GIS (Geographic Information System) data quick retrieval and visualization method fusing spatial-temporal characteristics

Abstract

The invention discloses a GIS data rapid retrieval and visualization method fusing spatio-temporal characteristics, and relates to the technical field of spatio-temporal geographic information systems.The method comprises the steps that a dynamic grid is divided according to data distribution entropy, and a spatio-temporal association index is constructed through a 3D-Hilbert algorithm; weighting the spatial-temporal characteristics by adopting an entropy weight method, and converting a query range into coding intervals for screening and sorting; dynamically adapting a visualization strategy according to the result density; the rendering precision is adjusted according to the camera distance and the screen error, and strategies are updated and switched in real time; newly-added data is inserted or a cube is newly built according to needs, index balance is maintained, and reconstruction is carried out; a space-time association network is constructed, and association rules are mined by means of a community discovery algorithm. Through innovative index construction, feature fusion retrieval and dynamic visualization, the GIS data retrieval efficiency and accuracy are improved, the visualization effect is optimized, data updating and deep analysis are supported, interactivity is enhanced, and the data processing value is improved in multi-field application.

Description

Technical Field

[0001] This invention relates to the field of spatiotemporal geographic information system technology, and in particular to a method for rapid retrieval and visualization of GIS data that integrates spatiotemporal features. Background Technology

[0002] With the widespread application of Geographic Information Systems (GIS) in smart cities, transportation and logistics, environmental monitoring, and other fields, GIS data is characterized by its multi-source nature, spatiotemporal dynamics, and massive growth. Traditional GIS data retrieval methods often employ single spatial indexing techniques, such as quadtrees and R-trees, which can only meet the query needs of static spatial data. When faced with dynamic data with a time dimension, such as traffic flow changes and disaster evolution processes, these methods cannot effectively integrate temporal and spatial characteristics, resulting in low retrieval efficiency and difficulty in quickly locating and obtaining the required data. For example, in urban traffic congestion analysis, traditional methods struggle to quickly retrieve the congestion evolution of a specific time period and area from a large amount of historical data.

[0003] Existing GIS data visualization methods also have significant shortcomings. For massive datasets, fixed visualization strategies are typically employed, such as simple point, line, and polygon plots or heatmap displays, without considering the spatiotemporal distribution and density differences of the data. In data-dense areas, visualization results are prone to information overlap and visual confusion; while in data-sparse areas, they fail to fully display data characteristics. Furthermore, traditional visualization lacks interactivity and adaptability; when users perform zooming, filtering, or other operations, visualization parameters cannot be adjusted in a timely manner, making it difficult for users to quickly extract useful information from complex data.

[0004] In terms of spatiotemporal data processing, existing technologies have limited capabilities for incremental updates and in-depth analysis. With the increasing frequency of real-time GIS data acquisition, efficient processing of incremental data has become crucial. However, traditional index structures often require significant computational resources to maintain index balance during data updates, leading to performance degradation. Furthermore, existing methods for mining spatiotemporal relationships between data largely rely on simple proximity calculations, failing to deeply analyze complex spatiotemporal patterns and evolutionary laws, thus hindering the development of in-depth GIS data applications. Summary of the Invention

[0005] The present invention proposes a method for rapid retrieval and visualization of GIS data that integrates spatiotemporal features, in order to solve the problems mentioned in the prior art.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A method for rapid retrieval and visualization of GIS data integrating spatiotemporal features includes the following steps: The steps for constructing a spatiotemporal fusion index are as follows: GIS data is spatially divided into dynamic grids, with the grid size determined based on the data distribution entropy. Grids in densely populated areas are automatically shrunk, while those in sparse areas are automatically enlarged. Spatiotemporal cubes are constructed by time slices, and the spatiotemporal centroid coordinates are mapped to one-dimensional codes using the 3D-Hilbert algorithm. A spatiotemporal association index structure is established to realize the association between spatial and temporal dimensions. Weighted feature fusion retrieval steps: Extract spatial and temporal features of GIS elements, calculate weight coefficients using the entropy weight method and fuse features; receive the spatiotemporal query range, convert it into Hilbert-coded intervals to filter candidate cubes, and sort candidate elements by combining spatiotemporal overlap rate and improved cosine similarity algorithm to generate retrieval result set; Dynamic adaptive visualization steps: Based on the spatiotemporal distribution density of the search results, dynamically adjust the visualization strategy: use hierarchical aggregation display for high density, display spatiotemporal trajectory lines for low density, and use heatmap visualization for medium density to adapt to different data density scenarios. Interactive-driven visualization optimization steps: Receive user operations, dynamically adjust rendering accuracy based on camera distance and feature screen space error, and update spatiotemporal distribution density and switch visualization strategies in real time during filtering operations. Incremental update and index optimization steps: When adding new data, calculate the spatiotemporal distance with adjacent cubes. If the distance is less than a threshold, insert it into an existing cube; otherwise, create a new cube. By maintaining an index balance factor, index reconstruction is triggered when the factor exceeds a threshold to ensure index efficiency. The steps of deep spatiotemporal correlation analysis are as follows: construct a spatiotemporal correlation network for the search results, with elements as nodes and spatiotemporal proximity between elements as edge weights. Use community discovery algorithms to identify related communities, mine spatiotemporal correlation rules for high-density communities, and reveal deep relationships in the data.

[0007] Furthermore, in the weighted feature fusion retrieval step, the spatiotemporal overlap rate influence factor for similarity calculation is dynamically adjusted according to the query type. The more query elements or the smaller the candidate set size, the stronger the influence factor's adjustment effect.

[0008] Furthermore, in the dynamic adaptive visualization step, the spatiotemporal distribution density threshold is determined based on the total amount of data and the spatial volume, supporting user zooming, panning, and filtering operations, and adjusting the rendering accuracy and visualization strategy in real time.

[0009] Furthermore, in the incremental update and index optimization steps, when creating a new spatiotemporal cube, the spatial grid and time slice parameters of the adjacent cubes are inherited, and dynamically adjusted according to the distribution characteristics of the newly added data.

[0010] Furthermore, in the spatiotemporal correlation depth analysis step, effective spatiotemporal correlation rules are filtered by setting thresholds for support exceeding 0.1 and confidence exceeding 0.7 to ensure rule reliability.

[0011] Furthermore, the spatiotemporal fusion index construction, weighted feature fusion retrieval, and dynamic adaptive visualization steps are implemented on the distributed computing framework Spark, and the spatiotemporal cube is allocated to different computing nodes for processing through a partitioning strategy.

[0012] Furthermore, in the spatiotemporal fusion index construction step, a Redis database is used to store the spatiotemporal index data, and the key-value pair storage characteristics are used to realize the mapping query between Hilbert encoding and spatiotemporal data.

[0013] Compared with existing technologies, the beneficial effects of this invention are: In data retrieval, an innovative spatiotemporal fusion indexing technology is employed to deeply link spatial and temporal dimensions. Compared to traditional single-spatial indexing methods, this significantly improves retrieval efficiency and enables rapid location of target information within massive amounts of spatiotemporal data. The weighted feature fusion retrieval strategy comprehensively considers both the spatial and temporal characteristics of the data, thereby enhancing retrieval accuracy and effectively addressing the challenge of handling complex spatiotemporal queries using traditional methods.

[0014] In terms of visualization, the dynamic adaptive visualization strategy automatically switches the display mode according to the spatiotemporal distribution density of the data. In densely populated areas, it uses hierarchical aggregation to avoid information congestion; in sparse areas, it uses trajectory lines to clearly present data changes, improving the clarity of the visualization effect. Interactive visualization optimization allows the system to adjust visualization parameters in real time when users perform zooming and filtering operations, significantly enhancing the convenience for users to obtain information.

[0015] For data updating and analysis, incremental updates and index optimization steps ensure the efficiency of the index when data changes dynamically, reducing index maintenance time. In-depth spatiotemporal correlation analysis can uncover complex spatiotemporal relationships and evolutionary patterns between data, providing more valuable decision support for applications such as urban planning and disaster early warning. This method improves GIS data processing performance while expanding the depth and breadth of data applications, possessing significant economic and social value. Attached Figure Description

[0016] Figure 1 This is a schematic block diagram of the GIS data rapid retrieval and visualization method that integrates spatiotemporal features proposed in this invention; Figure 2 A line graph showing the spatiotemporal distribution density and the corresponding visualization strategy; Figure 3 A bar chart comparing the user interaction performance of different methods. Detailed Implementation

[0017] 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.

[0018] The present invention will now be described in further detail with reference to the accompanying drawings.

[0019] Reference Figures 1 to 3 A method for rapid retrieval and visualization of GIS data integrating spatiotemporal features, comprising the following steps: The spatiotemporal fusion index construction steps are as follows: When spatially partitioning GIS data, the system adopts a dynamic grid strategy based on Shannon entropy. First, the target area is divided into initial grids, with grid sizes ranging from 100m x 100m to 10km x 10km depending on the estimated data density. By counting the number of features within each grid, the data distribution entropy value for that area is calculated to assess the data density. A higher data distribution entropy value indicates a denser data distribution in that area.

[0020] For data-intensive areas, such as commercial districts and transportation hubs in city centers, when the calculated entropy value is significantly higher than the sum of the global average entropy value and the standard deviation, the system will automatically trigger a grid subdivision mechanism. This mechanism will halve the current grid side length successively until a preset minimum grid granularity (e.g., 10 meters × 10 meters) is reached or the entropy value is reduced to a reasonable range. This subdivision operation ensures that the amount of data within each grid is maintained at an appropriate level, avoiding a decrease in indexing efficiency due to excessive data concentration.

[0021] Conversely, in sparsely populated areas such as suburbs, rural areas, or nature reserves, if the entropy value is lower than the difference between the global average entropy value and the standard deviation, the system will perform a grid merging operation to merge adjacent grids to expand the grid side length, up to a maximum of 50 km × 50 km, thereby reducing the number of empty grids and improving index storage efficiency.

[0022] In terms of time, the system divides time into fixed slice units based on data update frequency and business needs. For frequently updated data, such as traffic flow data and real-time monitoring data, smaller time slices are used, such as 15 minutes or 1 hour; for infrequently updated data, such as annual statistics and quarterly planning data, larger time slices are used, such as 1 day, 1 week, or 1 month.

[0023] Spatial grids and time slices intersect to form spatiotemporal cubes. Each spatiotemporal cube is mapped to a unique one-dimensional code using a 3D-Hilbert algorithm. Hilbert coding is a space-filling curve algorithm that maintains consistency between spatial proximity and coded proximity; that is, spatially adjacent spatiotemporal cubes also have similar coded values. When processing the time dimension, timestamps need to be converted to values ​​on the same scale as spatial coordinates. Specifically, the total time range of data collection is set (e.g., from start time t1 to end time t2), and any timestamp t is mapped proportionally to the [0,1] interval, i.e., through... The time-normalized value is calculated. This value is then combined with the spatial coordinates (x, y) to form the three-dimensional coordinates (x, y, t). norm ), where t norm This is a time-normalized value, thus achieving a unified representation of spatiotemporal dimensions. During the encoding process, the spatiotemporal centroid coordinates (including spatial and temporal coordinates) need to be normalized. During normalization, the spatial coordinates (x, y) are mapped to [0, 1] by dividing by the maximum range value of the corresponding dimension (e.g., the maximum length of the x-axis and the maximum length of the y-axis). The timestamp is normalized according to the aforementioned time range ratio method. The encoding order is selected based on the precision requirements of the application scenario: the higher the order, the finer the spatiotemporal division. For example, 10-order encoding can achieve approximately 1 meter of spatial precision and 1 second of temporal precision, suitable for city-level fine-grained queries; 8-order encoding corresponds to 4 meters of spatial precision and 4 seconds of temporal precision, suitable for regional-level macro-analysis.

[0024] The index data is stored in a Redis database, leveraging its key-value pair storage feature for efficient querying. Key naming conventions combine the spatiotemporal coding range and feature type; for example, "point:HCode:1000-2000" represents an index of point features with Hilbert codes between 1000 and 2000. Values ​​contain metadata and a data pointer: the metadata records the spatial range (represented by the coordinates of the lower left and upper right corners, such as x1, y1, x2, y2), the time interval (start and end times), and the number of features; the data pointer indicates the storage path of the original data in a distributed file system (such as an HDFS path).

[0025] To ensure precise alignment between spatial coordinates and image pixels, the system introduces a spatiotemporal calibration mechanism. A probabilistic graphical model is constructed using a Bayesian network to model and iteratively optimize the errors in positioning coordinates and camera extrinsic parameters. Specifically, the system utilizes sensor observation data and updates the posterior probability distributions of positioning coordinates and camera extrinsic parameters through Bayesian inference, thereby controlling the calibration error within ±0.1 meters to meet the requirements of high-precision visualization and analysis.

[0026] Weighted Feature Fusion Retrieval Steps: The system supports multi-dimensional feature extraction. For different types of GIS elements, spatial and temporal features can be extracted customizablely. Spatial features include, but are not limited to, the geometric area, perimeter, center point coordinates, distance from other elements, road grade, building height, etc.; temporal features include timestamps, frequency of element status changes, event occurrence time periods, data update cycles, etc.

[0027] The weighting calculation employs the entropy weighting method, an objective weighting method based on the inherent variability of the data. First, all feature values ​​are standardized, converting features of different dimensions into dimensionless values ​​within the [0,1] interval. The standardization method can be either max-min standardization or Z-score standardization, depending on the feature type. For example, for the numerical feature "area," the max-min standardization formula can be used: , where r ij Let x represent the dimensionless value of the j-th feature of the i-th element after max-min standardization. ij Let j be the feature value of the i-th element. and These are the minimum and maximum values ​​of the j-th feature among all elements, respectively. For features that follow a normal distribution (such as temperature and velocity), Z-score standardization is used, with the formula: , where μ j Let σ be the mean of the j-th feature among all factors. j The standard deviation of the j-th feature across all features. If the feature value is zero or the denominator is zero when normalized, a minimum value ϵ (e.g., 1 × 10⁻⁶) needs to be added to the original feature value. -6 Smoothing is performed, that is... Then substitute the values ​​into the formula to calculate and avoid errors.

[0028] After standardization, the entropy value of each feature is calculated. The entropy value reflects the uncertainty of feature information; the smaller the entropy value, the higher the information discrimination of that feature, and the greater its impact on the retrieval results. The formula for calculating the entropy value is: Where m is the total number of elements, e j To represent the entropy value of the j-th feature, This is the sum of the values ​​of all m features for the j-th feature, used for normalization calculations. i is the index of the feature, j is the index of the feature, and r... ij Let be the value of the i-th element on the j-th feature. Finally, calculate the feature weights based on the entropy values: Where k is the total number of features, w j Let the weight of the j-th feature be , For all k features Perform summation.

[0029] When a user initiates a spatiotemporal query, the system first converts the query range into the corresponding Hilbert-coded intervals, thereby filtering out spatiotemporal cubes that may contain the target element as a candidate set. When handling cross-interval queries, the system first converts the query range into multiple adjacent Hilbert-coded intervals (e.g., if the query area covers the code 1000-3000, it is split into two intervals: 1000-2000 and 2001-3000). It then uses Redis's batch range query function to obtain the index entries for all intervals, merges them to form a candidate set, and finally filters them based on spatiotemporal overlap and feature similarity to ensure complete retrieval results. The filtering process utilizes Redis's range query function to quickly locate index entries that match the coded intervals and obtain the candidate cube's metadata and data pointers.

[0030] Next, the candidate elements are ranked based on similarity. This ranking process combines two metrics: spatiotemporal overlap rate and improved cosine similarity. The spatiotemporal overlap rate measures the degree of spatiotemporal intersection between the query scope and the candidate spatiotemporal cube; a higher overlap rate indicates a higher spatiotemporal match between the candidate elements and the query requirements. Improved cosine similarity considers feature weights, evaluating the feature similarity between the query element feature vector and the candidate element feature vector by calculating the weighted cosine value.

[0031] To adapt to different query needs, the system introduces a dynamic adjustment factor to regulate the weights of spatiotemporal overlap rate and improved cosine similarity. The value of the dynamic adjustment factor is related to the number of query elements and the size of the candidate set. The dynamic adjustment factor is quantified by the number of query elements N and the size of the candidate set M: when N>50 or M<100, it is determined as "many query elements" or "small candidate set", and the weight of spatiotemporal overlap rate is increased; when N≤50 and M≥100, the weight of feature similarity is increased. The threshold is determined experimentally to balance retrieval efficiency and accuracy in different scenarios.

[0032] Dynamic adaptive visualization steps: The system automatically switches visualization strategies based on the spatiotemporal distribution density. The formula for calculating the spatiotemporal distribution density is: Where m is the number of features within the visualization area, and V is the spatiotemporal volume of the visualization area (equal to the product of spatial volume and time span). The system calculates the average density by statistically analyzing the density distribution of global data. and density standard deviation And thus set a density threshold: when the density is high: In this case, a hierarchical aggregation display strategy is adopted. For example, in the visualization of POI data in the city center, POI elements of the same category (such as dining, shopping, etc.) are aggregated into one icon, and the quantity statistics of elements of that category are displayed on the icon, avoiding visual clutter caused by a large number of elements being stacked. When medium density: When using heatmap visualization, a heatmap strategy is employed. GPU-accelerated technology is used to generate heatmaps, and density distribution is mapped using color gradients; darker colors indicate higher density, making it easier for users to intuitively identify data hotspots. When low density is present: At that time, a spatiotemporal trajectory line display strategy is adopted. For trajectory data, the movement path of the feature is drawn point by point, and the trajectory points are color-coded in chronological order. For example, from red to blue represents the time change from the past to the present. Users can play animations through timeline controls to analyze the dynamic change trend of the features.

[0033] To enhance the user experience, the system implements a real-time visualization optimization mechanism based on user actions. When a user zooms in, the system accurately senses changes in camera distance. If the user zooms in on the map, it means the camera distance has decreased. In this case, the system responds quickly, automatically loading higher-precision feature models and textures, making the details on the map clearer and more realistic, with building outlines and road signs displayed in exquisite detail. Conversely, when the user zooms out on the map, meaning the camera distance has increased, to avoid excessive rendering load affecting smoothness, the system simplifies the display of features, such as switching the originally detailed 3D models to simpler 2D point symbols. During panning operations, the system also dynamically adjusts the rendering precision based on the screen space error of the features. The screen space error is calculated using the camera focal length and pixel size. For example, when the camera is 10 meters away from the target, 1 meter of actual error corresponds to approximately 100 pixels of error. When the error exceeds 50 pixels, the precision is automatically increased. Simplification strategies include: using vertex clustering for 3D models, retaining key contour vertices (such as reducing secondary detail vertices); and adjusting texture resolution according to the object's screen ratio to ensure a stable rendering frame rate of over 30 FPS. Through this series of intelligent and efficient processing steps, the frame rate is ensured to remain stable at over 30 FPS, bringing users a smooth and comfortable interactive experience.

[0034] When users perform filtering operations, whether by feature type, defined time range, or specific attribute conditions, the system reacts quickly. It recalculates the spatiotemporal distribution density of the filtered data in real time and with high accuracy. This calculation is crucial because the system flexibly switches visualization strategies based on density changes. For example, when switching from displaying all vehicle trajectories to only showing trajectories of vehicles traveling at speeds exceeding 60 km / h, the system automatically abandons the hierarchical aggregation display method and switches to a heatmap display once the density changes from high to medium. Through such dynamic adjustments, the system ensures that the visualization effect is always perfectly adapted to the data density, allowing users to obtain clear, intuitive, and effective data presentations at all times.

[0035] This invention also includes incremental update and index optimization steps: When new data is added, the system first calculates the spatiotemporal distance between the new data and adjacent spatiotemporal cubes. The calculation of spatiotemporal distance comprehensively considers spatial distance and temporal distance, where the spatial distance uses the Euclidean distance formula, and the temporal distance is converted into an equivalent spatial distance by setting a time weighting coefficient. The specific calculation formula is as follows: Where D represents the spatiotemporal distance, This represents the coordinates of an object in space (corresponding to the coordinates of another object). ), t1 represents the time value of one object (the corresponding time value of another object is t2). Time weighting coefficient (unit: meters per second) 2 The time weighting coefficient can be set according to business needs. Adjust according to the business's time sensitivity: Real-time data (such as traffic flow) should be set to 1000 m / s. 2 Emphasis is placed on temporal proximity; low-frequency data (such as annual statistics) are taken as 100 m / s. 2 To mitigate the impact of time, the spatiotemporal distance threshold is set based on the data update frequency: 50 m / s for high-frequency data (updated every minute) and 500 m / s for low-frequency data (updated every day), ensuring that newly added data is properly categorized. The Hilbert code for the newly created cube is generated using the 3D-Hilbert algorithm and written to the Redis database.

[0036] The system monitors the balance of the index structure by maintaining an index balance factor. The index balance factor, essentially a ratio, is used to accurately measure the degree of balance in the distribution of features within the index structure. It is calculated as follows: ,in This indicates the number of elements contained in the cube. This refers to the number of features contained in the cube with the most features in the current index; while This factor represents the number of elements in the cube with the fewest elements. When the factor exceeds 1.5, index access latency increases significantly and storage efficiency decreases, triggering a local restructuring (migrating some elements from the cube with the most elements to neighboring low-density cubes). If the factor is ≥2.0, a global restructuring is performed, redistributing all elements using a genetic algorithm to balance the index load and maintain efficient system operation. By calculating this ratio, the system can clearly understand the differences in element distribution within the index structure, thus providing a strong basis for subsequent index structure optimization and ensuring efficient system operation.

[0037] When the index balancing factor exceeds a preset threshold (e.g., 1.5), the system triggers an index reconstruction mechanism. Reconstruction levels are categorized into local reconstruction and global reconstruction based on the factor size. when In this case, only the cube with the most elements is partially migrated, and elements exceeding the average number are migrated to adjacent empty cubes or low-density cubes to balance the index load of local areas.

[0038] when At that time, a full-range index reconstruction is performed, and all elements are redistributed to the spatiotemporal cube using optimization algorithms such as genetic algorithms. During the reconstruction process, the system ensures the continuity of query operations through a temporary caching mechanism to avoid service interruption due to reconstruction.

[0039] This invention also includes a spatiotemporal correlation depth analysis step: the system constructs a spatiotemporal correlation network using GIS elements as nodes and the spatiotemporal proximity between elements as edge weights. The spatiotemporal proximity is calculated using the reciprocal of the spatiotemporal distance, i.e. W ij Let D be the edge weight between element i and element j. ij Let i be the spatiotemporal distance between element i and element j. A smoothing factor (e.g., 0.01) is used to avoid cases where the denominator is zero. Furthermore, network edge weights can be combined with the semantic relevance of elements, such as calculating the semantic similarity of element category names using natural language processing techniques, to further enhance the network's expressive power.

[0040] The community detection algorithm employs the Louvain algorithm, which identifies tightly connected communities in a network by optimizing the modularity metric. Modularity measures the strength of the network's community structure, and its calculation formula is as follows: Where m is the number of network edges, k i and k j Let c be the degree of node i and node j, respectively. i and c j The community tag to which the node belongs, Q represents the modularity, and A represents the community tag to which the node belongs. ij For adjacency matrix elements, The indicator function takes a value of 1 when node i and node j belong to the same community, and 0 otherwise. The association rule thresholds (support > 0.1, confidence > 0.7) are general empirical values ​​applicable to most common scenarios. In high-density data areas (such as urban center business districts), where element associations are more concentrated, the thresholds can be increased to support > 0.15 and confidence > 0.8; in low-density data areas (such as suburbs or rural areas), they can be appropriately relaxed to support > 0.05 and confidence > 0.6 to adapt to the analytical needs of different domains and data distribution characteristics.

[0041] After identifying high-density communities, the system performs association rule mining on the elements within these communities. The reliability of the association rules is measured using two metrics: support and confidence. Support represents the frequency of a rule's occurrence in the dataset. The formula for calculating support is: ,in This represents the number of records that contain both element A and element B. This represents the total number of records in the dataset.

[0042] Confidence level represents the probability that a record containing element A also contains element B. The confidence level is calculated using the following formula: .

[0043] The system defines rules as valid if the support is greater than 0.1 and the confidence is greater than 0.7. For example, the rule "shopping mall → parking lot" mined from business data has a support of 0.15 and a confidence of 0.82.

[0044] In this invention, the core computing tasks of the distributed computing and storage system, such as spatiotemporal fusion index construction, weighted feature fusion retrieval, and dynamic adaptive visualization, are all implemented using the Spark distributed computing framework. Spark, with its powerful distributed processing capabilities, provides efficient computational support for the system. Through a carefully designed and reasonable partitioning strategy, the system can accurately allocate the spatiotemporal cube to different computing nodes. These computing nodes can perform parallel processing simultaneously, much like multiple workers working concurrently, greatly improving data processing efficiency, effectively shortening processing time, and accelerating data processing speed. This allows the system to quickly and accurately handle the complex computational needs of massive amounts of data, providing users with a smoother and more efficient user experience.

[0045] The partitioning strategy employs a composite approach of "spatiotemporal coding prefix hash + time range": Spatiotemporal coding prefix hash partitioning: Selecting the first few bits of the Hilbert code, such as the first 20 bits, as the hash key. Hilbert coding effectively maintains spatial proximity; using this as the hash key, spatiotemporal cubes can be accurately assigned to different partitions. The advantage of this approach is that it ensures that spatially adjacent cubes are highly likely to be distributed within the same or adjacent partitions. This significantly reduces cross-partition data interaction during data processing and querying. Because spatially adjacent data is often correlated, frequent data transfer between different partitions is unnecessary during related operations, improving data processing efficiency and reducing system overhead.

[0046] Time range partitioning: Time range partitioning divides data based on time slices. For example, different partitions can be created by day or month. This partitioning method greatly facilitates time range queries. When a user retrieves data within a specific time period, the system can quickly locate the corresponding partition and retrieve the data without having to traverse all the data. Furthermore, for incremental data updates, time range partitioning allows for more organized processing. New data can be accurately assigned to the corresponding partition based on its time attribute, ensuring the orderly storage and management of data.

[0047] When processing large-scale data, the parallel computing capabilities of the Spark framework can significantly improve processing efficiency. The Redis database is used to store spatiotemporal index data, leveraging its high-concurrency read / write features and memory caching mechanism to achieve fast mapping and querying between Hilbert codes and spatiotemporal data. To improve performance in high-concurrency scenarios, the system adopts the following optimization measures: Cluster Architecture: Deploying a RedisCluster cluster is a crucial foundation for ensuring efficient system operation. Through an advanced consistent hashing algorithm, the system intelligently and evenly distributes data across multiple master and slave nodes. The number of nodes is designed to store 1 million indexes per node, with shards 3-5 times the number of nodes. RDB snapshots are generated hourly, and AOF logs are synchronized every second, ensuring data security and query performance. This data distribution method achieves a perfect combination of data sharding and load balancing. In actual operation, different data is rationally allocated across various nodes, avoiding overloading of any single node. The entire cluster possesses powerful processing capabilities; a single cluster can easily support tens of thousands of query requests per second, ensuring stable and efficient system response even under high concurrency.

[0048] Data persistence: A persistence strategy combining RDB snapshots and AOF logs is employed. RDB snapshots periodically perform full backups of the index data, much like taking a complete photograph of the data at a specific point in time, recording the overall data structure at that moment. Meanwhile, the AOF log records write operations in real time, meticulously documenting every data write change. These two methods work together to maximize data integrity in the event of system failures or other unforeseen circumstances. Furthermore, data can be quickly restored based on backups and log records, allowing the system to rapidly return to normal operation.

[0049] Caching Strategy: For frequently accessed index data, such as the spatiotemporal cube index of the city center area, the system employs an LFU (Least Frequently Used) caching strategy. This strategy statistically analyzes the access frequency of data, gradually evicting the least frequently accessed data while retaining the frequently accessed data in memory. When users request this frequently accessed data again, it is retrieved directly from memory without needing to be read from disk, significantly reducing disk I / O operations. Actual verification shows that this caching strategy can reduce query latency to below 10 milliseconds, significantly improving system response speed and providing users with an extremely smooth user experience.

[0050] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for rapid retrieval and visualization of GIS data integrating spatiotemporal features, characterized in that, Includes the following steps: The steps for constructing a spatiotemporal fusion index are as follows: GIS data is spatially divided into dynamic grids, with the grid size determined based on the data distribution entropy. Grids in densely populated areas are automatically shrunk, while those in sparse areas are automatically enlarged. Spatiotemporal cubes are constructed by time slices, and the spatiotemporal centroid coordinates are mapped to one-dimensional codes using the 3D-Hilbert algorithm. A spatiotemporal association index structure is established to realize the association between spatial and temporal dimensions. Weighted feature fusion retrieval steps: Extract spatial and temporal features of GIS elements, calculate weight coefficients using the entropy weight method and fuse features; receive the spatiotemporal query range, convert it into Hilbert-coded intervals to filter candidate cubes, and sort candidate elements by combining spatiotemporal overlap rate and improved cosine similarity algorithm to generate retrieval result set; Dynamic adaptive visualization steps: Based on the spatiotemporal distribution density of the search results, dynamically adjust the visualization strategy: use hierarchical aggregation display for high density, display spatiotemporal trajectory lines for low density, and use heatmap visualization for medium density to adapt to different data density scenarios. Interactive-driven visualization optimization steps: Receive user operations, dynamically adjust rendering accuracy based on camera distance and feature screen space error, and update spatiotemporal distribution density and switch visualization strategies in real time during filtering operations. Incremental update and index optimization steps: When adding new data, calculate the spatiotemporal distance with adjacent cubes. If the distance is less than a threshold, insert it into an existing cube; otherwise, create a new cube. By maintaining an index balance factor, index reconstruction is triggered when the factor exceeds a threshold to ensure index efficiency. The steps of deep spatiotemporal correlation analysis are as follows: construct a spatiotemporal correlation network for the search results, with elements as nodes and spatiotemporal proximity between elements as edge weights. Use community discovery algorithms to identify related communities, mine spatiotemporal correlation rules for high-density communities, and reveal deep relationships in the data.

2. The method for rapid retrieval and visualization of GIS data integrating spatiotemporal features according to claim 1, characterized in that, In the weighted feature fusion retrieval step, the spatiotemporal overlap rate influence factor for similarity calculation is dynamically adjusted according to the query type. The more query elements or the smaller the candidate set size, the stronger the influence factor's adjustment effect.

3. The method for rapid retrieval and visualization of GIS data integrating spatiotemporal features according to claim 1, characterized in that, In the dynamic adaptive visualization step, the spatiotemporal distribution density threshold is determined based on the total amount of data and spatial volume, supporting user zooming, panning, and filtering operations, and adjusting rendering accuracy and visualization strategy in real time.

4. The method for rapid retrieval and visualization of GIS data integrating spatiotemporal features according to claim 1, characterized in that, In the incremental update and index optimization steps, when creating a new spatiotemporal cube, the spatial grid and time slice parameters of the adjacent cubes are inherited, and dynamically adjusted according to the distribution characteristics of the newly added data.

5. The method for rapid retrieval and visualization of GIS data integrating spatiotemporal features according to claim 1, characterized in that, In the spatiotemporal correlation depth analysis step, effective spatiotemporal correlation rules are filtered by setting thresholds for support exceeding 0.1 and confidence exceeding 0.7 to ensure rule reliability.

6. The method for rapid retrieval and visualization of GIS data integrating spatiotemporal features according to claim 1, characterized in that, The spatiotemporal fusion index construction, weighted feature fusion retrieval, and dynamic adaptive visualization steps are implemented on the distributed computing framework Spark, and the spatiotemporal cube is allocated to different computing nodes for processing through a partitioning strategy.

7. The method for rapid retrieval and visualization of GIS data integrating spatiotemporal features according to claim 1, characterized in that, In the spatiotemporal fusion index construction step, a Redis database is used to store spatiotemporal index data, and the key-value pair storage characteristics are used to realize the mapping query between Hilbert encoding and spatiotemporal data.

Images