Method and system for summarizing jagged pattern data based on remote sensing extraction

Abstract

The application discloses a sawtooth pattern data summarization method and system based on remote sensing extraction, and belongs to the technical field of remote sensing image processing, and comprises the following steps: based on screen pixels, a screen display scale and a coordinate system, a scale denominator is dynamically estimated and an initial tolerance T0 is generated, for latitude and longitude coordinates, degree and meter unit conversion is carried out through a central latitude, and a unified scale geometric processing threshold is obtained; for a graph spot boundary extracted by remote sensing image intelligent interpretation, a DP algorithm is used to preliminarily simplify the outer ring and each inner ring respectively, pixel-level micro-vibration and fine burrs are eliminated, the point number is reduced, and the closed ring head and tail are kept consistent; the graph spot boundary after thinning is smoothed at the ring level by using a PAEK polynomial approximation exponential kernel method, boundary smoothing is realized through small-step iteration and tolerance control, and overshoot or deviation is prevented; the smoothed graph spot boundary is simplified again, redundant points introduced by smoothing are removed, the boundary is reshaped, and the topological relationship of the pattern itself is kept correct.

Description

Technical Field

[0001] This invention belongs to the field of remote sensing image processing technology, specifically relating to a method and system for summarizing jagged graphic data extracted from remote sensing. Background Technology

[0002] Remote sensing image processing technology, generally employing artificial intelligence (AI) techniques such as deep learning and computer vision, can intelligently interpret massive amounts of remote sensing image data, significantly improving the efficiency and accuracy of information extraction. However, in intelligent interpretation of remote sensing images, the boundaries of land cover patches automatically extracted using techniques such as deep learning segmentation and traditional image segmentation commonly suffer from problems such as jagged edges, numerous spikes, high local noise, inconsistent scale, and easily disrupted topology. This is mainly due to the influence of the original image pixel grid during the extraction process, causing the generated vector boundaries to jitter along the pixel edges, filled with irregular details and high-frequency noise. These rough boundaries severely affect the cartographic representation of maps and the accuracy of subsequent spatial analysis.

[0003] To address these issues, existing technologies typically employ geometric simplification or smoothing algorithms. One type is the single-stage simplification algorithm, such as the Douglas-Peucker algorithm. While effectively reducing the number of boundary points, it tends to retain a "polyline feel" in densely jagged areas and may overly weaken key corner points or cause "corner drift." Another type is the pure curve smoothing algorithm, such as Bézier curve or spline smoothing. Although it can achieve visually smooth boundaries, it is prone to introducing "overshoot" or "offset," leading to geometric deformation and area distortion. Furthermore, it is highly susceptible to disrupting the original topological relationships when processing patches with complex holes or multi-ring structures. In addition, while the buffering method (expanding outwards and then contracting inwards) can suppress burrs, it may cause hole deformation or loss of connection for narrow slits or elongated patches.

[0004] In summary, existing single processing methods struggle to achieve a good balance between shape fidelity and visual smoothness, and generally lack effective preservation of topological consistency across multiple rings. Furthermore, the processing parameters of existing solutions are often disconnected from the screen display scale, screen pixels, and geographic coordinate units, resulting in inconsistent summarization of the same map patch at different display scales, failing to meet the needs of multi-scale map representation and automated processing. Therefore, there is an urgent need for a summarization method and system for jagged graphic data based on remote sensing extraction that can balance geometric accuracy, visual quality, and topological robustness. Summary of the Invention

[0005] The present invention addresses the above-mentioned problems and overcomes the shortcomings of the prior art by providing a method and system for summarizing jagged graphic data based on remote sensing extraction.

[0006] To achieve the above objectives, the present invention adopts the following technical solution.

[0007] In a first aspect, the present invention provides a method for summarizing jagged graphic data based on remote sensing extraction, comprising the following steps; S1. Tolerance dynamic estimation: Based on screen pixels, screen display scale and coordinate system, dynamically estimate the scale denominator and generate the initial tolerance T0. Convert the degree to meter unit through the center latitude in the latitude and longitude coordinate system to obtain the geometric processing threshold of the uniform scale. S2. First stage thinning: The boundaries of the patches extracted by intelligent interpretation of remote sensing images are initially simplified by using the Douglas-Peucker algorithm for the outer ring and each inner ring, eliminating pixel-level micro-jitter and fine burrs, reducing the number of points, and keeping the beginning and end of the closed loop consistent. S3. Iterative Smoothing: The PAEK polynomial approximation exponential kernel method is used to smooth the boundaries of the thinned patches at the ring level. Through small-step iteration and tolerance control, the boundaries are smoothed and overshoot or offset is prevented. S4. Second stage thinning: The smoothed patch boundaries are simplified a second time, redundant points introduced by smoothing are removed, the boundaries are shaped, and the correct topological relationship of the graphic itself is maintained. In the first stage of thinning in step S2, if the reduction rate of a certain ring after simplification is lower than a preset threshold, an adaptive thinning rollback strategy is automatically triggered, and the minimum distance deletion method is used to perform thinning again, with a tolerance Tmin of 0.5 times the initial tolerance T0.

[0008] Furthermore, the tolerance dynamic estimation in step S1 also includes: ring-level tolerance adaptive adjustment: the tolerance is dynamically adjusted according to the external / internal attributes and point density of the ring; when the Douglas-Peucker thinning point reduction rate is insufficient, the minimum distance point deletion of Tmin = 0.5 × T0 is automatically used for compensation.

[0009] Furthermore, the iterative smoothing in step S3 smooths the boundary through multiple small-step iterations, controls the tolerance to prevent geometric expansion or deformation, and maintains topological stability under complex holes and multi-ring structures.

[0010] Furthermore, after the second stage of thinning in step S4, topology preservation processing is performed to ensure that the outer ring and the hole do not cross boundaries or intersect, preventing topology damage during smoothing or simplification.

[0011] Furthermore, the method for summarizing jagged graphic data based on remote sensing extraction also includes a degenerate loop processing step: degenerate loops with few points or small areas are removed or merged according to a preset threshold to avoid abnormal loops affecting subsequent processing.

[0012] Furthermore, the summarization method supports tolerance conversion between geographic coordinate systems and projected coordinate systems, solves the problem of unified control between pixels, meters, and degrees, and achieves visual consistency of the summarization results of the same graphic at different scales for the boundaries of map patches.

[0013] Furthermore, the threshold for the drop rate of the adaptive thinning backoff strategy is set to 30%, and thinning is performed again when the drop rate is lower than this value.

[0014] Furthermore, the boundary of the patch is generated by intelligent interpretation of remote sensing images. The patch boundary includes patches generated by deep learning segmentation or traditional image segmentation, and its boundary has the characteristics of being jagged, having many burrs, and having large local noise.

[0015] Furthermore, after the second stage of thinning, the summarized patch data is output for multi-scale map representation or spatial analysis, while preserving the original form as much as possible, that is, preserving the geometric features of area and boundary orientation without distortion.

[0016] Secondly, the present invention also provides a summarization system for jagged graphic data based on remote sensing extraction, for implementing the summarization method for jagged graphic data based on remote sensing extraction, including a tolerance calculation module, a thinning module, a smoothing module, a secondary thinning module, an adaptive thinning module, and a topology preservation module. The tolerance calculation module is used to dynamically generate an initial tolerance based on screen pixels, screen display scale and coordinate system, and to perform ring-level tolerance adaptive adjustment. The thinning module is used to perform the first stage of thinning on the boundaries of the patches extracted by intelligent interpretation of remote sensing images, and uses the Douglas-Peucker algorithm to simplify the outer ring and each inner ring respectively. The smoothing module is used to iteratively smooth the boundaries of the thinned patches, using a polynomial approximation exponential kernel method. The secondary thinning module is used to perform a second stage of thinning on the smoothed patch boundary, removing redundant points and reshaping the boundary. The adaptive thinning module is used to trigger the minimum distance deletion of points for thinning again when the thinning point reduction rate in the first stage is less than a preset threshold. The topology preservation module is used to ensure the topological correctness of each inner and outer ring of the graph after the second stage of thinning, and to prevent self-intersection or boundary crossing.

[0017] Beneficial effects of this invention: Compared with existing technologies, this invention achieves significant beneficial effects through a phased, pipeline-like processing method of "thinning → iterative smoothing → further thinning," combined with an adaptive thinning strategy and a scale-driven tolerance dynamic estimation framework. Firstly, this method achieves a better balance between visual smoothness and geometric fidelity, significantly eliminating jagged edges and burrs while better preserving the original shape, area, and boundary orientation of the image patches. Secondly, through dynamic tolerance conversion between screen pixels and the scale, it ensures visual consistency in the generalization effect of the image patches at different display scales. Finally, by preserving topology during the simplification process, it effectively avoids topological errors such as self-intersection and hole deformation, ensuring the correctness of spatial relationships. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the overall process of the method for summarizing jagged graphic data based on remote sensing extraction according to the present invention.

[0020] Figure 2 This is a schematic diagram illustrating the processing flow of the outer ring and several inner rings of the method for summarizing jagged graphic data based on remote sensing extraction according to the present invention, including the thinning and smoothing sequence, closure, and topological constraints.

[0021] Figure 3 This is a schematic diagram of the structure of the jagged graphic data summarization system based on remote sensing extraction according to the present invention.

[0022] Figure 4 This is a schematic diagram of the tolerance calculation process of the tolerance calculation module of the remote sensing-based sawtooth graphic data summarization system of the present invention.

[0023] The diagram is labeled as follows: 100 is the summarization system for jagged graphic data extracted from remote sensing, 101 is the tolerance calculation module, 102 is the thinning module, 103 is the smoothing module, 104 is the secondary thinning module, 105 is the adaptive thinning module, and 106 is the topology preservation module. Detailed Implementation

[0024] To make the technical problems solved, the technical solutions, and the beneficial effects of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0025] like Figure 1 and 2 As shown in the figure, an embodiment of the present invention provides a method for summarizing jagged graphic data based on remote sensing extraction, which includes the following steps; S1. Tolerance dynamic estimation: Based on screen pixels, screen display scale and coordinate system, dynamically estimate the scale denominator and generate the initial tolerance T0. Convert the degree to meter unit through the center latitude in the latitude and longitude coordinate system to obtain the geometric processing threshold of the uniform scale. The aforementioned dynamic tolerance estimation also includes: ring-level tolerance adaptive adjustment: dynamically adjusting the tolerance based on the ring's external / internal attributes and point density; when the Douglas-Peucker thinning point reduction rate is insufficient, automatically using the minimum distance point deletion of Tmin=0.5×T0 as a remedy; Specifically, the dynamic estimation process is as follows: The system first obtains the bounding rectangle of the current map view, the screen resolution (DPI), and the geographic coordinate system; by inversely calculating the formula: scale denominator = (ground distance / map distance), and combining it with the screen pixel size, the scale denominator of the current view is estimated; for example, on a 96 DPI screen, the map distance corresponding to 1 pixel is approximately 0.264 millimeters. If this pixel represents 2 meters of ground, then the scale denominator is approximately 1:7,576; Degree to meter unit conversion: When the data source is a geographic coordinate system (such as WGS84 latitude and longitude coordinate system), the system will take the latitude φ of the current view center and use the formula 1 degree latitude ≈ 111,319.5 meters, 1 degree longitude ≈ 111,319.5 * cos(φ) meters to convert the degree-based tolerance to the meter-based tolerance, thereby achieving scale uniformity in cross-coordinate system processing; Ring-level tolerance adaptive adjustment: The system can be fine-tuned according to the ring's properties (outer rings usually require higher fidelity and can have slightly tighter tolerances; inner rings can have slightly looser tolerances) and vertex density (high-density regions may require stricter tolerances to preserve details); when the DP thinning rate is insufficient, the automatically triggered Tmin is a manifestation of this dynamic adjustment mechanism.

[0026] S2. First stage thinning: The boundaries of the patches extracted by intelligent interpretation of remote sensing images are initially simplified by using the Douglas-Peucker (DP) algorithm for the outer ring and each inner ring, eliminating pixel-level micro-jitter and fine burrs, reducing the number of points, and keeping the beginning and end of the closed loop consistent, so as to improve the efficiency of the subsequent iterative smoothing stage. In the first stage of thinning, if the reduction rate of a certain ring after simplification is lower than the preset threshold, an adaptive thinning backoff strategy is automatically triggered, and the minimum distance deletion method is used to perform thinning again. Its tolerance Tmin is 0.5 times the initial tolerance T0. This can more robustly remove fine serrated segments while maintaining the consistency of the beginning and end of the closed loop. Specifically, the tolerance setting for the DP algorithm is as follows: the initial tolerance T0 is dynamically calculated by step S1; For example, at a scale of 1:10,000, T0 can be set to a threshold equivalent to a distance of 2 meters from the ground. This threshold can effectively filter out pixel-level jaggedness caused by image resolution (such as 0.5 meters). Loop closure handling: Before the algorithm is executed, the system will check and ensure that the starting point and ending point coordinates of each loop are consistent; after the processing is completed, the closure of the loop will be checked again to prevent the loop from breaking due to point deletion; Adaptive thinning and backoff strategy: If the reduction rate ((original number of points - simplified number of points) / original number of points) of a ring after DP simplification is lower than the preset 30% threshold, it indicates that the jagged edges in the region are abnormally dense. At this time, the backoff strategy is automatically triggered, and the "minimum distance deletion method" is used to thin the ring again. This method traverses each point on the ring and calculates the perpendicular distance from it to the line connecting two adjacent points. If the distance is less than Tmin = 0.5 × T0, the point is deleted. This strategy can more robustly remove fine jagged segments that are difficult for the DP algorithm to handle, while maintaining the consistency of the beginning and end of the point sequence to keep the ring closed.

[0027] S3. Iterative Smoothing: The PAEK polynomial approximation exponential kernel method is used to smooth the boundaries of the thinned patches at the ring level. Through small-step iteration and tolerance control, the boundary is smoothed and overshoot or offset is prevented; it can balance smoothness and fidelity. The iterative smoothing adopts the PAEK polynomial approximation exponential kernel method, which smooths the boundary through multiple small-step iterations, controls the tolerance to prevent geometric expansion or deformation, and maintains topological stability under complex holes and multi-ring structures. Specifically, PAEK smoothing: The PAEK polynomial approximation exponential kernel method smooths the boundary through a local weighted average algorithm. A neighborhood window is defined for each boundary point, and each point in the window contributes a different weight to the new position of the point (determined by the kernel function, such as a Gaussian kernel), thereby preserving the original orientation as much as possible while smoothing. Small-step iteration control: Multiple small-step iterations are used (e.g., the number of iterations is set to 3 to 5 times) instead of a single large-scale smoothing; the step size of the moving point in each iteration is subject to tolerance constraints (usually set to a value close to or slightly smaller than T0), which can effectively prevent "overshoot" (i.e. the boundary is excessively straightened or pushed outside the true position) and "balloon effect" (the boundary expands outward as a whole). Topological stability guarantee: The smoothing process is carried out independently on each ring. When dealing with complex holes (inner rings), the system monitors the spatial relationship between the inner and outer rings in real time to ensure that the smoothed inner ring does not overflow outside the outer ring, thereby maintaining topological stability under complex structures.

[0028] S4. Second stage thinning: The smoothed patch boundaries are simplified a second time, redundant points introduced by smoothing are removed, and the boundaries are shaped to improve storage and rendering efficiency and maintain the correct topological relationship of the graphics themselves, that is, to ensure that the outer ring and holes do not cross the boundary or intersect. The specific implementation logic for maintaining the correct topological relationships of the graph is as follows: the system embeds topological checks during the simplification process; before deleting a vertex, self-intersection detection and inter-loop relationship detection are performed; if deleting the point would cause the loop to intersect itself or illegally overlap with adjacent loops (such as holes and shells), or the containment relationship would be broken, then the point is retained; this ensures that the final output patch is still geometrically "clean" and suitable for spatial querying and analysis.

[0029] In addition, the method for summarizing jagged graphic data based on remote sensing extraction also includes a degradation ring processing step: for degradation rings with few points or small areas, they are removed or merged according to a preset threshold to avoid abnormal rings affecting subsequent processing; specifically, for degradation rings with fewer than 4 points or an area smaller than the coverage area of ​​a single pixel, the system will automatically remove or merge them into adjacent rings according to a preset threshold to avoid these abnormal rings causing errors in subsequent processing.

[0030] The summarization method supports tolerance conversion between geographic coordinate systems and projected coordinate systems, solves the problem of unified control between pixels, meters, and degrees, and achieves visual consistency of the summary results of the same graphic at different scales. It realizes dynamic calculation of tolerance and drawing range, and dynamically associates the annotation content with processing parameters (such as tolerance). Only one set of parameter systems coupled with the display scale needs to be defined to adapt to different map sheets and view sizes. During the drawing or mapping process, the system automatically calculates the processing tolerance and drawing range of each geometric ring, eliminating the need for manual preset of the size of each component. This makes it more flexible and convenient to use, greatly improving the intelligence level of processing and the efficiency of engineering applications.

[0031] Preferably, the reduction rate threshold of the adaptive thinning backoff strategy is set to 30%. When the reduction rate is lower than this value, thinning is triggered again. That is, when the reduction rate is less than 30%, thinning is supplemented with a minimum distance threshold of Tmin = 0.5 × T0, which takes into account the robust noise reduction of fine jagged segments. This value of 30% is a balance point, which can effectively identify dense jagged areas, while avoiding unnecessary secondary operations on already relatively smooth boundaries.

[0032] Preferably, the map patch boundaries in step S2 are generated by intelligent interpretation of remote sensing images. The map patch boundaries include map patches generated by deep learning segmentation or traditional image segmentation, and their boundaries have the characteristics of being jagged, having many spikes, and having large local noise. After thinning in the second stage, the summarized map patch data is output for multi-scale map representation or spatial analysis, while maintaining the original shape as much as possible, that is, maintaining the geometric features of area and boundary orientation as much as possible without distortion.

[0033] like Figure 3 and 4 As shown in the figure, an embodiment of the present invention provides a summarization system for jagged graphic data extracted from remote sensing, used to implement the summarization method for jagged graphic data extracted from remote sensing. The summarization system 100 for jagged graphic data extracted from remote sensing includes a tolerance calculation module 101, a thinning module 102, a smoothing module 103, a secondary thinning module 104, an adaptive thinning module 105, and a topology preservation module 106. The tolerance calculation module 101 is used to dynamically generate an initial tolerance based on screen pixels, screen display scale, and coordinate system, and to perform ring-level tolerance adaptive adjustment. The thinning module 102 is used to intelligently interpret and extract the jagged graphic data from remote sensing images. The first stage of patch boundary thinning uses the Douglas-Peucker (DP) algorithm to simplify the outer ring and each inner ring separately. The smoothing module 103 is used to iteratively smooth the thinned patch boundaries using a polynomial approximation exponential kernel method. The second stage of thinning module 104 is used to perform a second stage of thinning on the smoothed patch boundaries, removing redundant points and reshaping the boundaries. The adaptive thinning module 105 is used to trigger minimum distance deletion for a second thinning when the point reduction rate in the first stage is insufficient to a preset threshold. The topology preservation module 106 is used to ensure the topological correctness of each inner and outer ring of the graphic after the second stage of thinning, preventing self-intersection or boundary overflow.

[0034] Specifically, the modules of the remote sensing-based jagged graphic data summarization system 100 work collaboratively: the thinning module 102 receives the original jagged patches and performs a first-stage Douglas-Peucker (DP) algorithm simplification; the smoothing module 103 receives the thinned patches and performs iterative smoothing using the PAEK polynomial approximation exponential kernel method; the secondary thinning module 104 performs a refined secondary simplification on the smoothed patches; the adaptive thinning module 105, as an enhancement component of the thinning module 102, intervenes when the reduction rate is insufficient; the tolerance calculation module 101 is the scheduling core of the entire system, providing dynamically calculated tolerance parameters for modules such as the thinning module 102 and the smoothing module 103; and the topology preservation module 106, as a quality assurance component, mainly performs final quality verification and repair on the output results after the re-thinning module 102 has worked.

[0035] The tolerance calculation module 101, thinning module 102, smoothing module 103, secondary thinning module 104, adaptive thinning module 105, and topology preservation module 106 can be integrated into a plugin of desktop GIS software (such as ArcGIS Pro or QGIS), or they can be used as a microservice architecture for cloud processing services, receiving processing requests and returning summarized patch data through a RESTful API.

[0036] In summary, the iterative smoothing method in the generalization method for jagged graphic data extracted from remote sensing in this invention significantly reduces the "pixel-like" appearance and jaggedness, resulting in more coherent boundary trends and easier interpretation of patch morphology, which is beneficial for users to quickly understand spatial structure and trends; this is particularly evident at small scales. This invention ensures that the outer rings do not intersect or cross boundaries with holes through multi-ring-level processing and topology preservation, preventing distortion and topological destruction caused by smoothing, and guaranteeing the correctness of subsequent spatial relationship determination and spatial processing. This invention specifically optimizes the boundary characteristics of patches generated through intelligent interpretation of remote sensing images, exhibiting excellent suppression capabilities for boundary jaggedness caused by pixel raster, greatly improving the clarity of multi-scale map representation and the reliability of spatial analysis.

[0037] It is understood that the above specific description of the present invention is only for illustrating the present invention and is not limited to the technical solutions described in the embodiments of the present invention. Those skilled in the art should understand that modifications or equivalent substitutions can still be made to the present invention to achieve the same technical effect; as long as the use needs are met, they are all within the protection scope of the present invention.

Claims

1. A method for summarizing jagged graphic data extracted from remote sensing, characterized in that: Includes the following steps; S1. Tolerance dynamic estimation: Based on screen pixels, screen display scale and coordinate system, dynamically estimate the scale denominator and generate the initial tolerance T0. Convert the degree to meter unit through the center latitude in the latitude and longitude coordinate system to obtain the geometric processing threshold of the uniform scale. S2. First stage thinning: The boundaries of the patches extracted by intelligent interpretation of remote sensing images are initially simplified by using the Douglas-Peucker algorithm for the outer ring and each inner ring, eliminating pixel-level micro-jitter and fine burrs, reducing the number of points, and keeping the beginning and end of the closed loop consistent. S3. Iterative Smoothing: The PAEK polynomial approximation exponential kernel method is used to smooth the boundaries of the thinned patches at the ring level. Through small-step iteration and tolerance control, the boundaries are smoothed and overshoot or offset is prevented. S4. Second stage thinning: The smoothed patch boundaries are simplified a second time, redundant points introduced by smoothing are removed, the boundaries are shaped, and the correct topological relationship of the graphic itself is maintained. In the first stage of thinning in step S2, if the reduction rate of a certain ring after simplification is lower than a preset threshold, an adaptive thinning rollback strategy is automatically triggered, and the minimum distance deletion method is used to perform thinning again, with a tolerance Tmin of 0.5 times the initial tolerance T0.

2. The method for summarizing jagged graphic data based on remote sensing extraction according to claim 1, characterized in that: The tolerance dynamic estimation in step S1 also includes: ring-level tolerance adaptive adjustment: the tolerance is dynamically adjusted according to the external / internal attributes and point density of the ring; when the Douglas-Peucker thinning point reduction rate is insufficient, the minimum distance point deletion of Tmin = 0.5 × T0 is automatically used for compensation.

3. The method for summarizing jagged graphic data based on remote sensing extraction according to claim 1, characterized in that: The iterative smoothing in step S3 smooths the boundary through multiple small-step iterations, controls the tolerance to prevent geometric expansion or deformation, and maintains topological stability in complex holes and multi-ring structures.

4. The method for summarizing jagged graphic data based on remote sensing extraction according to claim 1, characterized in that: After the second stage of thinning in step S4, topology preservation processing is performed to ensure that the outer ring and the hole do not cross the boundary or intersect, and to prevent topology damage during the smoothing or simplification process.

5. The method for summarizing jagged graphic data based on remote sensing extraction according to claim 1, characterized in that: It also includes a degenerate loop processing step: for degenerate loops with few points or small area, they are removed or merged according to a preset threshold to avoid abnormal loops affecting subsequent processing.

6. The method for summarizing jagged graphic data based on remote sensing extraction according to claim 1, characterized in that: The summarization method supports tolerance conversion between geographic coordinate systems and projected coordinate systems, solves the problem of unified control between pixels, meters, and degrees, and achieves visual consistency of the boundary of the same graphic at different scales.

7. The method for summarizing jagged graphic data based on remote sensing extraction according to claim 1, characterized in that: The threshold for the drop rate in the adaptive thinning rollback strategy is set to 30%. When the drop rate is lower than this value, thinning is performed again.

8. The method for summarizing jagged graphic data based on remote sensing extraction according to claim 1, characterized in that: The boundary of the image patch is generated by intelligent interpretation of remote sensing images. The boundary of the image patch includes image patches generated by deep learning segmentation or traditional image segmentation. The boundary of the image patch has the characteristics of being jagged, having many burrs, and having large local noise.

9. The method for summarizing jagged graphic data based on remote sensing extraction according to claim 1, characterized in that: After thinning in the second stage, the summarized patch data is output for multi-scale map representation or spatial analysis, while maintaining the original shape, that is, preserving the geometric features of area and boundary orientation without distortion.

10. A system for summarizing jagged graphic data extracted from remote sensing, characterized in that: It includes a tolerance calculation module, a thinning module, a smoothing module, a secondary thinning module, an adaptive thinning module, and a topology preservation module; The tolerance calculation module is used to dynamically generate an initial tolerance based on screen pixels, screen display scale and coordinate system, and to perform ring-level tolerance adaptive adjustment. The thinning module is used to perform the first stage of thinning on the boundaries of the patches extracted by intelligent interpretation of remote sensing images, and uses the Douglas-Peucker algorithm to simplify the outer ring and each inner ring respectively. The smoothing module is used to iteratively smooth the boundaries of the thinned patches, using a polynomial approximation exponential kernel method. The secondary thinning module is used to perform a second stage of thinning on the smoothed patch boundary, removing redundant points and reshaping the boundary. The adaptive thinning module is used to trigger the minimum distance deletion of points for thinning again when the thinning point reduction rate in the first stage is less than a preset threshold. The topology preservation module is used to ensure the topological correctness of each inner and outer ring of the graph after the second stage of thinning, and to prevent self-intersection or boundary crossing.

Images