation field.

[0068] In the coherent point drift algorithm, a Bayesian parameter estimation mechanism is introduced, which models the deformation parameters and noise parameters as random variables, uses the initial deformation field as the posterior inference initialization input, performs posterior inference processing on the joint point set input structure, and obtains the Bayesian deformation field, which includes the deformation field expectation and the deformation uncertainty distribution.

[0069] Based on the linear reference structure along the route, the joint point set input structure is divided into a set of segmented subsets along the route according to a fixed mileage segment length. For each segmented subset along the route, a conditional coherent point drift sub-model is constructed. The tangential direction of the geometric point set of the road centerline is used as the directional consistency constraint input of the sub-model, and the Bayesian deformation field is used as the deformation initialization input of the sub-model. Segmented deformation inference is performed and cross-segment continuous consistency processing is performed to obtain the consistent deformation field along the route and the segmented uncertainty distribution.

[0070] Based on the consistent deformation field and piecewise uncertainty distribution along the line, the central meridian parameter and the east pseudo-offset parameter in the engineering projection coordinate system are discretized and inverted to obtain the set of central meridian parameter values ​​and the set of east pseudo-offset parameter values. Based on the set of central meridian parameter values ​​and the set of east pseudo-offset parameter values, the building outline elements and building annotation elements are substantially reprojected to convert the engineering coordinates into projection coordinates suitable for web map services, and generate corrected building geometric data and corrected annotation data.

[0071] Based on the consistent deformation field along the line, coordinate mapping processing is performed on the correction annotation data. Within a unified coordinate domain, based on the spatial consistency results of the annotation point set and the building outline control point set in the consistent deformation field along the line, the annotation correspondence between the correction annotation data and the correction building geometry data is determined. Based on the annotation correspondence, the building identification information carried in the correction annotation data is attached to the corresponding building outline elements, and the annotation association confidence is generated. Based on the building identification information and the annotation association confidence, the number of building floors and the total building area are calculated, and building attribute data is generated. The building attribute data includes the number of building floors and the total building area.

[0072] Geometric standardization and topological consistency correction are performed on the corrected building geometric data. The corrected building geometric data and building attribute data are then written into the traffic digital twin spatial database to complete the extraction and storage of building spatial data along the route.

[0073] In this embodiment, the generation of the joint point set input structure includes:

[0074] Acquire CAD data of the projects along the route, perform layer parsing processing on the CAD data of the projects along the route, identify and extract building outline elements representing the outer boundary of the buildings and building annotation elements representing building text labels. The building outline elements are represented in a closed polyline geometric form, and the building annotation elements are represented in a point geometric form and carry building identification information.

[0075] Geometric discretization is performed on the building outline elements. Outline inflection points are extracted based on the connection relationship of each line segment in the building outline elements. Equidistant sampling is performed between adjacent inflection points at a fixed interval. The resulting set of sampling points is determined as the building outline control point set, which is used to characterize the geometric shape of the building outline elements.

[0076] To ensure that the building outline control point set remains consistent under different data parsing implementations, a fixed spacing is written as a preset sampling rule into the sampling execution condition: the outline inflection point of the building outline element is taken as the control point that must be retained, and equal-distance sampling is performed only when the fixed spacing sampling condition is met between adjacent outline inflection points. This ensures that the building outline control point set remains consistent at the inflection point position and avoids the generation of overly dense sampling points in short segments, which would affect the stability of the probability correspondence of subsequent point sets.

[0077] The architectural annotation elements are processed with a unified geometric representation. The annotation base point of the architectural annotation element in the engineering CAD coordinate domain is used as the spatial representation of the annotation point set. The architectural annotation elements are mapped to the annotation point set, so that the annotation point set and the architectural outline control point set are in the same engineering coordinate domain.

[0078] Acquire traffic digital twin base map data, extract road centerline elements representing the direction of the road center from the traffic digital twin base map data, and perform geometric discretization processing on the road centerline elements to convert the road centerline elements into a set of road centerline geometric points to characterize the linear direction characteristics of the space along the road.

[0079] A set of reference points is selected from the traffic digital twin base map data. The set of reference points consists of reference points that have stable spatial positions in the traffic digital twin base map data and can establish a spatial correspondence between the engineering CAD data and the traffic digital twin base map data. These reference points are used as spatial constraint inputs for subsequent point set deformation inference.

[0080] To ensure that the base map reference point set can be used for spatial correspondence constraints between engineering CAD data and traffic digital twin base map data, the selection of the base map reference point set adopts stability judgment processing: the coordinate values ​​of candidate reference points are read at different scaling levels of traffic digital twin base map data and consistency verification is performed. If the coordinate values ​​are consistent and the candidate reference point has a unique spatial location identifier in the base map, the candidate reference point is determined as a member of the base map reference point set and used for constraints of subsequent continuous deformation mapping relationships.

[0081] Based on the geometric point set of the road centerline, a linear reference structure is constructed along the line. Using the linear reference structure along the line as a unified spatial index framework, the point along the line is projected and positioned on the control point set and annotation point set of the building outline. The projection position of each point on the geometric point set of the road centerline is determined, and the position along the line of the control point set and annotation point set of the building outline is calibrated according to the cumulative mileage coordinates corresponding to the projection position.

[0082] The construction of the linear reference structure along the route specifically involves connecting the geometric point set of the road centerline in the road centerline elements in sequence to form a directed polyline structure, and taking the starting point of the polyline as the starting point of the mileage along the route. The spatial distance between adjacent geometric points of the road centerline is accumulated segment by segment, and a unique cumulative mileage coordinate is assigned to each geometric point of the road centerline, thereby establishing a one-to-one mapping relationship between the geometric point set of the road centerline and the cumulative mileage coordinate, thus forming a linear reference structure along the route.

[0083] The control point set and annotation point set of the building outline along the completed location, as well as the geometric point set of the road centerline and the base map reference point set, are uniformly organized to generate a joint point set input structure. The joint point set input structure is used as the input data basis for the subsequent coherent point drift algorithm to perform continuous deformation inference and segmented consistency processing.

[0084] In this embodiment, the generation of the initial deformation field includes:

[0085] The input structure of the joint point set is processed by dividing the point set into roles. The building outline control point set, the road centerline geometric point set, and the base map reference point set are marked as input point sets participating in deformation estimation. The building outline control point set and the road centerline geometric point set are used as point sets to be aligned, and the base map reference point set is used as a spatial constraint point set to stabilize the overall spatial position in the deformation inference process.

[0086] Based on the engineering coordinate domain and the traffic digital twin base map coordinate domain, the source point set and target point set of the coherent point drift algorithm are determined. The source point set is composed of the point set representation of the building outline control point set and the road centerline geometric point set in the engineering coordinate domain, and the target point set is composed of the point set representation of the base map reference point set in the traffic digital twin base map coordinate domain.

[0087] The engineering coordinate domain refers to a Cartesian coordinate system established based on engineering design benchmarks, used to express the spatial positions of building outline elements, building annotation elements, and road centerline elements in the CAD data of the project along the route. This engineering coordinate domain is based on the local engineering origin and does not include map projection parameters. The traffic digital twin base map coordinate domain refers to the map coordinate system used by the traffic digital twin base map data, used to express the spatial positions of road centerline elements and base map reference point sets in the traffic digital twin base map. This traffic digital twin base map coordinate domain is constructed based on preset map projection parameters and is used to support spatial visualization and spatial calculation in the traffic digital twin scenario.

[0088] In the coherent point drift algorithm, a point set probability correspondence is established between the source point set and the target point set. The source point set is regarded as a probability distribution sample generated from the target point set. Under the constraint of this probability model, continuous non-rigid deformation estimation is performed to obtain a continuous deformation mapping relationship from the engineering coordinate domain to the traffic digital twin base map coordinate domain.

[0089] The point set probability correspondence is established using the public probability matching mechanism of the coherent point drift algorithm: the matching probability between each source point and the target point is used as the iterative update object, and the error contribution of continuous non-rigid deformation estimation is weighted according to the matching probability in each iteration, so that the update of the continuous deformation mapping relationship is dominated by high confidence matching pairs and the disturbance of abnormal matching to the continuous deformation mapping relationship is reduced.

[0090] Based on the continuous deformation mapping relationship, the overall deformation calculation is performed on the building outline control point set and the road centerline geometric point set in the joint point set input structure to generate an initial deformation field that describes the initial correspondence between the engineering coordinate domain and the traffic digital twin base map coordinate domain. The initial deformation field serves as the deformation initialization input for subsequent Bayesian deformation inference and segmented consistency processing.

[0091] In this embodiment, the generation of the Bayesian deformation field includes:

[0092] In the continuous deformation mapping framework of the coherent point drift algorithm, a Bayesian parameter estimation mechanism is introduced to model the internal representation of the initial deformation field used to characterize the continuous deformation mapping relationship from the engineering coordinate domain to the traffic digital twin base map coordinate domain. The matching uncertainty formed by the joint point set input structure under the continuous deformation mapping of the coherent point drift algorithm is also modeled as a random variable, generating a random variable representation for Bayesian posterior inference.

[0093] The random variable modeling of the internal representation of the initial deformation field adopts the "spatial location association" modeling method: the deformation values ​​of the continuous deformation mapping relationship at each spatial location are regarded as updatable random quantities, and the matching fluctuations formed by the joint point set input structure under the action of continuous deformation mapping are regarded as noise random quantities. This allows the posterior inference to output the deformation uncertainty distribution while updating the continuous deformation mapping relationship, thereby ensuring that the Bayesian deformation field not only provides the deformation field expectation but also provides the source basis of the deformation uncertainty distribution.

[0094] Based on the joint point set input structure, a probabilistic observation model for point set matching is constructed in the coherent point drift algorithm that introduces a Bayesian parameter estimation mechanism. The spatial residual formed by the control point set of building contour and the geometric point set of road centerline under the action of continuous deformation mapping is used as the observation term in the probabilistic observation model for modeling. Prior constraints are introduced in this probabilistic observation model to constrain the spatial smoothness of continuous deformation, so as to limit the variation range of deformation mapping at adjacent spatial locations.

[0095] The observations of the probabilistic observation model adopt the open modeling method of spatial residuals: the coordinate difference between the location of the building outline control point set and the road centerline geometric point set after continuous deformation mapping and the location of their matching target point is used as the spatial residual input, and spatial smoothness constraints are introduced into the prior constraints to limit the deformation change amplitude at adjacent spatial locations, thereby realizing the continuous deformation spatial smoothness as a computable constraint and using it for posterior inference iterative update.

[0096] The initial deformation field is used as the initial input for posterior inference. Under the combined action of the probabilistic observation model and prior constraints, Bayesian posterior inference processing is performed on the joint point set input structure. In each iteration, the matching probability distribution of the joint point set input structure is updated based on the current random variable representation, and the continuous deformation mapping relationship is updated synchronously until the posterior inference result satisfies the convergence condition, and the converged posterior probability distribution is obtained.

[0097] The convergence condition adopts an implementable threshold determination method: during the posterior inference iteration process, the iterative change of the matching probability distribution and the iterative change of the continuous deformation mapping relationship are calculated respectively. When both types of changes are less than the preset threshold and the preset number of iterations is continuously met, the posterior inference result is determined to meet the convergence condition and the converged posterior probability distribution is output for subsequent statistical feature extraction processing of the Bayesian deformation field.

[0098] After completing the Bayesian posterior inference processing, statistical features of the continuous deformation mapping relationship are extracted based on the convergent posterior probability distribution to generate a Bayesian deformation field. The Bayesian deformation field includes the deformation field expectation used to characterize the average deformation trend, and the deformation uncertainty distribution used to characterize the degree of deformation uncertainty at the spatial location. The Bayesian deformation field serves as the input for subsequent segmented deformation inference and projection parameter inversion processing along the line.

[0099] The specific steps for extracting the statistical features of the continuous deformation mapping relationship are as follows: after the posterior probability distribution reaches a convergent state, statistical convergence processing is performed on the posterior distribution results corresponding to the continuous deformation mapping relationship at each spatial location. By weighting and summarizing different deformation mapping results in the posterior probability distribution according to their probability weights, the central trend of the continuous deformation mapping relationship at each spatial location is determined, forming the deformation field expectation used to characterize the overall continuous deformation behavior. At the same time, statistical calculations are performed on the dispersion of the deformation mapping results in the posterior probability distribution at each spatial location to obtain the spatial distribution features reflecting the uncertainty of the deformation mapping results, forming the deformation uncertainty distribution.

[0100] In this embodiment, the generation of the consistent deformation field and piecewise uncertainty distribution along the line includes:

[0101] Based on the linear reference structure along the line, the joint point set input structure is divided into segments along the line according to the preset fixed mileage segment length, resulting in a set of segments along the line arranged in the order along the line. Each segment in the set of segments along the line contains a set of building outline control points, a set of annotation points, and a set of road centerline geometric points within the corresponding mileage interval.

[0102] For each segment subset along the line, a corresponding conditional coherent point drift sub-model is constructed by combining the coherent point drift algorithm. The Bayesian deformation field is used as the deformation initialization input of the conditional coherent point drift sub-model, so that the conditional coherent point drift sub-model performs local deformation inference processing on the segment subset along the line in the initial state of continuous deformation mapping, generating the initial segment deformation result without introducing directional consistency constraints.

[0103] The local deformation inference of the conditional coherent point drift sub-model adopts a local iterative method consistent with the coherent point drift algorithm: a local point set probability correspondence is established in the segmented subset along the line and the local continuous deformation mapping relationship is updated based on the correspondence. The output of the local continuous deformation mapping relationship is determined as the initial segmented deformation result, so that the initial segmented deformation result can be directly modulated by the subsequent directional consistency constraint input to form the segmented deformation result.

[0104] In the conditional coherent point drift sub-model, based on the initial segmented deformation results, the tangential direction information corresponding to the direction along the road in the geometric point set of the road centerline is introduced as the direction consistency constraint input. The direction consistency modulation processing is performed on the initial segmented deformation results so that the local deformation inference results in the segmented subset along the road meet the continuous direction consistency requirement in the direction along the road, and the segmented deformation results corresponding to each segmented subset along the road are obtained.

[0105] The tangential direction of the road centerline geometric point set is obtained by the publicly available adjacent point difference calculation method: the coordinate difference of adjacent road centerline geometric points in the road centerline geometric point set is normalized to obtain the tangential direction, and the tangential direction is used as the input of the direction consistency constraint to the initial segment deformation result, so that the deformation change of the segment deformation result along the line direction remains continuous and consistent with the mileage direction of the linear reference structure along the line.

[0106] Based on the deformation uncertainty distribution at the corresponding spatial location in the segmented deformation result and the Bayesian deformation field, cross-segment continuous consistency processing is performed on the segmented deformation results between adjacent segmented subsets along the line. By adjusting the deformation mapping difference between adjacent segments, the segmented deformation results of adjacent segmented subsets along the line are made to remain continuous and consistent in the direction along the line, thus generating a consistent deformation field along the line.

[0107] The cross-segment continuous consistency processing adopts the boundary interval coupling implementation method: extract the boundary point set in the boundary mileage interval of adjacent segment subsets along the line and calculate the deformation difference of adjacent segment deformation results at the boundary point set. The deformation difference is used as the consistency adjustment object and the deformation uncertainty distribution at the corresponding position in the Bayesian deformation field is introduced as the adjustment weight. This increases the adjustment intensity at high uncertainty positions and suppresses cross-segment deformation abrupt changes, while reducing the adjustment intensity at low uncertainty positions and preserving local deformation details, thereby obtaining a consistent deformation field along the line.

[0108] After completing the cross-segment continuous consistency processing, the deformation uncertainty of each segment subset along the line during the consistency process is aggregated to form a segment uncertainty distribution corresponding to the consistent deformation field along the line. The segment uncertainty distribution is used to characterize the uncertainty level of deformation inference results in different intervals along the line.

[0109] In this embodiment, the generation of the corrected building geometry data and the corrected annotation data includes:

[0110] Based on the consistent deformation field and piecewise uncertainty distribution along the line, the discretized inversion search space of the central meridian parameter and the east pseudo offset parameter in the engineering projection coordinate system is determined. The discretized inversion search space of the central meridian parameter is composed of candidate values ​​within a preset longitude range, and the discretized inversion search space of the east pseudo offset parameter is composed of candidate values ​​within a preset translation range.

[0111] In the discretized inversion search space of the central meridian parameter and the discretized inversion search space of the east pseudo offset parameter, a set of parameter combinations is constructed. For each parameter combination, the building outline and building annotation elements are subjected to tentative reprojection processing in combination with the consistent deformation field along the line. The spatial deviation index of the building outline and building annotation elements in the coordinate domain of the traffic digital twin base map under the corresponding parameter combination is calculated.

[0112] The spatial deviation index adopts a two-factor consistency calculation method: the outline deviation value of the building outline element in the coordinate domain of the traffic digital twin base map after the trial reprojection process and the annotation deviation value of the building annotation element in the coordinate domain of the traffic digital twin base map are calculated separately. The outline deviation value and the annotation deviation value are used together as the evaluation input for the effectiveness of the parameter combination, so that the discretization inversion evaluation of the central meridian parameter and the east pseudo offset parameter simultaneously constrains the consistency between the building geometric position and the annotation position.

[0113] Based on the piecewise uncertainty distribution, uncertainty-aware weighting processing is performed on the spatial deviation index. The segments along the line located in the high uncertainty interval are given lower weights in the spatial deviation index calculation, while the segments along the line located in the low uncertainty interval are given higher weights in the spatial deviation index calculation, generating a comprehensive evaluation result for evaluating the effectiveness of the parameter combination.

[0114] The uncertainty perception weighting adopts a weighting method that is directly applicable: based on the segmented uncertainty distribution, segmented weight coefficients are assigned to each segmented subset along the line, so that the segmented subsets along the line with higher segmented uncertainty distributions correspond to smaller segmented weight coefficients, and the segmented subsets along the line with lower segmented uncertainty distributions correspond to larger segmented weight coefficients. Based on the segmented weight coefficients, the spatial deviation index is weighted and aggregated to generate a comprehensive evaluation result, so that the discretization inversion process is driven by the segmented uncertainty distribution and has a structured evaluation basis that is different from conventional error fitting.

[0115] Based on the comprehensive evaluation results, the optimal values ​​for the central meridian parameter and the east pseudo-offset parameter are determined from the parameter combination set, forming the central meridian parameter value set and the east pseudo-offset parameter value set.

[0116] Based on the set of parameters of the central meridian and the set of parameters of the east pseudo offset, a substantial reprojection process is performed on the building outline elements and building annotation elements to convert the engineering coordinates under the engineering projection coordinate system into projection coordinates suitable for online map services, and generate corrected building geometric data and corrected annotation data.

[0117] In this embodiment, the generation of the building attribute data includes:

[0118] Based on the consistent deformation field along the line, coordinate mapping processing is performed on the correction annotation data to map the corresponding annotation point set in the correction annotation data from the engineering projection coordinate domain to a unified coordinate domain consistent with the correction architectural geometry data, so as to obtain the annotation point set representation under the unified coordinate domain.

[0119] Within a unified coordinate domain, based on the spatial consistency results of the annotation point set representation and the building outline control point set in the consistent deformation field along the line, the spatial matching relationship between the annotation point set representation and the building outline control point set is calculated, and under the condition that the spatial matching relationship meets the preset consistency conditions, the one-to-one annotation correspondence between the corrected annotation data and the corrected building geometric data is determined.

[0120] The preset consistency condition adopts a reproducible judgment method: calculate the nearest distance from the annotation point set to the corresponding building outline element and calculate the local deformation consistency residual of the annotation point set under the action of the consistent deformation field along the line. When both the nearest distance and the local deformation consistency residual meet the preset threshold condition, a one-to-one annotation correspondence is determined. When the same annotation point set and multiple building outline elements simultaneously meet the threshold condition, only the annotation correspondence with the highest spatial consistency score is retained and the other annotation correspondences are rejected, thereby ensuring that the one-to-one constraint of the annotation correspondence can be implemented.

[0121] Based on the annotation correspondence, the spatial consistency score between the corrected annotation data and the corresponding building outline elements is calculated. Combined with the deformation uncertainty distribution at the corresponding spatial location in the consistent deformation field along the line, uncertainty-aware modulation processing is performed on the spatial consistency score to generate annotation association confidence.

[0122] The calculation of the annotation association confidence level adopts the method of "joint modulation of spatial consistency score and deformation uncertainty distribution": the spatial consistency score is used as the base value and the base value is suppressed and modulated according to the deformation uncertainty distribution, so that the annotation association confidence level decreases when the deformation uncertainty distribution increases, and the annotation association confidence level increases when the deformation uncertainty distribution decreases; the annotation association confidence level is used as a reference for the subsequent calculation of building number and total building area to reduce the impact of low confidence association on building attribute data;

[0123] When the confidence level of the annotation association meets the preset confidence threshold, the building identification information carried in the corrected annotation data is attached to the corresponding building outline elements to complete the association between the building outline elements and the building identification information.

[0124] Based on the confidence level of the association between building identification information and corresponding annotations, attribute calculation processing is performed on the building outline elements to determine the number of building floors and total building area of ​​the corresponding building, and to generate building attribute data, which includes the number of building floors and total building area.

[0125] In this embodiment, the data entry includes:

[0126] After completing the projection correction process and obtaining the corrected building geometric data, the corrected building geometric data is subjected to geometric standardization processing according to the preset geometric data storage specifications in the traffic digital twin spatial database. This process converts the geometric expression of the building outline elements into a standardized geometric representation that conforms to the geometric type constraints of the database.

[0127] The geometric standardization process adopts the public legality processing method of spatial database input: without changing the coordinate values ​​of the corrected building geometric data, the geometric expression of the building outline elements is converted into a standardized geometric representation that meets the geometric type constraints of the traffic digital twin spatial database, and vertex sequence legality verification is performed on the standardized geometric representation to ensure that the subsequent topology consistency correction process has a consistent geometric input format.

[0128] The traffic digital twin spatial database is a spatial database system used for unified storage, management and query of spatial element data in traffic digital twin scenarios. The traffic digital twin spatial database is used to store building geometric data and corresponding building attribute data, and is preset with geometric type constraints, coordinate reference constraints and topological legality constraints compatible with network map services to support the calling and display of spatial data in traffic digital twin scenarios.

[0129] Based on the topological constraint rules of the traffic digital twin spatial database, topological consistency correction processing is performed on the corrected building geometric data after geometric standardization to eliminate geometric anomalies in the building outline elements that do not meet the topological constraint conditions, including self-intersection, duplicate nodes, hanging boundaries and non-closed outlines, and generate corrected building geometric data that meets the topological consistency requirements.

[0130] The topology consistency correction process adopts the public correction method of the topology legality constraint of the spatial database: self-intersection elimination processing is performed on the self-intersection contours in the corrected building geometry data, node deduplication processing is performed on the duplicate nodes, boundary connection repair processing is performed on the suspended boundaries, and closure repair processing is performed on the non-closed contours. After each type of correction processing is completed, the verification is performed according to the topology legality constraint of the traffic digital twin spatial database until the corrected building geometry data meets the topology consistency requirements and enters the writing process.

[0131] The corrected building geometry data that meets the constraints of geometric standardization and topological consistency are structured together with the corresponding generated building attribute data to generate building space data records for writing to the spatial database.

[0132] According to the spatial indexing rules and attribute field constraints of the transportation digital twin spatial database, the building spatial data records are written into the transportation digital twin spatial database to complete the unified database entry process of building spatial data along the route.

[0133] Example 1:

[0134] To verify the feasibility of this invention in practice, it was applied to a real-world business scenario involving the integration and storage of spatial data of buildings along transportation infrastructure during digital construction. In this scenario, the CAD data of the projects along the route and the digital twin base map data of the transportation system come from diverse sources and have different production cycles. Significant differences exist between the two in terms of coordinate systems, projection parameters, local deformations, and annotation accuracy, leading to problems such as overall offset, local distortion, and disruption of continuity along the route in the spatial positions of building outline elements and building annotation elements. Existing methods often use a single affine or rigid registration method to perform overall correction of the CAD data, which is insufficient to simultaneously address the requirements of local nonlinear deformation and segmental continuity over long distances along the route. Consequently, even after the building geometric data is stored, problems such as outline misalignment, incorrect annotation mounting, and unstable attribute calculations still exist, directly affecting the usability and reliability of building spatial data in the transportation digital twin scenario.

[0135] In application scenarios, after inputting the CAD data of the engineering project along the route and the traffic digital twin base map data into the method of this invention, the system performs unified parsing processing on the building outline elements and building annotation elements in the engineering CAD data, and combines the road centerline elements extracted from the traffic digital twin base map data with the base map reference point set to construct a joint point set input structure. This joint point set input structure maintains a clear spatial order relationship along the route under the constraint of the linear reference structure along the route, providing a stable spatial index foundation for subsequent continuous deformation inference. Based on this input structure, a coherent point drift algorithm is introduced to establish a continuous deformation mapping relationship between the engineering coordinate domain and the traffic digital twin base map coordinate domain. This non-rigid method is used to perform overall correction of the building spatial data along the route, thereby avoiding the problem of insufficient adaptability of a single rigid model to complex along-route deformations.

[0136] In the continuous deformation inference process, a Bayesian parameter estimation mechanism is introduced to integrate the uncertainty of the continuous deformation mapping relationship and the point set matching error into the posterior inference framework. By modeling deformation parameters and noise parameters as random variables, the system obtains the corresponding deformation uncertainty distribution while inferring the deformation field, thus providing a reliable basis for subsequent segmented processing and parameter inversion. In sections with long distances and significant local data quality differences, the deformation uncertainty distribution can accurately reflect the stability of local inference results, avoiding unreliable deformation results from having an excessive impact on the overall correction.

[0137] Based on this, the system performs fixed-mileage segmentation processing on the joint point set input structure according to the linear reference structure along the route, and constructs a conditional coherent point drift sub-model for each subset of the route segments. The sub-model performs local deformation inference under Bayesian deformation field initialization conditions, while using the tangential direction of the road centerline geometric point set as the input for directional consistency constraints, ensuring that the deformation results within the route segments remain continuous and smooth along the route direction. Subsequently, through cross-segment continuous consistency processing, the deformation mapping between adjacent segments is coordinated and adjusted to generate a route-consistent deformation field covering the entire route range, and simultaneously forming a segment uncertainty distribution. This processing method effectively alleviates the breakage and jump problems caused by independent segment correction in traditional methods.

[0138] After generating a consistent deformation field along the route, the system performs discretization and inversion processing on the central meridian parameters and east pseudo-offset parameters in the engineering projection coordinate system, taking into account the piecewise uncertainty distribution. Through tentative reprojection in the candidate parameter space and selection of the optimal parameter combination based on the spatial deviation index weighted by uncertainty perception, a substantial reprojection transformation of the engineering coordinates to projection coordinates suitable for network map services is achieved, ultimately generating corrected building geometric data and corrected annotation data. Subsequently, coordinate mapping processing is performed on the corrected annotation data based on the consistent deformation field along the route. Spatial matching between the annotation point set and the building outline control point set is completed within a unified coordinate domain. Under the constraint of annotation association confidence, building identification information is stably mounted to the corresponding building outline elements. The number of building floors and total building area are further calculated, forming building attribute data that can be directly entered into the database. Finally, the system performs geometric standardization and topological consistency correction processing on the corrected building geometric data, writing the building spatial data that meets the constraints into the traffic digital twin spatial database, achieving unified database entry of building spatial data along the route.

[0139] In this application scenario, three different methods were compared for processing the same batch of CAD data for engineering projects along the route: the traditional overall rigid registration method, an improved method based on non-rigid registration but without considering uncertainties, and the method for extracting and storing spatial data of buildings along the route based on coherence point drift and Bayesian deformation inference proposed in this invention. The comparison results are summarized below.

[0140] Table 1. Comparison of methods for extracting and storing spatial data of buildings along the route

[0141] Performance indicators Traditional overall rigid registration method Non-rigid registration without uncertainty method Method of the present invention Average planar position error of building outline (meters) 1.32 0.86 0.58 Correct mounting rate of annotations (%) 81.4 85.6 87.2 Rate of disruption of the continuity of the profile of adjacent segments along the line (%) 6.8 3.1 1.9 Relative error in building area calculation (%) 5.4 3.7 2.6 Percentage of data manually corrected after being entered into the database (%) 12.5 6.9 4.3

[0142] As shown in Table 1, the traditional rigid registration method results in an average planar position error of over 1 meter for buildings along long distances with significant local deformation. The accuracy rate of annotations is low, and the continuity between adjacent segments along the line is significantly disrupted, leading to a high proportion of subsequent manual corrections. Introducing non-rigid registration reduces the overall error level, but because it does not model deformation uncertainty, local overfitting or incorrect corrections still easily occur in sections with uneven data quality, limiting further accuracy improvements.

[0143] In comparison, the method of this invention reduces the average planar position error of the building outline to 0.58 meters, increases the correct annotation mounting rate to 87.2%, controls the continuity disruption rate of adjacent segments along the line to a low level, and also reduces the relative error in building area calculation. This performance improvement does not stem from overly complex calculations, but rather from introducing a Bayesian parameter estimation mechanism in the continuous deformation inference stage to explicitly model deformation uncertainties. This allows the system to reasonably balance the data reliability of different segments during parameter inversion, segment consistency, and annotation association. Simultaneously, the introduction of a linear reference structure along the line and directional consistency constraints ensures that the deformation results remain stable and continuous along the line, fundamentally reducing the geometric breakage problem caused by segmentation processing.

[0144] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for extracting and storing spatial data of buildings along transportation routes based on digital twins, characterized in that, Includes the following steps: Acquire CAD data of engineering projects along the route and digital twin base map data of traffic, construct a linear reference structure along the route, and generate a joint point set input structure; The coherent point drift algorithm is invoked, and the joint point set input structure is used as the multi-source point set input to establish a continuous deformation mapping relationship from the engineering coordinate domain to the traffic digital twin base map coordinate domain, thereby generating the initial deformation field. In the coherent point drift algorithm, a Bayesian parameter estimation mechanism is introduced. The initial deformation field is used as the posterior inference initialization input, and the posterior inference processing is performed on the joint point set input structure to obtain the Bayesian deformation field. Based on the linear reference structure along the line, the joint point set input structure is divided into a set of segmented subsets along the line according to the fixed mileage segment length. A conditional coherent point drift sub-model is constructed, and segmented deformation inference and cross-segment continuous consistency processing are performed to obtain the consistent deformation field along the line and the segmented uncertainty distribution. Based on the consistent deformation field and piecewise uncertainty distribution along the line, discretization inversion processing and substantial reprojection processing are performed to generate corrected building geometry data and corrected annotation data. Based on the consistent deformation field along the line, coordinate mapping processing is performed on the correction annotation data to determine the annotation correspondence between the correction annotation data and the correction building geometry data, and building attribute data is calculated and generated. Geometric standardization and topological consistency correction are performed on the corrected building geometry data, and the data is written into the traffic digital twin spatial database along with the building attribute data.

2. The method for extracting and storing spatial data of buildings along a transportation route based on traffic digital twins according to claim 1, characterized in that, The generation of the joint point set input structure includes: Acquire CAD data of the project along the route, perform layer parsing processing, and identify and extract building outline elements representing the outer boundary of the building and building annotation elements representing building text labels; Geometric discretization and equidistant sampling are performed on the building outline elements to determine the set of building outline control points. Perform geometric representation unification processing on architectural annotation elements, mapping architectural annotation elements to annotation point sets; Obtain traffic digital twin base map data, extract road centerline elements representing the direction of the road center, perform geometric discretization processing, and convert them into a set of geometric points of the road centerline; Select a set of reference points from the digital twin base map data of the transportation system; Based on the geometric point set of the road centerline, a linear reference structure is constructed along the line as a unified spatial index framework to complete the position calibration of the building outline control point set and annotation point set along the line. The control point set and annotation point set of the building outline along the completed location, as well as the geometric point set of the road centerline and the base map reference point set, are organized in a unified manner to generate a joint point set input structure.

3. The method for extracting and storing spatial data of buildings along a transportation route based on traffic digital twins according to claim 1, characterized in that, The generation of the initial deformation field includes: The joint point set input structure is processed by point set role classification, and the building outline control point set, the road centerline geometric point set, and the base map reference point set are marked as input point sets participating in deformation estimation, respectively. Based on the engineering coordinate domain and the traffic digital twin base map coordinate domain, the source point set and target point set of the coherent point drift algorithm are determined. The source point set is composed of the point set representation of the building outline control point set and the road centerline geometric point set in the engineering coordinate domain, and the target point set is composed of the point set representation of the base map reference point set in the traffic digital twin base map coordinate domain. In the coherent point drift algorithm, a point set probability correspondence is established between the source point set and the target point set. The source point set is regarded as a probability distribution sample generated from the target point set. Continuous non-rigid deformation estimation is performed to obtain a continuous deformation mapping relationship. Based on the continuous deformation mapping relationship, the overall deformation calculation is performed on the control point set of the building outline and the geometric point set of the road centerline to generate the initial deformation field.

4. The method for extracting and storing spatial data of buildings along a transportation route based on traffic digital twins according to claim 1, characterized in that, The generation of the Bayesian deformation field includes: In the continuous deformation mapping framework of the coherent point drift algorithm, a Bayesian parameter estimation mechanism is introduced to model the internal representation of the initial deformation field as random variables and to model the matching uncertainty formed under the action of continuous deformation mapping as random variables, thereby generating random variable representations. Based on the joint point set input structure, a probabilistic observation model for point set matching is constructed, and prior constraints are introduced to constrain the smoothness of the continuous deformation space. Using the initial deformation field as the initial input for posterior inference, Bayesian posterior inference processing is performed on the joint point set input structure under the combined action of the probabilistic observation model and prior constraints. This updates the matching probability distribution of the joint point set input structure and simultaneously updates the continuous deformation mapping relationship, resulting in a converged posterior probability distribution. After completing the Bayesian posterior inference process, statistical features of the continuous deformation mapping relationship are extracted based on the convergent posterior probability distribution to generate the Bayesian deformation field.

5. The method for extracting and storing spatial data of buildings along a transportation route based on traffic digital twins according to claim 1, characterized in that, The generation of the uniform deformation field and piecewise uncertainty distribution along the line includes: Based on the linear reference structure along the route, the input structure of the joint point set is divided into segments along the route according to the preset fixed mileage segment length to obtain a set of segments along the route. For each segment subset along the line, a corresponding conditional coherence point drift sub-model is constructed by combining the coherence point drift algorithm. The Bayesian deformation field is used as the deformation initialization input to perform local deformation inference processing on the segment subset along the line and generate the initial segment deformation result. In the conditional coherent point drift sub-model, the tangential direction information corresponding to the direction along the road in the geometric point set of the road centerline is introduced as the direction consistency constraint input, and the direction consistency modulation processing is performed on the initial segmented deformation result to obtain the segmented deformation result; Based on the piecewise deformation results and the Bayesian deformation field, cross-segment continuous consistency processing is performed on the piecewise deformation results between adjacent segment subsets along the line to generate a consistent deformation field along the line. After completing the cross-segment continuous consistency processing, the deformation uncertainty of each segment subset along the line during the consistency process is aggregated to form a segment uncertainty distribution.

6. The method for extracting and storing spatial data of buildings along a transportation route based on traffic digital twins according to claim 1, characterized in that, The generation of the corrected architectural geometry data and corrected annotation data includes: Based on the consistent deformation field and piecewise uncertainty distribution along the line, the discretized inversion search space of the central meridian parameters and the east pseudo offset parameters in the engineering projection coordinate system is determined. In the discretized inversion search space, a set of parameter combinations is constructed. For each parameter combination, a tentative reprojection process is performed on the building outline elements and building annotation elements in combination with the consistent deformation field along the line, and the spatial deviation index is calculated. Based on the piecewise uncertainty distribution, uncertainty-perceived weighted processing is performed on the spatial deviation index to generate a comprehensive evaluation result. Based on the comprehensive evaluation results, the optimal values ​​for the central meridian parameter and the east pseudo-offset parameter are determined from the parameter combination set, forming the central meridian parameter value set and the east pseudo-offset parameter value set. Based on the set of parameters of the central meridian and the set of parameters of the east pseudo offset, a substantial reprojection process is performed on the building outline elements and building annotation elements to generate corrected building geometric data and corrected annotation data.

7. The method for extracting and storing spatial data of buildings along a transportation route based on traffic digital twins according to claim 1, characterized in that, The generation of the building attribute data includes: Based on the consistent deformation field along the line, coordinate mapping processing is performed on the correction annotation data to map it to a unified coordinate domain consistent with the correction building geometry data, thus obtaining the annotation point set representation; Within a unified coordinate domain, based on the spatial consistency results of the annotation point set representation and the building outline control point set in the consistent deformation field along the line, the spatial matching relationship is calculated to determine the one-to-one annotation correspondence between the corrected annotation data and the corrected building geometry data; Based on the annotation correspondence, the spatial consistency score between the corrected annotation data and the corresponding building outline elements is calculated, and uncertainty-aware modulation processing is performed in combination with the deformation uncertainty distribution to generate annotation association confidence. If the confidence level of the annotation association meets the preset confidence threshold, the building identification information carried in the corrected annotation data will be attached to the corresponding building outline elements. Based on the confidence level of the association between building identification information and corresponding annotations, attribute calculations are performed on building outline elements to generate building attribute data.

8. The method for extracting and storing spatial data of buildings along a transportation route based on traffic digital twins according to claim 1, characterized in that, The data import process includes: Based on the preset geometric data storage specifications in the transportation digital twin spatial database, geometric standardization processing is performed on the corrected building geometric data; Based on the topological constraint rules of the traffic digital twin spatial database, topological consistency correction is performed on the corrected building geometric data after geometric standardization to generate corrected building geometric data. The corrected building geometry data that meets the constraints of geometric standardization and topological consistency are structured and organized with the corresponding generated building attribute data to generate building space data records. According to the spatial indexing rules and attribute field constraints of the transportation digital twin spatial database, the building spatial data records are written into the transportation digital twin spatial database.

Images