Method and device for generating variable resolution seafloor terrain grid based on depth datum unification

Abstract

The present disclosure is a kind of based on depth datum unified variable resolution seabed terrain grid generation method and device, the method comprises: data preprocessing;Point density and slope calculation;According to the point density, the function relationship between the point density and the slope correction amount is established, and the slope is corrected;Based on the inverse distance weighted method, the corrected slope is smoothed;According to the slope threshold, the point cloud is filtered;Variable resolution triangular irregular grid is constructed.This embodiment effectively eliminates redundancy and noise by filtering repeated points of multi-source point cloud, ensures the accuracy and consistency of input data;Combined with point density and slope information, the terrain features are corrected and optimized, so that the terrain model can adaptively reflect the detail difference of different regions;Adaptive threshold screening improves the retention rate of key terrain features;Variable resolution TIN construction effectively reduces the overall data volume while maintaining important terrain details, balances modeling efficiency and storage cost.

Description

Technical Field

[0001] This invention relates to the field of seabed bathymetry data processing and terrain modeling technology, and in particular to a method for generating variable-resolution seabed terrain grids based on a unified depth benchmark. Background Technology

[0002] With the continuous development of marine scientific research and marine engineering activities, high-precision digital terrain models (DTMs) are of great significance in waterway surveying, marine resource development, ecological environment monitoring, and national defense security. Current seabed sounding mainly relies on multi-source sounding technologies such as multibeam sonar, single-beam sonar, and lidar. However, different data sources differ in spatial resolution, accuracy, and coverage, and direct fusion can easily generate redundant and noisy points, affecting the quality of subsequent modeling.

[0003] Currently, terrain modeling mainly employs two types of methods: Regular grid method (DEM): This method converts irregular point clouds into regular grid data through interpolation or resampling. It has the advantages of simple data structure and high computational efficiency, but it is prone to smoothing of feature details and loss of edge features when dealing with complex terrain, making it difficult to guarantee the precision of terrain representation.

[0004] Triangular Irregular Mesh (TIN) method: This method generates an irregular terrain surface model by triangulating the point cloud. TIN can preserve terrain details well, and is especially suitable for representing areas with large slope variations. However, when dealing with large amounts of data, constructing a TIN at the original resolution can lead to excessive storage overhead and decreased computational efficiency.

[0005] In contrast, DEM emphasizes regularity but sacrifices feature fidelity, while TIN focuses on accuracy but is inefficient in large-scale applications. Some existing studies attempt to combine point density or feature selection for data compression, but they generally lack systematic modeling of the relationship between point density and slope, making it difficult to balance data compression and feature preservation.

[0006] Therefore, it is necessary to improve one or more of the problems existing in the above-mentioned related technical solutions.

[0007] It should be noted that this section is intended to provide background or context for the technical solutions of this disclosure as set forth in the claims. The description herein does not constitute an admission that it is prior art simply because it is included in this section. Summary of the Invention

[0008] The purpose of this invention is to provide a method and apparatus for generating variable resolution seabed topography meshes based on a unified depth benchmark, thereby overcoming, to at least to some extent, one or more problems caused by the limitations and defects of related technologies.

[0009] This invention first provides a method for generating variable-resolution seabed topography meshes based on a unified depth benchmark, comprising: S1, Data Preprocessing: The experimental area is manually selected from the original multi-source seabed bathymetry point cloud data, and duplicate points are filtered to obtain the experimental point cloud. S2, Point Density and Slope Calculation: Based on the experimental point cloud, calculate the point density and slope of the point cloud; S3, Based on the point density, establish a functional relationship between the point density and the slope correction amount, and correct the slope; S4, Slope Smoothing: Based on the inverse distance weighting method, the corrected slope is smoothed; S5, Filter point cloud based on slope threshold: Based on the set slope threshold, retain points in the experimental point cloud with slope values ​​greater than the threshold, and delete points in the experimental point cloud with slope values ​​less than or equal to the threshold. S6. Based on the selected point cloud, construct a variable resolution triangular irregular mesh.

[0010] In this invention, the method further includes: S7, Accuracy Verification: A comparative regular grid is generated using an interpolation method. The accuracy of the variable resolution triangular irregular grid is verified by comparing the profile elevations of the real point paths and the interpolated point paths and calculating the root mean square error.

[0011] In this invention, in step S1, the repeat point filtering includes: S11, Divide the point cloud data according to a three-dimensional grid and establish a spatial hash index; S12, calculate the arithmetic mean of the elevations of each repeated point group; S13, use the arithmetic mean to generate a new point to replace the entire repeated point group, and finally retain the new point.

[0012] In this invention, the calculation process of point density in S2 includes: S21, a KDTree was constructed based on the experimental point cloud, and the space was recursively partitioned with the median as the dividing point; S22, based on KDTree, implements k-nearest neighbor search, using Euclidean distance as the metric; S23, For each point, define the local point density based on the average distance of its k nearest neighbors, and take a logarithmic transformation on the local point density to obtain the logarithmic density; S24 normalizes the logarithmic density to the [0,1] interval.

[0013] In this invention, the slope calculation process in S2 includes: S201, Delaunay triangulation is performed based on experimental point clouds, and incremental insertion method and edge flipping operation are used to ensure the empty circle property; S202, the local scalar field is calculated using linear interpolation within each triangular piece, and the gradient is solved; S203 defines the gradient of a vertex as the area-weighted average of the gradients of its adjacent triangles. S204 calculates the slope based on the gradient vector and converts the slope angle into degrees.

[0014] In this invention, the functional relationship between point density and slope correction in S3 is as follows:

[0015] in, It is the slope correction amount. This is the original calculated slope value. It is the standardized point density. It is the maximum correction factor, which controls the strength of the correction. .

[0016] In this invention, step S5, the process of filtering point clouds based on a slope threshold, includes: S51, based on the target data compression ratio, statistically sorts the slope values ​​of all points and automatically calculates the corresponding threshold; S52, retain points in the experimental point cloud with a slope greater than the threshold, and delete points in the experimental point cloud with a slope value less than or equal to the threshold.

[0017] The present invention further provides a variable resolution seabed topography mesh generation device based on a unified depth benchmark, comprising: Data preprocessing module: used to manually select experimental areas from raw multi-source seabed bathymetry point cloud data and filter duplicate points to obtain experimental point clouds; Point density and slope calculation module: used to calculate the point density and slope of the point cloud based on the experimental point cloud; The slope correction module is used to establish a functional relationship between the point density and the slope correction amount based on the point density, and to correct the slope. Slope smoothing module: used to smooth the corrected slope based on the inverse distance weighting method; Point cloud filtering module: Used to retain points in the experimental point cloud with a slope value greater than the set slope threshold, and delete points in the experimental point cloud with a slope value less than or equal to the set slope threshold; Variable resolution TIN building block: Used to build a variable resolution triangular irregular mesh based on the filtered point cloud.

[0018] In this invention, the device further includes: Accuracy verification module: Used to generate a comparative regular mesh using interpolation methods, and verify the accuracy of the variable resolution triangular irregular mesh by comparing the profile elevations of the real point paths and the interpolated point paths and calculating the root mean square error.

[0019] The technical solution provided by this invention may include the following beneficial effects: This invention presents a variable-resolution seabed topography mesh generation method based on a unified depth benchmark. By filtering duplicate points from multi-source point clouds, redundancy and noise are effectively eliminated, ensuring the accuracy and consistency of the input data. The method combines point density and slope information to correct and optimize topographic features, enabling the generated topographic model to adaptively reflect the detailed differences in different regions. Adaptive threshold filtering avoids the subjectivity of manual threshold setting, improving the retention rate of key topographic features. Variable-resolution TIN construction effectively reduces the overall data volume while maintaining important topographic details, achieving a balance between modeling efficiency and storage cost. Attached Figure Description

[0020] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0021] Figure 1 This is a flowchart of a variable resolution seabed topography mesh generation method based on a unified depth benchmark in an embodiment of the present invention; Figure 2 This is a flowchart of a variable resolution seabed topography mesh generation method based on a unified depth benchmark in another embodiment of the present invention; Figure 3 This is the software operation interface of the present invention; Figure 4 Point density map of the experimental point cloud; Figure 5 A slope diagram of the experimental point cloud; Figure 6 This is a schematic diagram of variable resolution TIN; Figure 7 A schematic diagram of a DEM with a resolution of 150 meters; Figure 8 This is a diagram of the menu bar; Figure 9 This is a path comparison chart; Figure 10 Elevation comparison chart of TEST1 route; Figure 11 A comparison chart of path differences for TEST1; Figure 12 Elevation comparison chart of TEST2 route; Figure 13 A comparison chart of path differences for TEST2; Figure 14 Elevation comparison chart of TEST3 path; Figure 15 A comparison chart of path differences for TEST3; Figure 16 Elevation comparison chart of TEST4 route; Figure 17 A comparison chart of path differences for TEST4; Figure 18 Elevation comparison chart of TEST5 route; Figure 19 A comparison chart of path differences for TEST5; Figure 20 Elevation comparison chart of TEST6 route; Figure 21 This is a comparison chart of the path differences for TEST6. Detailed Implementation

[0022] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that this disclosure will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0023] Furthermore, the accompanying drawings are merely illustrative diagrams of embodiments of this disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities.

[0024] This example implementation first provides a method for generating variable-resolution seabed topography meshes based on a unified depth benchmark. Please refer to [link / reference]. Figure 1 and Figure 2 This method may include: S1-S6. Specifically: S1, Data Preprocessing: The experimental area is manually selected from the original multi-source seabed bathymetry point cloud data, and duplicate points are filtered to obtain the experimental point cloud. S2, Point Density and Slope Calculation: Based on the experimental point cloud, calculate the point density and slope of the point cloud; S3, Based on the point density, establish a functional relationship between the point density and the slope correction amount, and correct the slope; S4, Slope Smoothing: Based on the inverse distance weighting method, the corrected slope is smoothed; S5, Filter point cloud based on slope threshold: Based on the set slope threshold, retain points in the experimental point cloud with slope values ​​greater than the threshold, and delete points in the experimental point cloud with slope values ​​less than or equal to the threshold. S6. Based on the selected point cloud, construct a variable resolution triangular irregular mesh.

[0025] In this embodiment, redundant points are effectively eliminated by filtering duplicate points from multi-source point clouds, ensuring the accuracy and consistency of the input data. The terrain features are corrected and optimized by combining point density and slope information, enabling the generated terrain model to adaptively reflect the detailed differences in different regions. Adaptive threshold filtering avoids the subjectivity of manual threshold setting and improves the retention rate of key terrain features. Variable resolution TIN construction effectively reduces the overall data volume while maintaining important terrain details, achieving a balance between modeling efficiency and storage cost.

[0026] The specific process of each step in the above embodiments will be described below.

[0027] S1, Data Preprocessing.

[0028] The duplicate point filtering includes: S11, Divide the point cloud data according to a three-dimensional grid and establish a spatial hash index; S12, calculate the arithmetic mean of the elevations of each repeated point group; S13, use the arithmetic mean to generate a new point to replace the entire repeated point group, and finally retain the new point.

[0029] In this embodiment, experimental areas are selected and duplicate points are filtered from multi-source seabed sounding point clouds to generate high-quality, non-redundant input data.

[0030] The user selects a local TXT file as the point cloud input file. This file contains three columns of data: longitude, latitude, and elevation, with each row representing one point. The user selects the experimental area based on the latitude and longitude range of the input data, entering the minimum and maximum values ​​to ensure the number of points in the area is between 1 and 2 million.

[0031] S2, point density and slope calculation.

[0032] The calculation process of point density mainly includes: S21, a KDTree was constructed based on the experimental point cloud, and the space was recursively partitioned with the median as the dividing point; S22, based on KDTree, implements k-nearest neighbor search, using Euclidean distance as the metric; S23, For each point, define the local point density based on the average distance of its k nearest neighbors, and take a logarithmic transformation on the local point density to obtain the logarithmic density; S24 normalizes the logarithmic density to the [0,1] interval.

[0033] Specifically, after constructing the KDTree, the local point density of each point is calculated. This invention employs a local density estimation method based on k-Nearest Neighbor (k-NN), which can accurately reflect the spatial distribution characteristics of the point cloud.

[0034] The point density calculation process is as follows: 1. Nearest neighbor search: For each point Using KDTree to perform k-nearest neighbor search, we obtain the set of its k nearest neighbors. And calculate the corresponding Euclidean distance. : in, For point In the m Dimensional coordinates.

[0035] 2. Calculation of local average distance For each point The average distance of the k nearest neighbors is taken:

[0036] Average distance Able to reflect points The sparseness or density of the surrounding neighborhood; the smaller the distance, the denser the local area.

[0037] 3. Definition of point density Point density Defined as average distance The reciprocal:

[0038] Density is inversely proportional to the average nearest neighbor distance; the smaller the distance, the greater the density, which aligns with the intuitive spatial distribution.

[0039] 4. Numerical transformation and standardization To reduce the impact of extreme values ​​and improve the stability of numerical calculations, a logarithmic transformation is performed on the density to obtain the logarithmic density. as follows:

[0040] Then, the logarithmic density of all points is linearly normalized to the [0,1] interval:

[0041] 5. Algorithm optimization and acceleration strategies: (1) KDTree accelerates search: By utilizing the hierarchical space partitioning and backtracking pruning strategy of KDTree, the efficiency of k-nearest neighbor search is improved, and global brute-force search is avoided.

[0042] (2) Batch processing: For large-scale point clouds, the point set can be divided into batch parallel computing densities, which significantly reduces memory pressure and computing time.

[0043] (3) Neighborhood adaptation: The k value can be dynamically adjusted according to the local point density. A small k value is used for dense areas and a large k value is used for sparse areas to improve the density estimation accuracy.

[0044] (4) Outlier handling: Filter or prune density extreme values ​​to avoid the impact of isolated points or noise on subsequent slope correction and TIN construction.

[0045] By following the steps above, the standardized local point density of each point can be obtained. This serves as the foundational information for subsequent slope correction, key point selection, and variable resolution TIN construction, such as... Figure 4 As shown.

[0046] In S2, the slope calculation process includes: S201, Delaunay triangulation is performed based on experimental point clouds, and incremental insertion method and edge flipping operation are used to ensure the empty circle property; S202, the local scalar field is calculated using linear interpolation within each triangular piece, and the gradient is solved; S203 defines the gradient of a vertex as the area-weighted average of the gradients of its adjacent triangles. S204 calculates the slope based on the gradient vector and converts the slope angle into degrees.

[0047] In this embodiment, after completing the point density calculation, the next step is to construct an initial TIN model based on Delaunay triangulation and calculate the gradient of each triangular patch to obtain a continuous slope field, such as... Figure 5 As shown in the figure, this slope field reflects the local topographic undulations and provides a basis for subsequent key topographic point selection and variable-resolution TIN construction. The specific method is as follows: 1. Construct Delaunay triangulation TIN Discrete terrain sampling point set Convert to a continuous triangular mesh to ensure topological correctness and geometric quality, supporting subsequent interpolation, analysis, and visualization.

[0048] 2. Calculate the gradient of the triangle. Each triangle The vertex elevation is treated as a scalar field and linear interpolation is used: in, Indicates elevation at x The rate of change of direction, i.e. z / x is a constant inside the triangle; It is the rate of change of elevation in the y-direction, that is z / y is a constant inside the triangle; It is the elevation (intercept) of the plane at x=0, y=0. , It is obtained by fitting the coordinates of the vertices of the triangle to the elevation.

[0049] The gradient of the triangle is constant:

[0050] 3. Continuous processing of vertex slope Since the vertices belong to multiple triangles, directly using a single-plane gradient would lead to discontinuities and instability. Therefore, the vertex gradient should be... Defined as the area-weighted average of the gradients of adjacent triangles: in The area of ​​the triangle is represented by the weight, and a larger weight indicates a stronger influence of the triangle on the local geometry.

[0051] 4. Slope Calculation and Unit Conversion Slope angle (in radians):

[0052] Convert to angle system: The above method can accurately reflect the terrain undulation features on discrete TINs and ensure the spatial continuity and robustness of the slope at the vertices, providing a reliable foundation for subsequent point density correction of slope, key point selection and variable resolution TIN construction.

[0053] S3 corrects the slope.

[0054] During terrain data acquisition, due to technical limitations, terrain complexity, and accessibility factors, point cloud data often exhibits an uneven distribution: high-density areas (such as flat areas) may have data redundancy, while low-density areas (such as steep slopes) may lack sufficient sampling points. This unevenness can cause systematic biases in slope calculations, interpolation analysis, and other processes. Low-density areas: Due to the sparse sampling points, the calculated slope value may be underestimated; High-density areas: Oversampling may amplify local noise.

[0055] To address the aforementioned issues, this step proposes a slope correction algorithm based on point density. The core idea is to establish a functional relationship between point density and slope correction, thereby amplifying the slope value in low-density areas and maintaining the original value in high-density areas, thus improving the accuracy and reliability of terrain analysis.

[0056] 1. Establish a slope correction model This application constructs a slope correction model, which adopts a linear correction form, as follows:

[0057] in This is the original calculated slope value. It is the standardized point density. It is the maximum correction factor, controlling the strength of the correction. The physical meaning of the correction factor is clear: when... When = 1 (high density), the correction factor is 1, and there is no correction; when (At low density) the correction factor is Maximum correction; when At (medium density), the correction factor is Moderate correction. Correction factor. The choice of α needs to balance the risks of over-correction and under-correction. An excessively large α may lead to an overemphasis on slope in low-density areas. Too small a value may not effectively improve accuracy in low-density areas. Depending on the terrain type and data characteristics, a value is typically taken as... The empirical value range. This application uses... This is a compromise value obtained from testing various terrain data, which can achieve good correction results in most cases.

[0058] This application analyzes the correction effect from a statistical perspective. The mean of the slope distribution before correction is... The corrected expected slope value is:

[0059] if (Ideal case) The corrected expected slope value is:

[0060] This indicates that the correction algorithm can systematically adjust the slope estimate and improve the estimation accuracy in low-density areas.

[0061] S4, with a smooth slope.

[0062] In TIN-based slope calculations, due to the non-uniformity of point cloud distribution and the presence of local noise, slope values ​​directly derived from gradients often exhibit high-frequency fluctuations or discontinuities. This application aims to improve the smoothness and robustness of the slope field by... Introducing based on the plane The Inverse Distance Weighting (IDW) method spatially smooths the initial slope. Specifically, for each vertex... Query its Let each of the following neighboring points be a vertex and its relationship to the vertex be a vertex. The planar distance is To avoid numerical instability in the denominator when the distance is zero, this application introduces a very small positive number into the distance term. Based on this, we define the weights:

[0063] The smoothed slope is obtained as follows:

[0064] in neighboring points The original slope, This is the smoothed slope. Parameter The choice affects the degree of smoothness: smaller... It can better preserve local details, but its noise resistance is weaker; while larger ones... While it can effectively suppress noise, it may lead to an overly smooth slope and loss of detail. Therefore, It is necessary to weigh and set the parameters based on data density and research needs.

[0065] This local neighborhood-based IDW smoothing method can effectively reduce local discrete noise while maintaining the overall trend of the slope field, thereby improving the stability and interpretability of slope estimation.

[0066] S5, the point cloud is filtered based on the slope threshold. The main process includes: S51, based on the target data compression ratio, statistically sorts the slope values ​​of all points and automatically calculates the corresponding threshold; S52, retain points in the experimental point cloud with a slope greater than the threshold, and delete points in the experimental point cloud with a slope value less than or equal to the threshold.

[0067] Slope thresholding is a crucial step in TIN generation, aiming to preserve important terrain features while reducing data redundancy. This method is based on a key topographic principle: significant terrain changes are typically accompanied by large slope values, while relatively flat areas have smaller slopes. By selectively retaining high-slope points, the main terrain features can be effectively reconstructed with fewer data points.

[0068] To quantify and make the filtering intensity more controllable, this study uses the data compression ratio as the basis for determining the slope threshold. Users can customize a target data compression ratio according to their actual application needs (such as the trade-off between data accuracy and efficiency). The algorithm then statistically sorts the slope of all points according to this ratio and automatically calculates the corresponding slope threshold—that is, the minimum slope value required to retain a specific proportion of high-slope points.

[0069] After confirming the slope threshold, filtering will be performed according to the slope threshold. This application defines the filtering condition as a binary mask function:

[0070] The reason for using absolute values ​​is to take into account the uncertainty of slope direction. Whether uphill or downhill, a large slope indicates an important topographic feature. After filtering, the point set is usually concentrated in areas such as ridgelines, valley lines, cliffs and steep slopes, and topographic turning points. These areas are the most critical parts of topographic analysis.

[0071] S6. Based on the selected point cloud, construct a variable resolution triangular irregular mesh.

[0072] Using the filtered keypoint set to generate a variable-resolution triangular irregular mesh (TIN), efficient and detailed terrain modeling is achieved, such as... Figure 6 As shown.

[0073] Users can click "Generate Variable Resolution TIN" on the software interface to display the results in the visualization area on the right. The menu bar shows the generation time and the number of TIN points, and the point data can be saved.

[0074] Users can input the required DEM resolution through the user interface and generate the corresponding raster DEM, such as... Figure 7 As shown.

[0075] Multiple resolution DEMs can be generated simultaneously, and a specific resolution DEM can be selected to be displayed or deleted in the view. The generation time and number of points for each generated DEM are displayed in the menu bar, and the data can be saved.

[0076] Furthermore, the seabed topography mesh generation method of this application also includes: S7, Accuracy Verification: A comparative regular mesh is generated using an interpolation method. The accuracy of the variable resolution triangular irregular mesh is verified by comparing the profile elevations of the real point paths and the interpolated point paths and calculating the root mean square error (RMSE).

[0077] For further experiments, this application generated variable resolution TIN data with 278,268 points, which is comparable to the number of points in a 50m resolution DEM.

[0078] To simulate the path planning scenario of ships or submarines in actual navigation, this application randomly set six sets of start-end pairs within the study area, generating three real paths and three interpolated paths (see...). Figure 9 ).

[0079] The real point path generation follows these rules: Starting from the starting point, under the constraint of a minimum step size (50 meters), the nearest neighbor point with the smallest angle to the current direction of travel is selected as the next path point. This process is iterated until the destination is reached, thus forming a continuous path.

[0080] The interpolation point path generation follows these rules: starting from the starting point, select points spaced 50 meters apart on a straight line connecting to the ending point to form a continuous path.

[0081] Detailed information on the types, starting and ending coordinates, elevation ranges, number of path points, and total length of the six test paths is shown in Table 1.

[0082] Furthermore, this application extracts the elevation values ​​of each path point through interpolation from the comparison of variable resolution TIN and three different resolution DEMs (50 meters, 100 meters, 150 meters, and 350 meters), and draws a path elevation comparison map and a path elevation difference comparison map. Figures 10-21 This is to intuitively reflect the differences in elevation representation from different data sources. To quantitatively evaluate the accuracy of each model in elevation reconstruction, this application calculates the root mean square error (RMSE) between the elevation values ​​based on TIN and DEM and the actual elevation values ​​of the experimental point cloud on each path. For n observation points, let the actual value be yi and the predicted value be yi, then the RMSE is defined as:

[0083] The smaller the RMSE value, the higher the prediction accuracy of the model, or the closer the image quality is to the original image. The results are shown in Table 2.

[0084] Specifically, the RMSE of the variable resolution TIN remained stable at a low level (1.797 m - 21.572 m), demonstrating its excellent and robust accuracy performance. In contrast, the error of the fixed resolution DEM systematically increased as the resolution decreased: the RMSE range for the 50m DEM was 9.814 m - 49.609 m; for the 100m DEM, it was 15.322 m - 53.687 m; for the 150m DEM, it was 21.733 m - 74.736 m; while the 350m DEM had the largest error, reaching 69.461 m - 217.745 m. On the most challenging TEST1 path, the accuracy of the variable resolution TIN was improved by approximately 63%, 69%, 78%, and 92% compared to the 50m, 100m, 150m, and 350m DEMs, respectively. This series of data fully demonstrates that variable resolution TIN can achieve more accurate depiction of complex terrain by dynamically adjusting the level of detail, while effectively controlling the amount of data, thus solving the problem of insufficient expressive ability of fixed resolution DEM in local terrain undulation areas.

[0085] Table 1. Relevant information for the six paths

[0086] Table 2. RMSE of different data in the six test paths

[0087] This application also provides a variable resolution seabed topography mesh generation device based on a unified depth benchmark, comprising: Data preprocessing module: used to manually select experimental areas from raw multi-source seabed bathymetry point cloud data and filter duplicate points to obtain experimental point clouds; Point density and slope calculation module: used to calculate the point density and slope of the point cloud based on the experimental point cloud; The slope correction module is used to establish a functional relationship between the point density and the slope correction amount based on the point density, and to correct the slope. Slope smoothing module: used to smooth the corrected slope based on the inverse distance weighting method; Point cloud filtering module: Used to retain points in the experimental point cloud with a slope value greater than the set slope threshold, and delete points in the experimental point cloud with a slope value less than or equal to the set slope threshold; Variable resolution TIN building block: used to build a variable resolution triangular irregular mesh based on the filtered point cloud; Accuracy verification module: Used to generate a comparative regular mesh using interpolation methods, and verify the accuracy of the variable resolution triangular irregular mesh by comparing the profile elevations of the real point paths and the interpolated point paths and calculating the root mean square error.

[0088] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0089] In summary, the method of this application has the following beneficial effects: (1) Depth benchmark unification and redundant point removal: By filtering duplicate points of multi-source point clouds, redundancy and noise are effectively eliminated, ensuring the accuracy and consistency of input data.

[0090] (2) Joint modeling of point density and slope: The terrain features are modified and optimized by combining point density and slope information, so that the generated terrain model can adaptively reflect the detailed differences in different regions.

[0091] (3) Adaptive threshold screening: The slope threshold is calculated by back-calculating the data compression ratio, avoiding the subjectivity of manual threshold setting and improving the retention rate of key terrain features.

[0092] (4) Variable resolution TIN construction: While maintaining important terrain details, it effectively reduces the overall data volume and achieves a balance between modeling efficiency and storage cost.

[0093] (5) High-precision verification mechanism: The accuracy of the generated model is verified by DEM interpolation and RMSE evaluation method, which ensures the reliability and engineering applicability of the mesh construction results.

[0094] It should be noted that although several modules of the system for executing actions are mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of the present invention, the features and functions of two or more modules described above can be embodied in one module. Conversely, the features and functions of one module described above can be further divided into multiple modules for embodiment. Components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of the present invention according to actual needs. Those skilled in the art can understand and implement this without any inventive effort.

[0095] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A method for generating variable-resolution seabed topography meshes based on a unified depth benchmark, characterized in that, include: S1, Data Preprocessing: The experimental area is manually selected from the original multi-source seabed bathymetry point cloud data, and duplicate points are filtered to obtain the experimental point cloud. S2, Point Density and Slope Calculation: Based on the experimental point cloud, calculate the point density and slope of the point cloud; S3, Based on the point density, establish a functional relationship between the point density and the slope correction amount, and correct the slope; S4, Slope Smoothing: Based on the inverse distance weighting method, the corrected slope is smoothed; S5, Filter point cloud based on slope threshold: Based on the set slope threshold, retain points in the experimental point cloud with slope values ​​greater than the threshold, and delete points in the experimental point cloud with slope values ​​less than or equal to the threshold. S6. Based on the selected point cloud, construct a variable resolution triangular irregular mesh.

2. The method for generating variable-resolution seabed topography mesh based on a unified depth benchmark according to claim 1, characterized in that, The method further includes: S7, Accuracy Verification: A comparative regular grid is generated using an interpolation method. The accuracy of the variable resolution triangular irregular grid is verified by comparing the profile elevations of the real point paths and the interpolated point paths and calculating the root mean square error.

3. The method for generating variable-resolution seabed topography meshes based on a unified depth benchmark according to claim 1, characterized in that, In S1, the duplicate point filtering includes: S11, Divide the point cloud data according to a three-dimensional grid and establish a spatial hash index; S12, calculate the arithmetic mean of the elevations of each repeated point group; S13, use the arithmetic mean to generate a new point to replace the entire repeated point group, and finally retain the new point.

4. The method for generating variable-resolution seabed topography meshes based on a unified depth benchmark according to claim 1, characterized in that, In S2, the calculation process of point density includes: S21, a KDTree was constructed based on the experimental point cloud, and the space was recursively partitioned with the median as the dividing point; S22, based on KDTree, implements k-nearest neighbor search, using Euclidean distance as the metric; S23, For each point, define the local point density based on the average distance of its k nearest neighbors, and take a logarithmic transformation on the local point density to obtain the logarithmic density; S24 normalizes the logarithmic density to the [0,1] interval.

5. The method for generating variable-resolution seabed topography meshes based on a unified depth benchmark according to claim 1, characterized in that, In S2, the slope calculation process includes: S201, Delaunay triangulation is performed based on experimental point clouds, and incremental insertion method and edge flipping operation are used to ensure the empty circle property; S202, the local scalar field is calculated using linear interpolation within each triangular piece, and the gradient is solved; S203 defines the gradient of a vertex as the area-weighted average of the gradients of its adjacent triangles. S204 calculates the slope based on the gradient vector and converts the slope angle into degrees.

6. The method for generating variable-resolution seabed topography meshes based on a unified depth benchmark according to claim 1, characterized in that, In S3, the functional relationship between point density and slope correction is as follows: in, It is the slope correction amount. This is the original calculated slope value. It is the standardized point density. It is the maximum correction factor, which controls the strength of the correction. .

7. The method for generating variable-resolution seabed topography meshes based on a unified depth benchmark according to claim 1, characterized in that, In S5, the process of filtering point clouds based on slope thresholds includes: S51, based on the target data compression ratio, statistically sorts the slope values ​​of all points and automatically calculates the corresponding threshold; S52, retain points in the experimental point cloud with a slope greater than the threshold, and delete points in the experimental point cloud with a slope value less than or equal to the threshold.

8. A variable-resolution seabed topography mesh generation device based on a unified depth benchmark, characterized in that, include: Data preprocessing module: used to manually select experimental areas from raw multi-source seabed bathymetry point cloud data and filter duplicate points to obtain experimental point clouds; Point density and slope calculation module: used to calculate the point density and slope of the point cloud based on the experimental point cloud; The slope correction module is used to establish a functional relationship between the point density and the slope correction amount based on the point density, and to correct the slope. Slope smoothing module: used to smooth the corrected slope based on the inverse distance weighting method; Point cloud filtering module: Used to retain points in the experimental point cloud with a slope value greater than the set slope threshold, and delete points in the experimental point cloud with a slope value less than or equal to the set slope threshold; Variable resolution TIN building block: Used to build a variable resolution triangular irregular mesh based on the filtered point cloud.

9. The apparatus for generating variable-resolution seabed topography meshes based on a unified depth benchmark according to claim 8, characterized in that, The device further includes: Accuracy verification module: Used to generate a comparative regular mesh using interpolation methods, and verify the accuracy of the variable resolution triangular irregular mesh by comparing the profile elevations of the real point paths and the interpolated point paths and calculating the root mean square error.

Images