Three-dimensional visualization method and system based on deformation data thinning and risk fusion

Abstract

The invention discloses a three-dimensional visualization method and system based on deformation data thinning and risk fusion, and relates to the crossing field of a remote sensing technology, synthetic aperture radar interferometry and a geographic information system. The method comprises the following steps: acquiring a deformation rate grid map and a coherence coefficient grid map formed by SBAS-InSAR processing, and setting a threshold value by using a coherence coefficient; the method comprises the following steps: calculating a deformation rate gradient through spatial convolution, setting a deformation gradient threshold, then dividing a whole monitoring area into a plurality of subareas in combination with a coherence coefficient threshold, adopting different thinning strategies based on the subareas, generating a non-uniformly distributed vector discrete monitoring point data set, and then generating a three-dimensional monitoring point. And then, integrating geological environment factors for each three-dimensional monitoring point, calculating a comprehensive risk index, importing the comprehensive risk index into a three-dimensional platform, and carrying out visualization according to the comprehensive risk index. The system is realized based on the method. According to the method, the enhanced expression of the essential characteristics of the deformation field and the visual display of the risk are realized.

Description

Technical Field

[0001] This invention relates to the intersection of remote sensing technology, synthetic aperture radar interferometry (InSAR), and geographic information systems (GIS), and specifically discloses a three-dimensional visualization method and system based on deformation data thinning and risk fusion. Background Technology

[0002] Small Baseline Set Interferometric Radar (SBAS-InSAR) technology is an important means of monitoring wide-area surface deformation. Its standard output—the deformation rate raster map—is massive in data (often containing tens of millions of pixels), posing a significant challenge to direct display in 3D scenes. Currently, there are two main approaches: one is to overlay the entire raster map as a texture map onto the 3D terrain, resulting in low rendering efficiency and a clunky fit between deformation information and terrain; the other is to convert all raster pixels into vector points, which solves the terrain fit problem but generates massive, uniformly distributed redundant data, leading to slow loading and lag in 3D platforms.

[0003] More importantly, current technologies only visualize deformation. Uniformly distributed point clouds fail to highlight the core features of the deformation field (such as the edge of a settlement funnel, landslide shear outlet, and potential sliding surface), which are precisely the focus of engineering geological analysis. Furthermore, the magnitude of deformation does not entirely equate to the level of risk. A small deformation on a steep slope may pose a far greater risk than an equivalent deformation in a plain. Existing methods cannot integrate deformation with factors such as slope and geological structure, failing to provide professionals with intuitive risk assessment criteria and still requiring secondary interpretation based on human experience, resulting in low levels of automation and intelligence. Therefore, there is an urgent need in this field for a three-dimensional visualization method that can intelligently extract deformation features, deeply integrate geological environmental information, and directly output risk insights. Summary of the Invention

[0004] The purpose of this invention is to provide a three-dimensional visualization method system based on deformation data thinning and risk fusion, so as to overcome the shortcomings of existing three-dimensional visualization methods in terms of efficiency, expressiveness and intelligence.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A 3D visualization method based on deformation data thinning and risk fusion includes the following steps: S1. Obtain the deformation rate raster and coherence coefficient raster generated by SBAS-InSAR processing, and set a threshold using the coherence coefficient; S2. Calculate the deformation rate gradient through spatial convolution, synthesize a deformation gradient modulus grid that reflects the severity of spatial deformation changes, set a deformation gradient threshold, and then combine it with the coherence coefficient threshold set in S1 to divide the entire monitoring area into several zones. S3. Based on the partitioning results in S2, perform thinning to generate a non-uniformly distributed two-dimensional vector discrete monitoring point dataset; S4. 3D Integration and Risk Fusion: The two-dimensional vector points generated in S3 are integrated with the digital elevation model, and each point is given a real elevation coordinate to generate a three-dimensional monitoring point. Then, geological environmental factors are integrated into each three-dimensional monitoring point, and the comprehensive risk index R is calculated through a preset risk assessment model. S5. Import the 3D monitoring point dataset with the comprehensive risk index into the 3D platform and visualize it based on the comprehensive risk index.

[0006] Furthermore, step S1 also includes: performing masking processing on the deformation rate grid to remove regions with low solution reliability.

[0007] Furthermore, in step S2, the deformation rate gradient in the corresponding directions is calculated according to the east-west and north-south directions.

[0008] Further, in step S2, two deformation gradient thresholds T_high and T_low are specifically set, and a coherence coefficient threshold C_min is set. T_high is taken as the 80th-95th percentile of the deformation gradient modulus value, T_low is taken as the 40th-60th percentile, and the coherence coefficient threshold C_min is taken as 0.3-0.4. The region where the deformation gradient is greater than T_high and the coherence is greater than C_min is divided into the deformation rapidly changing region, the region where the deformation rate gradient is between T_high and T_low and the coherence is greater than C_min is divided into the deformation gradually changing region, and the region where the deformation gradient is less than or equal to T_low or the coherence is less than or equal to C_min is divided into the stable region.

[0009] Furthermore, in step S3, an adaptive density thinning strategy is adopted to thin the deformation rate values ​​under the three partition masks according to different thinning distances. The thinning distance in the region of severe deformation is less than that in the region of gradual deformation, and the thinning distance in the region of gradual deformation is less than that in the region of stable deformation.

[0010] Furthermore, the geological environmental factors integrated at each three-dimensional monitoring point in step S4 include slope, lithology, and distance from the fault.

[0011] Furthermore, in step S5, the monitoring results of different risk areas are visualized by setting color mapping, symbol size, and dynamic flashing effects.

[0012] A system for implementing the above-mentioned 3D visualization method based on deformation data thinning and risk fusion includes: The data input and preprocessing module includes a data reading unit, a coordinate system unit, and a data validity filtering unit. The data reading unit reads the raster data file and parses its georeferenced information. The coordinate system unit converts all input data to a unified spatial coordinate system. The data validity filtering unit performs masking processing on the deformation rate raster based on the coherence coefficient threshold to remove unreliable data in low coherence areas. The feature recognition and partitioning module includes a gradient calculation unit and an intelligent partitioning unit. The gradient calculation unit calculates the gradient modulus of the deformation rate grid through a spatial convolution algorithm. The intelligent partitioning unit divides the monitoring area into a rapidly deforming zone, a slowly deforming zone, and a stable zone based on the deformation gradient threshold and the coherence coefficient threshold. The adaptive thinning processing module includes a thinning strategy configuration unit and a partition thinning execution unit. The thinning strategy configuration unit is used to set the thinning parameters corresponding to different partitions. The partition thinning execution unit executes the corresponding thinning algorithm on each partition and merges them to generate a two-dimensional vector discrete monitoring point dataset. The risk fusion calculation module includes a three-dimensional unit, a geological factor integration unit, and a risk calculation unit. The three-dimensional unit integrates two-dimensional monitoring points with the digital elevation model to provide elevation information. The geological factor integration unit integrates relevant geological environmental factors for each monitoring point. The risk calculation unit calculates the comprehensive risk index of each monitoring point through a risk assessment model. The 3D visualization rendering module includes a style configuration unit and a scene publishing unit. The style configuration unit configures visualization symbols and dynamic effects according to the risk level, while the scene publishing unit converts the data into a format supported by the 3D platform and loads and displays it on the 3D globe platform to achieve risk-oriented visualization.

[0013] Furthermore, it also includes: The data management module includes a database storage unit and a metadata management unit. The database storage unit is used to store various types of raster and vector data in a structured manner, while the metadata management unit is used to record the source, processing time, and parameter information of the data. The user interaction module includes a parameter configuration interface and a spatial query unit. The parameter configuration interface receives threshold parameters and weight coefficients input by the user, and the spatial query unit queries the attribute information of the monitoring points.

[0014] The beneficial effects of this invention are as follows: 1. Intelligent compression and feature preservation: By feature partitioning based on deformation gradient and adaptive thinning, the data volume is significantly compressed without losing key deformation information, which fundamentally solves the efficiency bottleneck of massive InSAR data in 3D rendering, while highlighting the core features of concern to engineering geology.

[0015] 2. From Deformation to Risk: By integrating multi-source geological environmental data and calculating a comprehensive risk index, the visualization content is upgraded from a simple "deformation field" to a "risk field," which greatly enhances the interpretive depth and application value of InSAR monitoring results and directly serves the early identification and warning of geological disasters.

[0016] 3. Automation and intelligence: The entire process, from deformation feature partitioning to risk calculation, can be completed automatically by algorithms, reducing reliance on human experience, improving the objectivity and efficiency of analysis results, and facilitating business-oriented operation.

[0017] 4. The presentation effect is intuitive and profound: Risk-oriented visualization allows analysts to quickly identify high-risk target areas without having to perform complex mental overlay analysis, which significantly improves the efficiency and effectiveness of results reporting and decision support. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the method of the present invention.

[0019] Figure 2 This is an architecture diagram of the system of the present invention. Detailed Implementation

[0020] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0021] like Figure 1 As shown, the present invention provides a three-dimensional visualization method based on deformation data thinning and risk fusion, including steps S1 to S5. S1: Data input and preprocessing. Obtain the deformation rate raster and coherence coefficient raster generated by SBAS-InSAR processing. Use the coherence coefficient to set a threshold and perform masking on the deformation rate raster to remove areas with low solution reliability.

[0022] Specifically, 46 Sentinel-1 images covering a mountain slope area over three years were acquired. These were then processed using SBAS-InSAR to obtain an annual average deformation rate raster (vel.tif, 30m pixel size) and a coherence coefficient raster (coh.tif). The coherence coefficient threshold C_min was set to 0.3-0.4; here, a value of 0.3 was used to mask the deformation rate raster, retaining only reliable data.

[0023] S2: Deformation field feature partitioning. First, the deformation rate gradient in the east-west and north-south directions is calculated using spatial convolution (such as the Sobel operator), and then a deformation gradient modulus grid reflecting the severity of spatial deformation changes is synthesized. Then, combining the coherence coefficient, two deformation gradient thresholds (T_high, T_low) and one coherence coefficient threshold (C_min) are set to scientifically and quantitatively divide the entire monitoring area into three distinct regions: Zones of dramatic deformation: Areas with large gradients and good coherence, usually corresponding to deformation boundaries, fault zones, or stress concentration zones, are sensitive areas with potential disasters.

[0024] Gradual deformation region: A region with moderate gradient and good coherence, representing the main deformation region.

[0025] Stable region: A region with small gradient or poor coherence, representing a region with no obvious deformation or unreliable observation.

[0026] This partitioning provides a logical basis for subsequent non-uniform processing.

[0027] Specifically, the GDAL library or ArcPy script is used to perform Sobel filtering on vel.tif and calculate and generate the deformation gradient modulus raster grad.tif.

[0028] Calculate the statistical values ​​of grad.tif. T_high is taken as the 80th-95th percentile of the deformation gradient modulus value, and T_low is taken as the 40th-60th percentile. Here, T_high is set as the 90th percentile of the deformation gradient and T_low is set as the 50th percentile of the deformation gradient.

[0029] Perform map algebra operations to partition the map: Zone_abrupt (region of severe deformation) = grad.tif > T_high and coh.tif > C_min Zone_gentle (gradient deformation zone) = T_low < grad.tif ≤ T_high and coh.tif > C_min Zone_stable (stable zone) = grad.tif ≤ T_low or coh.tif ≤ C_min, outputting binary mask files for the three partitions.

[0030] S3: Adaptive density thinning. Based on the partitioning results of S2, different thinning strategies are adopted to generate a non-uniformly distributed vector discrete monitoring point dataset. According to the Nyquist sampling theorem and spatial statistics principles in information theory and signal processing, in the drastic region, the deformation gradient is large and the changes are drastic, with many high-frequency signal components. In order not to lose key features, a higher sampling density must be used. Therefore, D1 is set to 0.5-2 times the pixel size. In the gradual deformation region, the deformation changes are relatively gentle, which is equivalent to a low-to-medium frequency signal. Appropriately reducing the sampling rate without losing the main deformation trend, and a thinning distance of 3-5 times can effectively reduce the amount of data while preserving the overall shape. The stable region exhibits minimal deformation and slow changes, equivalent to extremely low-frequency signals or DC components. Data can be compressed to the maximum extent by using a large-distance thinning method of 8-15 times because the information entropy in these regions is very low.

[0031] Specifically, the rasterio and geopandas libraries in Python are used to thin out the deformation rate values ​​under the three partition masks. Specifically, the thinning distance is D1 = 30 meters (1 pixel size) for Zone_abrupt (severe deformation zone); D2 = 90 meters (3 pixel size) for Zone_gentle (gradual deformation zone); and D3 = 300 meters (10 pixel size) for Zone_stable (stable zone).

[0032] The point datasets obtained after thinning the three partitions are merged into a single two-dimensional vector point file (GeoJSON format).

[0033] S4: 3D Integration and Risk Assessment. The aforementioned two-dimensional vector points are integrated with the Digital Elevation Model (DEM), assigning each point its true elevation coordinates to generate a three-dimensional monitoring point. Subsequently, another core innovation of this invention—risk integration—is implemented: multiple geological environmental factors (such as slope, lithology, and distance from faults) are integrated into each three-dimensional monitoring point, and a comprehensive risk index R is calculated using a pre-set, configurable risk assessment model (such as a multi-factor weighted superposition model). This elevates each monitoring point from a simple deformation observation value into an intelligent risk assessment unit integrating location, deformation, topography, and geological information.

[0034] Specifically, a 30-meter resolution SRTM DEM is used here to extract and add elevation attributes (Z value) to each two-dimensional monitoring point.

[0035] The following geological environmental factors are integrated for each point: Slope: The slope value (in degrees) calculated from the DEM.

[0036] Distance_to_fault: Distance (in meters) to the nearest fault obtained from the regional geological map.

[0037] Construct the risk assessment model: R = 0.6 * (|deformation rate| / Max|deformation rate|) + 0.3 *(Slope / MaxSlope) + 0.1 * (1 - Distance_to_fault / MaxDistance). All factors have been normalized.

[0038] Iterate through each monitoring point, calculate its comprehensive risk index R, and store it as a new attribute field in the GeoJSON file.

[0039] S5: Risk-oriented 3D visualization imports 3D monitoring point datasets with risk indices into 3D platforms such as Cesium and Google Earth. During visualization, the rendering basis shifts from "deformation rate" to "comprehensive risk index." By setting color mapping (e.g., red-yellow-green to represent high-medium-low risk), symbol size, and adding dynamic flashing effects to high-risk points, high-risk areas are automatically and prominently highlighted in the 3D scene, achieving intelligent visualization and early warning of risk areas in monitoring results.

[0040] Convert the final GeoJSON file to a 3D Tiles format supported by Cesium or Google Earth, or KML format.

[0041] Create a scene in Cesium or Google Earth and configure the visualization style: 0.0 ≤ R<0.3: Green dot; 0.3 ≤ R < 0.7: Yellow dot; R ≥ 0.7: Red dot, the pixel value of the dot increases with the R value, and a periodic scaling (pulse) animation effect is added.

[0042] A web service is published. Users can access it through a browser and see clearly that the risk areas on the slope, marked by red and yellow dots, are highly consistent with the hazard areas identified in the on-site investigation, thus achieving rapid and accurate risk location.

[0043] like Figure 2 As shown, the present invention also provides a system for implementing the above method, the system comprising: 1. Data Input and Preprocessing Module Function: Responsible for receiving, parsing, and cleaning the input SBAS-InSAR deformation rate raster and coherence coefficient raster; Specific components: The data reading unit is used to read raster data files and parse their georeferenced information. The coordinate system unit is used to transform all input data to a unified spatial coordinate system. The data validity filtering unit is used to perform masking processing on the deformation rate raster based on the coherence coefficient threshold, and to remove unreliable data in low coherence regions.

[0044] 2. Feature Recognition and Partitioning Module Function: Used to calculate the deformation gradient modulus and intelligently partition the monitoring area according to deformation characteristics and coherence; Specific components: The gradient calculation unit is used to calculate the gradient modulus of the deformation rate grid using a spatial convolution algorithm. The intelligent partitioning unit is used to divide the monitoring area into a rapidly changing deformation zone, a gradually changing deformation zone, and a stable zone based on the deformation gradient threshold and the coherence coefficient threshold.

[0045] 3. Adaptive thinning processing module Function: Used to generate non-uniformly distributed vector monitoring points by applying differentiated thinning strategies to different feature partitions; Specific components: The thinning strategy configuration unit is used to set the thinning parameters for different partitions; The partition thinning execution unit is used to execute the corresponding thinning algorithm on each partition and merge them to generate a two-dimensional vector discrete monitoring point dataset.

[0046] 4. Risk Fusion Calculation Module Function: Used to transform two-dimensional monitoring points into three dimensions and integrate geological environmental factors to calculate a comprehensive risk index; Specific components: Three-dimensional units are used to integrate two-dimensional monitoring points with digital elevation models, giving them elevation information; The geological factor integration unit is used to integrate relevant geological and environmental factors for each monitoring point; The risk calculation unit is used to calculate the comprehensive risk index for each monitoring point through a risk assessment model.

[0047] 5. 3D Visualization Rendering Module Function: Used to achieve risk-oriented visualization on a 3D globe platform; Specific components: The style configuration unit is used to configure visual symbols and dynamic effects according to the risk level; The scene publishing unit is used to convert data into a format supported by the 3D platform and load it for display.

[0048] The system described in this invention can be deployed in any computing device that includes memory, a processor, and a communication interface. The memory stores computer program instructions that perform the functions of the modules described above; the processor executes the instructions to implement the method described in this invention. This system can be a standalone desktop application or a distributed web application based on a client-server architecture.

[0049] 6. Data Management Module Function: Used to manage raw data, intermediate results, and final output data; Specific components: Database storage units are used for structured storage of various types of raster and vector data; The metadata management unit is used to record the source, processing time, and parameter information of data.

[0050] 7. User Interaction Module Functions: Provides a human-computer interaction interface, supporting parameter configuration and result query; Specific components: The parameter configuration interface is used to receive user input of threshold parameters and weight coefficients; The spatial query unit allows users to query monitoring point attribute information by clicking, selecting boxes, etc.

[0051] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of variations or substitutions within the technical scope disclosed in the present invention, and reasonable transformations of the specific values ​​in this specification should be covered within the scope of protection of the present invention.

Claims

1. A three-dimensional visualization method based on deformation data thinning and risk fusion, characterized in that, Includes the following steps: S1. Obtain the deformation rate raster and coherence coefficient raster generated by SBAS-InSAR processing, and set a threshold using the coherence coefficient; S2. Calculate the deformation rate gradient through spatial convolution, synthesize a deformation gradient modulus grid that reflects the severity of spatial deformation changes, set a deformation gradient threshold, and then combine it with the coherence coefficient threshold set in S1 to divide the entire monitoring area into several zones. S3. Based on the partitioning results in S2, perform thinning to generate a non-uniformly distributed two-dimensional vector discrete monitoring point dataset; S4. 3D Integration and Risk Fusion: The two-dimensional vector points generated in S3 are integrated with the digital elevation model, and each point is given a real elevation coordinate to generate a three-dimensional monitoring point. Then, geological environmental factors are integrated into each three-dimensional monitoring point, and the comprehensive risk index R is calculated through a preset risk assessment model. S5. Import the 3D monitoring point dataset with the comprehensive risk index into the 3D platform and visualize it based on the comprehensive risk index.

2. The three-dimensional visualization method based on deformation data thinning and risk fusion according to claim 1, characterized in that, Step S1 further includes: performing masking on the deformation rate grid to remove regions with low solution reliability.

3. The three-dimensional visualization method based on deformation data thinning and risk fusion according to claim 1, characterized in that: In step S2, the deformation rate gradient is calculated in the corresponding directions according to the east-west and north-south directions.

4. The three-dimensional visualization method based on deformation data thinning and risk fusion according to claim 1, characterized in that: In step S2, two deformation gradient thresholds T_high and T_low are specifically set, and a coherence coefficient threshold C_min is set. T_high is taken as the 80th-95th percentile of the deformation gradient modulus value, T_low is taken as the 40th-60th percentile, and the coherence coefficient threshold C_min is taken as 0.3-0.

4. The region with deformation gradient greater than T_high and coherence greater than C_min is divided into the deformation rapidly changing region, the region with deformation rate gradient between T_high and T_low and coherence greater than C_min is divided into the deformation gradually changing region, and the region with deformation gradient less than or equal to T_low or coherence less than or equal to C_min is divided into the stable region.

5. The three-dimensional visualization method based on deformation data thinning and risk fusion according to claim 4, characterized in that: In step S3, an adaptive density thinning strategy is adopted to thin the deformation rate values ​​under the three partition masks according to different thinning distances. The thinning distance of the severely deformed region is less than that of the slowly deformed region, and the thinning distance of the slowly deformed region is less than that of the stable region.

6. The three-dimensional visualization method based on deformation data thinning and risk fusion according to claim 1, characterized in that: In step S4, the geological environmental factors integrated for each three-dimensional monitoring point include slope, lithology, and distance from the fault.

7. The three-dimensional visualization method based on deformation data thinning and risk fusion according to claim 1, characterized in that: The risk assessment model preset in step S4 is a multi-factor weighted superposition model.

8. The three-dimensional visualization method based on deformation data thinning and risk fusion according to claim 1, characterized in that: In step S5, the monitoring results of different risk areas are visualized by setting color mapping, symbol size, and dynamic flashing effects.

9. A system for implementing the three-dimensional visualization method based on deformation data thinning and risk fusion as described in any one of claims 1-8, characterized in that, include: The data input and preprocessing module includes a data reading unit, a coordinate system unit, and a data validity filtering unit. The data reading unit reads the raster data file and parses its georeferenced information. The coordinate system unit converts all input data to a unified spatial coordinate system. The data validity filtering unit performs masking processing on the deformation rate raster based on the coherence coefficient threshold to remove unreliable data in low coherence areas. The feature recognition and partitioning module includes a gradient calculation unit and an intelligent partitioning unit. The gradient calculation unit calculates the gradient modulus of the deformation rate grid through a spatial convolution algorithm. The intelligent partitioning unit divides the monitoring area into a rapidly deforming zone, a slowly deforming zone, and a stable zone based on the deformation gradient threshold and the coherence coefficient threshold. The adaptive thinning processing module includes a thinning strategy configuration unit and a partition thinning execution unit. The thinning strategy configuration unit is used to set the thinning parameters corresponding to different partitions. The partition thinning execution unit executes the corresponding thinning algorithm on each partition and merges them to generate a two-dimensional vector discrete monitoring point dataset. The risk fusion calculation module includes a three-dimensional unit, a geological factor integration unit, and a risk calculation unit. The three-dimensional unit integrates two-dimensional monitoring points with the digital elevation model to provide elevation information. The geological factor integration unit integrates relevant geological environmental factors for each monitoring point. The risk calculation unit calculates the comprehensive risk index of each monitoring point through a risk assessment model. The 3D visualization rendering module includes a style configuration unit and a scene publishing unit. The style configuration unit configures visualization symbols and dynamic effects according to the risk level, while the scene publishing unit converts the data into a format supported by the 3D platform and loads and displays it on the 3D globe platform to achieve risk-oriented visualization.

10. The system according to claim 9, characterized in that, Also includes: The data management module includes a database storage unit and a metadata management unit. The database storage unit is used to store various types of raster and vector data in a structured manner, while the metadata management unit is used to record the source, processing time, and parameter information of the data. The user interaction module includes a parameter configuration interface and a spatial query unit. The parameter configuration interface receives threshold parameters and weight coefficients input by the user, and the spatial query unit queries the attribute information of the monitoring points.

Images