Oblique photograph three-dimensional surface model geological modeling system and method

Abstract

The present application relates to the technical field of three-dimensional geological modeling and geological mapping data processing, and particularly relates to a tilt photography three-dimensional surface model geological modeling system and method, comprising data acquisition, scene construction, boundary processing, geological body modeling, profile linkage and result output modules. The data acquisition module fuses tilt photography images and terrain data to generate a three-dimensional surface model through a motion recovery structure algorithm; the geological body modeling module constructs a closed geological body through implicit surface reconstruction and constrained triangulation; the profile linkage module calls a graph neural network to predict stratigraphic intersection trends, and realizes two-dimensional and three-dimensional real-time synchronization based on event driving and sparse voxel hash mapping; and the result output module performs data verification and generates a multi-dimensional geological report. The scheme eliminates the dependence on third-party tools, improves the efficiency of geological mapping, and solves the problem of profile linkage delay.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of three-dimensional geological modeling and geological mapping data processing, and particularly relates to a tilt photography three-dimensional surface model geological modeling system and method. BACKGROUND

[0002] The tilt photography measurement technology uses multi-angle tilt cameras to obtain high-resolution image data of the surface, integrates the spatial coordinates and pixel information of the overlapping images through a photogrammetry processing flow to construct a three-dimensional digital surface model; the model can accurately reflect the surface morphology and texture details in a GIS environment, and further support geological modeling work, and geologists can deduce the stratum distribution and geological structure trend according to the surface feature analysis and three-dimensional visualization of the model to optimize the decision basis for resource exploration and environmental assessment.

[0003] The existing geological mapping relies on manual field surveying and mapping and two-dimensional drawing paper drawing, and the data acquisition cycle is long and it is difficult to completely express the three-dimensional geological structure; the existing modeling software lacks the ability to originally support the tilt photography model, forcing the geological personnel to rely on third-party tools such as ArcGIS to perform data preprocessing and format conversion, resulting in a fragmented modeling process and complex operation; at the same time, the stratum boundary fitting precision is insufficient and the profile cutting function lacks a three-dimensional model dynamic feedback mechanism, for example, when analyzing the outcrop boundary of a certain area, the surface morphology needs to be corrected multiple times by switching software, or after editing the two-dimensional profile node, the three-dimensional fault model cannot be synchronized in real time, which significantly restricts the geological interpretation efficiency. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides a tilt photography three-dimensional surface model geological modeling system and method to solve the problems of low geological mapping efficiency, three-dimensional modeling relying on third-party tools and insufficient profile analysis dynamic linkage.

[0005] To solve the above technical problems, the specific technical solutions of the present application are as follows:

[0006] In a first aspect, the present application provides a tilt photography three-dimensional surface model geological modeling system, comprising:

[0007] A data acquisition module acquires tilt photography images and terrain data, and generates a three-dimensional surface model through a motion recovery structure algorithm;

[0008] A scene construction module receives the three-dimensional surface model, loads the model data and dynamically renders a three-dimensional scene according to the spatial coordinates and lithology attributes;

[0009] A boundary processing module receives the visual scene data of the scene construction module, registers the field measured points with the three-dimensional surface model and optimizes the collected stratum boundary;

[0010] a geological body modeling module receiving the stratigraphic boundary, generating a continuous stratigraphic interface and topologically processing a closed geological body structure;

[0011] a profile linkage module receiving the closed geological body, cutting the closed geological body to generate a profile map, and synchronizing the profile map with the closed geological body in real time through event driving;

[0012] an achievement output module receiving the closed geological body and attribute data, storing data and exporting standardized geological modeling achievements;

[0013] The data acquisition module outputs a three-dimensional surface model to the scene construction module, the scene construction module outputs visual scene data to the boundary processing module, the boundary processing module outputs optimized stratigraphic boundaries to the geological body modeling module, the geological body modeling module outputs closed geological bodies to the profile linkage module and the achievement output module, and the profile linkage module feeds back profile modification events to the geological body modeling module in real time to trigger three-dimensional model updates.

[0014] Further, the tilt photography three-dimensional surface model geological modeling system of the present application, the data acquisition module:

[0015] initial point cloud data is generated by dense matching through a multi-view stereo algorithm;

[0016] The initial point cloud data, the digital elevation model and the work area boundary are fused to construct a three-dimensional surface model;

[0017] The generated three-dimensional surface model is stored in blocks and automatically converted into a coordinate system.

[0018] Further, the tilt photography three-dimensional surface model geological modeling system of the present application, the scene construction module:

[0019] After loading the three-dimensional surface model, the resolution of the model is switched using a level of detail algorithm;

[0020] The lithology attribute data is analyzed, and lithology codes are mapped to a color space;

[0021] According to the distribution of the mapped lithology-color data, a stratigraphic lithology distribution heat map is displayed in real time.

[0022] Further, the tilt photography three-dimensional surface model geological modeling system of the present application, the boundary processing module:

[0023] Using the three-dimensional surface model, corrected point cloud data is generated by registering field measurement points through an iterative closest point algorithm;

[0024] For the corrected point cloud data, an optimized point set is generated by applying a random sample consensus algorithm to remove abnormal nodes;

[0025] The optimized point set is input as a curved surface fitting, and a genetic algorithm is used for optimization to generate a stratigraphic boundary.

[0026] Further, the oblique photography three-dimensional surface model geological modeling system, the geological body modeling module:

[0027] The optimized stratigraphic boundary is obtained, and a continuous stratigraphic interface is generated through an implicit curved surface reconstruction algorithm;

[0028] Based on the continuous stratigraphic interface, the top and bottom surfaces are constrained to construct the topological structure of the triangular subdivision;

[0029] After fusing the work area boundary data, the closed geological body is generated by stitching the side ring surface.

[0030] Further, the oblique photography three-dimensional surface model geological modeling system, the profile linkage module:

[0031] After receiving the geological body, a graph neural network is called to analyze the stratigraphic intersection trend in the cutting profile;

[0032] Based on the stratigraphic intersection trend, an image segmentation algorithm is used to extract the profile structural features;

[0033] When a profile editing event is detected, the three-dimensional geological body is updated in real time through sparse voxel hash mapping combined with the profile structural features.

[0034] Further, the oblique photography three-dimensional surface model geological modeling system, the result output module:

[0035] After receiving the geological body and attribute data, a consistency checking algorithm for stratigraphic interface and topographic data is executed and the checking result is output;

[0036] The lithology coding library in the attribute data is read to analyze the lithology pattern data;

[0037] Based on the checking result and the lithology pattern data, a column chart is generated;

[0038] The column chart and three-dimensional geological body model data are integrated, and a geological report with hyperlinks is output.

[0039] Further, the oblique photography three-dimensional surface model geological modeling system, the boundary processing module generates an optimized point set, which is filtered by a random sampling consensus algorithm after filtering out abnormalities, and outputs a clean stratigraphic boundary to the geological body modeling module;

[0040] The geological body modeling module completes implicit curved surface reconstruction based on the clean stratigraphic boundary, and generates stratigraphic interface curvature feature data;

[0041] The stratum interface curvature feature data is fed back to a genetic algorithm optimization link of the boundary processing module, and a curved surface fitting weight parameter is dynamically adjusted.

[0042] Further, the tilt photography three-dimensional surface model geological modeling system, a user drags a stratum boundary node in a two-dimensional profile view of the profile linkage module, triggers a topology reconstruction event captured by an event listener;

[0043] The topology reconstruction event carries a node displacement vector, and drives the geological body modeling module to update a three-dimensional stratum interface by calling a sparse voxel hash mapping method.

[0044] After the stratum morphology is corrected in real time based on the profile structure features, the synchronization delay is controlled to be in the order of milliseconds.

[0045] In the second aspect, the tilt photography three-dimensional surface model geological modeling method is applied to the tilt photography three-dimensional surface model geological modeling system, and includes the following steps:

[0046] Step 1, obtain tilt photography images and terrain data, and generate a three-dimensional surface model by a motion recovery structure algorithm;

[0047] Step 2, receive the three-dimensional surface model, load model data, and dynamically render a three-dimensional scene according to spatial coordinates and lithological properties;

[0048] Step 3, receive visual scene data of the scene construction module, register field measurement points and the three-dimensional surface model, and optimize collected stratum boundaries;

[0049] Step 4, receive the stratum boundaries, generate continuous stratum interfaces, and topologically process closed geological body structures;

[0050] Step 5, receive the closed geological bodies, cut the closed geological bodies to generate a profile map, and synchronize the profile map and the closed geological bodies in real time through event driving;

[0051] Step 6, receive the closed geological bodies and attribute data, store data, and export standardized geological modeling results.

[0052] The present application has the following advantages:

[0053] The present application directly generates an OSGB format three-dimensional surface model by native fusion processing of oblique photography images and terrain data, replaces manual field surveying and mapping, and compresses the data collection period; the built-in motion recovery structure algorithm and multi-view stereo matching engine realize integrated processing from the original image to the three-dimensional model, skip the format conversion link of third-party tools such as ArcGIS, avoid the fragmentation of the modeling process; combined with the random sampling consensus algorithm and genetic optimization, the stratigraphic boundary is automatically generated, which significantly reduces the amount of manual correction operation; through the sparse voxel hash mapping technology, local incremental updating is realized, and the event-driven mechanism feeds back the displacement of the profile node to the three-dimensional geological body in real time, millisecond-level synchronous delay breaks through the technical bottleneck of traditional two-dimensional and three-dimensional model separate updating, and improves the geological interpretation efficiency and model interaction experience. BRIEF DESCRIPTION OF DRAWINGS

[0054] In order to more clearly illustrate the technical solutions of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. Obviously, for those skilled in the art, other drawings can also be obtained from the drawings without creative labor.

[0055] Figure 1 The flowchart of the oblique photography three-dimensional surface model geological modeling method provided by the embodiments of the present application. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical scheme and advantages of the present application clearer, the following will combine the specific embodiments of the present application and the corresponding drawings to clearly and completely describe the technical scheme of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor belong to the scope of protection of the present application. The technical scheme provided by each embodiment of the present application will be described in detail below in combination with the drawings. In order to better understand the purpose of the present application, the present application will be further described in detail below.

[0057] In the first aspect, the present application provides an oblique photography three-dimensional surface model geological modeling system, comprising:

[0058] A data acquisition module acquires oblique photography images and terrain data, and generates a three-dimensional surface model through a motion recovery structure algorithm;

[0059] A scene construction module receives the three-dimensional surface model, loads model data and dynamically renders a three-dimensional scene according to spatial coordinates and lithological properties;

[0060] A boundary processing module receives the visual scene data of the scene construction module, registers the field measured points with the three-dimensional surface model, and optimizes the collected stratigraphic boundary;

[0061] The geological body modeling module receives the stratigraphic boundaries, generates continuous stratigraphic interfaces, and performs topological processing to close the geological body structure.

[0062] The profile linkage module receives the closed geological body, cuts the closed geological body to generate a profile view, and synchronizes the profile view and the closed geological body in real time through event-driven synchronization.

[0063] The output module receives the closed geological body and attribute data, stores the data, and exports standardized geological modeling results.

[0064] Specifically, the data acquisition module outputs a 3D surface model to the scene construction module, the scene construction module outputs visualized scene data to the boundary processing module, the boundary processing module outputs optimized stratigraphic boundaries to the geological body modeling module, the geological body modeling module outputs closed geological bodies to the profile linkage module and the results output module, and the profile linkage module provides real-time feedback on profile modification events to the geological body modeling module to trigger 3D model updates.

[0065] The data acquisition module first acquires surface image data using a drone platform equipped with a multi-angle tilt camera, while integrating digital elevation models and work area boundary vector information. A motion reconstruction structure algorithm is used to calculate the image pose and generate sparse point clouds, which are then used to construct high-density point cloud data through multi-view stereo dense matching. The point clouds undergo texture mapping to generate a 3D surface model in OSGB / S3C format, which is stored in multi-scale blocks using a quadtree index and automatically performs WGS84 / UTM / model coordinate conversion. This module directly outputs the textured 3D surface model to the scene construction module, eliminating the need for format conversion by third-party tools such as ArcGIS in traditional workflows.

[0066] After receiving 3D surface model data, the scene construction module dynamically schedules model tiles based on a chunked loading mechanism. It employs a level-of-detail algorithm to automatically switch model resolution based on viewpoint distance, achieving real-time lightweight rendering. The module reads the chronostratigraphic information recorded in the lithology coding attribute table, associates it with a predefined lithology-color mapping table, and maps different lithology codes to the HSV color space. A lithology distribution heatmap is generated through GPU parallel computing and overlaid on the 3D surface model to create a visualization of the spatial distribution of geological lithology. The visualized scene data output by this module includes spatial coordinates and lithology color mapping relationships, providing an interface for stratigraphic boundary acquisition.

[0067] The boundary processing module, based on visualized scene data, launches an interactive boundary drawing tool on the surface of a 3D model. It uses an iterative nearest-point algorithm to register the measured GPS point cloud with the oblique photogrammetry model surface point cloud, correcting spatial positioning deviations to generate corrected point cloud data. A random sampling consensus algorithm is applied to dynamically detect and remove abnormal nodes deviating from the trend line, outputting an optimized point set. A genetic algorithm is used to optimize the weights of NURBS surface control points, fitting and generating smooth, continuous stratigraphic boundaries and binding them with age codes and lithological attributes. The optimized stratigraphic boundary data is then transferred to the geological body modeling module, replacing the traditional manual two-dimensional map acquisition method.

[0068] After receiving the stratigraphic boundary point set, the geological body modeling module uses an implicit surface reconstruction algorithm to generate a continuous stratigraphic interface triangular mesh. It constructs the topological relationship between the top and bottom surfaces of the strata through constrained Delaunay triangulation, and integrates the polygonal data of the work area boundaries. Boolean operations are applied to trim redundant triangular facets and automatically stitch lateral toruses to form a closed geological body structure. A binary mesh file containing XML metadata is output, supporting cross-project reuse for complex structures such as anticlines and synclines. The generated closed geological body is synchronously transmitted to the profile linkage module and the results output module.

[0069] The profile linkage module performs arbitrary-direction cutting operations on closed geological bodies, and calls a graph neural network to analyze the stratigraphic intersection trends along the profile direction. The U-Net image segmentation algorithm is used to identify fault traces and lithological boundary features in the profile image. When the user drags stratigraphic intersection nodes in the 2D profile view, an event listener captures the displacement vector and triggers a topology reconstruction event. Sparse voxel hash mapping technology is used to update the triangulation mesh of the 3D geological body interface in real time, maintaining topological consistency between the 2D and 3D models. Profile modification data is fed back to the geological body modeling module in real time to drive model updates.

[0070] The output module performs elevation consistency checks between stratigraphic interfaces and topographic data, detecting and marking spatial conflict areas between geological interfaces and the DEM. It parses pattern template data from the lithology coding library and automatically fills in lithology symbols for columnar sections. It integrates the verification results, columnar section data, and 3D geological models to generate a PDF geological report with model hyperlinks. The work area directory and attribute library are managed in tandem, outputting standardized data in GRD grid and XYZ point cloud formats to support resource exploration and environmental assessment decision-making applications.

[0071] Specifically, the data acquisition module of the oblique photogrammetry three-dimensional surface model geological modeling system of the present invention includes:

[0072] Initial point cloud data is generated through dense matching using a multi-view stereo algorithm;

[0073] By integrating the initial point cloud data, digital elevation model, and work area boundary, a three-dimensional surface model is constructed.

[0074] The generated 3D surface model is stored in blocks and the coordinate system is automatically converted.

[0075] The initial point cloud data was generated by acquiring multi-angle surface images using a five-lens oblique photography camera. The image pose parameters were calculated using a structure-of-motion (SOR) algorithm to generate a sparse point cloud foundation. Based on the principle of multi-view stereo matching, pixel-level disparity was calculated for adjacent images, and a semi-global matching algorithm was used to optimize depth map generation, gradually expanding the data into high-density 3D point cloud data. This point cloud retains the original image texture information, providing a geometric and textural foundation for the construction of the 3D surface model.

[0076] The 3D surface model construction involves spatially overlaying and analyzing the initial point cloud data with a digital elevation model (DEM). Gaussian filtering is used to eliminate point cloud noise and match the terrain elevation trend. The boundary vector file of the work area is loaded, and the point cloud elevation and terrain data are fused using a Kriging spatial interpolation algorithm to generate a continuous surface. A topologically connected triangular mesh is constructed based on the Poisson surface reconstruction principle, and image texture coordinates are combined to complete the model mapping process, outputting a georeferenced 3D surface model in OSGB format.

[0077] The model storage and coordinate transformation process involves spatial octree partitioning of the 3D terrain model, dividing it into spatial tiles based on a preset resolution threshold. A pyramid-level hierarchical storage structure is used to organize tiles of different scales, and a spatial index file is created to record spatial location metadata. The coordinate transformation engine is invoked to parse the model's built-in coordinate system parameters, and a seven-parameter Bursa transformation model is used to automatically convert between WGS84 geodetic coordinates, UTM projected coordinates, and the local model coordinate system, supporting cross-coordinate system data retrieval requirements.

[0078] Specifically, the scene construction module of the oblique photogrammetry 3D surface model geological modeling system of the present invention includes:

[0079] After loading the three-dimensional surface model, the resolution of the model is switched using a level of detail algorithm;

[0080] The lithological attribute data is parsed and the lithological codes are mapped to a color space;

[0081] Based on the mapped lithology-color data distribution, a real-time thermal map of stratigraphic lithology distribution is displayed.

[0082] Dynamic model resolution management, after loading the OSGB format 3D terrain model, employs a level-of-detail algorithm to adaptively switch model resolution based on the spatial relationship between the viewpoint and the model surface. As the viewpoint distance increases, lower-resolution model tiles automatically replace higher-resolution tiles; conversely, as the viewpoint approaches the terrain, higher-resolution tiles are switched. A quadtree spatial index structure manages model tile files at different levels, and frustum pruning removes model data outside the field of view. This process utilizes GPU instantiation rendering technology to achieve a smooth transition in model resolution, optimizing GPU memory usage while ensuring visualization quality.

[0083] The lithological attribute data parsing process reads an attribute database that correlates geological age and lithology. This database stores the age code (e.g., P1 represents Permian) and lithology code (e.g., T003 represents sandstone) of stratigraphic units. Regular expressions are used to parse the attribute field structure and extract the mapping relationship between lithology codes and spatial coordinates. An HSV color space conversion model is employed to map lithology codes to predetermined hue ranges, where hue components correspond to lithology categories, saturation reflects lithology purity, and brightness correlates with stratigraphic age. A lookup table is established to map lithology codes to color vectors, providing a data foundation for 3D scene colorization.

[0084] The lithological heatmap is generated in real time by spatially overlaying color mapping results with vertex data from a 3D surface model. A fragment shader program interpolates lithological color values ​​onto the model surface. A Gaussian kernel density estimation algorithm is used to calculate the lithological distribution density per unit area, and color transparency and brightness are dynamically adjusted based on the density value. Screen space ambient occlusion technology enhances the visual depth of terrain undulations, ultimately generating a heatmap reflecting the spatial distribution characteristics of lithology in the 3D scene. This heatmap updates in real time with changes in viewing angle, allowing geologists to intuitively analyze regional lithological distribution patterns.

[0085] Specifically, the boundary processing module of the oblique photography three-dimensional surface model geological modeling system of the present invention includes:

[0086] Using the aforementioned three-dimensional surface model, calibration point cloud data is generated by registering field measurement points using an iterative nearest point algorithm;

[0087] For this corrected point cloud data, a random sampling consensus algorithm is applied to remove abnormal nodes and generate an optimized point set;

[0088] The optimized point set is used as input for surface fitting, and a genetic algorithm is used to optimize and generate stratigraphic boundaries.

[0089] After loading the 3D terrain model, dynamic resolution adjustment is implemented using a level-of-detail algorithm to manage model precision based on the distance between the viewpoint and the model surface. When the viewpoint moves away from the target area, low-resolution model tiles are used; when the viewpoint moves closer, high-resolution tiles are switched. Different levels of model data are organized using a quadtree spatial index, and tiles within the visible range are dynamically loaded using frustum pruning technology. GPU instantiation rendering technology is employed to achieve a smooth transition in model precision, optimizing system resource usage while maintaining visual quality.

[0090] The lithological attribute data conversion process involves reading an attribute database that correlates geological time with lithological characteristics and analyzing the mapping relationship between stratigraphic age codes and lithological codes. The HSV color model is used to convert lithological codes into visual parameters: hue components correspond to lithological types, saturation represents the purity of lithological components, and brightness is associated with the stratigraphic age sequence. A lookup table is established to map lithological codes to HSV vectors, generating color space mapping rules to provide data support for the surface coloring of the 3D model.

[0091] The lithological distribution heatmap generation overlays color mapping results onto the surface of a 3D terrain model. A fragment shader in the graphics rendering pipeline interpolates lithological color values ​​onto the model's triangular facets. A kernel density estimation algorithm is used to analyze the frequency of lithological distribution per unit area, dynamically adjusting color transparency and brightness contrast based on density values. Screen space ambient occlusion technology is combined to enhance the visual depth of undulating terrain areas, ultimately generating a heatmap reflecting the spatial distribution characteristics of lithology. This heatmap updates in real-time with changes in viewing angle, enabling geologists to intuitively identify lithological distribution patterns within a 3D scene.

[0092] Specifically, the geological modeling module of the oblique photography three-dimensional surface model geological modeling system of the present invention includes:

[0093] The optimized stratigraphic boundary is obtained, and a continuous stratigraphic interface is generated using an implicit surface reconstruction algorithm;

[0094] Based on the continuous stratigraphic interface, constrained triangulation is used to construct the topological structure of the top and bottom surfaces;

[0095] After integrating the boundary data of the work area, a closed geological body is generated by stitching the side annulus.

[0096] Optimized stratigraphic boundary point cloud data is obtained through continuous stratigraphic interface generation, and the discrete point set is processed using an implicit surface reconstruction algorithm. Local surface patches are fitted using the moving least squares principle, and adjacent surface patches are connected through radial basis function interpolation to generate a smooth and continuous stratigraphic interface triangular mesh. This process preserves the geological structural features of the stratigraphic boundaries, such as the geometry of anticline axial planes or fault planes, forming a three-dimensional surface structure that conforms to geological laws.

[0097] The topology construction of the top and bottom surfaces is based on a continuous stratigraphic interface triangular mesh, extracting the vertex set of the interface boundary as constraints. The constrained Delaunay triangulation algorithm is applied to generate a network of topologically connected triangular facets within the stratigraphic interface. The quality of the triangles is optimized using the minimum angle maximization criterion to construct the topological structure of the top and bottom surfaces of the stratigraphy. This topological network records the adjacency relationships of vertices, edges, and faces, providing a mathematical basis for representing the spatial relationships of geological bodies.

[0098] The closed geological body generates bounding vector polygon data of the work area boundary and projects it onto the horizontal reference plane where the stratigraphic interface is located. A polygon clipping algorithm is used to cut redundant triangular facets around the stratigraphic interface, aligning the interface boundary with the work area boundary. A boundary stitching algorithm is then used to generate lateral triangular facets along the cut edges, connecting the top and bottom faces to form closed annular side faces. Vertex normal consistency checks are used to correct the orientation of the triangular facets, ultimately outputting a closed geological body mesh model with a complete bounding box.

[0099] Specifically, the oblique photography three-dimensional surface model geological modeling system of the present invention includes the profile linkage module:

[0100] After receiving the geological body, a graph neural network is invoked to analyze the stratigraphic intersection trend in the cut profile;

[0101] Based on the stratigraphic intersection trend, an image segmentation algorithm is used to extract the profile structural features;

[0102] When a profile editing event is detected, the three-dimensional geological body is updated in real time by combining the profile structural features with a sparse voxel hash map.

[0103] The stratigraphic intersection trend prediction function generates geological profiles along user-defined cutting directions after receiving a closed geological body mesh model. It utilizes a graph neural network model, using stratigraphic intersection nodes as vertices and topological connections between nodes as edges to construct a profile structure graph. Through graph convolution operations, it aggregates the geological attribute features of neighboring nodes, learning the spatial correlation between stratigraphic strike and dip. The function outputs the predicted extension trend of stratigraphic intersections in unexposed areas, aiding in the construction of a complete geological profile.

[0104] The structural feature extraction of the cross-section is based on stratigraphic intersection trend data, and an image segmentation algorithm with an encoder-decoder structure is used to process the cross-section image. The encoder extracts multi-scale features such as fault lines and lithological boundaries from the cross-section image through convolutional layers; the decoder fuses shallow texture information and deep semantic features to output pixel-level structural partitioning results. Morphological optimization of the segmentation results is performed by incorporating prior geological knowledge to generate structural feature maps that annotate fault traces and lithological contact relationships.

[0105] A real-time 3D geological body update mechanism is established to capture user interactions on the 2D profile. When a dragging event of a stratigraphic intersection node is detected, the node displacement vector and associated structural feature data are extracted. The affected area of ​​the 3D geological body is located using sparse voxel hash mapping, and triangulation reconstruction is performed only on local voxel blocks. Parallel computing pipelines are used to update the vertex coordinates of the stratigraphic interface, maintaining the topology of unmodified areas. The 3D scene engine synchronously refreshes the visualization results, achieving real-time feedback through 2D / 3D interaction.

[0106] Specifically, the oblique photography three-dimensional surface model geological modeling system of the present invention includes the following output module:

[0107] After receiving the geological body and attribute data, execute the consistency verification algorithm between the stratigraphic interface and the topographic data and output the verification result;

[0108] Read the lithology coding library from the attribute data and parse the lithology pattern data;

[0109] Based on the verification results and the lithological pattern data, a columnar chart is generated;

[0110] Integrate the bar charts and 3D geological model data to output a geological report with hyperlinks.

[0111] After receiving the closed geological mesh model and topographic data, the data consistency verification process uses spatial overlay analysis to detect elevation differences between the vertices of the stratigraphic interfaces and the digital elevation model. A constrained Delaunay triangulation is used to construct the spatial topological relationship between the stratigraphic interfaces and the topography, and the vertical distance from the interface vertices to the topographic surface is calculated. Areas exceeding a preset threshold are marked as elevation conflicts, and verification result data containing the coordinates of the conflict locations and their deviation values ​​is generated. This process ensures the spatial consistency between the geological model and the surface morphology.

[0112] The lithological pattern data parsing process reads the lithological code library from the attribute database and matches the lithological codes with predefined pattern templates. A vector graphics parsing algorithm is used to convert the pattern templates into SVG format fill path data, preserving the geometric features and proportional relationships of the lithological symbols. A mapping table from lithological codes to vector patterns is established to provide standardized graphic resources for filling lithological symbols in columnar charts.

[0113] The column chart generation process determines the vertical stratification structure of the column chart based on the stratigraphic sequence information in the verification results. Vector lithological symbols are filled into the corresponding stratigraphic stratification intervals based on lithological pattern data. A red semi-transparent warning layer is overlaid on areas marked with elevation conflicts to visually highlight data anomalies. An automatic layout algorithm is used to adjust the column width and legend position, generating a standardized column chart that conforms to geological mapping specifications.

[0114] The integrated geological report output spatially correlates columnar sections with 3D geological body models, embedding spatial coordinate parameters of the 3D model at the stratigraphic stratification locations in the columnar sections. It integrates verification result reports, lithological legends, and hyperlinks to associated model coordinates using a PDF document engine. The output is a comprehensive geological report including text descriptions, 2D charts, and quick access paths to the 3D model, supporting one-click navigation to the corresponding 3D scene view of the geological body.

[0115] Specifically, in the oblique photogrammetry three-dimensional surface model geological modeling system of the present invention, the optimized point set generated by the boundary processing module is filtered for anomalies by a random sampling consensus algorithm, and clean stratigraphic boundaries are output to the geological body modeling module.

[0116] The geological body modeling module completes implicit surface reconstruction based on the clean stratum boundary, generating stratum interface curvature feature data.

[0117] The curvature feature data of the formation interface is fed back to the genetic algorithm optimization stage of the boundary processing module to dynamically adjust the surface fitting weight parameters.

[0118] The clean stratigraphic boundary generation process optimizes the point set output by the boundary processing module and applies a random sampling consensus algorithm for secondary filtering. By iteratively sampling subsets of the point set and evaluating the consistency of the internal point distribution, noise points deviating from the trend line are removed. The fitted plane equation is optimized based on the minimum median square criterion, retaining valid data points within the confidence interval, and outputting a spatially continuous point set that conforms to geological structural patterns, thus forming clean stratigraphic boundary data free of outliers.

[0119] The geological body modeling module, which extracts curvature features from stratigraphic interfaces, receives clean stratigraphic boundaries and generates smooth, continuous stratigraphic interfaces using an implicit surface reconstruction algorithm. Differential geometry algorithms are used to calculate the Gaussian curvature and mean curvature at each vertex of the interface, analyzing the surface's concavity, convexity, and degree of curvature. The distribution characteristics of curvature extrema and the curvature gradient are extracted to quantify the spatial geometric properties of stratigraphic fold morphology and fault inflection points, generating a curvature feature dataset containing a curvature scalar field and principal curvature directions.

[0120] The genetic algorithm dynamically adjusts its parameters, feeding back stratigraphic interface curvature feature data to the genetic algorithm optimization stage of the boundary processing module. It analyzes the geometric constraints within the curvature features and constructs a fitness function to evaluate the surface fitting quality. Based on curvature gradient changes, it dynamically adjusts the selection pressure coefficient of the genetic algorithm, increasing the weight of control points in areas of drastic curvature variation. An adaptive mutation operator optimizes the control point distribution density, ensuring that subsequent stratigraphic boundary fitting prioritizes matching the geological structural features of high-curvature areas.

[0121] Specifically, in the oblique photogrammetry three-dimensional surface model geological modeling system of the present invention, the user drags the stratigraphic boundary node in the two-dimensional profile view of the profile linkage module, triggering a topology reconstruction event captured by the event listener.

[0122] The topology reconstruction event carries node displacement vectors, driving the geological body modeling module to call the sparse voxel hash mapping method to update the three-dimensional stratigraphic interface.

[0123] After real-time correction of the stratigraphic morphology based on the aforementioned profile structural features, the synchronization delay is controlled to the millisecond level.

[0124] In the human-computer interface of the 2D profile view, a node displacement capture mechanism based on mouse event listening is established for topology reconstruction event capture. When the user drags a stratigraphic boundary node, the graphical interface system records the screen coordinate displacement in real time and calculates the 3D spatial displacement vector through the view matrix and projection matrix. The event listener encapsulates the displacement data into a topology reconstruction event object carrying the node ID, displacement direction, and displacement amount, and pushes it to the event processing queue of the geological body modeling module.

[0125] Upon receiving a topology reconstruction event, the sparse voxel hash mapping update geological body modeling module locates the affected local regions of the 3D geological body using a sparse voxel hash table. The displacement vector is decomposed into normal and tangential components, and the vertex positions are adjusted within the voxel block using a Laplacian coordinate offset algorithm. Local triangulation reconstruction is performed only on voxel blocks within the displacement influence radius, maintaining the topology of unmodified areas. A parallel computing pipeline updates the vertex buffer data of the affected voxel blocks, achieving incremental updates of the 3D stratigraphic interface.

[0126] Real-time stratigraphic morphology correction, combined with structural feature data extracted from the profile linkage module, applies geologically sound constraints to the displacement region. Vertex distribution within the neighborhood of moving nodes is optimized through curvature continuity and stratigraphic thickness constraints. A local parametric surface fitting algorithm smoothly transitions the modified boundaries, ensuring the updated 3D stratigraphic interface maintains consistency in geological structural features. Through rendering frame rate control and computational resource scheduling, the system compresses the end-to-end latency for model updates to visualization to millisecond-level response times.

[0127] Please see Figure 1 Secondly, the oblique photogrammetry three-dimensional surface model geological modeling method provided by the present invention, applied to the oblique photogrammetry three-dimensional surface model geological modeling system, includes:

[0128] Step 1: Acquire oblique photogrammetry images and terrain data, and generate a 3D surface model using the structure-reconstruction-motion algorithm;

[0129] Step 2: Receive the three-dimensional surface model, load the model data, and dynamically render the three-dimensional scene based on spatial coordinates and lithological properties;

[0130] Step 3: Receive the visualized scene data from the scene construction module, register the field measurement points with the three-dimensional surface model, and optimize the collected stratigraphic boundaries;

[0131] Step 4: Receive the stratigraphic boundary, generate a continuous stratigraphic interface, and perform topological processing to close the geological body structure;

[0132] Step 5: Receive the closed geological body, cut the closed geological body to generate a cross-sectional view, and synchronize the cross-sectional view and the closed geological body in real time through event-driven synchronization;

[0133] Step 6: Receive the closed geological body and attribute data, store the data and export the standardized geological modeling results.

[0134] This invention establishes a unified 3D surface model base through native fusion processing of oblique photogrammetric imagery and terrain data. The motion reconstruction structure algorithm directly generates textured OSGB format models, replacing traditional manual field surveying data acquisition methods. The boundary processing module integrates an iterative nearest-point algorithm to register measured point clouds, combined with a random sampling consensus algorithm to automatically remove outlier nodes, reducing manual correction operations by over 90%. A genetic algorithm optimizes the surface fitting process, enabling automatic generation and attribute binding of stratigraphic boundaries, thus shortening the geological mapping cycle.

[0135] The data acquisition module embeds a multi-view stereo dense matching and coordinate transformation engine, enabling integrated processing from raw imagery to 3D surface models. It employs a block storage strategy and quadtree index management for model tiles, natively supporting cross-platform calls to OSGB / S3C formats. The scene construction module directly loads the block models and associates them with a lithological attribute library. Through a level-of-detail algorithm, it dynamically schedules model accuracy, skipping format conversion and preprocessing steps required by third-party tools like ArcGIS, maintaining the continuity of the modeling workflow.

[0136] The profile linkage module combines geological body cutting operations with graph neural network prediction to establish a stratigraphic intersection trend analysis model. An event-driven mechanism captures the displacement vectors of 2D profile nodes and locates the affected areas of the 3D model through sparse voxel hash mapping. Incremental triangulation reconstruction is performed only on local voxel blocks, constraining geometric deformation based on profile structural features. Topology update data is fed back to the geological body modeling module in real time, with end-to-end synchronization latency controlled within milliseconds, overcoming the technical bottleneck of separate updates for 2D and 3D models in traditional software.

Claims

1. An oblique photography 3D surface modeling geological modeling system, characterized in that, include: The data acquisition module acquires oblique photogrammetric images and terrain data, and generates a 3D surface model using a motion reconstruction structure algorithm. Specifically, the data acquisition module generates initial point cloud data through dense matching using a multi-view stereo algorithm, calculates pixel-level disparity between adjacent images based on the multi-view stereo matching principle, and optimizes the depth map using a semi-global matching algorithm to generate high-density 3D point cloud data. It then fuses the initial point cloud data, digital elevation model, and work area boundary, using a Kriging space interpolation algorithm to fuse point cloud elevation and terrain data, and constructs a topologically connected triangular mesh using the Poisson surface reconstruction principle. Finally, it outputs a georeferenced OSGB format 3D surface model based on image texture coordinates. The generated 3D surface model is then stored in blocks and its coordinate system is automatically converted. The scene construction module receives the three-dimensional surface model, loads the model data, and dynamically renders the three-dimensional scene based on spatial coordinates and lithological properties. The boundary processing module receives the visualized scene data from the scene construction module, registers the field measurement points with the three-dimensional surface model, and optimizes the collected stratigraphic boundaries. The geological body modeling module receives the stratigraphic boundaries, generates continuous stratigraphic interfaces, and performs topological processing to close the geological body structure. The profile linkage module receives the closed geological body, cuts the closed geological body to generate a profile view, and synchronizes the profile view and the closed geological body in real time through event-driven synchronization. The output module receives the closed geological body and attribute data, stores the data, and exports standardized geological modeling results. Specifically, the data acquisition module outputs a 3D surface model to the scene construction module, the scene construction module outputs visualized scene data to the boundary processing module, the boundary processing module outputs optimized stratigraphic boundaries to the geological body modeling module, the geological body modeling module outputs closed geological bodies to the profile linkage module and the results output module, and the profile linkage module provides real-time feedback on profile modification events to the geological body modeling module to trigger 3D model updates.

2. The oblique photogrammetry three-dimensional surface modeling geological modeling system according to claim 1, characterized in that, The scenario construction module: After loading the three-dimensional surface model, the resolution of the model is switched using a level of detail algorithm; The lithological attribute data is parsed and the lithological codes are mapped to a color space; Based on the mapped lithology-color data distribution, a real-time thermal map of stratigraphic lithology distribution is displayed.

3. The oblique photography three-dimensional surface model geological modeling system according to claim 2, characterized in that, The boundary processing module: Using the aforementioned three-dimensional surface model, calibration point cloud data is generated by registering field measurement points using an iterative nearest point algorithm; For this corrected point cloud data, a random sampling consensus algorithm is applied to remove abnormal nodes and generate an optimized point set; The optimized point set is used as input for surface fitting, and a genetic algorithm is used to optimize and generate stratigraphic boundaries.

4. The oblique photography three-dimensional surface model geological modeling system according to claim 3, characterized in that, The geological body modeling module: The optimized stratigraphic boundary is obtained, and a continuous stratigraphic interface is generated using an implicit surface reconstruction algorithm; Based on the continuous stratigraphic interface, constrained triangulation is used to construct the topological structure of the top and bottom surfaces; After integrating the boundary data of the work area, a closed geological body is generated by stitching the side annulus.

5. The oblique photogrammetry three-dimensional surface modeling geological modeling system according to claim 4, characterized in that, The cross-section linkage module: After receiving the geological body, a graph neural network is invoked to analyze the stratigraphic intersection trend in the cut profile; Based on the stratigraphic intersection trend, an image segmentation algorithm is used to extract the profile structural features; When a profile editing event is detected, the three-dimensional geological body is updated in real time by combining the profile structural features with a sparse voxel hash map.

6. The oblique photogrammetry three-dimensional surface modeling geological modeling system according to claim 5, characterized in that, The output module: After receiving the geological body and attribute data, execute the consistency verification algorithm between the stratigraphic interface and the topographic data and output the verification result; Read the lithology coding library from the attribute data and parse the lithology pattern data; Based on the verification results and the lithological pattern data, a columnar chart is generated; Integrate the bar charts and 3D geological model data to output a geological report with hyperlinks.

7. The oblique photogrammetry three-dimensional surface modeling geological modeling system according to claim 6, characterized in that: After the optimized point set generated by the boundary processing module is filtered for anomalies by the random sampling consensus algorithm, a clean stratum boundary is output to the geological body modeling module. The geological body modeling module completes implicit surface reconstruction based on the clean stratum boundary, generating stratum interface curvature feature data. The curvature feature data of the formation interface is fed back to the genetic algorithm optimization stage of the boundary processing module to dynamically adjust the surface fitting weight parameters.

8. The oblique photogrammetry three-dimensional surface modeling geological modeling system according to claim 5, characterized in that, The user drags a stratigraphic boundary node in the two-dimensional profile view of the profile linkage module, triggering a topology reconstruction event captured by the event listener. The topology reconstruction event carries node displacement vectors, driving the geological body modeling module to call the sparse voxel hash mapping method to update the three-dimensional stratigraphic interface. After real-time correction of the stratigraphic morphology based on the aforementioned profile structural features, the synchronization delay is controlled to the millisecond level.

9. An oblique photogrammetry three-dimensional surface model geological modeling method, applied to the oblique photogrammetry three-dimensional surface model geological modeling system as described in any one of claims 1 to 8, characterized in that, include: Step 1: Acquire oblique photogrammetry images and terrain data, and generate a 3D surface model using the structure-reconstruction-motion algorithm; Step 2: Receive the three-dimensional surface model, load the model data, and dynamically render the three-dimensional scene based on spatial coordinates and lithological properties; Step 3: Receive the visualized scene data from the scene construction module, register the field measurement points with the three-dimensional surface model, and optimize the collected stratigraphic boundaries; Step 4: Receive the stratigraphic boundary, generate a continuous stratigraphic interface, and perform topological processing to close the geological body structure; Step 5: Receive the closed geological body, cut the closed geological body to generate a cross-sectional view, and synchronize the cross-sectional view and the closed geological body in real time through event-driven synchronization; Step 6: Receive the closed geological body and attribute data, store the data and export the standardized geological modeling results.

Images