A method and system for extracting river centerline from vector surface

Abstract

The present invention belongs to the field of water resources management, and specifically relates to a vector surface river centerline extraction method and system. The steps of the vector surface river centerline extraction method are as follows: acquisition of river boundary line data, input and preprocessing of river boundary line data, boundary tracking, medial axis transformation, tree branch selection, centerline optimization and result output. The system is composed of the following modules: river boundary line input and preprocessing module, boundary tracking module, medial axis transformation module, tree branch selection module, centerline optimization module and result output module. The present invention realizes efficient and accurate extraction of the centerline of a complex river system through a series of innovative technical steps, improves the quality and processing efficiency of data, and provides more reliable and efficient technical support for river analysis and management.

Description

Technical Field

[0001] The invention belongs to the field of water resource management, and in particular relates to a vector surface river centerline extraction method and system. Background Art

[0002] Rivers are a type of surface topographic factor with extremely high socio-economic and environmental protection value, and they have functions such as flood discharge, water storage for irrigation, hydroelectric power generation, transportation, fishery and ecological regulation. The analysis of river topographic characteristics is a key part of the geographic information system. By extracting the centerline of the river, we can not only simplify the map data, but also obtain the distribution information of the river system in a more targeted manner, providing basic data for applications in related professional fields.

[0003] In the existing technology, the methods for extracting river centerlines are mostly based on remote sensing images or digital elevation models (DEM). These methods include but are not limited to morphological processing, watershed algorithm and flow accumulation method. However, these methods have certain limitations, which are specifically manifested in the following aspects: (1) Low data processing accuracy: When processing input river boundary line data, traditional methods often have low pre-processing accuracy due to factors such as excessive noise and data impurities, which affects the accuracy of subsequent centerline extraction; (2) Inconsistent boundary tracking: Existing boundary tracking technology is difficult to handle complex river morphology, and it is easy to have discontinuous or inaccurate boundary lines, which affects the effect of central axis transformation; (3) Low efficiency of central axis transformation algorithm: Many central axis transformation algorithms are inefficient when dealing with large-scale and complex river systems. The extracted central axis does not match the actual river centerline and has low accuracy; (4) The central axis is not consistent with the actual river centerline. (5) Poor centerline optimization: The initially extracted centerline often contains redundant points and breakpoints. Traditional methods are not good at removing redundant points, smoothing, and connecting breakpoints. The quality of the final output centerline data is not high. (6) The output of results is not standardized: The centerline data output by many extraction methods is not in a unified format, which makes it difficult to use them directly for subsequent analysis and application, and adds additional data processing and conversion work.

[0004] In order to solve the above problems, the present invention proposes a vector surface river centerline extraction method and system, which aims to achieve efficient and accurate extraction of the centerline of complex river systems by improving the algorithm steps and optimizing the calculation process, improve the data quality and processing efficiency, and provide more reliable and efficient technical support for river analysis and management. Summary of the invention

[0005] The purpose of the present invention is to provide a method for extracting river center lines from a vector surface.

[0006] Another object of the present invention is to provide a vector surface river centerline extraction system for executing the above-mentioned extraction method.

[0007] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0008] The vector surface river centerline extraction method of the present invention comprises the following steps (the overall framework of the river centerline extraction method is shown in Figure 1 ):

[0009] Step 1: Acquisition of river boundary data

[0010] Collect remote sensing image data of the target river and digital elevation model data of the area, identify the river area after preprocessing, extract the river contour and edit and correct it to obtain the river boundary line data.

[0011] Step 2: River boundary data input and preprocessing

[0012] The river boundary line data are input into the river centerline extraction system, and the noise and irregular boundaries of the boundary line data are preprocessed.

[0013] Step 3: Boundary tracing (see the detailed process of boundary tracing Figure 2 )

[0014] After completing data preprocessing, the boundary tracing algorithm is used to identify the outer boundary of the river vector surface.

[0015] Step 4: Medial Axis Transformation

[0016] After completing the boundary tracing, the centerline of the river is extracted using the following medial axis transformation algorithm:

[0017] S1 initializes the distance field: creates a distance field on the river vector surface, where each point in the distance field represents the Euclidean distance to the nearest boundary;

[0018] S2 iteratively updates the distance field: iteratively updates the distance field by comparing the current distance value of each point with the distance value to the newly detected boundary point, and updates the distance value to the minimum distance to all boundary points;

[0019] S3 Extract the central axis: After the distance field is stabilized, the internal points with the maximum distance value are identified and connected to form the centerline of the river;

[0020] S4 centerline optimization: smooth the extracted centerline and remove isolated points and discontinuous points;

[0021] S5 Verification and adjustment: The extracted central axis is verified and adjusted by comparing with actual observation data or any one or more high-precision data sources such as high-resolution remote sensing images, lidar data and geographic information system data.

[0022] Step 5: Tree branch selection

[0023] The tree branch selection is performed according to the following tree branch selection algorithm:

[0024] S1 manually adds river sources and river sinks, and uses a topology-based pruning algorithm to retain the required trunks and branches;

[0025] S2 is automatically extracted according to parameter settings;

[0026] In the current solution, tree branch optimization selection is implemented through a specific algorithm. Alternative solutions can be considered:

[0027] (1) Machine learning optimization selection: Based on existing training data, develop and train machine learning models to automatically identify river branch points and paths. When processing new data, apply the trained model to automatically select and optimize river paths, thereby improving analysis efficiency and accuracy;

[0028] (2) Graph theory topology method: Establish a topological structure model of the river system, and regard the branches of the river as nodes and edges in the graph. By applying graph theory algorithms (such as the shortest path algorithm and the minimum spanning tree algorithm), analyze and optimize the river branches to achieve the optimal structure of the overall system;

[0029] (3) Artificial intelligence-assisted pruning: Design artificial intelligence algorithms (e.g., based on decision trees or neural networks) to identify and remove redundant or irrelevant branches in river systems. By continuously evaluating and adjusting pruning strategies, we can ensure that unnecessary complexity is removed while maintaining the integrity of the river system.

[0030] Step 6: Centerline optimization (see the centerline optimization process Figure 3 )

[0031] The extracted river centerline is optimized by removing redundant points, smoothing the centerline, connecting breakpoints, adjusting curvature, checking topological consistency, matching flow, fusing and correcting multi-source data, and dynamically adjusting the centerline according to seasonal changes or long-term changes in the river.

[0032] Ways to remove redundant points and connect breakpoints include but are not limited to:

[0033] (1) Curve fitting connection: Use curve fitting technology to connect breakpoints, and generate smooth curves through mathematical models to replace linear connection methods. This method can provide a smoother connection effect while retaining accuracy;

[0034] (2) Card connection alternative: In some application scenarios, two broken center lines can be spliced ​​together using a card connection, which can provide a fast and reliable connection.

[0035] Step 7: Output results

[0036] The optimized river centerline data is standardized and output into standard vector formats, including but not limited to Shapefile, GeoJSON, KML, and Geodatabase.

[0037] In the vector surface river centerline extraction method of the present invention, the process of step 1 is as follows:

[0038] S1 Collect any one or more types of high-resolution remote sensing image data with a spatial resolution better than 1 meter from Landsat series satellite images and Sentinel series satellite images of the target river and digital elevation model data of the area;

[0039] S2 preprocesses the collected remote sensing image data, including radiation correction, geometric correction and image registration;

[0040] S3 uses image processing technology to identify river areas from preprocessed images and extracts river contours by setting thresholds and parameters. The specific process is as follows: 1) Select appropriate band combinations: According to the characteristics of the target river, select the remote sensing image band combination that best reflects the surface characteristics of the river, including RGB combination, 760-900 nm near-infrared band, 1000-2500 nm short-wave infrared band, 450-495 nm blue band, 495-570 nm green band and 620-750 nm infrared band. Any one or more of the 10 nm red light bands; 2) Applying the adaptive threshold method: Using the adaptive threshold segmentation technology, the threshold is dynamically adjusted according to the local image features; 3) Introducing the deep learning model: Applying any one or more of the deep learning models in the convolutional neural network (CNN), recurrent neural network (RNN), generative adversarial network (GAN), U-Net model, ResNet model, SegNet model, to automatically identify the river area; 4) Post-processing optimization: Based on any one or more of the morphological operations in the opening operation, closing operation, morphological expansion, morphological erosion, morphological reconstruction, top hat transformation, bottom hat transformation, the extracted river contour is smoothed and the noise is removed to optimize the final river contour;

[0041] S4 uses a manual tracking method to superimpose the acquired river contours onto the image map. By comparing the extracted river contours with the actual river contours in the remote sensing image, the deviations are edited and corrected to ensure that the spatial matching degree between the two reaches more than 90% and the similarity of the contour shapes is not less than 85%, obtaining river boundary data with an error of no more than 2 meters.

[0042] In the vector surface river centerline extraction method of the present invention, the preprocessing in step 2 includes:

[0043] S1 verifies and screens river boundary data based on data quality standards and methods for determining erroneous data points.

[0044] The data quality criteria include:

[0045] (1) Position accuracy: whether the data is located at a reasonable river boundary position and whether the position is consistent with known river maps or historical data;

[0046] (2) Temporal consistency: whether the data reflect the current state of the river or the specific point in time, for example, whether seasonal changes or recent river diversions are taken into account;

[0047] (3) Data completeness: whether the data covers the entire river boundary and whether there are any missing parts;

[0048] (4) Logical consistency: whether the data are logically coherent, for example, whether the direction of the river boundary is reasonable, and whether there are unreasonable jumps or breakpoints.

[0049] The method for determining the erroneous data point is as follows:

[0050] (1) Data verification algorithm: Use a specific algorithm to check whether the data point meets the preset standards. For example, by calculating the distance or angle between the data point and the surrounding points, determine whether there are abnormal points;

[0051] (2) Manual inspection: Professionals conduct visual inspection of the data to identify unreasonable or abnormal data points;

[0052] (3) Comparison with other data sources: Compare the data with satellite images, aerial photographs, or other reliable data sources to identify inconsistencies;

[0053] (4) Comparison with historical data: Compare the data with historical data to determine whether there are points that do not conform to the laws of river changes.

[0054] S2 noise removal: Use a filtering algorithm to remove noise points in the boundary line data, and the filtering algorithm includes but is not limited to a median filtering algorithm, a mean filtering algorithm, a Gaussian filtering algorithm, a bilateral filtering algorithm, a Wiener filtering algorithm, and a Kalman filtering algorithm.

[0055] S3 Smoothing: Select a smoothing technique to implement smoothing, the smoothing technique includes but is not limited to moving average, local weighted regression, Gaussian filtering, and spline curve method.

[0056] S4 Boundary Correction: Use manual adjustment, automatic boundary correction algorithm, or any one or more topological correction techniques to correct the river boundary line, and obtain a high-quality river vector surface with geometric accuracy that meets national or industry standards, correct topological relationships, complete data, accurate attribute information, and high smoothness.

[0057] The manual adjustment steps are as follows:

[0058] (1) Visual inspection: First, the river boundary data is overlaid with satellite images, high-resolution aerial photography or topographic maps using geographic information system (GIS) software;

[0059] (2) Identification of problem areas: During the visual inspection, areas where the boundary line does not correspond to the actual river boundary are identified, especially those that are obviously deviated or missing;

[0060] (3) Adjustment: Use the editing tools of GIS software, such as moving points, adding points, or deleting erroneous points, to manually adjust the boundary line to make it closer to the actual river boundary;

[0061] (4) Repeated inspection and optimization: After adjustment, it is necessary to check and compare repeatedly to ensure the accuracy of the adjustment and make necessary fine-tuning to improve the accuracy and reliability of the data;

[0062] (5) Expert review: The final adjustment results should be reviewed by other professionals to ensure the objectivity and correctness of the modification.

[0063] The common methods and settings of the automatic adjustment boundary correction algorithm are as follows:

[0064] (1) Algorithm selection: Possible algorithms include edge detection algorithms based on image processing, pattern recognition algorithms based on machine learning, etc.

[0065] (2) Parameter settings:

[0066] Edge detection algorithms: such as Canny edge detection, parameters may include high and low threshold settings for identifying the strength of the boundary.

[0067] Machine learning algorithm: If you use a support vector machine (SVM) or deep learning model, parameter settings may include the number of training iterations, learning rate, etc.

[0068] Training data preparation: Provide sufficient training samples, including river boundary data with known boundaries, and corresponding satellite images or topographic maps.

[0069] Execute Algorithm: Run the algorithm through the entire dataset to automatically identify and adjust inaccurate boundary lines.

[0070] Result verification: Use professionals or actual ground survey data to verify the results of algorithm correction to ensure the accuracy of the algorithm.

[0071] Iterative optimization: Adjust algorithm parameters or perform algorithm iterations based on verification results to further optimize the correction effect.

[0072] The high quality standards of the high quality river vector surface include but are not limited to:

[0073] (1) Geometric accuracy: The geometric accuracy of the corrected river boundary line should comply with national or industry standards (for example, meet the accuracy requirements at a scale of 1:1000) to ensure that the spatial position of the river boundary is consistent with the actual situation;

[0074] (2) Topological consistency: The river vector surface should ensure the correctness of the topological relationship, including no self-intersection, no overlap, and correct connection relationship, to ensure the topological integrity of the data;

[0075] (3) Data integrity: The corrected river vector surface should contain complete river boundary information to ensure that all important river features are captured without omission;

[0076] (4) Accuracy of attribute information: The river vector surface should be accompanied by accurate attribute information, including the river name, drainage area, flow direction, flow velocity, etc., to ensure its effectiveness in subsequent applications;

[0077] (5) Smoothness standard: The corrected boundary line should be visually smooth and avoid too many jagged edges. The smoothed boundary should be evaluated by quantitative indicators (such as the curvature of the curve) to ensure a certain smoothness standard.

[0078] The above standards ensure that the obtained river vector surfaces are of high quality, reliability and practicality.

[0079] Preferably, the filtering algorithm of the present invention is a median filtering algorithm, and the specific steps are as follows:

[0080] (1) Select window size: Select an appropriate window size based on the characteristics of the data and the nature of the noise. A window that is too small may not effectively remove noise, while a window that is too large may result in loss of data details;

[0081] (2) Sliding window traversal: In the boundary area of ​​the vector data, use the selected window size to perform sliding traversal. For each point in the window, perform a median filter operation;

[0082] (3) Sorting and median selection: Sort all data points in the window, and then select the middle value as the output of the window center point. If the window size is an odd number, the median is the middle point after sorting; if the window size is an even number, the average of the two middle numbers is usually taken;

[0083] (4) Update data points: Replace the original value of the window center point with the calculated median value to remove noise points.

[0084] Preferably, the smoothing technique of the present invention is a spline curve method, and the specific steps are as follows:

[0085] (1) Select the spline type: Select the appropriate spline type based on the complexity of the river boundary and the required smoothness. Commonly used splines include Bezier curves, B-splines, and NURBS (non-uniform rational B-splines);

[0086] (2) Spline fitting: Use the selected spline type to fit the key points of the river boundary;

[0087] (3) Smoothing: After spline fitting, you may need to further adjust the spline parameters, such as adding or reducing control points, to achieve the desired smoothing effect. This step may require multiple iterations until the river boundary line looks smooth and natural;

[0088] (4) Result evaluation: Evaluate whether the river boundary line after smoothing meets the requirements. The quality of the result can be evaluated by comparing the original data with the processed data or by visual inspection.

[0089] In the vector surface river centerline extraction method of the present invention, the boundary tracking algorithm process described in step 3 is as follows:

[0090] S1 Initialize boundary point: select a boundary point from the river vector data as the tracking starting point;

[0091] S2 Tracking direction determination: Use river flow data or GIS tools to determine and set the tracking direction;

[0092] S3 boundary point tracking: starting from the initial point, the boundary points are tracked step by step along the set direction;

[0093] S4 loop tracking: find and record boundary points, and continue to track forward. When encountering branches or complex water system structures, select the main branches or follow the obvious flow direction according to preset rules.

[0094] The preset rules are as follows:

[0095] (1) Main branch selection: Priority is given to branches with large flow, wide width, or marked as main river channels in historical data. This decision is based on the physical properties of the river and historical data;

[0096] (2) Determination of apparent flow direction: The algorithm will use topographic data, water velocity, and flow direction information to determine the most apparent flow direction, which may include analyzing river slope, water flow direction, or using GIS tools to identify major flow paths;

[0097] (3) Data consistency: The algorithm will evaluate the consistency of the data and select the branch that best matches the existing data to maintain the coherence and consistency of the river network;

[0098] (4) Application of preset thresholds: The algorithm sets specific thresholds, such as minimum flow or width, and branches are considered primary branches only when they exceed these thresholds.

[0099] S5 Close Boundary: When tracing back to the initial point or reaching the preset termination condition, stop tracing and connect all recorded boundary points to form a closed boundary line.

[0100] The preset termination conditions are as follows:

[0101] (1) Reaching the preset end point: The algorithm has set a clear end point when it starts tracking. When the tracking reaches this point, the algorithm stops;

[0102] (2) Data interruption or anomaly: When the algorithm encounters missing data, data anomaly, or invalid data during the tracking process and cannot continue tracking, the algorithm will stop;

[0103] (3) Branch selection rules: When encountering complex branches or water system structures, the algorithm decides not to continue tracking a branch based on preset rules;

[0104] (4) Maximum number of tracking steps: To prevent infinite tracking, a maximum number of steps or distance is set. When this number of steps is reached, the algorithm automatically stops.

[0105] (5) Boundary round initial point: In the case of a closed boundary, the algorithm stops when the tracking boundary returns to the initial point.

[0106] S6 Boundary Line Optimization: Smooth the tracked boundary lines and remove noise points in the boundary lines.

[0107] In the vector surface river centerline extraction method of the present invention, the process of creating the distance field in step 4 is as follows:

[0108] S1 Define spatial grid: Select the spatial area where the distance field needs to be calculated. Here, it is the inner area of ​​the river, and divide it into a grid;

[0109] S2 Initialize distance value: Initialize the distance value of all points in the grid. The distance value of the points falling on the boundary line is set to 0, and the distance of the remaining points is set to infinity or a large value;

[0110] S3 calculates the distance field: using an appropriate distance calculation algorithm (such as Euclidean distance), calculates the distance from each grid point to the nearest river boundary line.

[0111] In the vector surface river centerline extraction method of the present invention, the specific process of manually adding river sources and river confluences in step 5 and using a topology-based pruning algorithm to retain the required trunks and branches is as follows:

[0112] S1 manually adds river sources and river confluences to build a topological structure model of the river system, considers river branches as nodes and edges in the graph, and marks key points at the breakpoints of the trunk and branches that need to be retained;

[0113] S2 retains the corresponding river source and river confluence paths based on the marked key points.

[0114] In the vector surface river centerline extraction method of the present invention, the automatic extraction according to parameters in step 5 is specifically to automatically extract and optimize river branches according to the parameters specified by the user, and the parameters include but are not limited to river width, flow rate, flow, river slope, river depth, surface runoff rate, riverbank morphology, sediment particle size distribution, precipitation, terrain undulation, watershed area, vegetation coverage, and water quality indicators. The specific process is as follows:

[0115] S1 branch identification: identify all branches generated by the medial axis transformation algorithm;

[0116] S2 Main stem confirmation: Determine the main stem of the river based on the widest part of the river with the largest flow;

[0117] S3 branch evaluation: evaluates the importance of a branch to the trunk based on any one or more of the branch's length, width, traffic data, and correlation with geographic information;

[0118] S4 Branch selection: Based on the evaluation results, the main river axis and important branches are retained, and minor branches are removed;

[0119] S5 branch optimization: smoothing, removing isolated points and discontinuous points, and adjusting the shape and position of branches;

[0120] S6 Verification and Adjustment: Verify and adjust the branch by comparing with actual observation data or any one or more high-precision data sources such as high-resolution remote sensing images, lidar data and geographic information system data.

[0121] The vector surface river centerline extraction system for executing the above extraction method described in the present invention is composed of the following modules: river boundary line input and preprocessing module, boundary tracking module, medial axis transformation module, tree branch selection module, centerline optimization module and result output module.

[0122] The river boundary line input and preprocessing module in the vector surface river centerline extraction system of the present invention is used to receive and preprocess river boundary line data to ensure data quality to support subsequent centerline extraction, and is composed of the following parts: (1) data receiving unit: receiving externally input river boundary line data, supporting multiple GIS data formats;

[0123] (2) Data verification and screening submodule: automatically verify data integrity and accuracy, and eliminate erroneous or non-compliant data points;

[0124] (3) Resolution adjustment submodule: automatically adjusts data resolution according to application requirements and optimizes data details through resampling or refinement operations;

[0125] (4) Data format conversion submodule: converts data into a unified format suitable for subsequent processing to ensure data compatibility and operability;

[0126] (5) Noise removal submodule: applies advanced filtering technology to effectively remove the noise generated during data acquisition;

[0127] (6) Smoothing submodule: Smoothing the boundary line data to reduce irregular fluctuations and enhance data visualization and analysis;

[0128] (7) Boundary correction submodule: Correct the boundary line automatically or manually to ensure that it accurately reflects the actual boundary of the river;

[0129] (8) Constructing a closed river vector surface module: After completing the preprocessing, the processed data is constructed into a closed river vector surface to further ensure the integrity and accuracy of the data and lay a solid foundation for the subsequent centerline extraction.

[0130] The boundary tracking module in the vector surface river centerline extraction system of the present invention is used to extract and track the outer boundary line of the river from the pre-processed vector surface data to ensure the accuracy and consistency of the river morphology, and is composed of the following parts:

[0131] (1) Initialization unit: responsible for identifying and selecting a significant boundary point from the river vector data as the starting point for tracking. Usually, the point where the river and non-river area are most obviously bounded is selected.

[0132] (2) Direction determination unit: Use river flow data or a dedicated library of GIS tools to determine and set the tracking direction to ensure that the tracking process is carried out along the flow direction of the river;

[0133] (3) Tracking execution unit: Starting from the initial point, it gradually tracks the boundary points along the set direction. This unit determines and determines the boundary points by comparing the attributes of adjacent points (such as elevation and slope) with the preset threshold;

[0134] (4) Decision-making processing unit: When encountering branches or complex water system structures, the main branch is selected or the most obvious flow direction is followed according to preset rules to ensure the continuity and accuracy of tracking;

[0135] (5) Closing processing unit: When the tracking returns to the initial point or meets the preset termination condition, the tracking is stopped and all recorded boundary points are connected to form a closed boundary line;

[0136] (6) Optimization processing unit: Smooth the tracked boundary line and remove noise points to improve the accuracy and smoothness of the boundary line.

[0137] The medial axis transformation module in the vector surface river centerline extraction system of the present invention is used to execute the medial axis transformation algorithm to extract the accurate centerline from the preprocessed river vector data, and is composed of the following parts:

[0138] (1) Distance field initialization unit: responsible for creating the initial distance field on the river vector surface, calculating the Euclidean distance from each internal point to the nearest boundary, and storing these distance values;

[0139] (2) Distance field update unit: Through an iterative process, the distance value in the distance field is continuously updated to ensure that the distance value of each internal point reflects the minimum distance to all boundary points;

[0140] (3) Central axis extraction unit: After the distance field is stabilized, the internal points with the maximum distance value are identified and extracted, and the center line of the river is formed by connecting these points;

[0141] (4) Centerline optimization unit: optimizes the extracted centerline, including smoothing, removing noise points, and ensuring continuity, to improve the quality and accuracy of the centerline;

[0142] (5) Verification and adjustment unit: Verify the optimized central axis and make necessary adjustments by comparing it with actual observation data or other high-precision data sources to ensure that the center line accurately reflects the actual center position of the river.

[0143] The tree branch selection module in the vector surface river centerline extraction system of the present invention is used to accurately extract and optimize the river centerline from the river data processed by the medial axis transformation to ensure the accuracy of the data and the analysis efficiency, and is composed of the following parts:

[0144] (1) Branch identification unit: automatically identifies and records all river branches generated by the medial axis transformation algorithm, which represent secondary channels or tributaries of the river;

[0145] (2) Mainstream confirmation unit: by analyzing the width, flow and other parameters of the river, the mainstream of the river, i.e. the main flow direction of the river, is determined and marked;

[0146] (3) Branch assessment unit: A comprehensive assessment of each branch is conducted, taking into account its length, width, flow, and correlation with geographic information to determine its importance to the overall flow and structure of the river;

[0147] (4) Branch selection unit: Based on the evaluation results, intelligently select and retain branches that have a significant impact on the river flow direction and structure, while removing unimportant branches;

[0148] (5) Branch optimization unit: adjust the shape and position of the selected branches to ensure the natural connection with the main trunk and the continuous smoothness of the river centerline;

[0149] (6) Verification and adjustment unit: Verify and adjust the selected branches by comparing with actual observation data or other high-precision data sources to ensure that the river centerline accurately reflects the actual flow direction and structure.

[0150] The centerline optimization module in the vector surface river centerline extraction system of the present invention is used to refine the initially extracted river centerline to improve its accuracy and applicability, and is composed of the following parts:

[0151] (1) Curvature optimization unit: The algorithm adjusts the position of the points on the center line to optimize the curvature of the curve, ensuring that the center line is consistent with the natural flow direction of the river and reducing unnecessary bends.

[0152] (2) Topology check unit: implements topology consistency check to ensure that the centerline is correctly connected in the river network and avoids branch errors or omissions;

[0153] (3) Flow matching unit: adjusts the width or weight of the center line according to the actual flow data to reflect the actual flow distribution of the river and enhance the accuracy of hydrological analysis;

[0154] (4) Multi-source data fusion unit: Integrates terrain data, satellite images, historical hydrological data, etc. to calibrate and optimize the centerline to improve its accuracy and reliability;

[0155] (5) Dynamic adjustment unit: Based on the available data, the centerline is dynamically adjusted to adapt to the river status in different time periods, taking into account the seasonal or long-term change trend of the river.

[0156] The result output module in the vector surface river centerline extraction system of the present invention is used to convert and output the optimized river centerline data into a standard vector format or other common geographic information data format for subsequent analysis and application, and is composed of the following parts:

[0157] (1) Format conversion unit: converts the processed river centerline data into a common vector data format or other common geographic information data formats, such as Shapefile or GeoJSON, to ensure data compatibility and readability;

[0158] (2) Data output interface: It provides data output function, allows users to select the output path and format, and supports direct output to GIS software or database to facilitate further data processing and analysis.

[0159] Beneficial effects of the present invention:

[0160] The present invention realizes efficient and accurate extraction of centerline in complex river systems through a series of technical steps. The technical effects are mainly reflected in the following aspects:

[0161] (1) Improved preprocessing accuracy: By removing noise, smoothing, correcting boundaries, and constructing closed river vector surface data for the input river boundary line data, the data quality is effectively improved, the errors caused by data impurities are reduced, and the accuracy of the extraction results is ensured. Traditional technologies often ignore the importance of data preprocessing, resulting in limited accuracy of centerline extraction. The present invention greatly improves the reliability of centerline extraction through more rigorous data preprocessing;

[0162] (2) Boundary tracking consistency: The use of advanced boundary tracking technology can accurately determine the outer boundary line of the river, maintain the consistency of the river morphology, and provide a reliable basis for subsequent central axis transformation. Compared with traditional methods, the consistency of boundary tracking in this invention significantly reduces uncertainty when dealing with complex river morphology, ensuring the accuracy of the final centerline;

[0163] (3) Medial axis transformation accuracy: By introducing an improved medial axis transformation algorithm, this algorithm can more efficiently and accurately extract the medial axis that is highly consistent with the centerline of the river through an accurate distance transformation method. Compared with traditional methods, this transformation method has significantly improved accuracy and efficiency when dealing with complex river morphology;

[0164] (4) Optimal selection of tree branches: In view of the tree branches that may appear in complex river systems, the present invention manually selects river sources and river confluences and uses a pruning algorithm; an automatic branch selection algorithm is used to retain the central axis of the main river and its important branches, ensuring that the final extracted center line can accurately reflect the main flow direction and structure of the river, and improving the adaptability to complex river morphology. This optimization selection mechanism is relatively lacking in traditional technologies, which makes the present invention have obvious advantages in dealing with complex rivers;

[0165] (5) Centerline optimization: After initially extracting the central axis and optimizing the tree branches, the present invention further performs optimization steps such as removing redundant points, smoothing, and connecting breakpoints to obtain continuous and accurate river centerline data, thereby improving the overall quality and applicability of the data. Traditional technologies often ignore these subsequent optimization steps, resulting in inaccurate final results;

[0166] (6) Standardized result output: The present invention uses a dedicated result output module to achieve standardized output of optimized river centerline data, which is convenient for further analysis and application, and provides high-quality data support for river research, management and application. This standardized output is extremely convenient and practical in practical applications.

[0167] In summary, the present invention has significant improvements over the prior art in terms of accuracy of river centerline extraction, processing efficiency, and adaptability to complex river morphology, and can provide more reliable and efficient technical support for river analysis and management. BRIEF DESCRIPTION OF THE DRAWINGS

[0168] Figure 1 This is the overall framework diagram of the vector surface river centerline extraction method;

[0169] Figure 2 Flowchart for boundary tracing;

[0170] Figure 3 Process flow chart for centerline optimization;

[0171] Figure 4 This is the image of the Songbaishan section of the Qingshui River (Tiantu image - superimposed river boundary line);

[0172] Figure 5 This is the image of the Songbaishan section of the Qingshui River (Jilin-1 satellite image - superimposed river boundary line);

[0173] Figure 6 Comparison diagram of river boundary line input and before and after preprocessing;

[0174] Figure 7 This is a comparison of the river vector surface before and after the median axis transformation;

[0175] Figure 8 Comparison of river centerlines before and after tree branch selection (without overlapping rivers);

[0176] Fig. 9 Comparison of river centerlines before and after selection for tree branching (fitting rivers);

[0177] Fig.10 Comparison diagram of river centerline before and after centerline optimization;

[0178] Fig.11 Output a schematic diagram for the result (output Shapefile);

[0179] Fig.12 Output a diagram of the result (output GeoJSON). DETAILED DESCRIPTION

[0180] The technical solution of the present invention is described in detail below in conjunction with specific embodiments. The following embodiments are only used for explanation and illustration, and do not constitute a limitation on the technical solution of the present invention.

[0181] Example 1

[0182] This example takes the Songbaishan section of Qingshui River, a typical river in the Wujiang River system in the Yangtze River Basin, as an example (the map range of the river section is shown in Figure 4 and Figure 5 ), describes in detail the specific operation process of how to extract the river centerline under the existing data conditions.

[0183] 1. Obtaining river boundary lines:

[0184] (1) First, use Tiandi Map and related high-definition satellite images to collect geographic information data of the Songbai Mountain section of Qingshui River. Since higher-resolution remote sensing images and digital elevation model (DEM) data are not available, this example will rely on existing remote sensing image data for analysis;

[0185] (2) Preprocess the collected remote sensing images, including radiation correction, geometric correction and image registration, to ensure the quality and accuracy of the image data for subsequent analysis and processing;

[0186] (3) Using image processing techniques, such as edge detection and threshold segmentation, the river area is identified from the preprocessed image, and the outline of the river is extracted by setting appropriate thresholds and parameters;

[0187] (4) The acquired river contours are superimposed on the image map by manual tracking for preliminary verification. The deviations are edited and corrected to obtain river boundary data with higher accuracy.

[0188] 2. River boundary line input and preprocessing:

[0189] After obtaining the river boundary data, the river boundary data is input into the "River Centerline Extraction System". The input boundary data is usually extracted from remote sensing images or map data, which may contain noise and irregular boundaries. The preprocessing steps include:

[0190] (1) Noise removal: Use a filtering algorithm (such as median filtering) to remove noise points in the boundary data. The following are the specific steps of the median filtering algorithm:

[0191] 1) Select the window size: Select an appropriate window size based on the characteristics of the data and the nature of the noise. A window that is too small may not effectively remove noise, while a window that is too large may result in loss of data details;

[0192] 2) Sliding window traversal: In the boundary area of ​​the vector data, use the selected window size to perform sliding traversal. For each point in the window, perform a median filter operation;

[0193] 3) Sorting and median selection: Sort all data points in the window, and then select the middle value as the output of the window center point. If the window size is an odd number, the median is the middle point after sorting; if the window size is an even number, the average of the two middle numbers is usually taken;

[0194] 4) Update data points: Replace the original value of the window center point with the calculated median value to remove noise points.

[0195] By removing noise points, data analysis and subsequent processing can be made more accurate, and the credibility and availability of data can be improved.

[0196] (2) Smoothing: Use the spline curve method to smooth the boundary line to reduce the sharpness and irregularity of the boundary line. The following is a detailed introduction to the steps of the spline curve method:

[0197] 1) Select the spline type: Select the appropriate spline type according to the complexity of the river boundary and the required smoothness. Commonly used splines include Bezier curves, B-splines, and NURBS (non-uniform rational B-splines);

[0198] 2) Spline fitting: Use the selected spline type to fit the key points of the river boundary;

[0199] 3) Smoothing: After spline fitting, you may need to further adjust the spline parameters, such as adding or reducing control points, to achieve the desired smoothing effect. This step may require multiple iterations until the river boundary line looks smooth and natural;

[0200] 4) Result evaluation: Evaluate whether the river boundary line after smoothing meets the requirements. The quality of the result can be evaluated by comparing the original data with the processed data or by visual inspection.

[0201] The effect comparison before and after treatment is shown in Figure 6 In the figure, Figure 1 shows the data of the river boundary line before preprocessing; Figure 2 shows the superimposed data before and after preprocessing; Figure 3 shows the data of the river boundary line after preprocessing.

[0202] (3) Boundary correction: Use topology correction technology to ensure the continuity and closure of the boundary line, and finally generate a high-quality river vector surface. In the basic platform FME corresponding to the "River Centerline Extraction System", a converter is provided to capture line segments within a given tolerance range and connect the line segments. The development kits provided by other GIS platforms also provide similar tools.

[0203] 3. Boundary tracking:

[0204] After completing data preprocessing, the boundary tracking algorithm is used to identify the outer boundary of the river vector surface. By tracking the outer boundary of the river, the consistency of the shape is ensured, providing reliable basic data for subsequent medial axis transformation. The specific steps of boundary tracking are as follows:

[0205] (1) Initialization unit: responsible for identifying and selecting a significant boundary point from the river vector data as the starting point for tracking. Usually, the point where the river and non-river area are most obviously bounded is selected.

[0206] (2) Direction determination unit: Use river flow data or GIS tools to determine and set the tracking direction to ensure that the tracking process is carried out along the flow direction of the river;

[0207] (3) Tracking execution unit: Starting from the initial point, it gradually tracks the boundary points along the set direction. This unit determines and determines the boundary points by comparing the attributes of adjacent points (such as elevation and slope) with the preset threshold;

[0208] (4) Decision-making processing unit: When encountering branches or complex water system structures, the main branch is selected or the most obvious flow direction is followed according to preset rules to ensure the continuity and accuracy of tracking;

[0209] (5) Closing processing unit: When the tracking returns to the initial point or meets the preset termination condition, the tracking is stopped and all recorded boundary points are connected to form a closed boundary line;

[0210] (6) Optimization processing unit: Smooth the tracked boundary line and remove noise points to improve the accuracy and smoothness of the boundary line.

[0211] Through the collaborative work of these components, this module can effectively handle complex river networks and provide high-quality data support for hydrological analysis and river management.

[0212] 4. Medial axis transformation:

[0213] By introducing an improved distance transformation method, it is particularly suitable for extracting the center line of the river. The algorithm is implemented by the following steps:

[0214] (1) Initialize the distance field: First, create a distance field on the river vector surface, where each point represents the distance to the nearest boundary. This distance field is initialized by calculating the Euclidean distance from each interior point to the nearest boundary point;

[0215] (2) Iteratively update the distance field: Based on the initial distance field, iteratively update to gradually refine the distance from each point to the boundary. In each iteration, check each internal point and update its distance value to the minimum distance to all boundary points. This process is achieved by comparing the current distance value with the distance to the newly detected boundary point;

[0216] (3) Extracting the central axis: When the distance field is stable, that is, when the distance value no longer changes significantly, the algorithm starts to extract the central axis. The points on the central axis are those internal points with the largest distance values, which represent the center of the river. By identifying these points and connecting them, the center line of the river can be formed;

[0217] (4) Centerline optimization: The extracted centerline may contain some irregular or noisy points, so it needs to be optimized. This includes smoothing, removing isolated points and discontinuous points, and ensuring the continuity and smoothness of the centerline;

[0218] (5) Verification and adjustment: Finally, the extracted central axis is verified to ensure that it accurately reflects the actual centerline of the river. This may involve comparison with actual observation data or other high-precision data sources, and necessary adjustments.

[0219] Through this improved medial axis transformation algorithm, high-quality river centerlines can be effectively extracted, providing accurate data support for applications such as river management, hydrological analysis, and environmental assessment.

[0220] Comparison before and after the central axis transformation Figure 7 In the figure, Figure 1 shows the data of the river vector surface before the median axis transformation; Figure 2 shows the centerline data extracted after the median axis transformation; Figure 3 shows the superposition of the data before and after the median axis transformation.

[0221] 5. Tree branch selection:

[0222] In view of the tree-like branching in the complex river system, the branch selection algorithm is used to process it. The specific steps are as follows:

[0223] (1) Branch identification: Identify the main river and its branches;

[0224] (2) Main branch retention: The algorithm selects and retains the central axis of the main river and its important branches, removes the secondary branches, and ensures that the centerline reflects the main flow direction and structure of the river;

[0225] See the comparison before and after tree branch selection. Figure 8 (Rivers not included), Fig. 9 (Overlapping rivers), in the figure, Figure 1 shows the data before tree branch selection; Figure 2 shows the superimposed data before and after tree branch selection; Figure 3 shows the data after tree branch selection.

[0226] 6. Centerline optimization:

[0227] The initially extracted central axis is optimized, including:

[0228] (1) Redundant point removal: Remove redundant points in the central axis to ensure the simplicity of the data;

[0229] (2) Smoothing: Smoothing the central axis to enhance its continuity;

[0230] (3) Connect breakpoints: Process the breakpoints in the central axis to ensure the continuity of the center line.

[0231] Comparison of river centerline before and after centerline optimization Fig.10 , Figure 1 shows the data before the optimization of the river centerline; Figure 2 shows the superimposed data before and after the optimization of the river centerline; Figure 3 shows the data after the optimization of the river centerline.

[0232] 7. Result output (select the result output format, see the attached Fig. 9 and attached Fig.10 Result output diagram):

[0233] The optimized river centerline is output in a standard vector format for further analysis and application. In the basic platform FME corresponding to the "River Centerline Extraction System", a dedicated result output module is provided to achieve standardized output of optimized river centerline data, which can output a variety of common standard vector formats, including but not limited to Shapefile, GeoJSON, KML, Geodatabase, and other common vector formats.

[0234] The result output is shown in the figure Fig.11 , Fig.12 , Fig.11 To output to Shapefile format; Fig.12 Output is in GeoJSON format.

[0235] In summary, the present invention realizes efficient and accurate extraction of the centerline of a complex river system through the above specific implementation methods. The improved technical solution has significantly improved the accuracy of river centerline extraction, processing efficiency, and adaptability to complex river morphology, providing more reliable and efficient technical support for river analysis and management.

[0236] Although the present invention has been described in detail above by means of general description, specific implementation methods and experiments, it is obvious to those skilled in the art that some modifications or improvements can be made to the present invention. Therefore, these modifications or improvements made without departing from the spirit of the present invention all fall within the scope of protection claimed by the present invention.

Claims

1. A method for extracting river centerlines from a vector surface, characterized in that: The steps include: Step 1: Obtain river boundary data: Collect remote sensing image data of the target river and digital elevation model data of the area, identify the river area after pre-processing, extract the river contour and edit and correct it to obtain the river boundary line data; Step 2: River boundary data input and preprocessing: The river boundary line data is input into the river centerline extraction system to pre-process the noise and irregular boundaries of the boundary line data; Step 3: Boundary tracking: After completing data preprocessing, the boundary tracing algorithm is used to identify the outer boundary of the river vector surface; Step 4: Medial axis transformation: After completing the boundary tracing, the centerline of the river is extracted using the following medial axis transformation algorithm: S1 Initialize the distance field: Create a distance field on the river vector surface. Each point in the distance field represents the Euclidean distance to the nearest boundary. The steps to create the distance field are as follows: ① Define the spatial grid: Select the internal area of ​​the river where the distance field needs to be calculated and divide it into a grid; ② Initialize the distance value: Set the distance value of the points falling on the boundary line to 0, and set the distance of the remaining points to an infinite value; ③ Calculate the distance field: Calculate the Euclidean distance of each grid point to the nearest river boundary line; S2 iteratively updates the distance field: iteratively updates the distance field by comparing the current distance value of each point with the distance value to the newly detected boundary point, and updates the distance value to the minimum distance to all boundary points; S3 Extract the central axis: After the distance field is stabilized, the internal points with the maximum distance value are identified and connected to form the centerline of the river; S4 centerline optimization: smooth the extracted centerline and remove isolated points and discontinuous points; S5 Verification and adjustment: Verify and adjust the extracted central axis by comparing it with actual observation data or any one or more high-precision data sources such as high-resolution remote sensing images, lidar data and geographic information system data; Step 5: Tree branch selection: The tree branch selection is performed according to the following tree branch selection algorithm: S1 manually adds river sources and river sinks, and uses a topology-based pruning algorithm to retain the required trunks and branches; S2 is automatically extracted according to parameter settings; Step 6: Centerline optimization: The extracted river centerline is processed by removing redundant points, smoothing the centerline, connecting breakpoints, adjusting curvature, checking topological consistency, matching flow, fusing and correcting multi-source data, and dynamically adjusting the centerline according to seasonal changes or long-term change trends of the river, so as to optimize the centerline; Step 7: Result output: The optimized river centerline data is standardized and output in standard vector format.

2. The method for extracting river centerlines from a vector surface according to claim 1, characterized in that: The specific process of step one is as follows: S1 Collect high-resolution remote sensing image data with a spatial resolution better than 1 meter from any one or two of the Landsat series satellite images and Sentinel series satellite images of the target river and digital elevation model data of the area; S2 preprocesses the collected remote sensing image data, including radiation correction, geometric correction and image registration; S3 uses image processing technology to identify river areas from preprocessed images and extracts river contours by setting thresholds and parameters; S4 uses a manual tracking method to superimpose the acquired river contours onto the image map. By comparing the extracted river contours with the actual river contours in the remote sensing image, the deviations are edited and corrected to ensure that the spatial matching degree between the two reaches more than 90% and the similarity of the contour shapes is not less than 85%, obtaining river boundary data with an error of no more than 2 meters.

3. The method for extracting river centerlines from a vector surface according to claim 1, characterized in that: The pretreatment in step 2 includes: S1 verifies and screens river boundary data according to data quality standards and the method for determining erroneous data points; S2 noise removal: Use filtering algorithm to remove noise points in boundary line data; S3 Smoothing: Select smoothing technology and implement smoothing; S4 Boundary Correction: Use any one or more topological correction techniques such as manual adjustment and automatic boundary correction algorithm to correct the river boundary line and obtain the river vector surface.

4. The method for extracting river centerlines from a vector surface according to claim 3, characterized in that: The filtering algorithm described in S2 is a median filtering algorithm; the smoothing technique described in S3 is a spline curve method.

5. The method for extracting river centerlines from a vector surface according to claim 1, characterized in that: The boundary tracking algorithm process described in step 3 is as follows: S1 Initialize boundary point: select a boundary point from the river vector data as the tracking starting point; S2 Tracking direction determination: Use river flow data or GIS tools to determine and set the tracking direction; S3 boundary point tracking: starting from the initial point, the boundary points are tracked step by step along the set direction; S4 loop tracking: find and record the boundary point, continue to track forward, and when encountering branches or complex water system structures, select the main branch or follow the obvious flow direction according to the preset rules; S5 Closed Boundary: When the tracking returns to the initial point or reaches the preset termination condition, the tracking stops and all recorded boundary points are connected to form a closed boundary line; S6 Boundary Line Optimization: Smooth the tracked boundary lines and remove noise points in the boundary lines.

6. The method for extracting river centerlines from a vector surface according to claim 1, characterized in that: The specific process of manually adding river sources and river sinks as described in step 5 and using the topology-based pruning algorithm to retain the required trunks and branches is as follows: S1 manually adds river sources and river confluences to build a topological structure model of the river system, considers river branches as nodes and edges in the graph, and marks key points at the breakpoints of the trunk and branches that need to be retained; S2 retains the corresponding river source and river confluence paths based on the marked key points.

7. The method for extracting river centerlines from a vector surface according to claim 1, characterized in that: The automatic extraction according to parameters described in step 5 is specifically to automatically extract and optimize river branches according to the parameters specified by the user, and the process is as follows: S1 branch identification: identify all branches generated by the medial axis transformation algorithm; S2 Main stem confirmation: Determine the main stem of the river based on the widest part of the river with the largest flow; S3 branch evaluation: evaluates the importance of a branch to the trunk based on any one or more of the branch's length, width, traffic data, and correlation with geographic information; S4 Branch selection: Based on the evaluation results, the main river axis and important branches are retained, and minor branches are removed; S5 branch optimization: smoothing, removing isolated points and discontinuous points, and adjusting the shape and position of branches; S6 Verification and Adjustment: Verify and adjust the branch by comparing with actual observation data or any one or more high-precision data sources such as high-resolution remote sensing images, lidar data and geographic information system data.

8. A vector surface river centerline extraction system for executing the extraction method according to any one of claims 1 to 7, characterized in that: The system consists of the following modules: river boundary line input and preprocessing module, boundary tracking module, medial axis transformation module, tree branch selection module, centerline optimization module and result output module, as follows: River boundary line input and preprocessing module: responsible for data reception and preprocessing, and obtaining river boundary line data; Boundary tracking module: responsible for executing the boundary tracking algorithm, extracting and tracking the outer boundary of the river from the river boundary data, and finally obtaining the closed outer boundary of the river; Medial axis transformation module: responsible for executing the medial axis transformation algorithm to extract the centerline of the river; Tree branch selection module: responsible for executing the tree branch selection algorithm on the river centerline obtained after the medial axis transformation processing to further extract the river centerline; Centerline optimization module: responsible for fine-tuning the extracted river centerline to obtain the final river centerline; Result output module: responsible for converting and outputting the final river centerline data into standard vector format.

9. The vector surface river centerline extraction system according to claim 8, characterized in that: The river boundary line input and preprocessing module includes the following submodules: Data receiving unit: responsible for receiving river boundary line data input from outside; Data verification and screening submodule: responsible for automatically verifying data integrity and accuracy, and eliminating erroneous or non-compliant data points; Resolution adjustment submodule: responsible for automatically adjusting data resolution and optimizing data details; Data format conversion submodule: responsible for converting data into a unified format suitable for subsequent processing to ensure data compatibility and operability; Noise removal submodule: responsible for applying filtering technology to remove the noise generated during data acquisition; Smoothing submodule: responsible for smoothing the boundary line data, reducing irregular fluctuations, and increasing data visualization and analysis; Boundary correction submodule: responsible for correcting the boundary line to ensure that it accurately reflects the actual boundary of the river; The module for constructing a closed river vector surface is responsible for constructing the processed data into a closed river vector surface; The boundary tracking module includes the following submodules: Initialization unit: responsible for identifying and selecting the starting point of boundary tracing from river vector data; Direction determination unit: responsible for determining and setting the tracking direction; Tracking execution unit: responsible for tracking the boundary points along the set direction starting from the initial point; Decision-making processing unit: responsible for making decisions when encountering branches or complex water system structures during boundary tracking to ensure the continuity and accuracy of tracking; Closing processing unit: responsible for forming a closed boundary line; Optimization processing unit: responsible for smoothing the tracked boundary line and removing noise points to improve the accuracy and smoothness of the boundary line; The central axis transformation module includes the following submodules: Distance field initialization unit: responsible for creating the initial distance field on the river vector surface, calculating the Euclidean distance from each internal point to the nearest boundary, and storing these distance values; Distance field update unit: responsible for updating the distance value in the distance field through an iterative process to ensure that the distance value of each internal point reflects the minimum distance to all boundary points; Central axis extraction unit: After the distance field is stabilized, the internal points with the maximum distance value are identified and extracted, and the center line of the river is formed by connecting these points; Centerline optimization unit: optimizes the extracted centerline, including smoothing, removing noise points, and ensuring continuity, to improve the quality and accuracy of the centerline; Verification and adjustment unit: Verify the optimized central axis and make necessary adjustments by comparing it with actual observation data or other high-precision data sources to ensure that the center line accurately reflects the actual center position of the river; The tree branch module includes the following submodules: Branch identification unit: responsible for automatically identifying and recording all river branches generated by the medial axis transformation algorithm; Main stem confirmation unit: responsible for identifying and marking the main stem of the river; Branch Assessment Unit: Responsible for conducting a comprehensive assessment of each branch to determine its importance to the overall flow and structure of the river; Branch selection unit: responsible for selecting and retaining branches that have a significant impact on the flow direction and structure of the river based on the evaluation results, while removing unimportant branches; Branch optimization unit: responsible for adjusting the shape and position of the selected branches to ensure the natural connection with the main trunk and the continuous smoothness of the river centerline; Verification and Adjustment Unit: Responsible for verifying and adjusting the selected branches to ensure that the river centerline accurately reflects the actual flow direction and structure; The centerline optimization module includes the following submodules: Curvature optimization unit: responsible for adjusting the position of points on the center line through algorithms, optimizing the curvature of the curve, and ensuring that the center line is consistent with the natural flow direction of the river; Topology Check Unit: Responsible for implementing topology consistency checks to ensure that the center lines are connected correctly in the river network and avoid branch errors or omissions; Flow matching unit: adjusts the width or weight of the center line according to the actual flow data to reflect the actual flow distribution of the river and enhance the accuracy of hydrological analysis; Multi-source data fusion unit: responsible for correcting and optimizing the center line to improve its accuracy and reliability; Dynamic adjustment unit: responsible for dynamically adjusting the center line to adapt to the river status in different time periods; The result output module includes the following submodules: Format conversion unit: responsible for converting river centerline data into standard vector data format to ensure data compatibility and readability; Data output interface: responsible for data output according to the output path and format.

Images