Spatial range search method and system based on adaptive grid and sdmr tree fusion

Abstract

The application discloses a spatial range search method and system based on adaptive grid and SDMR tree fusion, relates to the technical field of spatial data set retrieval and data management, and obtains the local region data distribution density of a spatial data set to be indexed, generates a dynamic adaptive grid based on the local region data distribution density of the spatial data set to be indexed, extracts each grid unit of the dynamic adaptive grid, obtains grid feature metadata, and the grid feature metadata at least includes minimum circumscribed rectangle information, screens candidate tree nodes of a pre-constructed SDMR tree index layer that have a spatial overlap relationship with a query range, obtains a first candidate set, filters the first candidate set based on spatial position similarity and grid density similarity, obtains a second candidate set, sorts the second candidate set, and outputs the spatial data set according to the sorting result as a spatial range search result.

Description

Technical Field

[0001] This invention relates to the field of spatial dataset retrieval and data management technology, specifically a spatial range search method and system based on the fusion of adaptive grid and SDMR tree. Background Technology

[0002] Existing spatial indexing methods mainly include tree-based spatial indexing, grid indexing, space-filled curve indexing, and the learning-based spatial indexing developed in recent years. Among them, R-tree is a classic dynamic spatial indexing structure that organizes multidimensional spatial objects using minimum bounding rectangles and supports spatial queries and dynamic updates, and has been widely used in spatial databases and geographic information systems. Guttman's R-tree organizes spatial objects through hierarchical minimum bounding rectangles, providing a basic indexing framework for multidimensional spatial data retrieval.

[0003] Existing spatial indexing methods are inefficient for range queries on large-scale spatial datasets because they cannot fully extract features such as density and distribution entropy from the spatial data. Furthermore, the fixed granularity of the fixed grid index is pre-set and cannot adaptively adjust the grid granularity according to the distribution of the spatial data. Summary of the Invention

[0004] To address the shortcomings mentioned in the background section, the present invention aims to provide a spatial range search method and system based on the fusion of adaptive grid and SDMR tree.

[0005] Firstly, the objective of this invention can be achieved through the following technical solution: a spatial range search method based on the fusion of adaptive mesh and SDMR tree, the method comprising the following steps: Obtain the local region data distribution density of the spatial dataset to be indexed, and generate a dynamic adaptive grid based on the local region data distribution density of the spatial dataset to be indexed, wherein the coarseness and fineness of the dynamic adaptive grid are inversely proportional to the magnitude of the local region data distribution density. Extract each grid cell of the dynamic adaptive grid to obtain grid feature metadata, which includes at least the minimum bounding rectangle information. Filter the pre-constructed SDMR tree index layer for candidate tree nodes that have spatial overlap with the query range to obtain the first candidate set. The SDMR tree index layer is constructed based on the minimum bounding rectangle information, and the tree nodes of the SDMR tree index layer are bound to the corresponding grid feature metadata. Similarity is calculated based on grid feature metadata to obtain spatial location similarity and grid density similarity. Based on spatial location similarity and grid density similarity, the first candidate set is filtered by feature similarity to obtain the second candidate set. The second candidate set is sorted, and the spatial dataset output according to the sorting result is used as the spatial range search result.

[0006] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the generation process of generating a dynamic adaptive grid based on the local region data distribution density of the spatial dataset to be indexed, comprising: Data distribution density in each local region of a statistical spatial dataset; The regions in the spatial dataset to be indexed whose local data distribution density is higher than a preset first density threshold are divided into fine-grained grids. Regions in the spatial dataset to be indexed whose local data distribution density is lower than a preset second density threshold are divided into coarse-grained grids or adjacent grids are merged. Wherein, the first density threshold is greater than or equal to the second density threshold.

[0007] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the SDMR tree index layer is an R-tree variant oriented towards spatial datasets, wherein leaf nodes store the minimum bounding rectangle of a single spatial dataset and its bound grid feature metadata, and intermediate nodes store the merge range of the minimum bounding rectangle of their child nodes and the corresponding feature statistics.

[0008] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the feature similarity is calculated by weighting and summing the spatial location similarity and the grid density similarity.

[0009] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: when the feature similarity is lower than a preset similarity threshold, removing the corresponding candidate tree node or candidate grid from the first candidate set to obtain a second candidate set.

[0010] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: during the process of sorting the second candidate set, a multidimensional weighted similarity is established by using spatial overlap rate similarity, grid density distribution similarity and data distribution feature similarity for sorting, and local outlier factors are combined to suppress abnormal candidate data.

[0011] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the local outlier factor is used to identify abnormal candidate data in the candidate results, and when the local outlier factor of the candidate data is higher than a preset outlier threshold, the weight of the candidate data in the ranking results is reduced.

[0012] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the grid feature metadata also includes grid center point coordinates, grid boundary range, data point density, data distribution entropy, covered dataset identifier list, spatial storage estimate, number of data points within the grid, data distribution statistics corresponding to the grid, and grid level information.

[0013] Secondly, in order to achieve the above objectives, this invention discloses a spatial range search system based on the fusion of adaptive grid and SDMR tree, comprising: The data acquisition module is used to acquire the local region data distribution density of the spatial dataset to be indexed, and generate a dynamic adaptive grid based on the local region data distribution density of the spatial dataset to be indexed, wherein the coarseness and fineness of the dynamic adaptive grid are inversely proportional to the magnitude of the local region data distribution density. The data processing module is used to extract each grid cell of the dynamic adaptive grid to obtain grid feature metadata. The grid feature metadata includes at least the minimum bounding rectangle information. The module filters candidate tree nodes that have spatial overlap with the query range in the pre-constructed SDMR tree index layer to obtain the first candidate set. The SDMR tree index layer is constructed based on the minimum bounding rectangle information, and the tree nodes of the SDMR tree index layer are bound to the corresponding grid feature metadata. The search output module is used to calculate spatial location similarity and grid density similarity based on grid feature metadata, filter the first candidate set based on spatial location similarity and grid density similarity to obtain the second candidate set, sort the second candidate set, and output the spatial dataset based on the sorting result as the spatial range search result.

[0014] The beneficial effects of this invention are: This invention uses a dynamic adaptive grid to characterize the spatial data distribution features, enabling the index structure to automatically adjust the grid granularity according to the different data characteristics of dense and sparse regions. This reduces empty and overloaded grids, improving the rationality of spatial partitioning. By binding grid feature metadata with SDMR tree nodes, traditional tree indexes can not only use MBR for spatial range filtering but also use features such as grid density, data distribution entropy, and covered dataset identifiers for candidate set compression, improving the quality of candidates after coarse filtering. A two-level progressive filtering mechanism reduces invalid candidate data from participating in subsequent ranking calculations, lowering computational overhead during the query process. By constructing a multi-dimensional weighted similarity ranking model and combining it with local outlier factors for outlier suppression, the stability and accuracy of Top-k search results are improved, avoiding excessive interference from abnormal spatial data on the final ranking results. It supports local incremental updates; when spatial data is inserted, deleted, or undergoes local changes, only the corresponding grid feature metadata and associated SDMR tree node information are updated, without global index reconstruction, thus improving update efficiency in dynamic spatial data scenarios. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of a two-layer index structure that integrates a dynamic adaptive grid and an SDMR tree, provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of the spatial dataset range search process provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of dynamic adaptive mesh partitioning provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the incremental index update process provided in an embodiment of the present invention; Figure 6 This is a diagram showing the average query time comparison of different indexing methods under different data scales provided in the embodiments of the present invention; Figure 7 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0016] 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.

[0017] Example 1: like Figure 1 As shown, the spatial extent search method based on the fusion of adaptive mesh and SDMR tree includes the following steps: S101: Obtain the local region data distribution density of the spatial dataset to be indexed, and generate a dynamic adaptive grid based on the local region data distribution density of the spatial dataset to be indexed, wherein the coarseness and fineness of the dynamic adaptive grid are inversely proportional to the magnitude of the local region data distribution density. In this context, a local region refers to a basic statistical unit region obtained according to a spatial partitioning strategy, the extent of which is determined by the initial spatial partitioning granularity. In this embodiment, an initial uniform partitioning region is first established based on the overall boundary of the spatial data, dividing the entire spatial range into several basic statistical units. The local data density is calculated by counting the number of data objects within each basic statistical unit. Furthermore, the statistical range is dynamically adjusted based on density changes in adjacent regions, allowing regions with obvious spatial clustering characteristics to form independent local regions.

[0018] The spatial dataset to be indexed undergoes preprocessing operations such as coordinate normalization, initial outlier removal, spatial boundary determination, and data density statistics to ultimately obtain a set of spatial data points, a global spatial boundary, and local regional data distribution density. Among these, the local regional data distribution density plays a dominant role in subsequent applications. Specifically, the spatial dataset is selected as the data source, and the data is uniformly converted to Cartesian coordinates according to geographic projection rules, while dirty data such as out-of-bounds coordinates and single-point anomalies is removed. The process of generating a dynamic adaptive grid based on the local region data distribution density of the spatial dataset to be indexed includes: like Figure 4 As shown, the data distribution density of each local region in the statistical spatial dataset; The regions in the spatial dataset to be indexed whose local data distribution density is higher than a preset first density threshold are divided into fine-grained grids. Regions in the spatial dataset to be indexed whose local data distribution density is lower than a preset second density threshold are divided into coarse-grained grids or adjacent grids are merged. Wherein, the first density threshold is greater than or equal to the second density threshold.

[0019] In this embodiment, the density threshold is determined based on the average density μ and density standard deviation σ of the spatial dataset, wherein the first density threshold T1 is set to μ + σ, and the second density threshold T2 is set to μ - 0.5σ. When the density ρ ≥ T1 in a local area, it is determined to be a high-density area, and fine-grained grids are used; when ρ ≤ T2, it is determined to be a low-density area, and coarse-grained grids or adjacent grid merging strategies are used; standard-grained grids are used for the remaining areas.

[0020] For regions where the data density is between two thresholds, a standard medium-granularity grid is used to generate an adaptive grid set with uneven density distribution.

[0021] S102: Extract each grid cell of the dynamic adaptive grid to obtain grid feature metadata, which includes at least the minimum bounding rectangle information. Filter the pre-constructed SDMR tree index layer for candidate tree nodes that have spatial overlap with the query range to obtain the first candidate set. The SDMR tree index layer is constructed based on the minimum bounding rectangle information, and the tree nodes of the SDMR tree index layer are bound to the corresponding grid feature metadata. Specifically, the grid feature metadata also includes the grid center point coordinates, the range of the four sides, the number of valid points inside the grid, the grid point density, the data distribution entropy, the grid-associated dataset ID list, the minimum bounding rectangle (MBR) of the dataset, and the grid storage estimate and grid level number information are supplemented as needed to achieve a complete set of feature metadata for each grid.

[0022] The SDMR tree index layer is a variant of the R-tree for spatial datasets. During its construction, the minimum bounding rectangle (MBR) for each spatial data object is first calculated, including the coordinates of its lower left corner (xmin, ymin) and upper right corner (xmax, ymax). Then, a bottom-up node aggregation method is used, selecting node merging objects based on the spatial distance of node MBRs and the cost of area expansion, forming leaf nodes and non-leaf nodes. During node insertion, a minimum area increment strategy is used to determine the insertion path. When the node capacity exceeds the preset maximum capacity M, a node splitting operation is performed, thus generating the SDMR tree index structure. Leaf nodes store the minimum bounding rectangle of a single spatial dataset and its bound mesh feature metadata, while intermediate nodes store the merging range of their child nodes' minimum bounding rectangles and corresponding feature statistics.

[0023] The tree nodes in the SDMR tree index layer are bound to the corresponding grid feature metadata of the grid cells to form a two-layer fused index structure, wherein, referring to Figure 2A two-layer index binding structure diagram is constructed based on the minimum bounding rectangle of each dataset to build an SDMR tree index. The SDMR tree is an improved variant of the R tree. The leaf nodes store the single dataset MBR and corresponding grid feature metadata, the intermediate nodes store the merged MBR of the subordinate child nodes and the feature statistical mean, and the root node stores the overall boundary information of the entire domain. A one-to-one binding relationship is established between the nodes of each level of the SDMR tree and the underlying grid metadata to complete the construction of the two-layer fused index.

[0024] Reference Figure 3 The retrieval process receives the query range R input by the user, traverses the SDMR tree from top to bottom, determines whether there is a spatial intersection between the minimum bounding rectangle of the node and the query range, directly prunes nodes that do not intersect, and summarizes nodes that have spatial overlap to form the first candidate set. S103: Calculate spatial location similarity and grid density similarity based on grid feature metadata. Filter the first candidate set based on feature similarity using spatial location similarity and grid density similarity to obtain the second candidate set. Sort the second candidate set and output the spatial dataset based on the sorting result as the spatial range search result.

[0025] Feature similarity is calculated by weighting and summing spatial location similarity and grid density similarity, as shown in the following formula: Sim=α·Sim pos +β·Sim den Among them, Sim pos Sim represents spatial similarity. den This represents the similarity of grid density, where α and β are weighting coefficients, and α+β=1.

[0026] When the feature similarity is lower than the preset similarity threshold, the corresponding candidate tree node or candidate grid is removed from the first candidate set to obtain the second candidate set.

[0027] In the process of ranking the second candidate set, a multi-dimensional weighted similarity is established by spatial overlap rate similarity, grid density distribution similarity and data distribution feature similarity for ranking, and local outlier factors are combined to suppress abnormal candidate data.

[0028] The comprehensive score of the candidate space dataset is calculated according to the following formula: Score=w1·S overlap +w2·S density + w3·S distribution Among them, S overlap S represents the spatial overlap similarity. density S represents the similarity of grid density distribution. distributionThe similarity of data distribution characteristics is represented by w1, w2, and w3, which are weight coefficients, and w1 + w2 + w3 = 1. The Local Outlier Factor (LOF) is used to identify anomalous candidate data in the candidate results. In this embodiment, the LEF is calculated based on the density reachability distance between candidate spatial data objects. First, the average distance between a candidate object and its k-neighboring objects is calculated. Then, the Local Reachability Density (LRD) is calculated based on the average neighborhood density. Finally, the LOF value is obtained by comparing the local reachability density of the current object with the average local reachability density of its neighboring objects. LOF(p) = avg(LRD(o) / LRD(p)) Where p represents the current candidate object, and o represents the k-neighbor objects of p. When LOF(p) > 1, it indicates that the object has more obvious abnormal characteristics compared to its neighboring objects, and its weight is reduced during the ranking stage.

[0029] Reference Figure 5 The update flowchart allows for the location of the corresponding grid based on the coordinates of the data points when data addition or deletion occurs. It also updates metadata such as the number of points, density, and distribution entropy within the grid and synchronously corrects the information of the bound SDMR leaf nodes and intermediate nodes. The entire process does not require a global reconstruction of the index.

[0030] To further verify the effectiveness of the spatial range search method based on the fusion of adaptive mesh and SDMR tree proposed in this invention, the performance of the method is verified below with specific experiments.

[0031] This embodiment uses a Python program to simulate the generation of urban spatial points of interest (POI) data, and tests the query performance of the method of the present invention under different spatial data scale conditions.

[0032] The experiment was set up with a two-dimensional spatial region of 10000m×10000m, generating a total of 100000 spatial objects, each containing two-dimensional spatial coordinate information (x,y).

[0033] To simulate the non-uniform distribution characteristics present in actual urban spatial data, the experimental data were generated using a combination of clustered and random distributions.

[0034] In some spatial areas, spatial objects are generated using a two-dimensional Gaussian distribution to create a high data density, which is used to simulate the concentrated distribution of spatial objects in the city center. In other areas, spatial objects are generated using a uniform random method to simulate the sparse distribution of spatial objects in the city periphery.

[0035] By using the above data generation method, the experimental data are made to have a spatially uneven distribution, thereby verifying the adaptability of the method of the present invention to regions with different spatial densities.

[0036] The experiment first preprocesses the generated spatial object data, calculates the spatial data distribution density based on the number of spatial objects in a local area, and completes adaptive mesh generation according to the density adjustment strategy proposed in this invention.

[0037] Subsequently, an SDMR tree index structure is constructed based on the minimum bounding rectangle (MBR) information corresponding to each grid cell, and spatial extent information, number of spatial objects, and local data density information are associated with the corresponding index nodes.

[0038] During the experiment, datasets with 10,000, 30,000, 50,000, and 100,000 spatial objects were constructed respectively.

[0039] For each data size, a spatial range query region is randomly generated, and comparative experiments are conducted using a fixed grid indexing method, a traditional SDMR tree indexing method, and the method proposed in this invention based on the fusion of adaptive grid and SDMR tree.

[0040] Each experiment performed 100 random spatial range queries, and the average query time was recorded.

[0041] Experimental results show that as the scale of spatial data increases, the average query time of various indexing methods all show an increasing trend.

[0042] However, since the method of the present invention first dynamically adjusts the mesh division granularity according to the spatial object distribution density, a finer granular index structure is used in high-density areas and a coarser granular index structure is used in low-density areas, thereby reducing access to invalid spatial areas.

[0043] Meanwhile, this invention further utilizes the SDMR tree structure to quickly filter irrelevant areas through the spatial relationship between the node MBR and the query range. Therefore, compared with the fixed grid index method and the traditional SDMR tree index method, it can maintain a low query time under different data scale conditions.

[0044] Specifically, the present invention will be further illustrated below through embodiments: Figure 2 This is a schematic diagram of the two-layer index structure that integrates the dynamic adaptive grid and SDMR tree of the present invention; the bottom dynamic adaptive grid layer stores metadata such as grid center point, density, and distribution entropy, while the upper SDMR tree index layer uses the dataset MBR as the node boundary, and the tree nodes are bound to the lower grid feature metadata; Figure 3This is a schematic diagram of the spatial dataset range search process of the present invention. The process is as follows: Start → Data Preprocessing → Dynamic Adaptive Grid Construction → SDMR Tree Coarse Filtering → Feature Similarity Fine Filtering → Multidimensional Weighted Similarity Ranking → LOF Outlier Suppression → Output Top-k → End; Figure 4 This is a schematic diagram of the global spatial adaptive meshing of the present invention, which includes two types of meshing forms: fine-grained mesh in dense regions and coarse-grained mesh in sparse regions. Figure 5 This is a schematic diagram of the incremental update process of the index in this invention. The steps are as follows: data insertion / deletion → locating the affected grid cell → updating the grid feature elements → synchronously updating the associated SDMR tree nodes → local update completed; Figure 6 This diagram illustrates the comparison of average query time for different indexing methods under different data scales, where the horizontal axis represents the spatial data scale and the vertical axis represents the average query time.

[0045] Example 2: To achieve the above objective, such as Figure 7 As shown, based on Embodiment 1, this invention discloses a spatial range search system based on the fusion of adaptive grid and SDMR tree, comprising: The data acquisition module 11 is used to acquire the local region data distribution density of the spatial dataset to be indexed, and generate a dynamic adaptive grid based on the local region data distribution density of the spatial dataset to be indexed, wherein the coarseness and fineness of the dynamic adaptive grid are inversely proportional to the magnitude of the local region data distribution density. Data processing module 12 is used to extract each grid cell of the dynamic adaptive grid to obtain grid feature metadata. The grid feature metadata includes at least the minimum bounding rectangle information. The pre-constructed SDMR tree index layer is used to filter candidate tree nodes that have spatial overlap with the query range to obtain the first candidate set. The SDMR tree index layer is constructed based on the minimum bounding rectangle information, and the tree nodes of the SDMR tree index layer are bound to the corresponding grid feature metadata. The search output module 13 is used to calculate spatial location similarity and grid density similarity based on grid feature metadata, filter the first candidate set based on spatial location similarity and grid density similarity to obtain the second candidate set, sort the second candidate set, and output the spatial dataset based on the sorting result as the spatial range search result.

[0046] Based on the same inventive concept, this invention also provides a computer device, comprising: one or more processors, and a memory for storing one or more computer programs; the programs include program instructions, and the processor executes the program instructions stored in the memory. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, used to implement one or more instructions, specifically for loading and executing one or more instructions stored in a computer storage medium to implement the above-described method.

[0047] It should be further explained that, based on the same inventive concept, the present invention also provides a computer storage medium storing a computer program, which, when executed by a processor, performs the above-described method. This storage medium can be any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In the present invention, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0048] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0049] The foregoing has shown and described the basic principles, main features, and advantages of this disclosure. Those skilled in the art should understand that this disclosure is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of this disclosure. Various changes and modifications can be made to this disclosure without departing from its spirit and scope, and all such changes and modifications fall within the scope of this disclosure as claimed.

Claims

1. A spatial range search method based on the fusion of adaptive mesh and SDMR tree, characterized in that, The method includes the following steps: Obtain the local region data distribution density of the spatial dataset to be indexed, and generate a dynamic adaptive grid based on the local region data distribution density of the spatial dataset to be indexed, wherein the coarseness and fineness of the dynamic adaptive grid are inversely proportional to the magnitude of the local region data distribution density. Extract each grid cell of the dynamic adaptive grid to obtain grid feature metadata, which includes at least the minimum bounding rectangle information. Filter the pre-constructed SDMR tree index layer for candidate tree nodes that have spatial overlap with the query range to obtain the first candidate set. The SDMR tree index layer is constructed based on the minimum bounding rectangle information, and the tree nodes of the SDMR tree index layer are bound to the corresponding grid feature metadata. Similarity is calculated based on grid feature metadata to obtain spatial location similarity and grid density similarity. Based on spatial location similarity and grid density similarity, the first candidate set is filtered by feature similarity to obtain the second candidate set. The second candidate set is sorted, and the spatial dataset output according to the sorting result is used as the spatial range search result.

2. The spatial range search method based on the fusion of adaptive mesh and SDMR tree according to claim 1, characterized in that, The process of generating a dynamic adaptive grid based on the local region data distribution density of the spatial dataset to be indexed includes: Data distribution density in each local region of a statistical spatial dataset; The regions in the spatial dataset to be indexed whose local data distribution density is higher than a preset first density threshold are divided into fine-grained grids. Regions in the spatial dataset to be indexed whose local data distribution density is lower than a preset second density threshold are divided into coarse-grained grids or adjacent grids are merged. Wherein, the first density threshold is greater than or equal to the second density threshold.

3. The spatial range search method based on the fusion of adaptive mesh and SDMR tree according to claim 1, characterized in that, The SDMR tree index layer is a variant of the R-tree oriented towards spatial datasets. Leaf nodes store the minimum bounding rectangle of a single spatial dataset and its bound grid feature metadata, while intermediate nodes store the merge range of the minimum bounding rectangle of their child nodes and the corresponding feature statistics.

4. The spatial range search method based on the fusion of adaptive mesh and SDMR tree according to claim 1, characterized in that, The feature similarity is calculated by weighting and summing the spatial location similarity and the grid density similarity.

5. The spatial range search method based on the fusion of adaptive mesh and SDMR tree according to claim 4, characterized in that, When the feature similarity is lower than a preset similarity threshold, the corresponding candidate tree node or candidate grid is removed from the first candidate set to obtain the second candidate set.

6. The spatial range search method based on the fusion of adaptive mesh and SDMR tree according to claim 1, characterized in that, In the process of sorting the second candidate set, a multi-dimensional weighted similarity is established by using spatial overlap rate similarity, grid density distribution similarity, and data distribution feature similarity to sort the data, and local outlier factors are combined to suppress abnormal candidate data.

7. The spatial range search method based on the fusion of adaptive mesh and SDMR tree according to claim 6, characterized in that, The local outlier factor is used to identify abnormal candidate data in the candidate results. When the local outlier factor of the candidate data is higher than the preset outlier threshold, the weight of the candidate data in the ranking results is reduced.

8. The spatial range search method based on the fusion of adaptive mesh and SDMR tree according to claim 1, characterized in that, The grid feature metadata also includes grid center point coordinates, grid boundary range, data point density, data distribution entropy, covered dataset identifier list, spatial storage estimate, number of data points within the grid, data distribution statistics corresponding to the grid, and grid level information.

9. A spatial range search system based on adaptive mesh and SDMR tree fusion, employing the spatial range search method based on adaptive mesh and SDMR tree fusion as described in any one of claims 1 to 8, characterized in that, include: The data acquisition module is used to acquire the local region data distribution density of the spatial dataset to be indexed, and generate a dynamic adaptive grid based on the local region data distribution density of the spatial dataset to be indexed, wherein the coarseness and fineness of the dynamic adaptive grid are inversely proportional to the magnitude of the local region data distribution density. The data processing module is used to extract each grid cell of the dynamic adaptive grid to obtain grid feature metadata. The grid feature metadata includes at least the minimum bounding rectangle information. The module filters candidate tree nodes that have spatial overlap with the query range in the pre-constructed SDMR tree index layer to obtain the first candidate set. The SDMR tree index layer is constructed based on the minimum bounding rectangle information, and the tree nodes of the SDMR tree index layer are bound to the corresponding grid feature metadata. The search output module is used to calculate spatial location similarity and grid density similarity based on grid feature metadata, filter the first candidate set based on spatial location similarity and grid density similarity to obtain the second candidate set, sort the second candidate set, and output the spatial dataset based on the sorting result as the spatial range search result.

10. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that, The memory stores a computer program that can run on the processor. When the processor loads and executes the computer program, it employs the spatial range search method based on the fusion of adaptive grid and SDMR tree as described in any one of claims 1 to 8.

Images