A map generation method and system based on multi-source heterogeneous geographic information processing

Abstract

This application provides a map generation method and system based on multi-source heterogeneous geographic information processing. It acquires three types of heterogeneous data: a geographic element attribute database, a geographic coordinate semantic database, and a geographic reference image database. Spatial spectrum reconstruction processing is then performed to generate a spatial spectrum cube. Belief propagation fusion processing is then used to construct a belief network from the spatial spectrum data. Combined with spatially constrained probability field optimization and D-S evidence synthesis, a geographic evidence chain is generated. Next, spatiotemporal field reconstruction processing is performed to separate the static and dynamic states of the geographic evidence chain and couple them into fields, constructing a hyperdimensional spatiotemporal map. Finally, holographic projection conversion processing is used to perform phase field compression and coherent diffraction reconstruction on the hyperdimensional spatiotemporal map, generating a holographic geographic projection map that supports 3D interaction. This method achieves the automatic generation of a structurally unified, highly reliable geographic map that supports dynamic behavioral reasoning and true 3D interaction from multi-source heterogeneous geographic data, significantly improving the decision-making value and cognitive efficiency of geographic information.

Description

Technical Field

[0001] This invention relates to the field of geographic information, and specifically to a map generation method and system based on multi-source heterogeneous geographic information processing. Background Technology

[0002] In modern traffic management, smart port scheduling, and the construction of digital twin cities, there is an urgent need to accurately represent the collaborative relationship between fixed facilities and mobile targets, including but not limited to the position matching of runways and taxiing aircraft, the real-time docking of berths and docked ships, and the dynamic interaction between roads and moving vehicles. These scenarios require map systems that can not only present static geographic features but also integrate the trajectory evolution and spatial constraints of dynamic targets in real time, providing a unified spatiotemporal benchmark for automated scheduling and emergency decision-making.

[0003] Existing technologies typically employ a phased, independent processing strategy. Geographic information systems (GIS) are responsible for parsing coordinates and topological relationships in a structured attribute database, natural language processing tools extract coordinate semantics from text logs, computer vision technology performs target detection and registration on remote sensing images, and finally, geographic information visualization is achieved through two-dimensional layer overlay or three-dimensional model annotation.

[0004] However, this approach has significant drawbacks. Multi-source heterogeneous geographic information suffers from inconsistent data structures when spatial modeling, reliable fusion, and dynamic display. Traditional methods cannot collaboratively process heterogeneous data such as structured attribute libraries (e.g., airport runway coordinates), unstructured text (e.g., ship navigation logs), and raster imagery (e.g., remote sensing maps). Information credibility is difficult to quantify, human experience is insufficient to assess the confidence level of dynamic targets such as aircraft positions, and spatial topological constraints are not integrated, making dynamic evolution difficult to express. Static maps cannot present the movement trajectory and spatiotemporal correlation of traffic targets in real time, resulting in insufficient decision-making basis and delayed dynamic response. Summary of the Invention

[0005] In view of the above-mentioned actual situation, this application proposes a map generation method and system based on multi-source heterogeneous geographic information processing, in order to solve the problems of inconsistent data structure, difficulty in quantifying information credibility, and difficulty in expressing dynamic evolution in the spatial modeling, reliable fusion and dynamic display of multi-source heterogeneous geographic information in the prior art.

[0006] A map generation method based on multi-source heterogeneous geographic information processing, the method comprising the following steps: S1, Obtain the data to be processed. The data to be processed includes a geographic feature attribute library, a geographic coordinate semantic library, and a geographic reference image library. The geographic feature attribute library is a database containing structured fields of geographic entities. The geographic coordinate semantic library is a collection of natural language texts carrying geographic coordinate descriptions. The geographic reference image library is a set of remote sensing images with geographic registration parameters. S2, perform spatial spectrum reconstruction processing on the data to be processed to generate a spatial spectrum cube. The spatial spectrum reconstruction processing is a three-level fusion mechanism of topological skeleton construction, semantic constraint binding and spectrum encoding. S3, perform confidence propagation fusion processing on the spatial spectrum cube to generate a geographic evidence chain. The confidence propagation fusion processing involves constructing a confidence network from the spatial spectrum data, combining it with probability field optimization based on spatial constraints, and completing the evidence synthesis processing through the DS framework. S4, perform spatiotemporal field reconstruction processing on the geographic evidence chain to generate a hyperdimensional spatiotemporal map. The spatiotemporal field reconstruction processing is to achieve unified spatiotemporal modeling through dynamic-static separation and field coupling mechanism. S5, perform holographic projection conversion processing on the hyperdimensional spatiotemporal map to generate a holographic geographic projection map. The holographic projection conversion processing is to achieve three-dimensional visualization through phase field compression and diffraction reconstruction.

[0007] Furthermore, step S2 includes the following sub-steps: S201, perform hierarchical topological decomposition on the geographic element attribute library to generate a spatial topological skeleton. The hierarchical topological decomposition is based on the geographic entity spatial relationship rule library to construct a node network and spatial constraint edges. S202, based on the geographic coordinate semantic library and combined with the spatial topology skeleton, semantic topology binding processing is performed to generate a semantic constraint graph. The semantic topology binding processing is to map text coordinate entities to topology nodes and generate a semantic relationship matrix through a graph neural network. S203, perform spectral fusion processing based on the geographic reference image library and the semantic constraint map to generate a spatial spectral cube. The spectral fusion is to extract image features through a spatial convolutional network and align them with the geographic coordinates of the semantic constraint map, and then use a spectral coding algorithm to generate a three-dimensional spatial spectral data block.

[0008] Furthermore, step S3 includes the following sub-steps: S301, perform confidence network construction processing on the spatial hierarchy cube to generate a spatial confidence network. The confidence network is constructed by mapping the spatial hierarchy data block to a probabilistic graphical model, where nodes represent geographic entities and edges represent spatial dependencies. S302, perform probability field optimization processing on the spatial confidence network based on spatial constraint propagation to generate an optimized probability field. The probability field optimization processing is to iteratively update the node confidence along the spatial constraint edge using a message passing algorithm. S303, perform evidence synthesis processing on the optimized probability field to generate a geographical evidence chain. The evidence synthesis processing involves mapping node confidence to the DS evidence framework and calculating the overall credibility by weighting the spatial consistency factor.

[0009] Furthermore, step S4 includes the following sub-steps: S401, The geographic evidence chain is processed by a field separation function to generate static field primitives and dynamic field primitives. The field separation function is to split the evidence chain into a static entity coordinate field and a dynamic trajectory vector field based on the entity motion attributes. S402, Motion constraint coupling processing is performed on static field primitives and dynamic field primitives to generate a hyperdimensional spatiotemporal map. The motion constraint coupling processing is to embed the dynamic trajectory vector field into the static entity coordinate field through the terrain-driven algorithm to construct a spatiotemporal coordinate system with joint spatial behavior and terrain constraints.

[0010] Furthermore, step S5 includes the following sub-steps: S501, perform holographic phase encoding processing on the hyperdimensional spatiotemporal diagram to generate compressed phase primitives. The holographic phase encoding processing is to convert the spatiotemporal coordinates into the phase distribution of light waves through a curvature phase mapping algorithm. S502, coherent diffraction reconstruction is performed on the compressed phase primitive to generate a holographic geographic projection map. The coherent diffraction reconstruction is to convert the phase distribution into a geographic projection that supports three-dimensional spatial interaction through optical wave interference.

[0011] Furthermore, the hierarchical topology decomposition process in S201 is based on a geographic entity spatial relationship rule base to construct a node network and spatial constraint edges, including entity topology relationship parsing and graph structure generation; the semantic topology binding process in S202 is to map text coordinate entities to topology nodes and generate a semantic relationship matrix through a graph neural network, including entity node alignment and semantic relationship reasoning; the spectral fusion in S203 is to extract image features through a spatial convolutional network and align them with the semantic constraint graph using geographic coordinates, and generate three-dimensional spatial spectral data blocks using a spectral coding algorithm, including geographic alignment feature extraction and spectral coding aggregation.

[0012] Furthermore, the confidence network construction in S301 involves mapping spatial spectral data blocks to a probabilistic graphical model, including entity node remapping and spatial dependency edge reconstruction; the probability field optimization in S302 involves iteratively updating node confidence along spatial constraint edges using a message passing algorithm, including confidence initialization and spatial message propagation; and the evidence synthesis in S303 involves mapping node confidence to a DS evidence framework and calculating comprehensive credibility using spatial consistency factors, including basic probability assignment and spatial weighted fusion.

[0013] Furthermore, the field separation function processing in S401 is to split the evidence chain into a static entity coordinate field and a dynamic trajectory vector field based on the entity motion attributes, including motion attribute quantization processing and primitive generation processing; the motion constraint coupling processing in S402 is to embed the dynamic trajectory vector field into the static entity coordinate field through a terrain-driven algorithm to construct a spatiotemporal coordinate system of joint spatial behavior and terrain constraints, including terrain constraint mapping processing and spatiotemporal coordinate synthesis processing.

[0014] Furthermore, the holographic phase encoding process in S501 converts spatiotemporal coordinates into light wave phase distribution through a curvature phase mapping algorithm, including spatiotemporal curvature calculation and phase field compression; the coherent diffraction reconstruction in S502 converts the phase distribution into a geographic projection that supports three-dimensional spatial interaction through light wave interference, including wavefront reconstruction and interferometric rendering.

[0015] Furthermore, this application also discloses a map generation system based on multi-source heterogeneous geographic information processing, characterized in that the system comprises: The acquisition unit is used to acquire data to be processed, which includes a geographic feature attribute library, a geographic coordinate semantic library, and a geographic reference image library. The geographic feature attribute library is a database containing structured fields of geographic entities, the geographic coordinate semantic library is a set of natural language text carrying geographic coordinate descriptions, and the geographic reference image library is a set of remote sensing images with geographic registration parameters. The spatial spectrum reconstruction unit is used to perform spatial spectrum reconstruction processing on the data to be processed, thereby generating a spatial spectrum cube. The spatial spectrum reconstruction processing is achieved through a three-level fusion mechanism of topological skeleton construction, semantic constraint binding, and spectrum encoding. The confidence propagation fusion unit is used to perform confidence propagation fusion processing on the spatial spectrum cube to generate a geographical evidence chain. The confidence propagation fusion processing involves constructing a confidence network from the spatial spectrum data, combining it with probability field optimization based on spatial constraints, and completing the evidence synthesis processing through the DS framework. The spatiotemporal field reconstruction unit is used to perform spatiotemporal field reconstruction processing on the geographic evidence chain, thereby generating a hyperdimensional spatiotemporal map. The spatiotemporal field reconstruction processing is to achieve unified spatiotemporal modeling through dynamic-static separation and field coupling mechanism. The holographic projection conversion unit is used to perform holographic projection conversion processing on the hyperdimensional spatiotemporal map to generate a holographic geographic projection map. The holographic projection conversion processing achieves three-dimensional visualization through phase field compression and diffraction reconstruction.

[0016] The map generation method and system proposed in this application based on multi-source heterogeneous geographic information processing realizes the automatic generation of geographic maps with unified structure, high credibility, and support for dynamic behavioral reasoning and true 3D interaction from multi-source heterogeneous geographic data, which significantly improves the decision-making value and cognitive efficiency of geographic information. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the method flow for a map generation method based on multi-source heterogeneous geographic information processing proposed in this application; Figure 2 This is a schematic diagram of the spatial hierarchy cube generation process of a map generation method based on multi-source heterogeneous geographic information processing proposed in this application. Figure 3 This is a schematic diagram of the geographic evidence chain generation process of a map generation method based on multi-source heterogeneous geographic information processing proposed in this application; Figure 4 A schematic diagram of a map generation system based on multi-source heterogeneous geographic information processing is provided in this application embodiment; Detailed Implementation

[0018] The simulation technology route in 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.

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] The features and performance of the present invention will be further described in detail below with reference to embodiments. Please refer to the appendix. Figure 1 As shown, a map generation method based on multi-source heterogeneous geographic information processing is described, the method comprising the following steps: S1, Obtain the data to be processed. The data to be processed includes a geographic feature attribute library, a geographic coordinate semantic library, and a geographic reference image library. The geographic feature attribute library is a database containing structured fields of geographic entities. The geographic coordinate semantic library is a collection of natural language texts carrying geographic coordinate descriptions. The geographic reference image library is a set of remote sensing images with geographic registration parameters. In some implementations, the data to be processed constitutes the basic data input layer of the map generation method. This processing involves the collection and standardization preprocessing of three types of heterogeneous geographic information sources. The geographic feature attribute database is stored using a relational database structure, with its schema defined as ordered triples. Where E i A represents the unique identifier of the i-th geographic entity. ij V represents the j-th structured attribute field of the entity (such as entity type, geometric boundary, elevation value). ijk Indicates attribute field A ij The k-th quantization value or classification label, and the topological relationship between entities is implicitly associated through foreign key constraints.

[0021] In this embodiment, the geographic coordinate semantic library is constructed as a set of mappings between natural language text and spatial coordinates. T m This refers to unstructured text paragraphs that contain references to geographical entities. The text anchors the geographic coordinates quadruple (x, y, z are spatial coordinates). (Time stamp). This library achieves coordinate extraction through a semantic parsing layer: it establishes a cascaded model of a bidirectional long short-term memory network and a conditional random field, with the model input being a sequence of word vectors. The output is a sequence of geographic coordinate entity labels. The loss function is defined as: ,in For the model parameter set, The coordinate values ​​are generated from the tagged text by the Named Entity Recognition (NER) module, which is used for regularization.

[0022] In some implementations, the georeferenced image library is organized as a multispectral remote sensing image set. For B-band image data matrix, G n The corresponding georegistration parameter set includes affine transformation matrices. With projection parameter set P n (e.g., EPSG coordinate system encoding, ellipsoid parameters). Image geographic coordinate transformation follows: ,in The row and column numbers are pixels. To output geodetic coordinates, the image database undergoes pre-normalization processing using a feature pyramid network, outputting multi-scale feature tensors. Feature dimensions and spatial resolution are aligned to a unified geographic grid.

[0023] In this embodiment, three types of data sources are forcibly associated through a spatial reference datum: entity location coordinates P in the geographic feature attribute database. e Parsing coordinates C with a geographic coordinate semantic library m Using the same geodetic coordinate system, WGS-84 in this embodiment, and the affine transformation matrix A of the georeferenced image library. n Map pixel coordinates to this unified coordinate system. The data coupling relationship is represented as: Attribute library entity E iThe image database establishes a membership relationship with semantic library text coordinates through spatial join operations; the image library uses spatial extent constraints. Spatial coverage association is established with attribute library entities. This step outputs standardized data triples. Its intrinsic correlation matrix R is defined as: ,in Using spatial proximity as the threshold, TF-IDF is used to calculate the semantic similarity of the text. This matrix provides cross-modal association constraints for subsequent spatial spectrum reconstruction.

[0024] S2, perform spatial spectrum reconstruction processing on the data to be processed to generate a spatial spectrum cube. The spatial spectrum reconstruction processing is a three-level fusion mechanism of topological skeleton construction, semantic constraint binding and spectrum encoding. For details, please refer to the appendix. Figure 2 As shown, this step includes the following sub-steps: S201, perform hierarchical topological decomposition on the geographic element attribute library to generate a spatial topological skeleton. The hierarchical topological decomposition is based on the geographic entity spatial relationship rule library to construct a node network and spatial constraint edges. In some implementations, the hierarchical topology decomposition process is based on a geographic entity spatial relationship rule base to construct a node network and spatial constraint edges, including entity topology relationship parsing and graph structure generation. Specifically, the entity topology relationship parsing process will decompose the geographic element attribute database D... a The structured entities in the code are deconstructed into hierarchical relationships based on a predefined spatial relationship rule base, which is defined as a set of predicate logic. ,in The mathematical form of the predicate representing the k-th type of spatial relation (such as containment, adjacency, intersection) is: The rule base uses a spatial predicate calculus engine to determine entity relationships, with the entity's geometric boundary field as the input. (e.g., WKT format polygon coordinates), the output is a relation determination matrix. .

[0025] In this embodiment, the graph structure generation process is based on the relationship determination matrix M. R Constructing a hierarchical topology skeleton The node network V is generated through an entity aggregation function: ,in To satisfy spatial inclusion relationship The index of the largest connected entity subset, and the nodes after aggregation. attribute vector a h Weighted average of sub-entity attributes: Spatial constraint edges The generation of an edge set is based on non-inclusion relation predicates (such as adjacent and intersecting), and the edge set is defined as follows: The edge weight matrix W is calculated using the spatial relation strength quantization function: ,in For relation type Semantic weight coefficients (such as adjacency relationships) Intersection relationship ).

[0026] It should be noted that hierarchical topology decomposition is implemented through a recursive partitioning algorithm: the root node is initialized as a global entity set. Iteratively perform the following operations until the depth threshold D is met. max Or minimum entity size S min :1) For the current node v h The internal entity performs Delaunay triangulation to generate a triangulated network. 2) Construct a local adjacency matrix based on the edge connection relationship of the triangulation network. 3) Merge Generate a global constraint edge set with the rule base decision matrix MR .

[0027] The spatial topological skeleton is ultimately represented as a weighted graph structure. , where the node feature matrix Each row is an aggregate entity attribute a h Adjacency matrix With weight matrix The skeleton co-encodes spatial constraints. It provides a structured spatial framework for subsequent semantic topological binding, with its node network representing hierarchical aggregation units of geographic entities and spatial constraint edges defining the strength of topological dependencies between units.

[0028] Preferably, the spatial relation predicate r in the rule base k This is achieved through OGC standard spatial operations. Furthermore, the entity aggregation function employs the OPTICS spatial clustering algorithm, with clustering parameters... Consistent with the spatial proximity threshold in step S1, ensuring data continuity.

[0029] In this embodiment, the mathematical representation of the hierarchical topological skeleton is as follows: , where ⊙ represents the Hadamard product, which encodes node attributes, topological connectivity and spatial constraint strength into a dense tensor structure that can be directly input into the graph neural network in step S202.

[0030] S202, based on the geographic coordinate semantic library and combined with the spatial topology skeleton, semantic topology binding processing is performed to generate a semantic constraint graph. The semantic topology binding processing is to map text coordinate entities to topology nodes and generate a semantic relationship matrix through a graph neural network. In some implementations, the semantic topology binding process maps text coordinate entities to topology nodes and generates a semantic relation matrix using a graph neural network, including entity node alignment processing and semantic relation inference processing. Specifically, the entity node alignment processing maps the geographic coordinate semantic database... Text coordinate entities and spatial topological skeleton The nodes undergo joint spatial semantic matching, and the input to this process is the node feature matrix generated in step S201. and coordinate text pairs parsed by the semantic library The output is the entity node mapping function. The mapping is implemented through a graph attention alignment network, and the network layers are defined as follows: , ,in Coordinate encoding function (using Gaussian kernel) , For learnable projection matrices, For attention vectors, For C m Center, radius The set of topological nodes within a spatial neighborhood.

[0031] In this embodiment, semantic relation reasoning processing generates a semantic relation matrix based on the alignment results. Where L is the number of predefined semantic relation categories (e.g., "located in", "connected to", "belongs to"). The processing employs a relational graph convolutional network, with the network input being: initial node features. Edge type tensor The spatial constraint edge E and the newly added semantic edge E sem Semantic edges are generated by concatenating elements and constructed using a text relation extraction model: for each pair of nodes with spatially constrained edges... Retrieve related text Input Bi-LSTM relation classifier: , The feature update formula for the R-GCN layer is: ,in For a set of relation types, This represents the set of neighbors r that have a relation r with node h. The output layer generates a semantic relation matrix through the relation decoder: ,in , These are learnable parameters.

[0032] It should be noted that the semantic constraint graph is defined as an augmented graph structure. The node set Includes the original topology node V and the newly added text entity node The edge set Integrate spatial constraint edges and semantic relation edges. Preferably, the text entity node feature v sem Initialize using a BERT pre-trained model: Furthermore, the relation type set R of R-GCN includes spatial relation category K (inherited from the S201 rule base) and semantic relation category L (such as "administrative affiliation" and "transportation connection"), realizing bimodal relation fusion.

[0033] In this embodiment, the semantic constraint graph is mathematically represented as a quadruple: ,in It is a node feature matrix (a concatenation of topological nodes and text entity nodes). It is an adjacency matrix (the union of spatial constraint edges and semantic relation edges). The edge weight matrix (spatial weight W and semantic confidence max) l S pql (Combined), where S is a semantic relation tensor. This structure extends the spatial topological skeleton into a unified graph representation that integrates semantic relations, providing a cross-modal constraint framework for S203 family fusion.

[0034] In other words, the essence of semantic topological binding is to achieve a triple mapping through graph neural networks, including the geometric mapping from spatial coordinates to topological nodes. Semantic mapping from natural language to relation categories Joint embedding of spatial constraints and semantic relations The process ensures that the text descriptions of the geographic coordinate semantic library are structurally bound to the spatial topological skeleton, generating an enhanced graph model with semantic interpretability.

[0035] S203, perform spectral fusion processing based on the geographic reference image library and the semantic constraint map to generate a spatial spectral cube. The spectral fusion is to extract image features through a spatial convolutional network and align them with the geographic coordinates of the semantic constraint map, and then use a spectral coding algorithm to generate a three-dimensional spatial spectral data block.

[0036] In some implementations, the spectral fusion involves extracting image features using a spatial convolutional network and aligning them with geographic coordinates using a semantic constraint graph. A spectral coding algorithm is then used to generate three-dimensional spatial spectral data blocks, including geographic alignment feature extraction and spectral coding aggregation. Specifically, the geographic alignment feature extraction process involves using a georeferenced image database... Multispectral image data and semantic constraint graphs The spatial nodes are precisely registered, and the input to this process is the image tensor from step S1. and its registration parameters and the node coordinate set generated in step S202 The coordinates of the topological nodes originate from the geometric center of entity S201, and the coordinates of the text entity nodes originate from D. s C m， The alignment is achieved through a differentiable bilinear sampling grid; specifically, an image-node coordinate mapping function is established under a unified geographic coordinate system. ,in Let be the affine transformation matrix. For the geodetic coordinates of the node, These correspond to the pixel coordinates of the image. Next, a spatial convolutional network (U-Net architecture) is used to extract node-centric feature blocks: The output is a node-aligned image feature vector. The size R is influenced by the radius of the node space. Determined dynamically.

[0037] In this embodiment, spectral coding aggregation processing fuses node-aligned image features and semantic constraint graph features to generate a spatial spectral cube. The processing performs the following operations: 1) Feature stitching and dimensionality reduction, specifically, image features... With node semantic features Splicing along the channel dimension, then compressing via a fully connected layer: 2) Graph Fourier Transform, specifically, the Laplace matrix based on semantically constrained graphs. Perform spectral domain projection: 3) Spectral coding compression, specifically, using a low-pass filter to retain the first Z dominant frequency components, and constructing a three-dimensional spectral basis tensor: in ,in Geographic grid coordinates Let k be the spatial frequency parameter of the spectral component.

[0038] It should be noted that the mathematical representation of the spatial spectrum cube is a five-dimensional tensor C, and its dimensions are defined as follows: It is a unified geographic grid spatial resolution (the grid step size is determined by the extreme values ​​of global coordinates). Z is the spectral dimension depth (taken as...) T is the time dimension (derived from D). s timestamp Discretized layering), D is the number of feature channels (D=d f The cube element values ​​are generated through spectral basis tensor interpolation: ,in Preferably, an adaptive sampling radius is used in the geographic alignment process. ,in r is the spatial influence radius of node h (derived from the S201 entity aggregation range). esnTo determine the spatial resolution of the image, the spectral coding algorithm further introduces a spatial-semantic weight factor. This is used to enhance the energy proportion of important spectral components.

[0039] In this embodiment, the generation logic of the spatial spectral cube embodies a triple fusion, including spatial alignment, feature coupling, and spectral compression. Spatial alignment achieves geometrically precise matching between image features and graph nodes through pixel-level coordinate mapping. Feature coupling involves concatenating the channel dimensions of image spectral features and graph semantic features. Spectral compression utilizes graph Fourier transform to project heterogeneous features onto a low-dimensional spectral space, forming computable data blocks. The cube provides structured input for subsequent S3 confidence propagation; its three-dimensional spatial dimensions (X, Y, Z) encode geospatial distribution and spectral features, the temporal dimension T supports dynamic analysis, and the feature dimension D carries multi-source fusion information.

[0040] In other words, the essence of genealogy fusion is to transform unstructured graph data. With rasterized image data A unified mapping is performed to a normalized five-dimensional tensor space, and an optimal compact representation of cross-modal features is achieved through spectral domain transformation. The spatial spectral cube C output by this step is mathematically defined as a mapping function from geographic coordinates to feature vectors: ,in This is a time window function that preserves both spatial topological relationships and semantic constraints.

[0041] S3, perform confidence propagation fusion processing on the spatial spectrum cube to generate a geographic evidence chain. The confidence propagation fusion processing involves constructing a confidence network from the spatial spectrum data, combining it with probability field optimization based on spatial constraints, and completing the evidence synthesis processing through the DS framework. For details, please refer to the appendix. Figure 3 As shown, this step includes the following sub-steps: S301, perform confidence network construction processing on the spatial hierarchy cube to generate a spatial confidence network. The confidence network is constructed by mapping the spatial hierarchy data block to a probabilistic graphical model, where nodes represent geographic entities and edges represent spatial dependencies. In some implementations, the confidence network is constructed to map spatial hierarchy data blocks to a probabilistic graphical model, including entity node remapping and spatial dependency edge reconstruction. Specifically, the entity node remapping process remaps the spatial hierarchy cube generated in S203. Backprojection to the geographic entity level defined in S201, the input of the process is the cube feature tensor C and the original entity set from step S201. Geometric center coordinates The remapping is achieved through trilinear interpolation and a fully connected layer: 1) Locating entity Ei In the spatial grid index of the cube: , ,in , 2) Extract the cubic spatial resolution parameters. Local spectral feature blocks centered on (3×3 neighborhood). 3) Decode the entity confidence vector through a deconvolutional layer: The output is the probability distribution of entity states. (C is the number of entity types), forming the node confidence matrix. .

[0042] In this embodiment, the spatial dependency edge reconstruction process reconstructs the probabilistic dependencies between entities based on the S201 spatial relationship rule base R. The process performs the following operations: 1) Edge set generation, specifically, the constraint edge set inheriting from the S201 topological skeleton. Filter weight is below the threshold The edge: 2) Dependency-based quantization, specifically, defining edge weights based on spatial spectral correlation: ,in Calculate the cosine similarity of eigenvectors. These are the original spatial constraint weights for S201.

[0043] It should be noted that the spatial belief network is defined as a Markov random field model. The node set Corresponding geographic entity, nodal potential function The edge set Define factor nodes, and parameterize the potential function using the spectral feature covariance matrix: , Indicates the connection to entity node E i With E j The potential function corresponding to the edge is used to quantify the spatial dependency strength between the states of two entities. The larger the value, the higher the spatial consistency between the states of the two entities. i with f j Defined as entity E i and E j Spectral eigenvectors: Where C is the spatial spectrum cube generated by S203, x i ,y i For entity E i The geographic coordinates, Z represents the spectral depth, and D represents the number of feature channels. The operation will move the cube to coordinates (x) i ,y i The Z×T×D subtensor at () is flattened into a one-dimensional vector. Represents entity E i With E j Covariance matrix of spectral feature differences ,in The dependency strength weights calculated for S301 (integrating spatial constraint weights and spectral feature similarity), where I is the identity matrix, and the tolerance for feature differences is determined by the dependency strength. Control; the larger the value, the greater the allowable variation. Exponential term. Essentially, it's Mahalanobis distance, used to measure the weighted distance between the differences in spectral feature vectors of two entities under spatial dependencies. Scaling the differences: Dependency strength The larger the value, the smaller the punishment.

[0044] Preferably, entity type probability The generation process introduces an attention mechanism: a spatial-channel dual attention transformation is performed on the local spectral block Ci to enhance the response of key features. Furthermore, the computation of edge weights adds temporal consistency constraints. In this embodiment, the mathematical representation of the spatial belief network is as follows: Where A is the adjacency matrix (A ij =1 if and only if ), Storage edge dependency strength , The four-dimensional tensor records the paired potential function values. This structure decodes the multidimensional features of the spatial spectrum cube into a probabilistic graphical model, where nodes represent entity state uncertainties and edges encode spatial dependency strengths.

[0045] In other words, the confidence network construction achieves a triple transformation: 1) geometric inverse mapping, specifically, back-projecting the regular grid cube C onto the irregular geographic entity E. i ;2) Feature decoding, specifically, converting spectral features into entity state probability distributions through neural networks;3) Relationship reconstruction, specifically, integrating the original spatial constraints with the spectral feature similarity quantization dependency strength.

[0046] The processing ensures that the high-dimensional spectral features generated by S203 are effectively transformed into a probabilistic inference framework, providing structured input for the probability field optimization of S302. The spatial belief network is mathematically a factor graph model, and its joint probability distribution is defined as: , where Z is a normalization constant, and this model reflects the state uncertainty and spatial dependence of geographic entities.

[0047] S302, perform probability field optimization processing on the spatial confidence network based on spatial constraint propagation to generate an optimized probability field. The probability field optimization processing is to iteratively update the node confidence along the spatial constraint edge using a message passing algorithm. In some implementations, the probability field optimization process involves iteratively updating the node confidence along the spatial constraint edges using a message passing algorithm, including confidence initialization and spatial message propagation. Specifically, the confidence initialization process initializes the node potential function of the spatial confidence network generated in step S301. Set as initial confidence distribution ,in ( ) represents entity E i The initial probability of belonging to category c. This initial probability originates from the feature decoding result of the spatial hierarchy cube C, and its value is related to the entity state probability distribution s from step S301. i Maintain consistency.

[0048] In this embodiment, the spatial message propagation process executes a confidence iterative update algorithm, which is implemented based on the LoopyBelief Propagation framework. The algorithm is defined as starting from entity node E in iteration round t. i Pass to neighbor node E j message vector Its computational fusion space depends on the weights. With potential function constraints: , where N(i) is entity E i The set of neighboring nodes (i.e., the set of nodes with spatial constraints) (node ​​index set) Exclude node E from the neighbor set j A subset of; It is from E in the t-th iteration i Transmitted to E j The message vector, For entity E i The initial class probabilities (derived from spectral feature decoding); Coding space dependency strength ( E represents i In category c i (The spectral eigenvectors below).

[0049] It should be noted that the node confidence update formula is defined as follows: ,in To ensure the normalization factor The iteration terminates when the change in the Frobenius norm of the confidence matrix between two iterations is less than a threshold. : Output the converged confidence matrix. .

[0050] The optimized probability field is mathematically represented as a binary tuple. ,in Storage-optimized entity state probability distribution The spatial dependency strength matrix is ​​defined for S301. This structure fully preserves the topological relationships of the spatial confidence network and corrects spatial inconsistencies in the initial confidence scores through a message propagation mechanism.

[0051] Preferably, a damping factor is introduced into message passing. Improve convergence stability: .

[0052] Furthermore, the boundary potential function Add category compatibility matrix : ,in Encoding prior knowledge of spatial co-occurrence among encoding categories, specifically manifested as: high compatibility ( Runways and taxiways for aircraft, berths and docked ships, roads and moving vehicles; low compatibility ( Aircraft and ocean areas, ships and runways, vehicles and off-road areas.

[0053] In this embodiment, the essence of spatial message propagation processing is to optimize the probability field through a three-stage mechanism: 1) local evidence injection, specifically, the node potential function. 1) Inject initial confidence based on spectral features; 2) Spatial constraint diffusion, specifically, along dependency edges. Message passing, weight 3) Global consistency convergence, specifically, iterative correction to ensure that the confidence levels of adjacent nodes satisfy spatial dependencies. This process integrates local observation evidence of geographic entities with global spatial topological constraints to generate an optimized probability distribution that satisfies spatial consistency. This probability field provides input to the S303 evidence synthesis process, and its output... It is directly used as the basic probability assignment for the DS evidence framework.

[0054] S303, perform evidence synthesis processing on the optimized probability field to generate a geographical evidence chain. The evidence synthesis processing involves mapping node confidence to the DS evidence framework and calculating the overall credibility by weighting the spatial consistency factor.

[0055] In some implementations, the evidence synthesis process maps node confidence levels to the DS evidence framework and calculates the overall credibility through spatial consistency factor weighting, including basic probability assignment and spatial weighted fusion processing. Specifically, the basic probability assignment process transforms the optimized probability field generated in S302... Entity state probability distribution This is converted into the basic probability assignment function of the DS evidence theory. The input to the process is the entity confidence matrix. (Where C = {runway, aircraft, berth, ship, road, vehicle}), the output is a set of entity evidence. The assignment function is defined as follows: ,in To identify the frame a subset of It includes all entity categories and global uncertain states.

[0056] In this embodiment, the spatial weighted fusion process is based on the spatial dependency strength between entities. Calculate the spatial consistency factor The process involves weighted synthesis of BPA. 1. Space factor calculation: Entity E i Its spatial consistency factor Quantify the strength of its topological constraints with neighboring entities: ,in The category compatibility matrix elements defined for S302 (e.g., runway-aircraft compatibility value of 0.9), and N(i) is the set of spatially adjacent entities.

[0057] 2. Weighted Dempster Synthesis: On the evidence set Sort by spatial factors in descending order to generate an ordered sequence of evidence. Execute the synthesis rules in sequence: ,in It is an identification framework that includes all entity categories and uncertain states. A, B, and C are... a subset of It is the spatial consistency factor of the kth piece of evidence (output from the spatial factor calculation step). The kth piece of evidence assigns a basic probability value to hypothesis B, and the conflict coefficient K is defined as follows: Iterative synthesis continues until all evidence is fused into a global BPA function m. g .

[0058] It should be noted that the aforementioned geographical evidence chain is ultimately represented as a sequence of quadruplets: ,in For entity E i The final category decision, For the confidence function, For entity timestamps (derived from the S1 geographic coordinate semantic library Ds).

[0059] The mathematical essence of the chain of evidence is a set of credibility distributions with spatiotemporal labels, whose decision rules satisfy: This rule inherits the spatial behavior constraints of traffic entities (such as the requirement that a runway must be adjacent to an aircraft entity).

[0060] Preferably, the spatial factor calculation incorporates dynamic weight decay: ,in For entity timestamps (derived from the S1 geographic coordinate semantic library Ds). Current system time or reference time base (such as the time of remote sensing image acquisition), T decay This is the time decay constant (unit: seconds), which controls the rate at which historical data weights decay. This enhances the credibility of recently high-dynamic entities (such as aircraft taking off and landing). Furthermore, the DS composite conflict coefficient K exceeds a threshold. At that time, a spatial relationship re-verification based on the S201 topological skeleton is triggered to ensure the geographical rationality of the evidence chain.

[0061] In this embodiment, the evidence synthesis process achieves two-stage knowledge fusion, including local uncertainty modeling and global spatial collaboration. The local uncertainty modeling preserves the cognitive uncertainty of entity state probabilities through the BPA function, and the global spatial collaboration utilizes spatial consistency factors. The joint decision-making of topologically constrained entities is strengthened. The processed output geographical evidence chain provides structured input for S4 spatiotemporal field reconstruction, and its confidence value directly drives the separation logic of static field primitives and dynamic field primitives.

[0062] S4, perform spatiotemporal field reconstruction processing on the geographic evidence chain to generate a hyperdimensional spatiotemporal map. The spatiotemporal field reconstruction processing is to achieve unified spatiotemporal modeling through dynamic-static separation and field coupling mechanism. Specifically, this step includes the following sub-steps: S401, The geographic evidence chain is processed by a field separation function to generate static field primitives and dynamic field primitives. The field separation function is to split the evidence chain into a static entity coordinate field and a dynamic trajectory vector field based on the entity motion attributes. In some implementations, the field separation function process involves splitting the evidence chain into a static entity coordinate field and a dynamic trajectory vector field based on entity motion attributes, including motion attribute quantization and primitive generation. Specifically, the motion attribute quantization process performs a dynamic evaluation of the entities in the geographic evidence chain generated in S303, and the input to the process is the entity decision category. and its spatiotemporal sequence data. The evaluation defines motion attribute scalars. : ,in The average velocity vector of the entity ( (Coordinates of time slice t) The maximum speed threshold is related to the category (e.g., 300 m / s for aircraft, 30 m / s for ships).

[0063] In this embodiment, the primitive generation process is based on motion properties. Implementing field separation: Specifically, the generation of static field primitives involves extracting... Entity constructs static coordinate field ,in The coordinates of the geometric center of the entity (derived from the S201 hierarchical topological decomposition). Static weights ( The importance coefficients are: runway = 1.0, road = 0.7); specifically, the dynamic field primitive generation extracts... Entity constructs dynamic trajectory vector field A single trajectory is defined as a time-space mapping function: , Trajectory vector v j By linear least squares fitting: .

[0064] It should be noted that static field inheritance is based on resource type definition of symbolic rules, while dynamic field implements real-time trajectory tracking, T j (t) Supports position interpolation calculations at any time. The mathematical representations of the static and dynamic field primitives are as follows: , A j b is the trajectory slope vector (velocity direction). j This is the intercept vector (initial position).

[0065] Preferably, static entity aggregation uses the spatial clustering algorithm DBSCAN, with parameters... (S1 proximity threshold), merge distance less than Similar entities are used to simplify rendering. Furthermore, dynamic trajectory segmentation introduces acceleration constraints: when... The time segment is divided into new trajectory segments to ensure that the motion model conforms to physical laws, such as the ship's turning rate limit, where a j It is entity E j The instantaneous acceleration vector, in its physical sense, represents the rate of change of the trajectory velocity. Change in velocity , Let be the velocity vector at time tk. Time interval, timestamps from the S303 evidence chain , It is a preset maximum acceleration threshold, with a range of values. In this embodiment, the maximum acceleration threshold for the aircraft is 3 m / s². 2 The speed of the ship is 0.3 m / s2 The vehicle speed is 2.5 m / s 2 .

[0066] In this embodiment, the field separation function processing achieves triple technical enhancement, including dynamic precise quantization, static field optimization, and dynamic field modeling. The dynamic precise quantization is combined with category priors ( ) and measured speed ( The static field optimization is achieved through importance weights s. i To achieve the required "differentiated labeling based on importance"; the dynamic field modeling is a linear trajectory function T j (t) Meets the computational efficiency requirements of "real-time trajectory tracking". The output provides input for S402 motion constraint coupling, and its separation logic directly supports the collaborative visualization of "fixed facilities and moving targets" in airport and port scenarios.

[0067] S402, Motion constraint coupling processing is performed on static field primitives and dynamic field primitives to generate a hyperdimensional spatiotemporal map. The motion constraint coupling processing is to embed the dynamic trajectory vector field into the static entity coordinate field through the terrain-driven algorithm to construct a spatiotemporal coordinate system with joint spatial behavior and terrain constraints.

[0068] In some implementations, the motion constraint coupling process involves embedding a dynamic trajectory vector field into a static entity coordinate field using a terrain-driven algorithm. Constructing a spatiotemporal coordinate system with joint spatial behavior and terrain constraints includes terrain constraint mapping and spatiotemporal coordinate synthesis. Specifically, the terrain constraint mapping process establishes the motion dependency between the dynamic trajectory and the static entity, and the input to this process is the static field primitive generated by S401. and dynamic field elements The mapping is achieved through spatial and semantic association functions: ,in Based on the current position of the trajectory Center, radius A static collection of entities within. It is divided into dynamic trajectory types (aircraft / ships / vehicles) and static entity types (runways / berths / roads). These are elements of the category compatibility matrix defined in S302. Static entity weights. Output is trajectory-terrain constraint weights. This characterizes the traction strength of a dynamic target under the influence of a static facility.

[0069] In this embodiment, the spatiotemporal coordinate synthesis process generates a hyperdimensional spatiotemporal graph. Its construction process is as follows: 1. Static node generation: Each static entity Mapped to nodes Attribute vector: A zero at the end indicates a static property; 2. Dynamic node generation: Each dynamic trajectory Discretized into a spatiotemporal node sequence Single node attributes: The last t k Mark dynamic timestamps; 3. Construction of constraint edges: when When creating dynamic nodes To static nodes Constraint edges: edge weight Integrating spatial correlation strength and temporal decay, among which A directed edge from the dynamic node (j,k) to the static node i represents the dynamic target at time t. k Constrained by the spatial behavior of static facilities. Terrain constraint weights for dynamic trajectory j Constraint activation threshold Dynamic target j at time t k The spatial coordinates are derived from the linear interpolation function of the S401 dynamic field primitive. The geometric center coordinates of static entity i are derived from the entity aggregation coordinates of the S201 hierarchical topological decomposition. The maximum spatial connection distance represents the effective spatial radius of the constraint; exceeding this distance disconnects the constraint. k The trajectory node timestamps originate from the S303 evidence chain, t now System current time, remote sensing image acquisition time / user operation time, T decay It is the time decay constant.

[0070] It should be noted that the mathematical representation of the hyperdimensional spacetime graph is as follows: , where N s N is the static number of entities. d K represents the number of dynamic trajectories, and K represents the number of nodes in the trajectory segment. Each row of the node attribute matrix V corresponds to a six-dimensional attribute vector. This data structure simultaneously encodes spatial constraints, behavioral coupling, and temporal evolution; the spatial constraints are the static facility location (x...). i ,y i ,z i The behavioral coupling is the dynamic target location; by Driven; the time evolution timestamp t k Supports real-time interpolation.

[0071] Preferably, the terrain-driven algorithm is compatible with heterogeneous scenes: when the static field lacks associated entities (such as ships in open water), it automatically downgrades to an absolute coordinate system. ( This ensures that spatiotemporal nodes can still be constructed for targets without static constraints, such as ship trajectories. Furthermore, the motion equation correction module adjusts the equations of motion based on constraint weights. Adjusting trajectory prediction: High weight ( ): The trajectory is locked to the associated static entity (such as the aircraft being forced to conform to the runway centerline during landing). low weight ( ): The trajectory direction deflects towards the static entity (e.g., a vehicle's orientation towards the road is corrected). In this embodiment, this step achieves triple technical enhancement, including cross-scenario compatibility, real-time guarantee, and behavioral physicality. The cross-scenario compatibility is achieved through conditional branching. It supports heterogeneous environments such as airports (with runways), ports (with berths), and highways (without fixed facilities); the real-time guarantee is a spatiotemporal node. Attribute vector pre-computation Linear interpolation during rendering: The physical behavior is the constraint edge weight. The driving dynamic targets conform to spatial behavior rules (such as prohibiting ships from crossing runways). The hyperdimensional spatiotemporal map provides input for the S5 holographic projection, and its six-dimensional attribute vector (spatial coordinates + static weights + dynamic constraints + timestamps) fully embodies the unified display capability of dynamic and static integrated resources.

[0072] S5, perform holographic projection conversion processing on the hyperdimensional spatiotemporal map to generate a holographic geographic projection map. The holographic projection conversion processing is to achieve three-dimensional visualization through phase field compression and diffraction reconstruction. Specifically, this step includes the following sub-steps: S501, perform holographic phase encoding processing on the hyperdimensional spatiotemporal diagram to generate compressed phase primitives. The holographic phase encoding processing is to convert the spatiotemporal coordinates into the phase distribution of light waves through a curvature phase mapping algorithm. In some implementations, the holographic phase encoding process involves converting spatiotemporal coordinates into light wave phase distributions using a curvature phase mapping algorithm, including spatiotemporal curvature calculation and phase field compression. Specifically, the spatiotemporal curvature calculation process converts the six-dimensional attribute vectors of the hyperdimensional spatiotemporal graph nodes... Convert to geometric curvature features, where spatial coordinates x, y, z define the target position, and static weights or dynamic constraints. The system characterizes entity importance or motion coupling strength, uses a category identifier (c) to distinguish entity types, and a timestamp (t) to record temporal states; the curvature calculation employs a gradient-driven hyperbolic tangent transform. , here This is the wavelength coefficient, and its value is strictly bound to the entity type: for example, for an airport runway... aircraft take Ship take This coefficient adjusts the curvature response sensitivity of static facilities and dynamic targets. The phase field compression process maps curvature characteristics to the phase distribution of light waves: first, complex phase primitives are generated. ,in The wavelength λ of the light wave is fixed at 532 nanometers. The normal vector n is assigned a value based on the dynamic nature of the entity—static entities take the vertical upward vector (0,0,1), while dynamic entities take the vector of their motion direction. , The trajectory direction angle, spatial coordinates r=(x,y,z) directly inherit node attributes; subsequently, spatial compression is implemented through a one-dimensional convolutional coding network: ,in For learnable projection matrix (D f =256 (number of feature channels) To correct the linear unit activation function max(0,x), It is a one-dimensional convolutional encoder (three convolutional kernels, size 3, stride 2). To output a compressed phase vector, the convolutional network is configured with three convolutional kernels (3×3 size, stride 2, number of channels 64-128-256), reducing the input dimension from the original number of nodes N (approximately...). Compression to logarithmic scale (on the order of magnitude) (Typical value 20-dimensional), the output compressed phase primitive is represented as a three-dimensional complex tensor. Its spatial dimensions Based on geographic grid sampling step size Meter determination (calculation formula) The phase channel depth Dc stores compressed complex phase information, and the phase value of each grid cell is the arithmetic mean of the phase primitives of the nodes in the spatial cell.

[0073] Preferably, the curvature calculation includes a time dimension modulation term. Dynamic entity oscillation amplitude And frequency Set by type (0.5 Hz for aircraft, 0.1 Hz for ships), static entity retention. This design generates periodic pulsations in the phase field of targets such as aircraft and ships, enhancing the dynamic recognition of holographic projection. Furthermore, the output layer of the compressed network employs a complex number separation encoding strategy: amplitude component modulus... Direct output, phase component The data is converted into a Hough-coded bitstream, ultimately achieving a data compression ratio of 83:1, ensuring the transmission efficiency of real-time holographic rendering.

[0074] In this embodiment, the processing achieves the mapping of geospatial data to optical phase through a three-stage transformation: 1. Physical modeling: curvature term The corresponding wavefront radius of curvature reciprocal, normal vector term Encodes the direction of wave vector propagation; 2. Domain Adaptation: Wavelength Coefficient and oscillation parameters Customized according to the physical characteristics of the transportation entity; 3. Computational optimization: The convolutional compression network reduces tens of millions of nodes to a complex number of 20 dimensions, meeting the requirements for millisecond-level real-time encoding.

[0075] The compressed phase primitive P provides a standardized input for S502 coherent diffraction reconstruction, and its tensor structure fully carries the spatial distribution characteristics of static facilities and the motion trajectory information of dynamic targets.

[0076] S502, coherent diffraction reconstruction is performed on the compressed phase primitive to generate a holographic geographic projection map. The coherent diffraction reconstruction is to convert the phase distribution into a geographic projection that supports three-dimensional spatial interaction through optical wave interference.

[0077] In some implementations, the coherent diffraction reconstruction involves converting the phase distribution into a geographic projection that supports three-dimensional spatial interaction through optical wave interference, including wavefront reconstruction processing and interferometric rendering processing. Specifically, the wavefront reconstruction processing converts the compressed phase primitives generated in S501... The decoding process yields a three-dimensional light wave field distribution. The input to this process is a compressed complex phase tensor P, and the output is a complex amplitude wavefront. The reconstruction follows the Fresnel diffraction integral principle: Where z is the reconstruction depth (dynamically calculated from the user's viewpoint position), and i is the imaginary unit satisfying... , Phase delay factor, here wave number ( ), The input compressed phase primitive originates from step S501. Phase element plane coordinates, Reconstruct the spatial target point coordinates (z is the user's viewpoint depth), and cover the phase primitive space grid with the integral domain. The integral is calculated more quickly using a Fast Fourier Transform: first, the Fourier transform of the phase primitives is calculated: Where F represents the two-dimensional Fourier transform operator, and then multiply by the transfer function: Then, the inverse Fourier transform is performed to obtain the wavefront: The transfer function The physical essence of this is the propagation response of light waves in free space, where F is the Fourier transform operator and H is the transfer function. These are spatial frequency coordinates.

[0078] In this embodiment, the interferometric rendering process converts the complex amplitude wave field into an interactive holographic geographic projection map. The processing involves three steps: first, calculating the intensity distribution. Secondly, the reference light interference fringes are superimposed. ,in The light intensity was set to a constant value of 10.0. Let arg(U) be the phase of the object light. The reference optical phase (based on the user's parallax angle) calculate: Finally, it is output through four RGB-D channels: RGB channels (Tone mapping phase angle, brightness mapping intensity), D channel (The depth channel records the location of the maximum amplitude.)

[0079] It should be noted that the 3D interactive function is implemented in conjunction with the gesture recognition module through the depth channel D(x,y): when the user touches the projection surface, the spatial coordinates (x,y) are... u ,y u Trigger entity query: ,in This is the depth weighting factor (default 0.7), which returns the nearest entity. The attribute information (type, speed, confidence level, etc.) is displayed in an overlay. Preferably, the interference fringes dynamically adapt to the user's viewing angle: reference light incident angle. , The viewing angle is set to ensure true 3D stereoscopic effect through binocular parallax. Furthermore, wavefront reconstruction introduces an adaptive focusing algorithm: focusing on the center of the user's gaze region (x... c ,y c Improve depth resolution The outer area was reduced to Optimize the allocation of computing resources.

[0080] In this embodiment, the processing outputs the map generation process: the holographic geographic projection map retains static facility symbolic annotations (runway icons, berth numbers) while dynamically rendering moving targets such as aircraft landing trajectories and ship entry paths. Users can rotate and zoom in to view details of any feature using gestures. The projection supports terminals such as helmet displays and holographic screens, and its RGB-D data stream is rendered in real time through the OpenGL pipeline, with a frame rate ≥30fps to meet the requirements for smooth interaction, forming a complete closed-loop technology chain from multi-source data fusion to 3D interactive visualization.

[0081] Based on the description of the map generation method embodiments based on multi-source heterogeneous geographic information processing above, this application also discloses a map generation system based on multi-source heterogeneous geographic information processing. The map generation system based on multi-source heterogeneous geographic information processing can be a computer program (including program code) that runs the aforementioned map generation method based on multi-source heterogeneous geographic information processing. Please see the appendix. Figure 4 As shown, a map generation system based on multi-source heterogeneous geographic information processing can run the following units: The acquisition unit 110 is used to acquire data to be processed. The data to be processed includes a geographic feature attribute library, a geographic coordinate semantic library, and a geographic reference image library. The geographic feature attribute library is a database containing structured fields of geographic entities. The geographic coordinate semantic library is a set of natural language texts carrying geographic coordinate descriptions. The geographic reference image library is a set of remote sensing images with geographic registration parameters. The spatial spectrum reconstruction unit 120 is used to perform spatial spectrum reconstruction processing on the data to be processed, thereby generating a spatial spectrum cube. The spatial spectrum reconstruction processing is achieved through a three-level fusion mechanism of topological skeleton construction, semantic constraint binding and spectrum encoding. The confidence propagation fusion unit 130 is used to perform confidence propagation fusion processing on the spatial spectrum cube to generate a geographical evidence chain. The confidence propagation fusion processing involves constructing a confidence network from the spatial spectrum data, combining it with probability field optimization based on spatial constraints, and completing the evidence synthesis processing through the DS framework. The spatiotemporal field reconstruction unit 140 is used to perform spatiotemporal field reconstruction processing on the geographic evidence chain, thereby generating a hyperdimensional spatiotemporal map. The spatiotemporal field reconstruction processing is to achieve unified spatiotemporal modeling through dynamic-static separation and field coupling mechanism. The holographic projection conversion unit 150 is used to perform holographic projection conversion processing on the hyperdimensional spatiotemporal map to generate a holographic geographic projection map. The holographic projection conversion processing achieves three-dimensional visualization through phase field compression and diffraction reconstruction.

[0082] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the concept described herein through the above teachings or related technologies or knowledge. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. A map generation method based on multi-source heterogeneous geographic information processing, characterized in that, The method includes the following steps: S1, Obtain the data to be processed. The data to be processed includes a geographic feature attribute library, a geographic coordinate semantic library, and a geographic reference image library. The geographic feature attribute library is a database containing structured fields of geographic entities. The geographic coordinate semantic library is a collection of natural language texts carrying geographic coordinate descriptions. The geographic reference image library is a set of remote sensing images with geographic registration parameters. S2, perform spatial spectrum reconstruction processing on the data to be processed to generate a spatial spectrum cube. The spatial spectrum reconstruction processing is a three-level fusion mechanism of topological skeleton construction, semantic constraint binding and spectrum encoding. S3, perform confidence propagation fusion processing on the spatial spectrum cube to generate a geographic evidence chain. The confidence propagation fusion processing involves constructing a confidence network from the spatial spectrum data, combining it with probability field optimization based on spatial constraints, and completing the evidence synthesis processing through the DS framework. S4, perform spatiotemporal field reconstruction processing on the geographic evidence chain to generate a hyperdimensional spatiotemporal map. The spatiotemporal field reconstruction processing is to achieve unified spatiotemporal modeling through dynamic-static separation and field coupling mechanism. S5, perform holographic projection conversion processing on the hyperdimensional spatiotemporal map to generate a holographic geographic projection map. The holographic projection conversion processing is to achieve three-dimensional visualization through phase field compression and diffraction reconstruction.

2. The map generation method based on multi-source heterogeneous geographic information processing according to claim 1, characterized in that, Step S2 includes the following sub-steps: S201, perform hierarchical topological decomposition on the geographic element attribute library to generate a spatial topological skeleton. The hierarchical topological decomposition is based on the geographic entity spatial relationship rule library to construct a node network and spatial constraint edges. S202, based on the geographic coordinate semantic library and combined with the spatial topology skeleton, semantic topology binding processing is performed to generate a semantic constraint graph. The semantic topology binding processing is to map text coordinate entities to topology nodes and generate a semantic relationship matrix through a graph neural network. S203, perform spectral fusion processing based on the geographic reference image library and the semantic constraint map to generate a spatial spectral cube. The spectral fusion is to extract image features through a spatial convolutional network and align them with the geographic coordinates of the semantic constraint map, and then use a spectral coding algorithm to generate a three-dimensional spatial spectral data block.

3. The map generation method based on multi-source heterogeneous geographic information processing according to claim 1, characterized in that, Step S3 includes the following sub-steps: S301, perform confidence network construction processing on the spatial hierarchy cube to generate a spatial confidence network. The confidence network is constructed by mapping the spatial hierarchy data block to a probabilistic graphical model, where nodes represent geographic entities and edges represent spatial dependencies. S302, perform probability field optimization processing on the spatial confidence network based on spatial constraint propagation to generate an optimized probability field. The probability field optimization processing is to iteratively update the node confidence along the spatial constraint edge using a message passing algorithm. S303, perform evidence synthesis processing on the optimized probability field to generate a geographical evidence chain. The evidence synthesis processing involves mapping node confidence to the DS evidence framework and calculating the overall credibility by weighting the spatial consistency factor.

4. A map generation method based on multi-source heterogeneous geographic information processing according to any one of claims 1-3, characterized in that, Step S4 includes the following sub-steps: S401, The geographic evidence chain is processed by a field separation function to generate static field primitives and dynamic field primitives. The field separation function is to split the evidence chain into a static entity coordinate field and a dynamic trajectory vector field based on the entity motion attributes. S402, Motion constraint coupling processing is performed on static field primitives and dynamic field primitives to generate a hyperdimensional spatiotemporal map. The motion constraint coupling processing is to embed the dynamic trajectory vector field into the static entity coordinate field through the terrain-driven algorithm to construct a spatiotemporal coordinate system with joint spatial behavior and terrain constraints.

5. A map generation method based on multi-source heterogeneous geographic information processing according to claim 4, characterized in that, Step S5 includes the following sub-steps: S501, perform holographic phase encoding processing on the hyperdimensional spatiotemporal diagram to generate compressed phase primitives. The holographic phase encoding processing is to convert the spatiotemporal coordinates into the phase distribution of light waves through a curvature phase mapping algorithm. S502, coherent diffraction reconstruction is performed on the compressed phase primitive to generate a holographic geographic projection map. The coherent diffraction reconstruction is to convert the phase distribution into a geographic projection that supports three-dimensional spatial interaction through optical wave interference.

6. A map generation method based on multi-source heterogeneous geographic information processing according to claim 2, characterized in that, The hierarchical topology decomposition process in S201 is based on the geographic entity spatial relationship rule base to construct a node network and spatial constraint edges, including entity topology relationship parsing and graph structure generation. The semantic topology binding process in S202 maps text coordinate entities to topology nodes and generates a semantic relationship matrix through a graph neural network, including entity node alignment processing and semantic relationship reasoning processing; the spectral fusion in S203 extracts image features through a spatial convolutional network and aligns them with the geographic coordinates of the semantic constraint graph, and generates a three-dimensional spatial spectral data block using a spectral coding algorithm, including geographic alignment feature extraction processing and spectral coding aggregation processing.

7. A map generation method based on multi-source heterogeneous geographic information processing according to claim 3, characterized in that, The confidence network construction in S301 involves mapping spatial spectral data blocks to a probabilistic graphical model, including entity node remapping and spatial dependency edge reconstruction. The probability field optimization in S302 involves iteratively updating node confidence along spatial constraint edges using a message passing algorithm, including confidence initialization and spatial message propagation. The evidence synthesis in S303 involves mapping node confidence to a DS evidence framework and calculating comprehensive credibility using spatial consistency factors, including basic probability assignment and spatial weighted fusion.

8. A map generation method based on multi-source heterogeneous geographic information processing according to claim 4, characterized in that, The field separation function in S401 processes the evidence chain into a static entity coordinate field and a dynamic trajectory vector field based on the entity's motion attributes, including motion attribute quantization and primitive generation. The motion constraint coupling process in S402 involves embedding the dynamic trajectory vector field into the static entity coordinate field through a terrain-driven algorithm to construct a spatiotemporal coordinate system for joint spatial behavior and terrain constraints, including terrain constraint mapping processing and spatiotemporal coordinate synthesis processing.

9. A map generation method based on multi-source heterogeneous geographic information processing according to claim 5, characterized in that, The holographic phase encoding process in S501 is to convert spatiotemporal coordinates into light wave phase distribution through a curvature phase mapping algorithm, including spatiotemporal curvature calculation and phase field compression. The coherent diffraction reconstruction in S502 is a process of converting the phase distribution into a geographic projection that supports three-dimensional spatial interaction through optical wave interference, including wavefront reconstruction processing and interferometric rendering processing.

10. A map generation system based on multi-source heterogeneous geographic information processing, characterized in that, The system includes: The acquisition unit is used to acquire data to be processed, which includes a geographic feature attribute library, a geographic coordinate semantic library, and a geographic reference image library. The geographic feature attribute library is a database containing structured fields of geographic entities, the geographic coordinate semantic library is a set of natural language text carrying geographic coordinate descriptions, and the geographic reference image library is a set of remote sensing images with geographic registration parameters. The spatial spectrum reconstruction unit is used to perform spatial spectrum reconstruction processing on the data to be processed, thereby generating a spatial spectrum cube. The spatial spectrum reconstruction processing is achieved through a three-level fusion mechanism of topological skeleton construction, semantic constraint binding, and spectrum encoding. The confidence propagation fusion unit is used to perform confidence propagation fusion processing on the spatial spectrum cube to generate a geographical evidence chain. The confidence propagation fusion processing involves constructing a confidence network from the spatial spectrum data, combining it with probability field optimization based on spatial constraints, and completing the evidence synthesis processing through the DS framework. The spatiotemporal field reconstruction unit is used to perform spatiotemporal field reconstruction processing on the geographic evidence chain, thereby generating a hyperdimensional spatiotemporal map. The spatiotemporal field reconstruction processing is to achieve unified spatiotemporal modeling through dynamic-static separation and field coupling mechanism. The holographic projection conversion unit is used to perform holographic projection conversion processing on the hyperdimensional spatiotemporal map to generate a holographic geographic projection map. The holographic projection conversion processing achieves three-dimensional visualization through phase field compression and diffraction reconstruction.

Images