A method and system for dynamic assessment of regional ecological value based on remote sensing data
By constructing a hybrid resolution ecological adjacency map in remote sensing data, the problem of balancing large-scale and local accuracy in remote sensing data evaluation is solved, enabling accurate identification and dynamic analysis of ecological patch changes.
Patent Information
- Application Number
- CN202611269241.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-20
- Publication Date
- 2026-11-10
AI Technical Summary
Existing technologies, when using remote sensing data for regional ecological value assessment, struggle to balance the processing efficiency of large-scale remote sensing data with the accuracy of identifying local ecological connectivity structures, leading to inaccurate identification of ecological patch splitting, merging, and continuous changes.
By introducing high-resolution remote sensing data to construct a hybrid resolution ecological adjacency map in a local area, the splitting and merging changes of ecological patches are identified. Combined with conservative and loose ecological masks, dynamic analysis results of ecological spatial integrity are generated.
It improves the objective and clear representation of ecological spatial structure and its dynamic changes, reduces false connections or disconnections caused by single spatial resolution, and improves the accuracy of assessment.
Smart Images

Figure CN123071852A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of regional ecological analysis technology, and in particular to a method and system for dynamic assessment of regional ecological value based on remote sensing data. Background Technology
[0002] With the continuous advancement of ecological environment monitoring and land space management, the dynamic assessment of regional ecological value using remote sensing data has become an important technical means. Remote sensing data is characterized by its wide coverage, stable observation period, and ease of multi-time comparison. By identifying the spatial distribution and changes of natural ecological surfaces such as forests, grasslands, and wetlands, it can provide data support for the protection, restoration, and management of regional ecological spaces. Current assessment work typically extracts natural ecological surface patches from remote sensing images and analyzes the overall condition of regional ecological spaces by combining the area, number, continuity, and spatial variation of these patches.
[0003] In practical remote sensing data processing, the spatial resolution of an image affects the representation of natural ecological surface boundaries and local connectivity structures. When there are narrow vegetation belts between adjacent ecological patches, or when there are narrow gaps such as roads or bare land between two ecological surfaces, a single remote sensing pixel may simultaneously cover multiple surface types. If the connectivity between ecological patches is directly determined based on classification results with lower spatial resolution, some actual connections may not be fully identified, and some actual gaps may be weakened in the pixel representation process. For multi-temporal remote sensing data, the aforementioned local differences may further affect the identification results of ecological patch splitting, merging, and continuity changes.
[0004] While high-resolution remote sensing imagery can provide more detailed information about the Earth's surface, acquiring and processing high-resolution data over large areas and multiple observation periods increases the amount of data acquired, stored, and computationally intensive. Therefore, in the dynamic assessment of regional ecological value, balancing the processing efficiency of large-scale remote sensing data with the accuracy of identifying local ecological connectivity structures, and ensuring a more accurate reflection of spatial changes in ecological patches across different observation periods, still requires further research and improvement. Summary of the Invention
[0005] To address the aforementioned technical issues, this invention proposes a method and system for dynamic assessment of regional ecological value based on remote sensing data. By introducing high-resolution remote sensing data into local areas that may affect the connectivity of ecological patches and constructing a hybrid resolution ecological adjacency map, the method achieves dynamic characterization of changes in ecological patch splitting and merging at different acquisition periods, taking into account both regional scope processing and local spatial structure identification.
[0006] To achieve the above objectives, the first aspect of the present invention provides a method for dynamic assessment of regional ecological value based on remote sensing data, comprising: Multiple sets of first-resolution remote sensing data of the target area are acquired and target ecological surface masks are extracted respectively. Each target ecological surface mask is processed by shrinking and expanding to generate conservative ecological masks and loose ecological masks. Multiple ecological kernel blocks are identified based on a conservative ecological mask. Multiple topological control regions are extracted based on a relaxed ecological mask and a conservative ecological mask. Multiple local kernel windows are determined based on the topological control regions and the ecological kernel blocks they contact. Acquire second-resolution remote sensing data covering the local verification window, generate a local ecological surface mask based on the second-resolution remote sensing data, and replace the first raster unit in each local verification window with the second raster unit. Establish the adjacency relationship between the first grid cell and the second grid cell, and construct a hybrid resolution ecological adjacency map of the target area for each set of first resolution remote sensing data; The corrected ecological patches of each mixed-resolution ecological adjacency map are extracted, and the splitting and merging positions of the corrected ecological patches corresponding to multiple sets of first-resolution remote sensing data are determined to generate dynamic analysis results of the ecological spatial integrity of the target area.
[0007] Preferably, each target ecological surface mask is subjected to shrinkage and expansion treatments to generate conservative and loose ecological masks, including: Each set of first-resolution remote sensing data includes multiple first grid cells, and each target ecological surface mask includes multiple first grid cells marked as the target ecological surface; In each target ecological surface mask, the first grid cell that is marked as the target ecological surface and has a common edge with at least one first grid cell that is not marked as the target ecological surface is determined as the inner boundary grid cell. The inner boundary grid cell is excluded from the ecological surface range defined by the target ecological surface mask to generate a conservative ecological mask. The first grid cell that is marked as a non-target ecological surface and has a common edge with at least one first grid cell marked as a target ecological surface is determined as the outer boundary grid cell. The outer boundary grid cell is included within the ecological surface range defined by the target ecological surface mask to generate a loose ecological mask.
[0008] Preferably, multiple ecological kernel blocks are identified based on a conservative ecological mask, multiple topological control regions are extracted based on a relaxed ecological mask and a conservative ecological mask, and multiple local kernel windows are determined based on the topological control regions and the ecological kernel blocks they contact, including: Connectivity regions are marked for the first grid cell in each conservative ecological mask that is identified as the target ecological surface, and each connected region is defined as an ecological kernel block. The first grid cell in the loose ecological mask that is marked as the target ecological surface and not marked as the target ecological surface in the conservative ecological mask is defined as the boundary uncertain grid cell; Connectivity regions are marked for multiple boundary uncertain grid cells to obtain multiple boundary uncertain blocks. Ecological kernel blocks that have a common edge adjacency with each boundary uncertain block are identified, and boundary uncertain blocks that have a common edge adjacency with at least two ecological kernel blocks are identified as topology control areas. Each topology control region is expanded, and the intersection of the expanded topology control region with each of the contacting ecological kernel blocks is calculated to obtain the anchoring region corresponding to each ecological kernel block. The topology control region is merged with the corresponding multiple anchoring regions, and the minimum bounding rectangle of the merged region is determined as the local kernel window.
[0009] Preferably, establishing the adjacency relationship between the first raster unit and the second raster unit, and constructing a hybrid resolution ecological adjacency map of the target area for each set of first resolution remote sensing data includes: Each set of second-resolution remote sensing data includes multiple second raster units, and each local ecological surface mask includes multiple second raster units marked as the target ecological surface. The first raster unit outside the local verification window that is marked as the target ecological surface, and the second raster unit inside the local verification window that is marked as the target ecological surface, are respectively identified as graph nodes; When the ground cover areas corresponding to two graph nodes have a common boundary segment with a length greater than zero, a graph edge is established between the two graph nodes, and a mixed-resolution ecological adjacency graph is generated based on multiple graph nodes and graph edges.
[0010] Preferably, the process of extracting corrected ecological patches from each mixed-resolution ecological adjacency map and determining the splitting and merging locations of the corrected ecological patches corresponding to multiple sets of first-resolution remote sensing data includes: Connectivity components are marked for each hybrid resolution ecological adjacency graph, and the ground coverage areas corresponding to the graph nodes in each connected component are merged to obtain multiple corrected ecological patches corresponding to each set of first resolution remote sensing data. According to the acquisition time sequence of the first resolution remote sensing data, the correction ecological patches corresponding to adjacent acquisition times are selected in turn. When there is a spatial overlap area greater than zero between the correction ecological patch located at the previous acquisition time and the correction ecological patch located at the next acquisition time, a cross-time correspondence is established between the corresponding correction ecological patches. For a correction ecological patch that establishes a cross-temporal correspondence between at least two correction ecological patches in the previous collection time and the next collection time, the disappearance region in the correction ecological patch that is not covered by any correction ecological patch in the next collection time is identified, the disappearance region is marked as a connected region, and the disappearance connected region that has a common boundary with at least two correction ecological patches with a length greater than zero is identified as the split position. For a correction ecological patch that establishes a cross-temporal correspondence with at least two correction ecological patches in the later collection time, identify the new area in the correction ecological patch that is not covered by any correction ecological patch in the previous collection time, mark the new area as a connected region, and determine the new connected region that has a common boundary with at least two correction ecological patches with a length greater than zero as the merge location.
[0011] Preferably, the expansion process for each topology control region includes: expanding each topology control region outward by one layer of first grid cells to obtain the expanded topology control region.
[0012] A second aspect of the present invention provides a regional ecological value dynamic assessment system based on remote sensing data, comprising: The ecological mask generation module is used to acquire multiple sets of first-resolution remote sensing data of the target area and extract the target ecological surface mask respectively. Each target ecological surface mask is processed by shrinking and expanding to generate a conservative ecological mask and a loose ecological mask. The verification window determination module is used to identify multiple ecological kernel blocks based on a conservative ecological mask, extract multiple topological control regions based on a relaxed ecological mask and a conservative ecological mask, and determine multiple local verification windows based on the topological control regions and the ecological kernel blocks they contact. The local raster refinement module is used to acquire second-resolution remote sensing data covering the local verification window, generate a local ecological surface mask based on the second-resolution remote sensing data, and replace the first raster unit in each local verification window with the second raster unit. The ecological adjacency graph construction module is used to establish the adjacency relationship between the first grid cell and the second grid cell, and to construct a mixed-resolution ecological adjacency graph of the target area for each set of first-resolution remote sensing data; The ecological space dynamic analysis module is used to extract the corrected ecological patches of each mixed resolution ecological adjacency map, determine the splitting and merging positions of the corrected ecological patches corresponding to multiple sets of first resolution remote sensing data, and generate dynamic analysis results of the ecological space integrity of the target area.
[0013] The present invention has the following beneficial effects: This invention filters out local areas that may affect the connectivity of ecological patches from first-resolution remote sensing data, and introduces second-resolution remote sensing data into these local areas to construct a hybrid-resolution ecological adjacency map. This allows narrow local ecological connectivity zones and non-ecological intervals to participate in the expression of the overall connectivity structure of the target area, without the need to uniformly use high-resolution data for the entire area. By correcting the spatial correspondence between ecological patches at different times to determine the location of patch splitting and merging, it can reduce false connections or false breaks caused by a single spatial resolution, and provide a more objective and clear characterization of the regional ecological spatial structure and its dynamic changes. Attached Figure Description
[0014] Figure 1 This is a flowchart of a method for dynamic assessment of regional ecological value based on remote sensing data, as described in an embodiment of the present invention.
[0015] Figure 2 This is a structural diagram of a regional ecological value dynamic assessment system based on remote sensing data, as described in an embodiment of the present invention. Detailed Implementation
[0016] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention.
[0017] Please see Figure 1 This invention provides a method for dynamic assessment of regional ecological value based on remote sensing data, comprising the following steps: Step S1: Obtain multiple sets of first-resolution remote sensing data of the target area and extract the target ecological surface mask respectively. Perform shrinkage and expansion processing on each target ecological surface mask to generate a conservative ecological mask and a loose ecological mask.
[0018] In this embodiment, the dynamic assessment of regional ecological value is mainly characterized by the spatial continuity of the natural ecological surface within the target area and its structural changes over time. Multiple sets of first-resolution remote sensing data correspond to different acquisition times within the target area and are arranged in chronological order. The first-resolution remote sensing data can be satellite remote sensing imagery, aerial remote sensing imagery, or land cover data based on such remote sensing imagery. To ensure that spatial objects acquired at different times correspond to each other, each set of first-resolution remote sensing data can be pre-converted to the same projection coordinate system and reference grid, ensuring that each set of data has a consistent grid size, grid origin, and row / column arrangement.
[0019] Each set of first-resolution remote sensing data includes multiple first-grid units arranged according to a preset grid. Based on the ecological spatial analysis requirements of the target area, target ecological surfaces are identified from each set of first-resolution remote sensing data, and the corresponding first-grid units are assigned target ecological surface labels, thereby forming target ecological surface masks. The target ecological surfaces may include one or more of the following natural or semi-natural surfaces: woodland, shrubland, natural grassland, wetlands, etc. First-grid units not identified as target ecological surfaces retain their non-target ecological surface labels.
[0020] In one optional implementation, the target ecological surface can be identified based on the spectral information, texture information, or existing land cover category information of each first raster unit in the first resolution remote sensing data. For example, supervised classification, unsupervised classification, remote sensing image semantic segmentation, or existing land cover products can be used to determine the land cover category corresponding to each first raster unit, and the first raster unit belonging to the preset target ecological surface category can be selected from the identification results to generate a target ecological surface mask.
[0021] The above-described remote sensing surface classification and mask generation process can be implemented using commonly used remote sensing image processing methods in this field. This embodiment does not limit the specific classification model or classification algorithm. For multiple sets of first-resolution remote sensing data, the same target ecological surface category range and mask marking rules are used. Specifically, the first raster unit identified as the target ecological surface is included in the target ecological surface mask, while the first raster unit not identified as the target ecological surface is not included in the target ecological surface mask.
[0022] Considering that the target ecological surface boundary is typically represented by discretely arranged first grid cells, the first grid cells located at the boundary may directly affect the connectivity between adjacent ecological regions in the grid space. This embodiment adjusts one layer of the first grid cells from both the inner and outer sides of the target ecological surface boundary to form two expressions of the target ecological surface connectivity range: a conservative ecological mask and a loose ecological mask. The conservative ecological mask excludes the connectivity portion directly dependent on the boundary grid, while the loose ecological mask incorporates the outer grid cells immediately adjacent to the target ecological surface boundary.
[0023] In one alternative implementation, for any target ecological surface mask, the first grid cells marked as target ecological surfaces are traversed; when a first grid cell marked as target ecological surface shares a common edge with at least one first grid cell not marked as target ecological surface, that first grid cell is identified as an inner boundary grid cell. All inner boundary grid cells are excluded from the target ecological surface mask, thereby generating a conservative ecological mask.
[0024] Accordingly, the first grid cells not marked as target ecological surfaces in the original target ecological surface mask are traversed; when a first grid cell not marked as target ecological surface shares a common edge with at least one first grid cell marked as target ecological surface, that first grid cell is identified as an outer boundary grid cell. All outer boundary grid cells are included in the target ecological surface mask, thereby generating a loose ecological mask.
[0025] Using the above method, conservative and loose ecological masks are extracted from the target ecological surface mask, and both are preserved as derived masks distinct from the original target ecological surface mask. For target ecological surface regions connected only by a single boundary grid layer, the connecting portion is removed in the conservative ecological mask; for adjacent target ecological surface regions separated by no more than one first grid unit, the separated portion is included in the loose ecological mask. Thus, the conservative and loose ecological masks can reflect the spatial connectivity of the target ecological surface from different boundary expression ranges.
[0026] Step S2: Identify multiple ecological kernel blocks based on the conservative ecological mask, extract multiple topological control regions based on the relaxed ecological mask and the conservative ecological mask, and determine multiple local kernel windows based on the topological control regions and the ecological kernel blocks they contact.
[0027] In this embodiment, the conservative ecological mask excludes the first layer of grid cells inside the target ecological surface boundary, and the remaining ecological surface area reflects the internal connectivity structure that is not maintained by the boundary grid. The loose ecological mask further incorporates the first layer of grid cells outside the target ecological surface boundary to represent the possible connection range that may be formed when the boundary extends outward. By comparing the spatial differences between the two, local locations that may change the ecological surface connectivity can be selected from the target area.
[0028] Specifically, the first grid cell included in each conservative ecological mask is marked as a connected region. Connected region marking uses a common edge adjacency method; that is, two first grid cells sharing a grid edge are considered adjacent, while those only connected by a corner point are not considered connected. Multiple first grid cells continuously connected by a common edge form a connected region, and each connected region is defined as an ecological kernel block. This results in multiple independent ecological kernel blocks.
[0029] In one alternative implementation, the first grid cell that is included in the loose ecological mask but not in the corresponding conservative ecological mask is defined as the boundary uncertain grid cell. The boundary uncertain grid cell includes both the inner boundary grid cells excluded by the conservative ecological mask and the outer boundary grid cells further included by the loose ecological mask, thus covering the grid range near the target ecological surface boundary that may affect connectivity.
[0030] Multiple boundary-uncertain raster cells are labeled as connected regions according to a common edge adjacency method, and interconnected boundary-uncertain raster cells are combined into boundary-uncertain blocks. Subsequently, for each boundary-uncertain block, ecological kernel blocks with common edge adjacency are found, and the corresponding ecological kernel block set is recorded.
[0031] When an uncertain boundary block contacts only one ecological core block, its inclusion or exclusion typically only alters the local outer edge of the corresponding ecological core block without establishing a connection between the two independent ecological core blocks. When an uncertain boundary block contacts at least two independent ecological core blocks simultaneously, the ecological surface distribution within the uncertain boundary block may determine whether the ecological core blocks are interconnected; therefore, the uncertain boundary block is identified as a topological control area.
[0032] For example, when there is a column of raster cells with uncertain boundaries between two ecological kernel blocks, if this column of cells forms a continuous ecological surface in a finer spatial representation, the two ecological kernel blocks can be connected to each other through this location; if there is a penetrating road, bare land, or other non-target ecological surface at this location, the two ecological kernel blocks remain disconnected. Thus, the topology control area defines the local area that needs further verification using second-resolution remote sensing data, without requiring fine processing of the entire boundary area of the target region.
[0033] After determining the topological control area, a first grid cell is expanded outward from each topological control area, and the expanded topological control area is then intersected with each of the ecological kernel blocks it contacts. The first grid cell located within the intersection area forms the anchoring region of the corresponding ecological kernel block. The anchoring region is located within the ecological surface defined by the conservative ecological mask and is adjacent to the topological control area. It can be used as the determining end point in subsequent local high-resolution processing to determine whether different ecological kernel blocks are interconnected through the topological control area.
[0034] Each topology control region is spatially merged with all its corresponding anchored regions to form the area to be verified, and the smallest bounding rectangle that can completely cover this area is determined. To facilitate the subsequent replacement of the first grid cell with the second grid cell, the boundary of the smallest bounding rectangle can be adjusted to the grid boundary of the adjacent first grid cell, so that the local verification window is composed of the complete coverage area of the first grid cell. The adjusted smallest bounding rectangle is the corresponding local verification window.
[0035] When multiple local verification windows corresponding to the same set of first-resolution remote sensing data overlap or share common edges, the corresponding local verification windows can be merged first, and the merged rectangular area can be used as a unified local verification window, thereby avoiding repeated acquisition or processing of second-resolution remote sensing data for the same spatial range. After the above processing, each local verification window covers at least one topological control area and a local area of the ecological kernel block associated with that topological control area.
[0036] Step S3: Obtain the second-resolution remote sensing data covering the local verification window, generate a local ecological surface mask based on the second-resolution remote sensing data, and replace the first grid cell in each local verification window with the second grid cell.
[0037] In this embodiment, for each local verification window corresponding to the first resolution remote sensing data, second resolution remote sensing data covering the corresponding local verification window is acquired. The second resolution remote sensing data can be high-resolution satellite imagery, aerial orthophotos, UAV orthophotos, or other high-resolution surface images with geographic location information, and has the same or similar acquisition time as the corresponding first resolution remote sensing data, so as to reflect the surface cover of the target area at the corresponding time.
[0038] Each set of second-resolution remote sensing data includes multiple second grid cells arranged according to a preset grid. The ground cover size corresponding to a single second grid cell is smaller than the ground cover size corresponding to a single first grid cell. For example, when the ground cover size of the first grid cell is 10 meters × 10 meters and the ground cover size of the second grid cell is 2 meters × 2 meters, the ground area corresponding to one first grid cell can be represented by 25 second grid cells.
[0039] After acquiring the second-resolution remote sensing data, it is converted to the projection coordinate system used for the corresponding first-resolution remote sensing data and cropped according to the geographic extent of the local verification window. The second-resolution and first-resolution remote sensing data use a unified geographic coordinate reference, enabling the second raster cell to be located at its corresponding ground position within the first raster grid. The boundary of the local verification window preferably coincides with the raster boundary of the first raster cell, so that the two types of rasters inside and outside the window can establish a clear spatial adjacency relationship at the window boundary.
[0040] For the cropped second-resolution remote sensing data, surface identification is performed according to the target ecological surface range consistent with the target ecological surface mask to generate a local ecological surface mask. The local ecological surface mask consists of multiple mask grids corresponding to the second-resolution remote sensing data. Among them, second grid cells belonging to woodland, shrubland, natural grassland, wetland, or other pre-determined target ecological surfaces are included in the local ecological surface mask, while other second grid cells are not included in the local ecological surface mask.
[0041] When generating a local ecological surface mask, remote sensing image classification, land cover segmentation, or existing high-resolution land cover data can be used for determination. The specific surface identification method used in this step can be selected based on the data type of the second-resolution remote sensing data, but the target ecological surface category range corresponding to the first-resolution and second-resolution remote sensing data should remain consistent to avoid changes in the local verification results due to different surface category definitions.
[0042] Furthermore, for each local verification window, all the first raster cells it covers are determined. When subsequently constructing the hybrid resolution ecological adjacency map, the first raster cells are no longer used as spatial representation units within the local verification window; instead, the same ground extent is represented by second raster cells in the local ecological surface mask. Specifically, the second raster cells included in the local ecological surface mask are used to form graph nodes in the subsequent hybrid resolution ecological adjacency map, while the second raster cells not included in the local ecological surface mask are only used to represent non-target ecological surface intervals within the local verification window and do not form graph nodes in the ecological adjacency map.
[0043] It should be noted that replacing the first raster unit with the second raster unit means changing the spatial representation granularity used within the local verification window when constructing the spatial connectivity structure of the target area. Specifically, multiple second raster units replace the larger first raster units within the local verification window. The original first raster units in the first-resolution remote sensing data are not deleted, nor are the land surface category labels of each first raster unit in the original target ecological surface mask changed. The first raster units and their original labels are retained for data retrieval and analysis results traceability at different acquisition times.
[0044] After the above processing, the target area is still spatially represented using the first raster unit outside the local verification window, while within the local verification window, it is spatially represented using a smaller second raster unit. Narrow ecological connectivity areas or non-target ecological surface intervals that may be represented by a single first raster unit in the first-resolution remote sensing data can be further distinguished by multiple second raster units.
[0045] Step S4: Establish the adjacency relationship between the first grid cell and the second grid cell, and construct a hybrid resolution ecological adjacency map of the target area for each set of first resolution remote sensing data.
[0046] In this embodiment, a hybrid resolution ecological adjacency map is constructed for each set of first-resolution remote sensing data. Specifically, the first raster cell outside the local verification window and marked as the target ecological surface in the corresponding target ecological surface mask is determined as a first type of graph node; the second raster cell inside the local verification window and marked as the target ecological surface in the corresponding local ecological surface mask is determined as a second type of graph node. The first raster cell inside the local verification window is no longer used as a graph node, and the first and second raster cells not marked as target ecological surfaces are also not used as graph nodes, so that roads, bare land, construction land and other non-target ecological surfaces are represented as spatial intervals between adjacent ecological surfaces in the graph structure.
[0047] For each map node, the geographic coverage, raster type, and corresponding remote sensing data acquisition time of the corresponding raster cell are recorded. Specifically, the first type of map node corresponds to the ground coverage of the first raster cell, and the second type of map node corresponds to the ground coverage of the second raster cell. The adjacency relationships between different types of map nodes are determined based on the actual geographic coverage, rather than directly matching based on the row and column numbers of the two raster data types.
[0048] In one optional implementation, the spatial relationship between the ground cover areas corresponding to any two graph nodes is determined sequentially. When the ground cover areas corresponding to two graph nodes have a common boundary line segment with a length greater than zero, a graph edge is established between the two graph nodes. Using the above method, adjacency relationships between first-type graph nodes, between second-type graph nodes, and between first-type and second-type graph nodes can be uniformly established.
[0049] For two adjacent first grid cells outside the local verification window, when both are marked as target ecological surfaces and share a grid edge, a graph edge is established between the corresponding first-class graph nodes; for two adjacent second grid cells within the local verification window, when both are marked as target ecological surfaces and share a grid edge, a graph edge is established between the corresponding second-class graph nodes.
[0050] At the boundary of the local verification window, a grid edge of a first grid cell may simultaneously connect to multiple smaller second grid cells. In this case, the common boundary line segment between the ground coverage area corresponding to the first grid cell and each second grid cell is determined. For second grid cells with a common boundary line segment length greater than zero, graph edges are established between the corresponding first-class graph nodes and second-class graph nodes.
[0051] Preferably, when the grid boundary of the second grid cell is not fully aligned with the grid boundary of the first grid cell, the common boundary segment can be calculated based on the actual geographic coordinates of each grid cell without resampling the second grid cell as the first grid cell.
[0052] Based on all the first-class graph nodes, second-class graph nodes, and corresponding graph edges, a mixed-resolution ecological adjacency graph is generated for each set of first-resolution remote sensing data. Each graph node corresponds to a raster cell marked as the target ecological surface, and each graph edge indicates that there is a common edge connection between two target ecological surface raster cells.
[0053] In this way, narrow target ecological surface connection areas identified by the second-resolution remote sensing data can form connection paths through continuous second-type map nodes; while through-type non-target ecological surface intervals identified by the second-resolution remote sensing data have their connection paths severed due to the lack of corresponding map nodes. The corresponding local connections or disconnections can be incorporated into the overall adjacency structure of the target area through cross-resolution map edges, without having to downsample the local high-resolution results to the first resolution.
[0054] Step S5: Extract the corrected ecological patches from each mixed-resolution ecological adjacency map, determine the splitting and merging positions of the corrected ecological patches corresponding to multiple sets of first-resolution remote sensing data, and generate dynamic analysis results of the ecological spatial integrity of the target area.
[0055] In this embodiment, since the graph nodes in the mixed resolution ecological adjacency graph have been connected according to the actual common edge relationship of the target ecological surface, the range of interconnected ecological surfaces at the corresponding collection time can be determined by the connected components in the graph structure.
[0056] Specifically, connected component labeling is performed on each hybrid resolution ecological adjacency graph. Connected component labeling can be implemented using depth-first search, breadth-first search, or other graph traversal methods commonly used in this field. For any graph node, all graph nodes reachable from that node along graph edges are assigned to the same connected component, and there are no continuously reachable graph edge paths between different connected components.
[0057] For each connected component, the ground cover areas corresponding to each graph node are spatially merged to obtain a corrected ecological patch. The ground cover area can include either the first raster cell outside the local verification window or the second raster cell within the local verification window. The corrected ecological patch formed after spatial merging is no longer affected by the resolution differences of the graph nodes, but is represented as a continuous ecological surface extent in a unified geographic coordinate space. Each mixed-resolution ecological adjacency map can yield multiple corrected ecological patches corresponding to the corresponding acquisition time.
[0058] After extracting the corrected ecological patches for each acquisition time, spatial comparisons were performed on the corrected ecological patches corresponding to two adjacent acquisition times, following the acquisition time sequence of the first-resolution remote sensing data. For ease of explanation, the earlier acquisition time among two adjacent acquisition times is referred to as the previous acquisition time, and the later acquisition time is referred to as the next acquisition time.
[0059] The corrected ecological patches from the previous collection time are overlaid with those from the subsequent collection time onto the same geographic coordinate space. A temporal correspondence is established between corrected ecological patches from different collection times when there is a spatial overlap area greater than zero between them. No temporal correspondence is established between two corrected ecological patches that only touch at their boundary lines or corners but do not have actual area overlap.
[0060] The aforementioned intertemporal correspondence is used to reflect the spatial continuity of corrected ecological patches at adjacent acquisition times, without requiring matching thresholds such as overlap area ratio, patch center distance, or morphological similarity. When a corrected ecological patch from a previous acquisition time establishes an intertemporal correspondence with only one corrected ecological patch from a subsequent acquisition time, it can be considered that the corresponding ecological patch maintains single spatial continuity in the subsequent acquisition time; when it establishes an intertemporal correspondence with at least two corrected ecological patches from subsequent acquisition times, it is further determined whether patch splitting exists.
[0061] For a correction ecological patch that establishes a cross-temporal correspondence between at least two correction ecological patches in the previous collection time and the next collection time, the ground area covered by the correction ecological patch in the previous collection time is subtracted from the ground area covered by all correction ecological patches in the next collection time to obtain the corresponding disappearance area, which represents the spatial range that belonged to the target ecological surface in the previous collection time but was not covered by any correction ecological patch in the next collection time.
[0062] The aforementioned disappearing regions are marked as connected regions, dividing the continuous disappearing space into multiple disappearing connected regions. Then, the boundary relationships between each disappearing connected region and at least two subsequent time-corrected ecological patches are determined. When a disappearing connected region simultaneously shares a common boundary with at least two corresponding subsequent time-corrected ecological patches of greater than zero length, the disappearing connected region is identified as a split location.
[0063] For example, when a corrected ecological patch from a previous data collection time becomes two independent corrected ecological patches located on either side of a road in a subsequent data collection time, the space occupied by the newly added road or other non-target ecological surface forms a disappearing region. If both sides of this disappearing region are connected to two corrected ecological patches from the subsequent data collection time, then this disappearing region constitutes the actual splitting point where the original corrected ecological patch changes from one connected range to multiple connected ranges. Ordinary boundary reduction regions located only at the outer edge of the original corrected ecological patch and connected to only one corrected ecological patch from the subsequent data collection time are not identified as splitting points.
[0064] Accordingly, when a corrected ecological patch from a later acquisition time establishes a cross-temporal correspondence with at least two corrected ecological patches from a previous acquisition time, it is further determined whether patch merging exists. Specifically, the ground area covered by the corrected ecological patch from the later acquisition time is subtracted from the ground area covered by all corrected ecological patches from the previous acquisition time to obtain the corresponding newly added area, representing the spatial range of a corrected ecological patch that was not covered by any corrected ecological patch in the previous acquisition time but was included in the corresponding corrected ecological patch in the later acquisition time.
[0065] The newly added areas are marked as connected regions, dividing the contiguous new space into multiple newly added connected regions. The boundary relationships between each newly added connected region and at least two previously collected time-corrected ecological patches are determined. When a newly added connected region has a common boundary with at least two corresponding previously collected time-corrected ecological patches of greater than zero length, the newly added connected region is determined as the merged location.
[0066] For example, when two separate calibration ecological patches from a previous acquisition time are connected to each other by a newly added continuous target ecological surface at a later acquisition time, the newly added target ecological surface forms a new region. When this new region is connected to each of the two previously independent calibration ecological patches, and both are classified into the same connected component at the later acquisition time, the new region constitutes the actual merged location of the corresponding calibration ecological patches. New regions that are only located at the outer edge of a single calibration ecological patch and are not connected to other calibration ecological patches from the previous acquisition time are not identified as merged locations.
[0067] For cases involving three or more sets of first-resolution remote sensing data, the above-described cross-temporal correspondence, split location, and merge location processes are repeated for every two adjacent acquisition times, following the acquisition time sequence. Each analysis records the previous acquisition time, the next acquisition time, the involved corrected ecological patches, and the corresponding split or merge locations, thus forming a record of ecological spatial structure changes arranged chronologically.
[0068] Finally, the corrected ecological patches corresponding to each collection time, the temporal correspondence between adjacent collection times, the splitting location, the merging location, and the relevant collection time are correlated to generate dynamic analysis results of the ecological spatial integrity of the target area. These dynamic analysis results can be represented using spatial distribution maps, time series variation maps, or event log tables. Each splitting event includes at least the splitting location, the corrected ecological patch before the split, and the corrected ecological patch after the split; each merging event includes at least the merging location, the corrected ecological patch before the merging, and the corrected ecological patch after the merging.
[0069] Please see Figure 2 This invention provides a regional ecological value dynamic assessment system based on remote sensing data, comprising: The ecological mask generation module is used to acquire multiple sets of first-resolution remote sensing data of the target area and extract the target ecological surface mask respectively. Each target ecological surface mask is processed by shrinking and expanding to generate a conservative ecological mask and a loose ecological mask. The verification window determination module is used to identify multiple ecological kernel blocks based on a conservative ecological mask, extract multiple topological control regions based on a relaxed ecological mask and a conservative ecological mask, and determine multiple local verification windows based on the topological control regions and the ecological kernel blocks they contact. The local raster refinement module is used to acquire second-resolution remote sensing data covering the local verification window, generate a local ecological surface mask based on the second-resolution remote sensing data, and replace the first raster unit in each local verification window with the second raster unit. The ecological adjacency graph construction module is used to establish the adjacency relationship between the first grid cell and the second grid cell, and to construct a mixed-resolution ecological adjacency graph of the target area for each set of first-resolution remote sensing data; The ecological space dynamic analysis module is used to extract the corrected ecological patches of each mixed resolution ecological adjacency map, determine the splitting and merging positions of the corrected ecological patches corresponding to multiple sets of first resolution remote sensing data, and generate dynamic analysis results of the ecological space integrity of the target area.
[0070] The above are merely specific embodiments of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art. Parts not described in detail in this specification are prior art known to those skilled in the art.
Claims
1. A method for dynamic assessment of regional ecological value based on remote sensing data, characterized in that, include: Multiple sets of first-resolution remote sensing data of the target area are acquired and target ecological surface masks are extracted respectively. Each target ecological surface mask is processed by shrinking and expanding to generate conservative ecological masks and loose ecological masks. Multiple ecological kernel blocks are identified based on a conservative ecological mask. Multiple topological control regions are extracted based on a relaxed ecological mask and a conservative ecological mask. Multiple local kernel windows are determined based on the topological control regions and the ecological kernel blocks they contact. Acquire second-resolution remote sensing data covering the local verification window, generate a local ecological surface mask based on the second-resolution remote sensing data, and replace the first raster unit in each local verification window with the second raster unit. Establish the adjacency relationship between the first grid cell and the second grid cell, and construct a hybrid resolution ecological adjacency map of the target area for each set of first resolution remote sensing data; The corrected ecological patches of each mixed-resolution ecological adjacency map are extracted, and the splitting and merging positions of the corrected ecological patches corresponding to multiple sets of first-resolution remote sensing data are determined to generate dynamic analysis results of the ecological spatial integrity of the target area.
2. The method for dynamic assessment of regional ecological value based on remote sensing data according to claim 1, characterized in that, Each target ecological surface mask is subjected to shrinkage and expansion treatments to generate conservative and loose ecological masks, including: Each set of first-resolution remote sensing data includes multiple first grid cells, and each target ecological surface mask includes multiple first grid cells marked as the target ecological surface; In each target ecological surface mask, the first grid cell that is marked as the target ecological surface and has a common edge with at least one first grid cell that is not marked as the target ecological surface is determined as the inner boundary grid cell. The inner boundary grid cell is excluded from the ecological surface range defined by the target ecological surface mask to generate a conservative ecological mask. The first grid cell that is marked as a non-target ecological surface and has a common edge with at least one first grid cell marked as a target ecological surface is determined as the outer boundary grid cell. The outer boundary grid cell is included within the ecological surface range defined by the target ecological surface mask to generate a loose ecological mask.
3. The method for dynamic assessment of regional ecological value based on remote sensing data according to claim 2, characterized in that, Multiple ecological kernel blocks are identified based on a conservative ecological mask. Multiple topological control regions are extracted based on both a relaxed and conservative ecological mask. Multiple local kernel windows are determined based on the topological control regions and their contacting ecological kernel blocks, including: Connectivity regions are marked for the first grid cell in each conservative ecological mask that is identified as the target ecological surface, and each connected region is defined as an ecological kernel block. The first grid cell in the loose ecological mask that is marked as the target ecological surface and not marked as the target ecological surface in the conservative ecological mask is defined as the boundary uncertain grid cell; Connectivity regions are marked for multiple boundary uncertain grid cells to obtain multiple boundary uncertain blocks. Ecological kernel blocks that have a common edge adjacency with each boundary uncertain block are identified, and boundary uncertain blocks that have a common edge adjacency with at least two ecological kernel blocks are identified as topology control areas. Each topology control region is expanded, and the intersection of the expanded topology control region with each of the contacting ecological kernel blocks is calculated to obtain the anchoring region corresponding to each ecological kernel block. The topology control region is merged with the corresponding multiple anchoring regions, and the minimum bounding rectangle of the merged region is determined as the local kernel window.
4. The method for dynamic assessment of regional ecological value based on remote sensing data according to claim 3, characterized in that, Establish the adjacency relationship between the first and second raster cells, and construct a mixed-resolution ecological adjacency map of the target area for each set of first-resolution remote sensing data, including: Each set of second-resolution remote sensing data includes multiple second raster units, and each local ecological surface mask includes multiple second raster units marked as the target ecological surface. The first raster unit outside the local verification window that is marked as the target ecological surface, and the second raster unit inside the local verification window that is marked as the target ecological surface, are respectively identified as graph nodes; When the ground cover areas corresponding to two graph nodes have a common boundary segment with a length greater than zero, a graph edge is established between the two graph nodes, and a mixed-resolution ecological adjacency graph is generated based on multiple graph nodes and graph edges.
5. The method for dynamic assessment of regional ecological value based on remote sensing data according to claim 4, characterized in that, The corrected ecological patches are extracted from each mixed-resolution ecological adjacency map, and the splitting and merging locations of the corrected ecological patches corresponding to multiple sets of first-resolution remote sensing data are determined, including: Connectivity components are marked for each hybrid resolution ecological adjacency graph, and the ground coverage areas corresponding to the graph nodes in each connected component are merged to obtain multiple corrected ecological patches corresponding to each set of first resolution remote sensing data. According to the acquisition time sequence of the first resolution remote sensing data, the correction ecological patches corresponding to adjacent acquisition times are selected in turn. When there is a spatial overlap area greater than zero between the correction ecological patch located at the previous acquisition time and the correction ecological patch located at the next acquisition time, a cross-time correspondence is established between the corresponding correction ecological patches. For a correction ecological patch that establishes a cross-temporal correspondence between at least two correction ecological patches in the previous collection time and the next collection time, the disappearance region in the correction ecological patch that is not covered by any correction ecological patch in the next collection time is identified, the disappearance region is marked as a connected region, and the disappearance connected region that has a common boundary with at least two correction ecological patches with a length greater than zero is identified as the split position. For a correction ecological patch that establishes a cross-temporal correspondence with at least two correction ecological patches in the later collection time, identify the new area in the correction ecological patch that is not covered by any correction ecological patch in the previous collection time, mark the new area as a connected region, and determine the new connected region that has a common boundary with at least two correction ecological patches with a length greater than zero as the merge location.
6. The method for dynamic assessment of regional ecological value based on remote sensing data according to claim 3, characterized in that, The expansion process for each topology control region includes: expanding each topology control region outward by one layer of first grid cells to obtain the expanded topology control region.
7. A dynamic assessment system for regional ecological value based on remote sensing data, characterized in that, The system is used to implement the method for dynamic assessment of regional ecological value based on remote sensing data as described in any one of claims 1 to 6, comprising: The ecological mask generation module is used to acquire multiple sets of first-resolution remote sensing data of the target area and extract the target ecological surface mask respectively. Each target ecological surface mask is processed by shrinking and expanding to generate a conservative ecological mask and a loose ecological mask. The verification window determination module is used to identify multiple ecological kernel blocks based on a conservative ecological mask, extract multiple topological control regions based on a relaxed ecological mask and a conservative ecological mask, and determine multiple local verification windows based on the topological control regions and the ecological kernel blocks they contact. The local raster refinement module is used to acquire second-resolution remote sensing data covering the local verification window, generate a local ecological surface mask based on the second-resolution remote sensing data, and replace the first raster unit in each local verification window with the second raster unit. The ecological adjacency graph construction module is used to establish the adjacency relationship between the first grid cell and the second grid cell, and to construct a mixed-resolution ecological adjacency graph of the target area for each set of first-resolution remote sensing data; The ecological space dynamic analysis module is used to extract the corrected ecological patches of each mixed resolution ecological adjacency map, determine the splitting and merging positions of the corrected ecological patches corresponding to multiple sets of first resolution remote sensing data, and generate dynamic analysis results of the ecological space integrity of the target area.
