A regional accessibility evolution method and system based on local topology mask and dynamic spatial pruning

Abstract

The application relates to a region reachability evolution method and system based on a local topological mask and dynamic space pruning, which comprises the following steps: loading a topological connected point set, reconstructing a global free space topology based on the topological connected point set, extracting an external topological boundary of a converged true region set as an active front, generating a local search domain mask, and performing local pruning in the global free space topology; constructing a local visibility graph in the local search domain mask, and switching an adaptive solving algorithm according to the density of obstacle line segments in the local search domain mask, so as to perform shortest path relaxation calculation on the local visibility graph to obtain a shortest path graph; and performing three-dimensional multiple pruning and state iteration updating on the shortest path graph to obtain a final accurate reachable true region range. The application aims to solve the problems of easy overflow, low precision and slow calculation under large-scale data.

Description

Technical Field

[0001] This invention relates to the fields of geographic information systems and spatial analysis technology, and in particular to a method and system for regional accessibility evolution based on local topological masking and dynamic spatial pruning. Background Technology

[0002] Regional accessibility analysis is a core spatial analysis technique in fields such as land spatial planning, transportation network planning, spatial allocation of natural resources, emergency rescue route planning, and ecological corridor connectivity assessment. Its core is to solve for the geodesic distance and precise reachability range (i.e., the "true area") between source and target elements under geographical obstacle constraints. Current mainstream accessibility analysis methods rely on global raster cost distance algorithms from commercial GIS platforms. They primarily employ two technical approaches: Euclidean distance calculation and distance accumulation (solving the global distance field using the Eikonal equation). These methods have basic applicability in small- to medium-scale environments with simple obstacles.

[0003] However, in practical engineering applications involving large-scale provincial and municipal levels, spatial elements exceeding one million, and complex linear obstacles and topological connectivity constraints, the following significant limitations exist: First, the memory overflow problem is prominent. Traditional tools adopt a global raster one-time calculation paradigm, which requires the pre-construction of a global high-resolution resistance raster and distance raster. The data scale increases geometrically with the increase of region area and resolution (sampling accuracy). When processing large-scale irregular patches or high-precision raster, it is very easy to trigger the memory overflow (OOM) limit, resulting in calculation interruption.

[0004] Secondly, the accuracy of obstacle topology processing is insufficient. Traditional rasterization processing uses pixel-level polyline approximation for linear obstacles, and the simple Euclidean distance cannot handle complex "flow around" connection problems, making it difficult to accurately represent microscopic "topological openings" such as bridges and culverts.

[0005] Third, there is serious computational redundancy. Existing technologies mostly adopt a full-domain scanning mechanism, which requires the target area to participate in the calculation regardless of whether it is connected to the source point. This results in exponential computational redundancy when processing massive amounts of data, making it impossible to achieve fast convergence.

[0006] Fourth, there is a severe loss of topological accuracy. The rasterization calculation process leads to the truncation of geographic feature boundaries and the loss of floating-point precision, which cannot meet the stringent requirements for the accuracy of true area boundaries in land spatial planning, ecological red line delineation, and ownership determination.

[0007] Therefore, there is an urgent need for a regional accessibility evolution method and system based on local topological masks and dynamic spatial pruning to solve the problems of easy overflow, low accuracy and slow computation of traditional methods under large-scale data. Summary of the Invention

[0008] To address the problems of existing technologies, this invention aims to provide a regional accessibility evolution method and system based on local topological masks and dynamic spatial pruning. By transforming the obstacle-laden spatial accessibility problem into an obstacle-laden Euclidean shortest path (ESPO) problem, the global pathfinding problem is reduced to a multi-round iterative solution within a locally constrained space. A dynamic active front is constructed to achieve local evolution, and spatial extent is limited by local topological mask constraints. Redundancy is eliminated through three-dimensional multi-pruning in the spatial, semantic, and computational cost domains. Vector-level topological connectivity is achieved between linear obstacles and topologically connected points. The solution algorithm is adaptively switched based on obstacle density. Dynamic memory management is employed to release temporary data in real time, thereby achieving rapid convergence and high-precision vector-level true region extraction for regional accessibility in large-scale, complex obstacle environments at the provincial and municipal levels. This invention can provide accurate spatial accessibility support for scenarios such as land spatial planning, ecological protection, emergency rescue, and traffic service assessment, effectively improving the computational efficiency, topological accuracy, and engineering applicability of regional accessibility analysis.

[0009] To achieve the above objectives, the present invention provides the following solution: A method for regional accessibility evolution based on local topological masking and dynamic spatial pruning includes: Load the set of topological connected points, reconstruct the global free space topology based on the set of topological connected points, extract the external topological boundary of the converged true region set as the active frontier, generate a local search domain mask, and perform local pruning within the global free space topology; A local visibility map is constructed within the local search domain mask. At the same time, an adaptive solution algorithm is switched according to the density of obstacle line segments within the local search domain mask to perform shortest path relaxation calculation on the local visibility map and obtain the shortest path map. The shortest path graph is subjected to three-dimensional multi-pruning and state iteration update to obtain the final accurate reachable true region range.

[0010] Optionally, reconstructing the global free space topology based on the set of topological connected points includes: A local buffer mask is constructed based on the set of topological connected points. The local buffer mask is used to perform vector-level Boolean difference operations on the set of linear obstacles to erase linear obstacle elements and generate a set of effective constraints that conform to the physical passage rules. The global free space topology is reconstructed using the set of effective constraints; the global free space topology represents a continuous traversable space that satisfies the physical traversal rules after removing the set of effective constraints within the study region.

[0011] Optionally, performing the partial cropping includes: The external topological boundary of the converged true region set is extracted as the active frontier, and the Minkowski sum operation is performed according to the preset geodesic distance threshold to generate the local search domain mask. The local search domain mask is used to locally prune global features within the global free space topology, retaining only the set of candidate patches and the subset of obstacle features within the mask, thus completing the first round of spatial domain pruning.

[0012] Optionally, constructing a local visibility map within the local search domain mask includes: Within the local search domain mask, the local visibility graph is constructed using the endpoints of line segments from the obstacle feature subset, feature points of the active frontier of the true region, and feature points of candidate patches falling within the mask as graph node sets; the local visibility graph includes: graph node set and edge set; During the construction of the local visibility graph, when the interior of a line segment connecting any node in the graph node set does not intersect with any obstacle line segment in the obstacle element subset, it is determined to be a visible edge introduced into the edge set.

[0013] Optionally, obtaining the shortest path graph includes: An adaptive solution algorithm is switched based on the density of obstacle line segments within the local search domain mask. The total number of line segments in the subset of obstacle elements within the local search domain mask or its ratio to the mask area is calculated to obtain the calculated density, which is then compared with a preset density threshold. When the calculated density is higher than the density threshold, the current mask region is determined to be in a dense edge mode, and a sweep line algorithm is used for enumeration to handle high-density edge cases, obtaining the shortest geodesic distance value of each graph node relative to the active frontier of the true region, and constructing the shortest path graph accordingly. When the calculated density is lower than the density threshold, the current mask region is determined to be in a sparse edge mode, and a priority queue-optimized Dijkstra algorithm is used to perform shortest path relaxation calculation on the local visibility graph, and constructing the shortest path graph accordingly.

[0014] Optionally, the method further includes: The state transition equation for performing the shortest path relaxation calculation is as follows: ; in, Represents the set of true regions to nodes geodesic distance, Represents a node The set of adjacent nodes, Represents a node To the node The Euclidean distance.

[0015] Optionally, performing three-dimensional multiple pruning and state iteration update on the shortest path graph includes: Spatial domain pruning, semantic domain pruning, and computational cost pruning are performed on the shortest path graph to obtain a simplified feature set. The geodesic distance between the candidate patches and the real region is calculated on the simplified feature set. Candidate patches whose geodesic distance is not greater than the distance threshold are merged into the real region, and the real region set and active frontier are updated. The spatial domain pruning includes: limiting the retrieval range through local topological masks and performing spatial calculations only on regions adjacent to the active frontier; the semantic domain pruning includes: marking the status bits of patches that have been incorporated into the true region and prohibiting repeated topological intersections and distance calculations; the computational cost pruning includes: dynamically pruning obstacle elements and reducing the number of vertices participating in the visibility graph construction to a constant level.

[0016] Optionally, obtaining the final precisely reachable true region range includes: Determine whether there are candidate patches to be merged into the true region. If there are no candidate patches to be merged into the true region, output the final accurate reachable true region range. If there are candidate patches to be merged into the true region, destroy the buffer operator, visibility map and intermediate calculation data generated in this round, retain only the updated vector boundary and status identifier, and rebuild the local buffer mask until no new patches are merged into the true region, and output the final accurate reachable true region range.

[0017] To achieve the above objectives, the present invention also provides a regional accessibility evolution system based on local topological masking and dynamic spatial pruning, comprising: The dynamic mask generation module is used to load the set of topological connected points, reconstruct the global free space topology based on the set of topological connected points, extract the external topological boundary of the converged true region set as the active frontier, generate a local search domain mask, and perform local pruning within the global free space topology. The local ESPO solution module is used to construct a local visibility map within the local search domain mask, and simultaneously switch adaptive solution algorithms according to the density of obstacle line segments within the local search domain mask, in order to perform shortest path relaxation calculation on the local visibility map to obtain the shortest path map; The three-dimensional pruning control module is used to perform three-dimensional multiple pruning and state iteration updates on the shortest path graph to obtain the final accurate reachable true region range.

[0018] The beneficial effects of this invention are as follows: This invention employs a local topological mask iterative calculation and an instant memory release mechanism. Memory usage is only linearly related to the perimeter of the active frontier boundary and is independent of the global geographic area. It can stably support accessibility calculations at the scale of large scale and tens of millions of features, avoiding the memory explosion overflow of traditional global raster calculations.

[0019] This invention supports vector-level preprocessing of linear obstacles and topologically connected openings such as bridges, culverts, and tunnels. It solves the real geodesic distance using the ESPO algorithm, accurately restoring the geographic spatial traffic rules. The reachable range closely matches the real area boundary, and the accuracy is higher than that of traditional grid obstacle bypass.

[0020] This invention reduces the dimensionality of the global indiscriminate search to local iterative computation through triple pruning of the spatial domain, semantic domain, and computational cost domain. Combined with the obstacle density adaptive algorithm switching, it balances the computational efficiency of densely obstacle areas in cities and sparsely obstacle areas in rural areas, and the overall convergence speed is significantly improved compared with traditional tools.

[0021] This invention uses vector geographic features as the computational basis, avoiding feature boundary truncation and floating-point precision loss caused by rasterization. The true area extraction results can be directly used in engineering scenarios with stringent requirements for topological accuracy, such as land spatial planning and emergency rescue.

[0022] This invention can be widely applied to scenarios such as land and space planning, transportation facility service circle assessment, ecological corridor connectivity analysis, emergency evacuation range delineation, and natural resource supervision. It takes into account the three core requirements of large scale, high precision, and complex obstacles, and fills the gap in the field of large-scale high-precision accessibility analysis in existing technologies. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart of a regional accessibility evolution method based on local topological masking and dynamic spatial pruning according to an embodiment of the present invention. Figure 2 This is a schematic diagram of a regional accessibility evolution system based on local topological masking and dynamic spatial pruning, according to an embodiment of the present invention. Detailed Implementation

[0025] 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, and 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.

[0026] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0027] like Figure 1 As shown, this embodiment discloses a regional accessibility evolution method based on local topological masks and dynamic spatial pruning, including: loading a set of topological connected points; constructing a local buffer mask based on the set of topological connected points to form a computational base map that conforms to physical access rules; simultaneously extracting the external topological boundary of the converged true region set as an active frontier to generate a local search domain mask and performing local pruning; constructing a local visibility map within the local search domain mask; and simultaneously switching adaptive solution algorithms according to the density of obstacle line segments within the local search domain mask to perform shortest path relaxation calculations on the local visibility map to obtain a shortest path map; and performing three-dimensional multi-pruning and state iterative updates on the shortest path map to obtain the final accurate reachable true region range.

[0028] Specifically, this embodiment discloses a region accessibility evolution method based on local topological masking and dynamic spatial pruning, the method comprising the following steps: Step S1: Obstacle Space Preprocessing and Topology Connectivity: Collect the initial true region seed set, candidate patch set, linear obstacle set, and topological connected point set; construct a local buffer mask based on the topological connected points, perform Boolean difference operation on the linear obstacles, erase the connected opening positions on the obstacles, generate an effective constraint set that conforms to the physical passage rules, reconstruct the free space topology, and eliminate the false blockage caused by data overlap.

[0029] Step S2: Dynamic active front extraction and local mask generation: Extract the external topological boundary of the currently converged true region set as the active front; based on the preset distance threshold, generate a local search domain mask through Minkowski sum operation, retain only the candidate patches and obstacle elements within the mask, complete the first round of spatial domain pruning, and reduce the global search to a local restricted space solution.

[0030] Step S3: Local ESPO solution and adaptive algorithm switching: Within the local mask area, a visibility map is constructed based on obstacle vertices and source points; the solution algorithm is adaptively switched according to the obstacle line segment density within the mask. In dense obstacle areas, the sweep line algorithm is used for enumeration and judgment, while in sparse obstacle areas, the Dijkstra algorithm with priority queue optimization is used for path relaxation; the shortest path graph (SPM) is constructed to achieve fast and accurate local geodesic distance calculation.

[0031] Step S4: 3D Multiple Pruning and State Iteration Update: Execute a triple pruning mechanism of spatial domain pruning, semantic domain pruning, and computational cost domain pruning to eliminate redundant elements and redundant calculations, decouple the size of obstacle vertices involved in the solution from the global data growth, and strictly reduce and control it within the constant range; determine the geodesic distance between the candidate map patch and the current real area, and merge the candidate map patches that meet the distance threshold into the real area set, and update the active frontier and the state of the real area.

[0032] Step S5: Dynamic memory management and iterative convergence: After each iteration, the temporarily generated buffer operator, visibility graph and intermediate calculation data are immediately destroyed, and only the true region vector boundary and status identifier are retained; when no new map features are added to the true region in this iteration, the algorithm terminates and the final output can accurately reach the true region range.

[0033] Further, step S1 includes the following steps: Step S11: Load the initial true region seed set A, the candidate patch set B, and the linear obstacle set. and the set of topological connected points ; where sets A and B are both composed of vector polygons that support spatial indexing; Step S12: Construct an influence radius based on the set D of topologically connected points. Local buffer mask ; using the above For linear obstacle sets Perform vector-level Boolean difference operations ( Erasing linear obstacle elements generates a valid set of constraints that conform to physical traffic rules. .

[0034] Step S13: Based on the effective constraint set Reconstruct the global free space topology F, which represents the continuous traversable space that satisfies the physical traversal rules after removing the effective constraint set within the study area.

[0035] Further, step S2 includes the following steps: Step S21: Extract the set of converged true regions in the current t-th iteration. The external topological boundary as an active frontier ; ; in, For each candidate image patch, For the set of candidate image patches, To initialize the reachability status flags of the candidate polygons.

[0036] Step S22: Based on the preset geodesic distance threshold The local search domain mask is generated using the Minkowski Sum operation. Its mathematical expression is ; Step S23: Utilize the local search domain mask Within the global free space topology F, perform local clipping on global features, retaining only the set of candidate patches within the mask. and obstacle element subset The first round of spatial domain pruning was completed.

[0037] Furthermore, step S3 includes the following sub-steps: Step S31: Masking the local search domain Inside, a subset of obstacle elements The endpoints of line segments, the active frontier of the true region The feature points and the feature points of the candidate patches that fall within the mask are used as the graph node set. Construct a local visibility graph For graph node sets Any two nodes in and If and only if the line segment The internal non-obstacle element subset When any obstacle line segment intersects, determine Add a visible edge to the edge set. Its mathematical condition is: ; Step S32: Based on the local search domain mask The density of inner obstacle line segments is adaptively calculated using a different algorithm, specifically: calculating the mask. Internal barrier element subset The total number of line segments or its ratio to the mask area is calculated and compared with a preset density threshold. When the calculated density is higher than the preset threshold, the current mask area is determined to be a dense edge mode. A sweep line algorithm is used for enumeration to handle high-density edge cases, obtaining the shortest geodesic distance value of each graph node relative to the active frontier of the real area, and constructing a shortest path graph accordingly. When the calculated density is lower than the preset threshold, the current mask area is determined to be a sparse edge mode. A priority queue-optimized Dijkstra algorithm is used to optimize the constructed visibility graph. Shortest path relaxation calculations are performed, and a shortest path graph is constructed based on this. The state transition equation is as follows: ; in Represents the set of true regions to nodes geodesic distance, Represents a node The set of adjacent nodes, Represents a node To the node The Euclidean distance; Step S33: When it is necessary to solve the mask from the active front of the true region to the local search domain. When calculating the geodesic distance field of any continuous spatial point, a shortest path graph is constructed based on the solutions obtained in the above steps; the shortest path graph maps local masks through shortest path mapping. The corresponding planar space is divided into multiple regional units, and all continuous points within each regional unit have the same "predecessor vertex". Thus, when querying the reachability distance of any continuous point within the mask, the system can locate the regional unit to which the point belongs and calculate its distance to the predecessor vertex, achieving a fast solution with constant time complexity.

[0038] Further, step S4 includes the following sub-steps: Step S41: Spatial domain pruning: Limit the search range by using a local topological mask, and perform spatial calculations only on the regions adjacent to the active front, that is, limit the calculation range to the neighborhood of the active front and exclude irrelevant elements outside the mask; Step S42: Semantic domain pruning: Mark the status bits of the patches that have been merged into the real region, and prohibit repeated topological intersection and distance calculation. That is, remove the patches that have been merged into the real region from the candidate patches to obtain the subset of candidate patches that really need to be determined for reachability in this round. Step S43: Calculate cost pruning: Dynamically prune obstacle features to reduce the number of vertices participating in the visibility graph construction to a constant level, that is, reduce the number of obstacle vertices participating in the visibility graph construction to a constant level, and obtain a simplified feature set; Step S44: Calculate the geodesic distance between the candidate patch and the true region on the simplified feature set, merge the patches with a distance ≤ threshold L into the true region, and update the true region set and active frontier.

[0039] Further, step S5 includes the following steps: Step S51: After each iteration, immediately destroy the local search domain mask, visibility map, and intermediate computation data generated in this round, retaining only the updated vector boundaries and state identifiers. The intermediate computation data includes the temporary computation products generated during the ESPO solution in step S3. The state identifier is updated to the patch that has been merged into the true region, and its semantic state bit is flipped from "pending" to "converged / merged into true region". In step S44, updating the true region set and active frontier involves performing a vector polygon merging operation to update the vector boundaries of the true region set, updating the state identifier of the corresponding patch from the pending state to the converged state, and simultaneously updating the active frontier.

[0040] Step S52: Determine whether any new patches have been added to the true region in this iteration. If so, proceed to iteration t+1 and return to step S2 until no new patches are added to the true region. If not, the algorithm converges and outputs the final vector-level high-precision true region result.

[0041] In one possible implementation, this embodiment provides a region accessibility evolution method based on local topological masking and dynamic spatial pruning, aiming to solve the problems of memory overflow, severe computational redundancy, and insufficient topological accuracy in existing technologies when dealing with region accessibility analysis in large-scale, complex obstacle environments. This invention reduces the global pathfinding problem to a local iterative solution and combines it with a three-dimensional multi-pruning strategy to achieve efficient and high-precision extraction of the "true region".

[0042] like Figure 1 As shown, the method specifically includes the following steps: Step S1: Load basic data through the GIS engine, including: an initial set of true area seeds A, a set of candidate polygons B, a set of linear obstacles C (such as highways, rivers, etc.), and a set of topologically connected points D (such as bridges, culverts, tunnel entrances, etc.). Sets A and B are both vector polygons that support R-tree or quadtree spatial indexing. For handling linear obstacles and connected pathways, this step constructs a local buffer mask with an influence radius of ϵ based on the set of topologically connected points D. Its mathematical expression is: ; Using the local buffer mask Perform vector-level Boolean difference operations on the linear obstacle set C to generate the effective constraint set. The calculation formula is: This operation erases connecting points such as bridges and culverts on linear obstacles, opening up free space that was previously blocked by obstacles. Finally, based on the effective constraint set... Reconstructing the global free space topology F, defined as By eliminating false blocking elements caused by data overlap, a computational base map that conforms to physical access rules is formed.

[0043] Step S2: In each iteration, first extract the set of currently converged true regions. The external topological boundary as an active frontier Based on a preset geodesic distance threshold L, a local search domain mask is generated using Minkowski summation. This process reduces the global search to a local constrained space solution, and the mathematical expression for mask generation is: ; Using the generated mask Within the global free space topology F, perform local clipping on global features, retaining only the set of candidate patches within the mask. and obstacle element subset The first round of spatial domain pruning was completed. The formula for generating the local set of features to be solved is: ; Step S3: Apply a local mask Within the specified range, obstacle vertices, source points, and feature points of candidate patches are used as nodes to construct a local visibility map. Vertex set Depend on The endpoints and related boundary points constitute the boundary, any two points There are connected edges The mathematical condition is that the line segment does not intersect any local constraint line. The determination formula is: ; An adaptive algorithm switching is implemented based on the density of obstacle line segments within the local mask: when the density is higher than a preset threshold, it is determined to be a dense edge mode, and a sweep line algorithm is used for enumeration and judgment; when the density is lower than the preset threshold, it is determined to be a sparse edge mode, and a priority queue-optimized Dijkstra algorithm is used for path relaxation. If a continuous distance field needs to be calculated, a shortest path graph is constructed, and the plane is divided into multiple region units with the same predecessor vertex through shortest path mapping, realizing fast calculation of the distance field from a single source to any continuous point within the mask.

[0044] Step S4: This step implements a triple pruning mechanism to eliminate redundant elements and duplicate calculations: Spatial domain pruning is achieved by limiting the search range through a local topological mask, performing spatial calculations only on regions adjacent to the active frontier; semantic domain pruning is achieved by marking the status bits of map features already incorporated into the real region and prohibiting repeated topological intersections and distance calculations; and obstacle elements are dynamically pruned, reducing the number of vertices participating in the visibility graph construction to a constant level. Subsequently, the geodesic distance between the candidate map features and the real region is calculated, and the constructed map is then... The shortest path relaxation solution is performed, and the state transition equation is: ; Step S5: After each iteration, immediately destroy the buffer operators, visibility maps, and intermediate computation data generated in this round, retaining only the updated vector boundaries and status markers. This mechanism ensures that memory usage is linearly related only to the perimeter of the active frontier boundary and is independent of the global geographic area. Determine if any new polygons have been added to the true region in this iteration; if so, proceed to the next iteration; otherwise, the algorithm converges and outputs the final vector-level high-precision true region result.

[0045] like Figure 2As shown, this embodiment discloses a regional accessibility evolution system based on local topological masking and dynamic spatial pruning, including: The dynamic mask generation module is used to load the set of topological connected points, construct a local buffer mask based on the set of topological connected points, form a computational base map that conforms to the physical passage rules, and extract the external topological boundary of the converged true region set as the active frontier to generate a local search domain mask and perform local pruning. The local ESPO solution module is used to construct a local visibility map within the local search domain mask, and simultaneously switch adaptive solution algorithms according to the density of obstacle line segments within the local search domain mask, in order to perform shortest path relaxation calculation on the local visibility map to obtain the shortest path map; The three-dimensional pruning control module is used to perform three-dimensional multiple pruning and state iteration updates on the shortest path graph to obtain the final accurate reachable true region range.

[0046] Specifically, this embodiment discloses a regional accessibility evolution system based on local topological masks and dynamic spatial pruning, including: The topology preprocessing module is used to load the initial true region seed set, candidate patch set, linear obstacle set, and topological connected point set, perform obstacle erasure and topology connection operations, and construct a free space computational base map that conforms to the physical access rules; The dynamic mask generation module is used to extract the active frontier of the current true region set, generate a local search domain mask based on a preset distance threshold, and complete the spatial range constraint and the first round of spatial pruning. The local ESPO solution module is used to construct a visibility map within a local mask, adaptively switch the shortest path solution algorithm according to the obstacle density, and complete the solution of geodesic distance and reachability relationship with obstacles. The 3D pruning control module is used to perform triple pruning in the spatial domain, semantic domain, and computational cost domain, eliminating redundant elements and repeated calculations, and reducing the time and space complexity of the algorithm. The iterative evolution execution module is used to determine the reachability of candidate patches, merge newly added true region patches that meet the threshold, and update the active frontier and the set of true regions. The dynamic memory management module is used to release temporary computational data during the iteration process in real time, control the peak memory usage, and ensure that there is no memory overflow in large-scale data computation.

[0047] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for regional accessibility evolution based on local topological masking and dynamic spatial pruning, characterized in that, include: Load the set of topological connected points, reconstruct the global free space topology based on the set of topological connected points, extract the external topological boundary of the converged true region set as the active frontier, generate a local search domain mask, and perform local pruning within the global free space topology; A local visibility map is constructed within the local search domain mask. At the same time, an adaptive solution algorithm is switched according to the density of obstacle line segments within the local search domain mask to perform shortest path relaxation calculation on the local visibility map and obtain the shortest path map. The shortest path graph is subjected to three-dimensional multi-pruning and state iteration update to obtain the final accurate reachable true region range.

2. The region accessibility evolution method based on local topological masking and dynamic spatial pruning according to claim 1, characterized in that, Reconstructing the global free space topology based on the aforementioned set of topological connected points includes: A local buffer mask is constructed based on the set of topological connected points. The local buffer mask is used to perform vector-level Boolean difference operations on the set of linear obstacles to erase linear obstacle elements and generate a set of effective constraints that conform to the physical passage rules. The global free space topology is reconstructed using the set of effective constraints; the global free space topology represents a continuous traversable space that satisfies the physical traversal rules after removing the set of effective constraints within the study region.

3. The region accessibility evolution method based on local topological masking and dynamic spatial pruning according to claim 1, characterized in that, Performing the local cropping includes: The external topological boundary of the converged true region set is extracted as the active frontier, and the Minkowski sum operation is performed according to the preset geodesic distance threshold to generate the local search domain mask. The local search domain mask is used to locally prune global features within the global free space topology, retaining only the set of candidate patches and the subset of obstacle features within the mask, thus completing the first round of spatial domain pruning.

4. The region accessibility evolution method based on local topological masking and dynamic spatial pruning according to claim 3, characterized in that, Constructing a local visibility map within the local search domain mask includes: Within the local search domain mask, the local visibility graph is constructed using the endpoints of line segments from the obstacle feature subset, feature points of the active frontier of the true region, and feature points of candidate patches falling within the mask as graph node sets; the local visibility graph includes: graph node set and edge set; During the construction of the local visibility graph, when the interior of the line segment connecting any node in the graph node set does not intersect with any obstacle line segment in the obstacle element subset, it is determined as a visible edge and introduced into the edge set.

5. The region accessibility evolution method based on local topological masking and dynamic spatial pruning according to claim 3, characterized in that, The shortest path graph is obtained by: An adaptive solution algorithm is switched based on the density of obstacle line segments within the local search domain mask. The total number of line segments in the subset of obstacle elements within the local search domain mask or its ratio to the mask area is calculated to obtain the calculated density, which is then compared with a preset density threshold. When the calculated density is higher than the density threshold, the current mask region is determined to be in a dense edge mode, and a sweep line algorithm is used for enumeration to handle high-density edge cases, obtaining the shortest geodesic distance value of each graph node relative to the active frontier of the true region, and constructing the shortest path graph accordingly. When the calculated density is lower than the density threshold, the current mask region is determined to be in a sparse edge mode, and a priority queue-optimized Dijkstra algorithm is used to perform shortest path relaxation calculation on the local visibility graph, and constructing the shortest path graph accordingly.

6. The region accessibility evolution method based on local topological masking and dynamic spatial pruning according to claim 5, characterized in that, The method also includes: The state transition equation for performing the shortest path relaxation calculation is as follows: ； in, Represents the set of true regions to nodes geodesic distance, Represents a node The set of adjacent nodes, Represents a node To the node The Euclidean distance.

7. The region accessibility evolution method based on local topological masking and dynamic spatial pruning according to claim 1, characterized in that, The three-dimensional multiple pruning and state iteration update of the shortest path graph includes: Spatial domain pruning, semantic domain pruning, and computational cost pruning are performed on the shortest path graph to obtain a simplified feature set. The geodesic distance between the candidate patches and the real region is calculated on the simplified feature set. Candidate patches whose geodesic distance is not greater than the distance threshold are merged into the real region, and the real region set and active frontier are updated. The spatial domain pruning includes: limiting the retrieval range through local topological masks and performing spatial calculations only on regions adjacent to the active frontier; the semantic domain pruning includes: marking the status bits of patches that have been incorporated into the true region and prohibiting repeated topological intersections and distance calculations; the computational cost pruning includes: dynamically pruning obstacle elements and reducing the number of vertices participating in the visibility graph construction to a constant level.

8. The region accessibility evolution method based on local topological masking and dynamic spatial pruning according to claim 7, characterized in that, The range of the final precisely reachable true region includes: Determine whether there are candidate patches to be merged into the true region. If there are no candidate patches to be merged into the true region, output the final accurate reachable true region range. If there are candidate patches to be merged into the true region, destroy the buffer operator, visibility map and intermediate calculation data generated in this round, retain only the updated vector boundary and status identifier, and rebuild the local buffer mask until no new patches are merged into the true region, and output the final accurate reachable true region range.

9. A regional accessibility evolution system based on local topological masking and dynamic spatial pruning, implemented according to any one of claims 1-8, characterized in that, include: The dynamic mask generation module is used to load the set of topological connected points, reconstruct the global free space topology based on the set of topological connected points, extract the external topological boundary of the converged true region set as the active frontier, generate a local search domain mask, and perform local pruning within the global free space topology. The local ESPO solution module is used to construct a local visibility map within the local search domain mask, and simultaneously switch adaptive solution algorithms according to the density of obstacle line segments within the local search domain mask, in order to perform shortest path relaxation calculation on the local visibility map to obtain the shortest path map; The three-dimensional pruning control module is used to perform three-dimensional multiple pruning and state iteration updates on the shortest path graph to obtain the final accurate reachable true region range.

Images