Intelligent management and control method and system for national space planning based on GIS and cloud platform

Abstract

This invention discloses a method and system for intelligent management and control of land and space planning based on GIS and a cloud platform, belonging to the field of intelligent management and control technology for land and space planning. The method includes acquiring multi-temporal remote sensing image data and related spatial data to construct a dataset with a unified spatial benchmark; extracting changed areas and forming an initial set of changed areas; determining the boundary crossing positions within the neighborhood of boundary pixels to generate a sub-pixel-level boundary point set; forming standardized patch data through global topological association, curve reshaping, and topological repair; and then overlaying and analyzing the data with the land and space planning management boundary to output compliance judgment results and synchronize them to the cloud platform. This invention can reduce jagged boundaries, self-intersections, and elongated abnormal structures, improving the quality of changed patches and the reliability of dynamic management and control.

Description

Technical Field

[0001] This invention relates to the field of intelligent management and control technology for land and space planning, specifically to intelligent management and control methods and systems for land and space planning based on GIS and cloud platforms. Background Technology

[0002] With the increasing demand for refined and dynamic management of national land spatial planning, spatial change monitoring and control technologies based on GIS and cloud platforms have gradually become mainstream methods. In particular, these technologies rely on remote sensing data to automatically extract surface changes and generate change patches to support planning compliance analysis and supervision. However, existing technologies, in the process of converting raster change results into vector patches, typically extract contours directly based on pixel boundaries. This is easily affected by the discrete characteristics of low-to-medium resolution images, resulting in jagged boundaries in the generated patches, accompanied by geometric distortions such as self-intersection and fine burrs. This not only affects the spatial accuracy of the patches but also causes topological errors, leading to inaccurate area calculations, failed spatial analysis, and even abnormal data entry. Furthermore, existing methods lack detailed characterization of continuous boundary changes and directional consistency constraints, making it difficult to balance boundary smoothness and morphological accuracy, and thus failing to meet the application requirements of high-precision intelligent national land spatial management. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for intelligent management and control of land and space planning based on GIS and cloud platform, so as to solve the shortcomings of the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligent management and control of land spatial planning based on GIS and cloud platform, comprising: S1. Acquire multi-temporal remote sensing image data and related spatial data, perform registration, correction and unified processing on the data, and construct a dataset with a unified spatial reference. S2. Perform change detection based on the dataset, generate binary raster data of the changed regions, and form an initial set of changed regions through connectivity constraints; S3. Extract boundary pixels for the initial set of change regions, construct a continuous change trend representation in the local neighborhood of each boundary pixel, establish change transition descriptions along the boundary normal direction and tangential direction respectively, and determine the boundary crossing position inside the pixel accordingly, generating a sub-pixel level boundary point set with directional consistency. S4. Perform global topological association on the sub-pixel level boundary point set, and reconstruct the sequence based on the spatial continuity and directional consistency between boundary points to form a closed and non-intersecting boundary chain structure. S5. Based on the boundary chain structure, perform constrained curve reshaping to make the boundary stable and continuous while maintaining the overall morphological characteristics, thereby obtaining a smooth boundary curve. S6. Construct variation patches based on smooth boundary curves, and use topological consistency constraints to verify and repair the patches, eliminating self-intersections and slender abnormal structures to obtain standardized patch data. S7. Overlay and analyze the standardized map data with the land and space planning control boundary, output the compliance judgment results of the changed map, and synchronize the standardized map and the judgment results to the cloud platform for multi-source data collaborative updates and dynamic management.

[0005] Preferably, S1 includes: acquiring remote sensing image data of the same area to be managed at different time phases, and acquiring the land spatial planning boundary, administrative division, land type status and planning approval data corresponding to the area to be managed; performing radiometric correction, geometric correction and orthorectification on the remote sensing image data, and performing coordinate transformation, topological checking and field standardization processing on the relevant spatial data; registering and fusing the processed remote sensing image data and relevant spatial data according to a unified coordinate reference, a unified projection method and a unified spatial range to form a dataset with a unified spatial reference.

[0006] Preferably, S2 includes: The difference intensity of corresponding pixels in remote sensing images of each time phase in the dataset is calculated, and the continuity of change transition is constructed in the local neighborhood to obtain change response distribution data; The change threshold is determined based on the continuous spatial change trend of the change response distribution data, and the change response distribution data is segmented to generate initial binary raster data. An orientation consistency constraint is applied to the initial binary raster data, and adjacent changed units are aggregated based on connectivity analysis to form an initial set of changed regions.

[0007] Preferably, S3 includes: Identify the changed pixels adjacent to the non-changed pixels from the initial set of changed regions as boundary pixels, and construct a local neighborhood around each boundary pixel; Extract the spatial distribution sequence of change intensity within the local neighborhood, and determine the main change direction at the boundary pixel based on the spatial distribution sequence of change intensity. The stable transition interval from the non-changing region to the changing region is identified along the change transition path corresponding to the main change direction, and the boundary crossing position inside the pixel is determined within the stable transition interval to form a sub-pixel level boundary point set.

[0008] Preferably, forming a sub-pixel-level boundary point set further includes: The intensity of change along the path is sequentially rearranged according to the change transition path to form a continuous sequence of intensity of change along the path direction. A monotonicity constraint is applied to the change intensity sequence to eliminate abnormal change segments formed by local reverse fluctuations and to determine the stable transition interval; Within the stable transition interval, the search range for locations of abrupt changes in intensity is gradually reduced, and the determined spatial locations are mapped to the interior of the corresponding pixels to obtain the boundary crossing points. The boundary crossing points are screened and corrected for directional consistency to form a sub-pixel level boundary point set.

[0009] Preferably, S4 includes: Taking each boundary point in the sub-pixel level boundary point set as an object, candidate neighboring points are filtered within a limited neighborhood, and the orientation attributes of each boundary point are recorded. Based on the spatial distance and orientation attributes between candidate adjacent points, consistency matching is performed, and boundary point pairs with continuous orientation changes and small spatial distances are connected first to form an initial connection relationship. The initial connection relationships are traversed globally, and connection paths with intersection or loop anomalies are disconnected, reconnected, and have their endpoints completed to form a closed and non-intersecting boundary chain structure.

[0010] Preferably, S5 includes: Traverse the boundary points in the connection order of the boundary chain structure, and identify high-change and low-change segments based on the directional deflection between adjacent boundary points; In low-change sections, the positions of boundary points are continuously adjusted, while in high-change sections, the original distribution of boundary points is maintained to preserve the overall morphological characteristics of the changed patches. The adjusted boundary point sequence is subjected to overall consistency verification to limit the variation between adjacent boundary points, and a smooth boundary curve is constructed based on the verified boundary point sequence.

[0011] Preferably, S6 includes: The initial change pattern is generated by closing the beginning and end of the smooth boundary curve and filling the area inside the closed boundary. Perform topological consistency detection on the initial changed patches to identify self-intersections, overlapping boundaries, unclosed regions, and elongated anomalous regions, and record the corresponding anomalous locations; Based on the abnormal locations, the corresponding boundary segments are locally rearranged, adjacent to each other adjusted, and compressed and corrected. The overall consistency of the corrected changed patches is then checked to obtain standardized patch data.

[0012] Preferably, S7 includes: The standardized map patch data is spatially aligned with the land and space planning control boundary data, and a spatial mapping relationship is established between map patch identifiers and control boundary identifiers; Based on spatial mapping relationships, the standardized map data and the land spatial planning control boundary are overlaid and analyzed to extract the overlapping range, boundary distance and corresponding control zone category; Based on the control rules, the compliance judgment results of the changed patches are generated, and the standardized patch data, overlay analysis data and compliance judgment results are synchronized to the cloud platform according to a unified data structure.

[0013] This invention also provides an intelligent management and control system for land spatial planning based on GIS and a cloud platform, comprising: Data acquisition and preprocessing module: acquires multi-temporal remote sensing image data and related spatial data, performs registration, correction and unified processing on the data, and constructs a dataset with a unified spatial reference. Change detection module: Performs change detection based on the dataset, generates binary raster data of the changed regions, and forms an initial set of changed regions through connectivity constraints; Sub-pixel boundary extraction module: Extracts boundary pixels for the initial set of changing regions, constructs a continuous change trend representation in the local neighborhood of each boundary pixel, establishes a change transition description along the boundary normal direction and tangential direction respectively, and determines the boundary crossing position inside the pixel accordingly, generating a sub-pixel level boundary point set with consistent direction. Boundary chain reconstruction module: Performs global topological association on sub-pixel level boundary point set, and performs sequence reconstruction based on the spatial continuity and directional consistency between boundary points to form a closed and non-intersecting boundary chain structure; Curve reshaping module: Based on the boundary chain structure, it performs constrained curve reshaping, so that the boundary tends to be stable and continuous while maintaining the overall morphological characteristics, thereby obtaining a smooth boundary curve; The normalization module for the map features constructs varied map features based on smooth boundary curves and uses topological consistency constraints to verify and repair the map features, eliminating self-intersections and slender abnormal structures to obtain normalized map feature data. Planning control and cloud synchronization module: It overlays and analyzes standardized map data with the boundaries of land and space planning control, outputs compliance judgment results of changed map features, and synchronizes standardized map features and judgment results to the cloud platform for multi-source data collaborative updates and dynamic control.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: 1. This invention differs from existing methods that directly convert raster to vector along pixel edges. Its core lies in utilizing the physical characteristic that surface changes within a pixel are not abrupt but rather a continuous transition along spatial directions. It establishes normal transition and tangential continuity constraints within the neighborhood of the boundary pixel, determining the actual boundary crossing position within the pixel and forming a sub-pixel-level boundary point set. This avoids directly using the edges of 10-meter to 30-meter pixel squares as map boundaries, reducing jagged boundaries, boundary drift, and minor overlays caused by pixel discreteness, making the generated change boundaries closer to the actual surface change transition positions.

[0015] 2. This invention performs global topological association of sub-pixel-level boundary points based on spatial continuity and directional consistency. It then performs constrained curve reshaping and topological consistency repair on the boundary chain, eliminating self-intersections, overlapping boundaries, and fine spikes while maintaining realistic morphological features such as corners and elongated features. This method suppresses the generation of distorted structures from the geometric formation process of map features themselves, ensuring that standardized map features possess stable topological relationships during area calculation, planning boundary overlay, and cloud platform data storage. This improves the reliability of compliance judgments and the consistency of multi-source data collaborative updates. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a flowchart of the method of the present invention.

[0018] Figure 2 This is a flowchart of the system modules of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1, please refer to Figure 1 As shown in this embodiment, the intelligent management and control method for land spatial planning based on GIS and cloud platform includes: S1. Acquire multi-temporal remote sensing image data and related spatial data, perform registration, correction and unified processing on the data, and construct a dataset with a unified spatial reference.

[0021] In this embodiment, multi-temporal remote sensing image data and related spatial data corresponding to the area to be controlled during different monitoring periods are first acquired. The multi-temporal remote sensing image data includes optical remote sensing images, radar remote sensing images or UAV images of the same area collected at least in two different periods. The related spatial data includes land space planning boundary data, administrative division data, land type status data, topographic elevation data, construction land boundary data, ecological protection red line data, and existing cadastral or planning approval data.

[0022] To ensure the accuracy of subsequent change detection and patch generation, the acquired data undergoes unified preprocessing, specifically including: radiometric correction, atmospheric correction, geometric correction, and orthorectification of remote sensing images to eliminate errors caused by differences in imaging time, sensing conditions, terrain undulations, and shooting angles; spatial registration of images from different time periods to ensure that the same ground feature corresponds in a unified spatial location in images from different periods; and coordinate transformation, topological checking, and field standardization of vector spatial data to ensure that it maintains the same coordinate reference, projection method, and spatial resolution expression as the remote sensing images.

[0023] After calibration and registration, the remote sensing images are cropped according to a unified range, and a unified spatial index is established based on the boundary of the area to be managed. At the same time, metadata identifiers such as data source, acquisition time, spatial resolution, coordinate system, and processing status are assigned to various types of data.

[0024] After the above processing, a dataset is formed that covers the area to be controlled and has a unified spatial benchmark, unified data format and unified time identifier, which serves as the basic data for subsequent change area extraction, boundary refinement processing, change patch generation and planning control analysis.

[0025] S2. Perform change detection based on the dataset, generate binary raster data of the changed regions, and form an initial set of changed regions through connectivity constraints.

[0026] The pixel-level difference intensity was calculated for each temporal remote sensing image after centralized registration. Pixels at the same spatial location were used as the processing object, and relevant index features of vegetation, water bodies, and built-up land were extracted in the visible light band, near-infrared band, and normalized vegetation, water body, and built-up land bands. The difference between the feature values ​​of each time phase and the corresponding feature values ​​of the previous time phase was calculated, and the absolute difference was then weighted and summed according to data quality. The weights were determined based on image cloud cover, shadow ratio, and noise level, with higher data quality resulting in higher weights. The summarized results were uniformly stretched to the range of 0 to 1 to form pixel-level difference intensity. Subsequently, a 3x3, 5x5, or 7x7 local neighborhood was selected centered on each pixel, and the direction and magnitude of the change in difference intensity of adjacent pixels within the neighborhood were statistically analyzed. Pixel segments with continuous changes and abrupt changes of less than 0.15 were considered as effective transition segments, thus constructing a characterization of the continuity of the change transition. Isolated high-response pixels were attenuated to obtain change response distribution data.

[0027] The change threshold is determined based on the continuous spatial trend of the change response distribution data, and initial binary raster data is generated. First, grayscale distribution statistics are performed on the change response distribution data to identify the transition interval between low-response and high-response concentration areas. When a clear bimodal distribution exists, the position with the lowest response quantity and relatively stable spatial continuity between the two concentration areas is selected as the change threshold. When a clear bimodal distribution does not exist, the change threshold is determined jointly by the median response level and the local fluctuation level, where the local fluctuation level is determined by the average change in the response difference between adjacent pixels. The change threshold is limited to the range of 0.35 to 0.75; values ​​below 0.35 are taken as 0.35, and values ​​above 0.75 are taken as 0.75.

[0028] Pixels whose change response is greater than or equal to the change threshold are assigned as changed pixels, and the remaining pixels are assigned as non-changed pixels, resulting in initial binary raster data in which the change area is continuously distributed.

[0029] An orientation consistency constraint is applied to the initial binary raster data to obtain structurally continuous binary raster data. First, the positions of adjacent changing pixels and non-changing pixels in the initial binary raster data are extracted as boundary unstable regions. Then, the main extension direction is determined within the boundary unstable regions. The main extension direction is obtained by statistically analyzing the arrangement direction and continuous length of adjacent changing pixels.

[0030] For a segment of pixel change that is broken on both sides of the main extension direction, if the break interval is no more than 2 pixels and the change response at both ends of the break is no less than 80% of the change threshold, the break position is filled with a pixel change. For a hole located inside the change region, with an area of ​​no more than 4 pixels and surrounded by a majority of change pixels in the neighborhood, the hole position is corrected to a pixel change. For a small, protruding region that is deviated from the main extension direction and has a continuous width of 1 pixel, if the change response around it is less than 70% of the change threshold, it is corrected to a non-change pixel, thus obtaining a structurally continuous binary raster data.

[0031] Connectivity analysis is performed on structurally continuous binary raster data to form an initial set of variation regions. Taking the variation pixels in the structurally continuous binary raster data as the object, the spatial connectivity relationship between pixels is determined using the 8-neighborhood connectivity rule; that is, two variation pixels are considered connected when they are adjacent in the horizontal, vertical, or diagonal directions. All connected variation pixels are assigned the same connectivity identifier, forming independent variation units.

[0032] Based on spatial adjacency, adjacent variation units with a spacing of no more than one pixel and consistent boundary directions are aggregated, and isolated variation units with an area smaller than a preset area are removed. The preset area is determined according to the image spatial resolution and is 3 to 10 times the actual area of ​​a single pixel. After aggregation and removal, variation units with closed boundaries, internal continuity, and consistent spatial positions are retained to form an initial set of variation regions.

[0033] S3. Extract boundary pixels from the initial set of change regions, construct a continuous change trend representation in the local neighborhood of each boundary pixel, establish change transition descriptions along the boundary normal direction and tangential direction respectively, and determine the boundary crossing position inside the pixel accordingly, generating a sub-pixel level boundary point set with directional consistency.

[0034] Boundary pixels are extracted from the initial set of changed regions, and local neighborhoods are constructed to obtain the spatial distribution sequence of change intensity. Boundary identification is performed on each changed unit in the initial set of changed regions, and changed pixels adjacent to non-changed pixels are marked as boundary pixels. For each boundary pixel, a local neighborhood of a fixed scale is constructed with it as the center. The neighborhood scale is determined according to the image spatial resolution. When the resolution is 10 meters to 30 meters, the neighborhood size is selected as a 5x5 or 7x7 pixel range.

[0035] Further, the change intensity values ​​corresponding to each pixel in the neighborhood are extracted. The change intensity is derived from the aforementioned change response distribution data and arranged sequentially according to the positional relationship between the pixels and the center pixel, such as sorting by row priority or polar angle order, thus forming a spatial distribution sequence of change intensity with spatial positional correspondence. At the same time, the change intensity in the neighborhood is normalized and linearly mapped to the interval between 0 and 1 to eliminate the influence of amplitude differences between different regions and provide a unified scale basis for subsequent direction determination.

[0036] The main change direction at the boundary pixel is determined based on the spatial distribution sequence, and the change transition path and continuous change constraints are established.

[0037] For the local neighborhood of each boundary pixel, calculate the magnitude of change in intensity in different directions within the neighborhood, divide the neighborhood into several directional intervals, for example, divide a directional interval every 15 degrees within the range of 0 degrees to 180 degrees, and count the cumulative change in intensity in each directional interval; select the direction with the largest difference in intensity as the main change direction, which reflects the main transition direction from the non-changing region to the changing region.

[0038] Based on the main change direction, a direction perpendicular to it is constructed as the change transition path direction, and a pixel sequence passing through the center pixel is selected in this direction as the change transition path.

[0039] Meanwhile, the variation trend of adjacent boundary pixels is extracted along the boundary extension direction, and the difference in the main variation direction of adjacent boundary pixels is controlled within a range of no more than 30 degrees to form a continuous variation constraint, so that the change transition path direction is consistent with the boundary extension direction and avoids abrupt changes in local direction.

[0040] Determine the boundary crossing points inside the pixel on the changing transition path.

[0041] First, the intensity of change on the path is reordered using the changing transition path as a constraint, so that it is arranged in spatial order from one end of the path to the other to form a continuous change sequence; the path length is determined according to the neighborhood size, usually 5 to 7 pixels.

[0042] Monotonicity analysis is performed on continuously changing sequences. Starting from the path start point, the difference in change intensity between adjacent positions is compared point by point. When three or more consecutive adjacent differences maintain the same trend, the interval is identified as a stable transition interval. Local fluctuation segments with repeated changes in direction and amplitude less than 0.1 are regarded as noise and removed.

[0043] Within the stable transition range, a subdivision positioning process is performed, dividing the range into multiple equidistant sub-ranges. For example, the distance between two adjacent pixels is divided into four equal parts. By performing linear interpolation calculation on the change intensity within each sub-range, the range of the change range is gradually reduced. When the difference in change intensity between adjacent sub-ranges reaches a preset abrupt change threshold, the subdivision stops. The abrupt change threshold ranges from 0.15 to 0.25.

[0044] The spatial coordinates corresponding to the mutation location are mapped to the interior of the boundary pixel, and their positions are recorded in floating-point form as boundary crossing points.

[0045] Boundary crossing points are screened and corrected for directional consistency to form a sub-pixel level boundary point set. All boundary crossing points are initially arranged according to their corresponding boundary pixels, and the directional difference between each boundary crossing point and its neighboring points is calculated. The directional difference is obtained by comparing the angle between the direction of the line connecting the two points and the main change direction of the corresponding boundary pixel. When the angle is greater than 45 degrees, the boundary crossing point is determined to deviate from the main change direction and is marked as an anomaly.

[0046] For outliers, first check the directional distribution of other boundary crossing points in their neighborhood. If more than half of the boundary crossing points in the neighborhood have the same direction, adjust the position of the outlier along the main direction of the neighborhood to realign it. If it cannot be adjusted, remove the outlier directly.

[0047] Subsequently, the continuity of the remaining boundary crossing points is checked. When the distance between adjacent points exceeds a preset threshold, interpolation is performed to supplement the distance. This distance threshold is set to 1 to 2 pixels in length. After screening and correction, a set of boundary points with continuous spatial distribution, consistent orientation, and sub-pixel accuracy is obtained, which is used for subsequent boundary chain construction and curve reshaping.

[0048] S4. Perform global topological association on the sub-pixel level boundary point set, and reconstruct the sequence based on the spatial continuity and directional consistency between boundary points to form a closed and non-intersecting boundary chain structure.

[0049] Spatial adjacency relationships are established for the sub-pixel level boundary point set, and candidate neighboring points are selected. A limited neighborhood is constructed with each sub-pixel level boundary point as the center. The radius of the neighborhood is set to the actual distance corresponding to 1 to 3 pixels based on the image spatial resolution. Other boundary points are searched within this neighborhood, and boundary points whose distance does not exceed the radius of the neighborhood are selected as candidate neighboring points.

[0050] The spatial distance between boundary points is obtained by calculating the horizontal and vertical distances between the two points in the plane coordinate system and then taking the combined distance.

[0051] For each boundary point, its direction attribute is recorded. The direction attribute is derived from the main change direction determined in the previous step, expressed in the form of angles and uniformly converted to the range of 0 to 180 degrees.

[0052] For candidate neighboring points, their directional differences are further calculated, that is, the angle difference between the directional attributes of two boundary points is compared. When the directional difference is less than 30 degrees, it is retained as a valid candidate neighboring point; otherwise, it is eliminated, thus forming a set of candidate neighboring points that includes spatial adjacency relationships and directional constraints.

[0053] Candidate neighboring points are matched for consistency based on directional attributes and spatial distance to form initial connection relationships. For each boundary point, the comprehensive matching degree is calculated from its set of candidate neighboring points. The matching degree is determined by both spatial distance and directional difference, with smaller spatial distances and smaller directional differences indicating a higher matching degree. Specifically, spatial distance is normalized to the range of 0 to 1, and directional difference is normalized to a maximum allowable difference of 30 degrees. Then, a weighted sum is applied, with spatial distance accounting for 0.6 and directional difference accounting for 0.4. For each boundary point, the candidate point with the highest matching degree and uniqueness is selected as the connection object, and a connection relationship is established. If multiple candidate points have similar matching degrees and differences less than 0.05, the point with the smaller spatial distance is prioritized for connection.

[0054] By using the above method, connections are established for all boundary points one by one, forming an initial set of connections consisting of multiple boundary point pairs.

[0055] The initial connection relationships are traversed globally to eliminate intersection and loop anomalies. Each established connection relationship is checked to determine whether any two connecting line segments intersect in space. Intersection determination is achieved by comparing the relative positions of the endpoints of the two line segments. When two line segments cross each other on the plane, they are considered to have intersected.

[0056] For detected intersections, compare the matching degree of the connections involved in the intersection, retain the connection with the higher matching degree, disconnect the connection with the lower matching degree, and reconnect by selecting the second-best matching point from the candidate neighbor set of the disconnection point.

[0057] For loop anomalies, i.e., when the connection path repeatedly passes through the same boundary point or forms a closed loop with a length of less than 3 points, first identify all connections in the loop path, then disconnect them in order of matching degree from low to high, and re-perform adjacency matching to form a new connection path. Through multiple traversals and corrections, all connection relationships gradually tend towards a structure without intersections and loops.

[0058] The corrected connection paths are then subjected to closure checks and endpoint completion to form a complete boundary chain structure. First, each connection path is traversed, and the number of connections for each boundary point is counted. A boundary point is considered an endpoint if it is connected to only one path, and a normal connection point is considered if it is connected to two paths.

[0059] For endpoints, firstly, search for unconnected candidate boundary points within their neighborhood. If there are candidate points that meet the conditions of a distance of no more than 2 pixels and an directional difference of less than 30 degrees, then establish supplementary connection relationships. If there are no suitable candidate points, then interpolate along the current path extension direction to generate new boundary points. The interpolation interval is set to 0.5 to 1 times the average interval of the original boundary points, and the newly generated boundary points are included in the connection path.

[0060] After completing the endpoint completion, a closure check is performed on each path. When the distance between the starting point and the ending point of the path is less than a preset threshold, they are directly connected to form a closed structure. The threshold value is 1 pixel in length. When the distance is greater than the threshold, the path is gradually extended and intermediate connection points are introduced to achieve closure.

[0061] Ultimately, all boundary points are connected sequentially in spatial order to form a boundary chain structure that is closed at both ends and has no intersection relationship, providing a foundation for subsequent curve reshaping.

[0062] S5. Based on the boundary chain structure, perform constrained curve reshaping to make the boundary stable and continuous while maintaining the overall morphological characteristics, thereby obtaining a smooth boundary curve.

[0063] The boundary points in the boundary chain structure are traversed sequentially, and the degree of change between adjacent boundary points is calculated. Taking the boundary points that are already connected in spatial order in the boundary chain structure as the processing objects, starting from any closed boundary point, the spatial coordinates of the previous boundary point, the current boundary point, and the next boundary point are read in sequence along the same direction to form a continuous three-point combination.

[0064] The directional deflection at the current boundary point is determined based on the direction of the line connecting the previous boundary point to the current boundary point, and the direction of the line connecting the current boundary point to the next boundary point. The larger the angle between the two connecting directions, the higher the degree of change at that location. To avoid misjudgment caused by a single outlier, two boundary points can be selected before and after the current boundary point. The degree of change of multiple consecutive three-point combinations within this range is statistically analyzed, and the average result is taken as the degree of change of the current boundary point. The threshold for judging the degree of change is determined based on the image spatial resolution and the average spacing between boundary points. When the image resolution is 10 meters to 30 meters, the threshold can be set to 25 degrees to 45 degrees. Locations with a value greater than this threshold are marked as high-change segments, and locations with a continuous length of no less than three boundary points and a value less than or equal to this threshold are marked as low-change segments.

[0065] High-change sections typically correspond to corners of feature boundaries, zigzag turns, or abrupt changes in the actual edge, while low-change sections typically correspond to areas with straight and extended boundaries or areas with gentle curves.

[0066] In low-change sections, continuity constraints are applied to gradually adjust the positions of boundary points, while maintaining the original distribution of boundary points in high-change sections. For low-change sections, boundary points to be adjusted are selected sequentially according to their spatial order, and two to three boundary points before and after each adjusted boundary point are used as reference points. The average position of the reference point coordinates and the direction of the line connecting the reference points are calculated, thereby determining the smooth reference position of the boundary point to be adjusted.

[0067] During adjustment, the boundary point to be adjusted is not moved directly to the smoothing reference position, but is moved in a step-by-step approximation manner. The distance of each movement is limited to 20% to 50% of the distance between the boundary point and the smoothing reference position. After each movement, it is checked whether the distance between the boundary point and the adjacent boundary point is kept within the range of 0.5 to 1.5 times the average boundary point spacing. If it exceeds this range, the movement distance is reduced.

[0068] For high-change sections, the original boundary point coordinates are retained, and the boundary points at both ends of the high-change section are used as constraint endpoints, ensuring that adjacent low-change sections do not exceed the local directional range formed by the constraint endpoints during adjustment. In this way, minor jitters in straight or gently curved areas are weakened, while the actual corners and abrupt changes in boundary morphology are not flattened.

[0069] The adjusted boundary point sequence is subjected to overall consistency verification, and local mutations and elongated abnormal structures are eliminated. First, the distance changes between adjacent boundary points are checked point by point along the closed boundary chain. When the distance between a boundary point and its two adjacent boundary points is significantly greater than the average distance between boundary points, the boundary point is marked as a local mutation point; the criterion for "significantly greater" is that it exceeds 2 to 3 times the average distance between boundary points.

[0070] For local abrupt change points, if the degree of change between the point and its adjacent boundary points is less than the threshold for high-change segments, the point is adjusted to the vicinity of the line connecting the adjacent boundary points. If the degree of change reaches the threshold for high-change segments, the point is retained, but its connection length with adjacent points is limited. Next, elongated anomalous structures are examined. Elongated anomalous structures are defined as continuous boundary segments extending outward from the main boundary, with a width not exceeding 1 to 2 times the average distance between boundary points and a length exceeding 3 times the width. For these structures, it is determined whether their internal change response is lower than the average change response of the adjacent main boundary region. If it is lower than 70% of the average change response of the adjacent main boundary region, the boundary segment is compressed back towards the main boundary direction; if it is not lower than this proportion, it is retained as a true elongated change portion. After completing the above processing, it is confirmed again that the boundary point sequence does not intersect and that the directional changes between adjacent boundary points are continuous.

[0071] A continuous boundary curve is constructed based on the verified boundary point sequence. The verified boundary point sequence is then segmented and connected according to the original order of the boundary points. For low-change segments, a segmented smooth connection method is used to ensure the continuity of the transition direction between adjacent boundary points with the direction of the preceding and following boundary points. For high-change segments, a polyline connection method is used to preserve boundary corners and avoid morphological shifts caused by curve processing.

[0072] During the construction process, each continuous boundary curve is controlled by the corresponding boundary point. The curve must not deviate from the original boundary chain by more than a preset distance. The preset distance is set to the actual distance corresponding to 0.3 to 0.8 pixels according to the image resolution. When the curve deviates by more than this distance, the curve segment is readjusted to the vicinity of the original boundary chain.

[0073] After completing the connection of all segments, the connection points at the beginning and end are closed to form a continuous transition between the starting and ending boundary points, and the degree of change of the closing position is checked to see if it is consistent with the adjacent segments.

[0074] After the above processing, a continuous boundary curve is obtained. This continuous boundary curve is consistent with the boundary chain structure in the overall spatial range, reduces sawtooth fluctuations in the local morphology, and provides a boundary basis for the construction of subsequent change patches.

[0075] S6. Construct variation patches based on smooth boundary curves, and use topological consistency constraints to verify and repair the patches, eliminating self-intersections and slender abnormal structures to obtain standardized patch data.

[0076] The initial variation patch is generated by closing the beginning and end of the smooth boundary curve and filling the internal area. Taking the already formed smooth boundary curve as the processing object, the spatial coordinates of the start and end points of the curve are read first, and it is determined whether the distance between them is less than the closure tolerance. The closure tolerance is set according to the image spatial resolution as the actual distance corresponding to 0.5 to 1 pixel. When the distance between the start and end points is not greater than the closure tolerance, the start and end points are directly connected to form a closed boundary. When the distance between the start and end points is greater than the closure tolerance, adjacent boundary segments are searched inward along the local extension direction of the two ends. Connecting segments are added according to the principle of the shortest distance and direction difference not exceeding 20 degrees until a closed boundary is formed.

[0077] For smooth boundary curves with inner rings, they are first grouped according to the inclusion relationship between the outer and inner boundaries. The outer boundaries are recorded in clockwise order, and the inner boundaries are recorded in counterclockwise order to distinguish between filled areas and void areas.

[0078] After the beginning and end are closed, the internal region of the boundary is determined by scanning and filling. The method is to cross the boundary row by row or column by column within the range enclosed by the smooth boundary curve, record the intersection point of each scan line with the boundary, and determine the region between pairs of intersection points as the internal region of the patch. When there is an inner boundary, the range enclosed by the inner boundary is not filled.

[0079] After the filling is completed, the boundary and its internal area are uniformly transformed into planar geometric objects and assigned a unique identifier to obtain the initial changed patch.

[0080] Perform topological consistency checks on the initial changed patches to identify self-intersecting, overlapping, and unclosed regions, and record the locations of anomalies.

[0081] The outer and inner boundaries of the initial changed patch are divided into sequentially connected boundary segments. Each segment is checked to see if there is an intersection between any two non-adjacent boundary segments. The intersection is determined by comparing the relative positions of the endpoints of the two boundary segments in the direction of each other's extension. When the two boundary segments cross each other in the plane and the intersection point is not located at the common endpoint, it is considered a self-intersection, and the boundary segment number and spatial coordinates corresponding to the intersection point are marked as abnormal positions.

[0082] The identification of boundary overlap adopts the method of matching the overlap length. The boundary segments with similar directions and a distance of less than 0.2 pixels are matched. When two boundary segments overlap in position within a continuous interval and the overlap length reaches more than 50% of the average length of the boundary segments, it is determined to be a boundary overlap, and the start and end positions of the overlap are recorded.

[0083] The identification of unclosed regions first checks the connection relationship between the beginning and end of each boundary segment in each boundary ring. When there is an endpoint that is not connected by an adjacent boundary segment, or when the distance between the beginning and end points is greater than the closure tolerance, it is determined to be an unclosed region, and the positions of the two ends of the gap are recorded.

[0084] After the detection is completed, the abnormal locations are classified and saved according to the abnormality type to provide a basis for subsequent boundary reconstruction.

[0085] Based on the location of the anomaly, the corresponding boundary segments are reconstructed. Self-intersecting structures are eliminated by rearranging local boundaries and adjusting adjacency relationships. At the same time, slender anomaly regions are compressed and corrected to meet the requirements of spatial continuity.

[0086] For self-intersection anomalies, first extract 2 to 4 boundary segments before and after the intersection point as the reconstruction range. Compare the connection order and directional continuity of each boundary segment, and select the set of boundary segments that makes the boundary direction smoothest and the total connection length shortest. Reorder the original cross connections and change them to non-cross connections. After the reordering is completed, check whether new local loops are formed within the reconstruction range. If a small loop with a length of less than 3 boundary segments appears, delete the small loop and merge its boundary into the main boundary. For boundary overlap anomalies, take the overlap interval as the center and extend 1 to 2 boundary point spacings before and after the boundary as the adjustment interval. Merge two overlapping boundary segments into one boundary segment according to the inner and outer adjacency relationship, and retain the boundary position located outside the patch. If the overlapping boundary segments come from the outer boundary and the inner boundary respectively, prioritize keeping the boundary segment with the correct inclusion relationship, and adjust the other boundary segment inward or outward according to the principle of minimum displacement.

[0087] For unclosed areas, first determine the directional difference and spacing between the two ends of the gap. When the directional difference does not exceed 25 degrees and the spacing does not exceed the actual distance of 2 pixels, directly supplement and connect the boundary segments. When the directional difference exceeds 25 degrees or the spacing exceeds the above range, insert transition boundary points between the two ends of the gap. The number of transition boundary points is determined according to the ratio of the gap length to the average boundary point spacing, so that the spacing between adjacent transition boundary points is kept within the range of 0.8 to 1.2 times the average boundary point spacing.

[0088] For elongated anomalous regions, first calculate the maximum length in the main extension direction and the average width in the vertical direction. A region is considered elongated when the maximum length is 3 to 8 times the average width and the region tapers at its tail. Further compare the change response and boundary continuity between this region and adjacent main regions. If the average change response of the elongated anomalous region is less than 70% of the average change response of adjacent main regions, or if the effective width at the connection point between this region and the main region is less than 1 times the average boundary point spacing, then perform compression correction on this region along the normal direction of the main boundary. Gradually shrink the elongated protrusion until the maximum length does not exceed 2 times the average width, and the width at the connection point is restored to more than 1 times the average boundary point spacing. After completing the above processing, refill the boundary interior area to obtain the corrected variation patch.

[0089] The corrected changed patches undergo an overall consistency check to ensure boundary continuity, structural stability, and compliance with preset topological rules, resulting in standardized patch data. The overall consistency check comprises three parts: geometric continuity check, area stability check, and topological rule check. In the geometric continuity check, the connection between adjacent boundary segments is examined segment by segment along the corrected changed patch boundary to confirm the absence of breakpoints, duplicate points, and reverse reversals. When the directional change of adjacent boundary segments exceeds 45 degrees at non-corner locations, this location is marked as a discontinuity and corrected again using the aforementioned local boundary rearrangement method.

[0090] In the area stability verification, the area difference between the initial changed patch and the corrected changed patch is compared. The area difference is judged according to the ratio of the difference between the two areas to the area of ​​the initial changed patch. When the area difference does not exceed 10%, it is considered that the correction process has not changed the main range of the changed patch. When the area difference exceeds 10%, the concentrated area of ​​area change is checked back. If the change is concentrated near the recorded abnormal position, the correction result is retained. If the change is concentrated in the non-abnormal position, the corresponding boundary segment is restored to the state before correction and then verified again.

[0091] In the topology rule verification, it is confirmed that each changed patch satisfies the following conditions: closed boundary, no self-intersection within the same patch, no unexpected overlap between different patches, and the inner boundary is completely located inside the corresponding outer boundary and does not contact the outer boundary. At the same time, it is checked whether the slender abnormal regions have been eliminated. The judgment criterion is that the length of any protruding region within the patch does not exceed 2 to 3 times its average width.

[0092] After all verifications are passed, the boundary coordinates, inner boundary information, area, perimeter, anomaly correction records, and corresponding identifiers of the changed patches are uniformly organized to form standardized patch data, which will be used for subsequent compliance judgment and synchronous processing on the cloud platform.

[0093] S7. Overlay and analyze the standardized map data with the land and space planning control boundary, output the compliance judgment results of the changed map, and synchronize the standardized map and the judgment results to the cloud platform for multi-source data collaborative updates and dynamic management.

[0094] The standardized map patch data is spatially aligned with the pre-processed land spatial planning control boundary data, and a spatial mapping relationship is established. The standardized map patch data includes map patch boundary coordinates, inner boundary information, area, perimeter, change time, change type identifier, and anomaly correction records; the land spatial planning control boundary data includes isal boundary data such as ecological protection red lines, permanent basic farmland, urban development boundaries, construction land control boundaries, and land use control zoning boundaries.

[0095] During processing, the coordinate reference, projection method, and spatial resolution of the standardized patch data and control boundary data are first read. When the coordinate references of the two are inconsistent, one type of data is converted to a unified coordinate reference. When the density of boundary points differs significantly, the boundary points are densified or thinned to keep the distance between adjacent boundary points within 0.5 to 2 times the image spatial resolution. Subsequently, using the bounding rectangle of the standardized patch as the search range, control boundaries that intersect or are adjacent to its spatial range are extracted, and a mapping table between patch identifiers and control boundary identifiers is established. The mapping table includes at least a unique patch identifier, control boundary category, control boundary identifier, initial spatial relationship state, and data time identifier, which are used for subsequent overlay calculations.

[0096] Based on spatial mapping relationships, the standardized map features and control boundaries are overlaid to identify the spatial relationships between the map features and different control zones, and the overlap range and location attributes are extracted. For each set of map features and control boundaries in the mapping table, the existence of an intersection between the bounding rectangle of the standardized map feature and the bounding rectangle of the control boundary is first compared; if there is no intersection, it is marked as spatially separated; if there is an intersection, the inclusion, intersection, overlap, or adjacency relationships between the standardized map feature boundary and the control boundary are further calculated.

[0097] In the overlay calculation, the area jointly covered by the standardized map patch and the control boundary is extracted as the overlapping range, and the area, perimeter, center location, administrative region, corresponding control zone category, and boundary distance information of the overlapping range are recorded. The boundary distance information is obtained by measuring the shortest distance from the map patch boundary point to the nearest control boundary line, and the distance unit is meters. To avoid misjudgment due to minor boundary deviations, a spatial tolerance can be set, which is 0.2 to 0.5 times the image spatial resolution. When the distance between the map patch boundary and the control boundary does not exceed the spatial tolerance, it can be considered as a boundary contact state, and not directly identified as overlay.

[0098] Based on preset control rules, spatial relationships are determined, compliance results are generated for each changed feature, and the determination category is identified. The preset control rules consist of control boundary categories, permitted change types, prohibited change types, area control conditions, and location control conditions.

[0099] For example, when a changed map patch overlaps with an ecological protection red line, and the change type is new construction, bare land expansion, or surface hardening, it is judged as suspected non-compliance; when a changed map patch is located within an urban development boundary, and the change type matches the permitted construction use, it is judged as compliance pending verification; when a changed map patch overlaps with permanent basic farmland, and the overlapping area reaches more than 5% of the map patch area or more than 100 square meters, it is judged as key inspection area.

[0100] For map patches that only touch the control boundary but do not form an effective overlap, they are marked as boundary proximity. For map patches that involve multiple control zones, they are judged according to the priority order of ecological protection red line, permanent basic farmland, urban development boundary, and other use control zones, and the overlap records corresponding to each control zone are retained. The compliance result includes at least five judgment categories: compliant, pending verification, suspected non-compliance, key inspection, and boundary proximity, and is associated with the unique identifier of the map patch.

[0101] Standardized map data and compliance results are packaged according to a unified data structure and uploaded to the cloud platform to achieve collaborative updates of multi-source data and dynamic applications for management and control business.

[0102] The unified data structure includes spatial geometric data, map feature attribute data, overlay analysis data, compliance result data, and processing log data. Spatial geometric data stores the outer boundary coordinates, inner boundary coordinates, and spatial reference information of standardized map features; map feature attribute data stores area, perimeter, change time, change type, and data source; overlay analysis data stores the control boundary category, overlapping range, overlapping area, boundary distance, and administrative region; compliance result data stores the judgment category, judgment basis, relevant control rules, and verification status; and processing log data stores the data processing time, version number, and anomaly correction records. Before uploading, a version identifier is created for each map feature, formed by combining the monitoring batch, administrative region code, and unique map feature identifier.

[0103] After uploading, the cloud platform determines whether it is a newly added, updated, or historical map feature based on the version identifier, and writes the corresponding data into the spatial data table and business results table. This ensures that remote sensing change data, planning and control boundary data, and compliance results are kept updated, providing a data foundation for map display, conditional retrieval, statistical summarization, and control and disposal.

[0104] Example 2, to verify the improvement effect of the proposed method on the geometric quality and control judgment results of changed patches under medium and low resolution remote sensing image conditions, based on the method described in Example 1, selected a rural-urban transition zone in a certain city as the test area. This area includes a construction land expansion area, a farmland edge area, a road construction area, and an area adjacent to the ecological control boundary, with a test area of ​​48 square kilometers. The test data used two phases of multispectral remote sensing images of the same area, the first phase was acquired in June 2023, and the second phase was acquired in June 2024, with a spatial resolution of 10 meters; at the same time, the land spatial planning control boundary data of the area was also introduced, including the ecological protection red line, the permanent basic farmland boundary, the urban development boundary, and the construction land control boundary.

[0105] Various types of data are processed according to steps S1 to S7 in Example 1 to form standardized map patch data and compliance results. Example 1 has already described the complete processing flow of the method of this application, including the acquisition of change response distribution data, the formation of the initial change area set, the generation of sub-pixel level boundary point set, the construction of boundary chain structure, the generation of smooth boundary curve, the formation of standardized map patch data, and the process of superimposing and determining with the land and space planning control boundary, which can be used as the processing basis for this example.

[0106] In this embodiment, for comparative verification, two processing methods are set up: Comparative Example 1 and the method of this application. Comparative Example 1 adopts the conventional raster-to-vector method, that is, the binary raster data after change detection is directly subjected to boundary tracking, and the pixel boundaries are converted into vector patches. This method only performs conventional area filtering and simple boundary smoothing, without performing sub-pixel-level boundary crossing point localization, direction consistency screening, or constrained curve reshaping based on the boundary chain structure. The method of this application is executed according to S2 to S7 of Example 1. In particular, after the initial set of change regions, it continues to generate a set of sub-pixel-level boundary points with direction consistency, and forms a closed and non-intersecting boundary chain structure through global topological association. Then, constrained curve reshaping and topological consistency repair are performed on the boundary chain structure, and finally, normalized patch data is obtained.

[0107] This embodiment uses manually interpreted map features as reference data. The manually interpreted map features are jointly determined by two technicians with remote sensing interpretation experience, based on the same temporal image, high-resolution auxiliary image, and planned boundary data. When the two technicians have different opinions on the map feature boundaries, the boundary jointly confirmed by both is used as the reference boundary. Evaluation indicators include the number of map features, the number of valid map features, the number of invalid geometric features, the average boundary deviation, the relative area deviation, the number of slender and elongated anomalous structures, the number of consistent compliance judgments, and the number of failed data entry.

[0108] Among them, invalid geometry refers to map features that have self-intersections, are not closed, have repeated boundaries, or have incorrect inner and outer loop relationships; average boundary deviation refers to the average distance from the boundary point of the map feature to be evaluated to the nearest position of the reference boundary; relative area deviation refers to the ratio of the difference between the area of ​​the map feature to be evaluated and the area of ​​the reference map feature to the area of ​​the reference map feature; slender abnormal structures refer to protruding areas whose length is more than 3 times the average width and whose width at the connection with the main change area is less than the average boundary point spacing; the number of compliance judgments consistent refers to the number of map features consistent with the results of manual verification and control; the number of failed data entry refers to the number of map features that cannot be written to the spatial data table due to topological errors, invalid geometry, or inability to associate fields.

[0109] Table 1. Basic Information on Experimental Data project Data content test area 48 square kilometers Remote sensing image phase June 2023 and June 2024 Image spatial resolution 10 meters Involving control boundaries Ecological protection red line, permanent basic farmland, urban development boundary, and construction land control boundary Number of manual reference patches 126 Main types of change New construction, expansion of bare land, road construction, changes in farmland edges After processing according to Comparative Example 1, a total of 184 changed patches were obtained, of which 143 were valid patches and 41 were invalid geometric patches. After comparison with manually referenced patches, 58 of these patches were identified as unstable patches caused by jagged boundaries, fragmented connectivity, or elongated spikes. Comparative Example 1 had an average boundary deviation of 6.8 meters, an average relative area deviation of 13.6%, and 37 elongated anomalous structures. During spatial data import, 19 patches failed to import due to self-intersection or loop structure errors, requiring manual repair before entering subsequent control and judgment. After processing using the method in this application, a total of 132 changed patches were obtained, of which 128 were valid patches and 4 were invalid geometric patches; the average boundary deviation was 3.1 meters, the average relative area deviation was 5.4%, and 6 elongated anomalous structures were found. One patch failed to import. These results indicate that the standardized patch data generated by the method in this application is closer in quantity to the manually referenced patches, and has better topological validity and boundary stability.

[0110] Table 2 Comparison results of image patch generation quality Evaluation Project Comparative Example 1 This application method Output the number of changed patches 184 132 Number of valid patches 143 128 Number of invalid geometric patches 41 4 Mean boundary deviation 6.8 meters 3.1 meters Average relative deviation of area 13.6% 5.4% Number of slender anomalous structures 37 places 6 places Number of failed inbound shipments 19 1 Furthermore, the control judgment results were verified. Manual verification showed that among the 126 reference patches, 61 were compliant, 18 were adjacent to the boundary, 21 were pending verification, 17 were suspected non-compliant, and 9 were under key inspection. In Comparative Example 1, due to some patches having jagged boundaries and local burrs, when superimposed with ecological protection red lines and permanent basic farmland boundaries, small overlapping areas were easily formed, causing some adjacent patches to be misjudged as suspected non-compliant. Simultaneously, some areas that should have been continuous were fragmented into multiple patches, resulting in multiple control records for the same change event. Statistically, the number of patches in Comparative Example 1 consistent with the manual verification results was 101, with a consistency rate of 80.2%. The method in this application, through sub-pixel-level boundary crossing point positioning and constrained curve reshaping, makes the patch boundaries closer to the actual change transition positions, and eliminates self-intersection and elongated anomalous structures through topological consistency repair, making the overlapping area and boundary distance more stable during the superposition judgment process. According to statistics, the number of image patches that are consistent with the method of this application and the results of manual verification is 118, with a consistency rate of 93.7%.

[0111] Table 3 Comparison of Control and Management Judgment Results Determine the category Number of manual verifications Comparative Example 1: Consistent Quantity The number of methods consistent with this application Compliance 61 53 59 Border proximity 18 pcs 10 16 Pending verification 21 17 20 Suspected non-compliance 17 14 16 Key Verification 9 7 7 total 126 101 118 To further illustrate the impact of this application's method on boundary-adjacent scenarios, 12 altered land parcels near the permanent basic farmland boundary were selected for individual verification. These parcels are close to the control boundary, and if the parcel boundaries have jagged or burr-like features, small areas of false overlap can easily occur. In Comparative Example 1, 5 parcels had boundary burrs that crossed the permanent basic farmland boundary, forming overlapping areas ranging from 42 square meters to 186 square meters, of which 3 parcels were mistakenly identified as key areas for verification. After processing with this application's method, the boundaries of parcels in the same location were screened for directional consistency and curve reshaping, and elongated anomalous structures were compressed and corrected. Only 1 of the 12 parcels still had valid overlap, and this overlap was manually verified as genuine occupation. The remaining parcels were marked as boundary-adjacent or pending verification. This result demonstrates that in areas adjacent to the land spatial planning control boundary, this application's method can reduce false overlap caused by geometric distortions.

[0112] Table 4 Verification Results of Boundary Proximity Scenarios project Comparative Example 1 This application method Number of test patches near the boundary 12 12 Number of accidentally pressed cover patches 5 0 Number of items mistakenly identified as key targets for inspection 3 0 Number of real overlapping patches identified 1 1 Average boundary distance deviation 4.9 meters 1.8 meters During synchronous verification on the cloud platform, the data generated by the two methods were written to the same spatial data table and business result table, respectively. Comparative Example 1, due to the large number of invalid geometric features, required manual repair or skipping before writing, resulting in some features failing to form complete version records. The standardized feature data generated by the method in this application includes boundary coordinates, inner boundary information, area, perimeter, anomaly correction records, compliance results, and version identifiers. The writing process maintains a one-to-one correspondence between spatial geometric data and business judgment data. Statistics show that the initial write success rate for Comparative Example 1 was 89.7%, while the initial write success rate for the method in this application was 99.2%. In Comparative Example 1, 14 business judgment records needed to be re-associated due to changes in identifiers after feature splitting or geometric repair, while in the method in this application, 1 record required manual verification.

[0113] Table 5 Comparison of cloud platform synchronization effects Evaluation Project Comparative Example 1 This application method First write success rate 89.7% 99.2% Number of patches requiring manual repair 19 1 Number of business records that need to be re-linked 14 articles 1 item Image version record completeness rate 88.9% 98.5% As demonstrated by the above embodiments, under low-to-medium resolution remote sensing imagery conditions, the method of this application, by determining the boundary crossing positions within pixels before vectorizing the changing region, forms a sub-pixel-level boundary point set. Combined with directional consistency, global topological association, constrained curve reshaping, and topological consistency repair, this reduces jagged boundaries, self-intersections, and elongated anomalous structures generated by direct raster-to-vector conversion, making the standardized patch data more suitable for overlay analysis with land spatial planning and control boundaries. Experimental data shows that the method of this application achieves quantifiable improvements in patch geometric legitimacy, area stability, consistency of control judgments, and the integrity of cloud platform synchronization, and can be used to support the beneficial effects of the technical solution described in this application.

[0114] Example 3, please refer to Figure 2 As shown in this embodiment, the intelligent management and control system for land spatial planning based on GIS and cloud platform includes: Data acquisition and preprocessing module: acquires multi-temporal remote sensing image data and related spatial data, performs registration, correction and unified processing on the data, and constructs a dataset with a unified spatial reference. Change detection module: Performs change detection based on the dataset, generates binary raster data of the changed regions, and forms an initial set of changed regions through connectivity constraints; Sub-pixel boundary extraction module: Extracts boundary pixels for the initial set of changing regions, constructs a continuous change trend representation in the local neighborhood of each boundary pixel, establishes a change transition description along the boundary normal direction and tangential direction respectively, and determines the boundary crossing position inside the pixel accordingly, generating a sub-pixel level boundary point set with consistent direction. Boundary chain reconstruction module: Performs global topological association on sub-pixel level boundary point set, and performs sequence reconstruction based on the spatial continuity and directional consistency between boundary points to form a closed and non-intersecting boundary chain structure; Curve reshaping module: Based on the boundary chain structure, it performs constrained curve reshaping, so that the boundary tends to be stable and continuous while maintaining the overall morphological characteristics, thereby obtaining a smooth boundary curve; The normalization module for the map features constructs varied map features based on smooth boundary curves and uses topological consistency constraints to verify and repair the map features, eliminating self-intersections and slender abnormal structures to obtain normalized map feature data. Planning control and cloud synchronization module: It overlays and analyzes standardized map data with the boundaries of land and space planning control, outputs compliance judgment results of changed map features, and synchronizes standardized map features and judgment results to the cloud platform for multi-source data collaborative updates and dynamic control.

[0115] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for intelligent management and control of land spatial planning based on GIS and cloud platform, characterized in that: include: S1. Acquire multi-temporal remote sensing image data and related spatial data, perform registration, correction and unified processing on the data, and construct a dataset with a unified spatial reference. S2. Perform change detection based on the dataset, generate binary raster data of the changed regions, and form an initial set of changed regions through connectivity constraints; S3. Extract boundary pixels for the initial set of change regions, construct a continuous change trend representation in the local neighborhood of each boundary pixel, establish change transition descriptions along the boundary normal direction and tangential direction respectively, and determine the boundary crossing position inside the pixel accordingly, generating a sub-pixel level boundary point set with directional consistency. S4. Perform global topological association on the sub-pixel level boundary point set, and reconstruct the sequence based on the spatial continuity and directional consistency between boundary points to form a closed and non-intersecting boundary chain structure. S5. Based on the boundary chain structure, perform constrained curve reshaping to make the boundary stable and continuous while maintaining the overall morphological characteristics, thereby obtaining a smooth boundary curve. S6. Construct variation patches based on smooth boundary curves, and use topological consistency constraints to verify and repair the patches, eliminating self-intersections and slender abnormal structures to obtain standardized patch data. S7. Overlay and analyze the standardized map data with the land and space planning control boundary, output the compliance judgment results of the changed map, and synchronize the standardized map and the judgment results to the cloud platform for multi-source data collaborative updates and dynamic management.

2. The intelligent management and control method for land spatial planning based on GIS and cloud platform according to claim 1, characterized in that, S1 includes: Remote sensing image data of the same area to be managed at different time phases are acquired, along with the corresponding land space planning boundaries, administrative divisions, land use status, and planning approval data. Radiometric, geometric, and orthorectified corrections are performed on the remote sensing image data, and coordinate transformation, topological checks, and field standardization are conducted on the relevant spatial data. The processed remote sensing image data and relevant spatial data are registered and fused according to a unified coordinate reference, a unified projection method, and a unified spatial range to form a dataset with a unified spatial reference.

3. The intelligent management and control method for land spatial planning based on GIS and cloud platform according to claim 1, characterized in that, S2 include: The difference intensity of corresponding pixels in remote sensing images of each time phase in the dataset is calculated, and the continuity of change transition is constructed in the local neighborhood to obtain change response distribution data; The change threshold is determined based on the continuous spatial change trend of the change response distribution data, and the change response distribution data is segmented to generate initial binary raster data. An orientation consistency constraint is applied to the initial binary raster data, and adjacent changed units are aggregated based on connectivity analysis to form an initial set of changed regions.

4. The intelligent management and control method for land spatial planning based on GIS and cloud platform according to claim 1, characterized in that, S3 includes: Identify the changed pixels adjacent to the non-changed pixels from the initial set of changed regions as boundary pixels, and construct a local neighborhood around each boundary pixel; Extract the spatial distribution sequence of change intensity within the local neighborhood, and determine the main change direction at the boundary pixel based on the spatial distribution sequence of change intensity. The stable transition interval from the non-changing region to the changing region is identified along the change transition path corresponding to the main change direction, and the boundary crossing position inside the pixel is determined within the stable transition interval to form a sub-pixel level boundary point set.

5. The intelligent management and control method for land spatial planning based on GIS and cloud platform according to claim 4, characterized in that, Forming a sub-pixel level boundary point set further includes: The intensity of change along the path is sequentially rearranged according to the change transition path to form a continuous sequence of intensity of change along the path direction. A monotonicity constraint is applied to the change intensity sequence to eliminate abnormal change segments formed by local reverse fluctuations and to determine the stable transition interval; Within the stable transition interval, the search range for locations of abrupt changes in intensity is gradually reduced, and the determined spatial locations are mapped to the interior of the corresponding pixels to obtain the boundary crossing points. The boundary crossing points are screened and corrected for directional consistency to form a sub-pixel level boundary point set.

6. The intelligent management and control method for land spatial planning based on GIS and cloud platform according to claim 1, characterized in that, S4 include: Taking each boundary point in the sub-pixel level boundary point set as an object, candidate neighboring points are filtered within a limited neighborhood, and the orientation attributes of each boundary point are recorded. Based on the spatial distance and orientation attributes between candidate adjacent points, consistency matching is performed, and boundary point pairs with continuous orientation changes and small spatial distances are connected first to form an initial connection relationship. The initial connection relationships are traversed globally, and connection paths with intersection or loop anomalies are disconnected, reconnected, and have their endpoints completed to form a closed and non-intersecting boundary chain structure.

7. The intelligent management and control method for land spatial planning based on GIS and cloud platform according to claim 1, characterized in that, S5 include: Traverse the boundary points in the connection order of the boundary chain structure, and identify high-change and low-change segments based on the directional deflection between adjacent boundary points; In low-change sections, the positions of boundary points are continuously adjusted, while in high-change sections, the original distribution of boundary points is maintained to preserve the overall morphological characteristics of the changed patches. The adjusted boundary point sequence is subjected to overall consistency verification to limit the variation between adjacent boundary points, and a smooth boundary curve is constructed based on the verified boundary point sequence.

8. The intelligent management and control method for land spatial planning based on GIS and cloud platform according to claim 1, characterized in that, S6 include: The initial change pattern is generated by closing the beginning and end of the smooth boundary curve and filling the area inside the closed boundary. Perform topological consistency detection on the initial changed patches to identify self-intersections, overlapping boundaries, unclosed regions, and elongated anomalous regions, and record the corresponding anomalous locations; Based on the abnormal locations, the corresponding boundary segments are locally rearranged, adjacent to each other adjusted, and compressed and corrected. The overall consistency of the corrected changed patches is then checked to obtain standardized patch data.

9. The intelligent management and control method for land spatial planning based on GIS and cloud platform according to claim 1, characterized in that, S7 includes: The standardized map patch data is spatially aligned with the land and space planning control boundary data, and a spatial mapping relationship is established between map patch identifiers and control boundary identifiers; Based on spatial mapping relationships, the standardized map data and the land spatial planning control boundary are overlaid and analyzed to extract the overlapping range, boundary distance and corresponding control zone category; Based on the control rules, the compliance judgment results of the changed patches are generated, and the standardized patch data, overlay analysis data and compliance judgment results are synchronized to the cloud platform according to a unified data structure.

10. A GIS- and cloud platform-based intelligent management and control system for land spatial planning, used to implement the GIS- and cloud platform-based intelligent management and control method for land spatial planning as described in any one of claims 1-9, characterized in that, include: Data acquisition and preprocessing module: acquires multi-temporal remote sensing image data and related spatial data, performs registration, correction and unified processing on the data, and constructs a dataset with a unified spatial reference. Change detection module: Performs change detection based on the dataset, generates binary raster data of the changed regions, and forms an initial set of changed regions through connectivity constraints; Sub-pixel boundary extraction module: Extracts boundary pixels for the initial set of changing regions, constructs a continuous change trend representation in the local neighborhood of each boundary pixel, establishes a change transition description along the boundary normal direction and tangential direction respectively, and determines the boundary crossing position inside the pixel accordingly, generating a sub-pixel level boundary point set with consistent direction. Boundary chain reconstruction module: Performs global topological association on sub-pixel level boundary point set, and performs sequence reconstruction based on the spatial continuity and directional consistency between boundary points to form a closed and non-intersecting boundary chain structure; Curve reshaping module: Based on the boundary chain structure, it performs constrained curve reshaping, so that the boundary tends to be stable and continuous while maintaining the overall morphological characteristics, thereby obtaining a smooth boundary curve; The normalization module for the map features constructs varied map features based on smooth boundary curves and uses topological consistency constraints to verify and repair the map features, eliminating self-intersections and slender abnormal structures to obtain normalized map feature data. Planning control and cloud synchronization module: It overlays and analyzes standardized map data with the boundaries of land and space planning control, outputs compliance judgment results of changed map features, and synchronizes standardized map features and judgment results to the cloud platform for multi-source data collaborative updates and dynamic control.

Images