A geographic information collection method and system

Abstract

The application discloses a kind of geographic information acquisition method and system, it is related to geographic information acquisition technical field, to initial space grid remote sensing image set is carried out image projection consistency processing, output image slice set, calculate comprehensive quality score, construct fusion model, construct sample set and establish ground surface change evolution model, output virtual ground surface evolution sequence, calculate change intensity atlas and identify change area, extract change feature, generate structured semantic label, the present application is through multi-source remote sensing image fusion and ground surface evolution modeling, realize the prediction and reconstruction of occlusion area and unobserved period ground surface state, comprehensive meteorological data, solar elevation angle and occlusion rate assess image quality, improve the effectiveness of fusion image, construct ground surface change evolution model and change driving classification mechanism, generate structured semantic label, automatically update geographic information system database, enhance the integrity, timeliness and applicability of geographic information.

Description

Technical Field

[0001] This invention relates to the field of geographic information acquisition technology, and in particular to a geographic information acquisition method and system. Background Technology

[0002] With the development of remote sensing technology, geographic information acquisition based on satellite imagery has been widely applied in fields such as environmental monitoring, disaster assessment, and land use change analysis. Existing methods typically rely on remote sensing images from a single sensor, extracting surface information through image classification or change detection techniques. However, these methods are affected by factors such as cloud cover, inconsistent imaging times, differences in image resolution, and inconsistencies in spatial reference.

[0003] Currently, Chinese invention patent application number CN202411110877.6 discloses a geographic information mapping system based on remote sensing imagery, comprising: a first remote sensing image acquisition module to acquire primary geographic information; a feature marking module to process the primary geographic information, identify and mark feature characteristics; a map drawing module to calculate a hazard index and an importance index, and draw a hazard distribution map and an importance map; a safe area selection module to select the best safe area and establish a rescue location; a UAV path planning module to plan the optimal path based on node location, node weight, and distance between nodes; a second remote sensing image acquisition module to acquire and process remote sensing images using a UAV to obtain secondary geographic information; a cloud computing module to acquire and process geographic information to obtain a geographic information mapping map; and a geographic information display module to display the geographic information mapping map.

[0004] The aforementioned technologies are insufficient to achieve continuous evolution reconstruction of land surface conditions and identification of driving types when multi-source remote sensing data are obstructed, have inconsistent resolutions, or have time discontinuities. Summary of the Invention

[0005] The technical problem solved by this invention is that existing technologies are unable to achieve continuous evolution reconstruction of land surface conditions and identification of driving types when multi-source remote sensing data have occlusion, inconsistent resolution, and time discontinuities.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: A geographic information collection method includes the following steps: Step S1: Perform image projection consistency processing on the initial spatial raster remote sensing image set and output an image tile set; Step S2: Calculate the overall quality score, construct the fusion model, and generate multi-dimensional remote sensing fusion image; Step S3: Construct a sample set and establish a land surface change evolution model, outputting a virtual land surface evolution sequence; Step S4: Calculate the change intensity map and identify the change area, and output the intent recognition label; Step S5: Extract change features and generate structured semantic labels; Step S6: Input structured semantic tag data and update the geographic information system database.

[0007] Preferably, step S1 includes the following sub-steps: Step S101: Call the satellite remote sensing image database to obtain optical remote sensing images, radar remote sensing images and infrared remote sensing images within the set area; Step S102: Perform image projection consistency processing on optical remote sensing images, radar remote sensing images and infrared remote sensing images to unify spatial resolution and coordinate system, and output the initial spatial raster remote sensing image set. Step S103: Classify and index the images according to timestamps and image types, and output the image slice set.

[0008] Preferably, step S2 includes the following sub-steps: Step S201: Evaluate the quality score of remote sensing images based on meteorological data, solar altitude angle, and image occlusion rate. The meteorological data includes cloud cover data, visibility data, and rainfall intensity data. Calculate the initial quality score according to a preset weather impact factor calculation strategy. In the weather impact factor calculation strategy, solar altitude angle and image occlusion rate are negatively correlated with the corresponding scores. Normalize the meteorological data, solar altitude angle, and image occlusion rate. Calculate the comprehensive quality score corresponding to the initial spatial raster remote sensing image set according to the weather impact factor calculation strategy. Step S202: Set a time-fitting factor based on time proximity. The logic for setting the time-fitting factor is that time proximity is positively correlated with the corresponding time-fitting factor. An angle complementarity factor is set based on the angle coverage complementarity. The setting logic of the angle complementarity factor is that the angle coverage complementarity and the angle complementarity factor are positively correlated. An image quality factor is set based on image sharpness, and the setting logic of the image quality factor is that image sharpness and image quality factor are positively correlated. Step S203: Construct a fusion model based on the comprehensive quality score, event fitting factor, angle complementarity factor, and image quality factor, and perform fusion calculation on the initial spatial raster remote sensing image set to generate a multi-dimensional remote sensing fusion image.

[0009] Preferably, step S3 includes the following sub-steps: Step S301: Select image sequences containing the same observation area at different times from historical remote sensing image data, and the area has known surface state change annotations at different time nodes. Construct a sample set based on the image time series and annotation information. The sample set includes image sequences, change correspondences and surface classification information. Step S302: Establish a land surface change evolution model based on the sample set. The land surface change evolution model consists of a spatial feature extraction module, a time series modeling module, and a land surface state identification module. Step S303: Input the multi-dimensional remote sensing fusion image and target time into the land surface change evolution model. The output of the land surface change evolution model is a land surface state prediction layer corresponding to multiple time nodes. The land surface state prediction layer contains land surface type and change degree identifiers. Output the land surface state prediction layer as a virtual land surface evolution sequence. Step S304: Perform continuity and consistency detection on the virtual land performance sequence, correct non-contiguous areas between land state prediction layers, fill in boundary anomalies and undefined areas according to the land state of the nearest time node, and output the corrected land performance result.

[0010] Preferably, step S4 includes the following sub-steps: Step S401: Calculate the change intensity map based on each adjacent layer in the virtual performance sequence; Step S402: Identify regions in the intensity change map that exceed a set threshold and mark their spatial range; Step S403: Call the historical change type database, and determine the driving type of the change based on the morphological characteristics, duration and spatial location of the labeled change area through a preset classification model, and output the corresponding intent recognition label.

[0011] Preferably, step S5 includes the following sub-steps: Step S501: Extract change features based on the intent recognition label and the corresponding change area location. The change features include change period, change direction, change range and change frequency. Step S502: Generate structured semantic tags based on change characteristics. The structured semantic tags include geographic feature type, change type, spatiotemporal attributes, and multilingual description fields.

[0012] Preferably, step S6 includes the following sub-steps: Step S601: Input the structured semantic tag data into the geographic information system database according to the spatial range and time index method; Step S602: Compare the geographic information data before and after the update, identify the changes in added, adjusted or removed geographic elements and update the database records. Step S603: Synchronize the updated data to the map platform and data service platform through the geographic information system interface.

[0013] Preferably, the surface state filling in step S304 includes: selecting the surface state of the same area in the time node adjacent to the area to be filled as the filling basis, using the dominant category consistency strategy in the area to specify the missing state, and generating an updated virtual performance layer.

[0014] Preferably, the driving type in step S403 includes natural evolution type and human intervention type, which are classified by historical map matching and deep model judgment. The deep model is based on change contour, change rate and environmental context for identification.

[0015] A geographic information acquisition system, applied in the aforementioned geographic information acquisition method, includes an image tiling module, an image fusion module, a geographic performance module, an intent recognition module, a feature extraction module, and a data update module; The image slicing module is used to perform image projection consistency processing on the initial spatial raster remote sensing image set and output an image slice set. The image fusion module is used to calculate the comprehensive quality score, construct the fusion model, and generate multi-dimensional remote sensing fusion images. The land evolution module is used to construct a sample set and establish a land surface change evolution model, and output a virtual land evolution sequence; The intent recognition module is used to calculate the change intensity map and identify the change area, and output the intent recognition label; The feature extraction module is used to extract change features and generate structured semantic labels; The data update module is used to input structured semantic tag data and update the geographic information system database.

[0016] The beneficial effects of this invention are as follows: This invention achieves prediction and reconstruction of the surface state in occluded areas and unobserved periods by fusing multi-source remote sensing images and performing modeling of the land surface. It comprehensively evaluates image quality by integrating meteorological data, solar altitude angle and occlusion rate, improves the effectiveness of fused images, constructs a land surface change evolution model and change-driven classification mechanism, generates structured semantic tags, automatically updates the geographic information system database, and enhances the completeness, timeliness and applicability of geographic information. Attached Figure Description

[0017] Figure 1 A flowchart illustrating the steps of a geographic information acquisition method according to an embodiment of the present invention; Figure 2 This is a basic flowchart of a geographic information acquisition system provided in one embodiment of the present invention. Detailed Implementation

[0018] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0019] Example 1, referring to Figure 1 This paper provides a geographic information collection method, which includes the following steps: Step S1: Perform image projection consistency processing on the initial spatial raster remote sensing image set and output an image tile set.

[0020] Step S2: Calculate the overall quality score, construct the fusion model, and generate multi-dimensional remote sensing fusion image.

[0021] Step S3: Construct a sample set and establish a land surface change evolution model, and output a virtual land surface evolution sequence.

[0022] Step S4: Calculate the intensity change map and identify the change area, and output the intent recognition label.

[0023] Step S5: Extract change features and generate structured semantic labels.

[0024] Step S6: Input structured semantic tag data and update the geographic information system database.

[0025] Step S1 includes the following sub-steps: Step S101: Call the satellite remote sensing image database to obtain optical remote sensing images, radar remote sensing images and infrared remote sensing images within the set area.

[0026] Step S101 acquires optical remote sensing images, radar remote sensing images, and infrared remote sensing images within the target area to provide a multi-source input data basis for subsequent processing.

[0027] Step S102: Perform image projection consistency processing on optical remote sensing images, radar remote sensing images and infrared remote sensing images to unify spatial resolution and coordinate system, and output an initial spatial raster remote sensing image set.

[0028] Step S102 performs projection consistency processing on the multi-source remote sensing images to unify their spatial resolution and coordinate system, ensuring the consistency of data in spatial dimensions and outputting a standardized initial spatial raster remote sensing image set.

[0029] Step S103: Classify and index the images according to timestamps and image types, and output the image slice set.

[0030] Step S103 classifies and indexes the initial spatial raster remote sensing image set according to timestamps and image types, forming a structured image tile set, thereby improving the efficiency of data retrieval and subsequent fusion.

[0031] Step S1 achieves unified processing of multi-source remote sensing images, completes the spatial standardization and structured organization of image data, and provides a standardized and alignable set of image slices for subsequent image fusion and analysis.

[0032] Step S2 includes the following sub-steps: Step S201: Evaluate the quality score of the remote sensing image based on meteorological data, solar altitude angle, and image occlusion rate. The meteorological data includes cloud cover data, visibility data, and rainfall intensity data. Calculate the initial quality score according to a preset weather impact factor calculation strategy. In this strategy, solar altitude angle and image occlusion rate are negatively correlated with the corresponding scores. and image occlusion rate After normalization, the comprehensive quality score corresponding to the initial spatial raster remote sensing image set is calculated according to the weather impact factor calculation strategy. The occlusion rate data represents the loss of effective pixel proportion.

[0033] Step S201 combines cloud data Visibility data and rainfall intensity data Based on the weather impact factor calculation strategy, the remote sensing images are quality assessed, and the comprehensive quality score corresponding to each group of images is output as the basis for subsequent weight allocation.

[0034] Cloud data Visibility data Rainfall intensity data Solar altitude angle and image occlusion rate After normalization, we get: ; ; ; ; ; In the weather impact factor calculation strategy, where solar altitude angle and image occlusion rate are negatively correlated with their corresponding scores, this invention explicitly uses a penalty term to express this negative correlation: Solar geometry penalty term That is, the lower the sun's altitude, the greater the punishment.

[0035] Obstruction penalty items In other words, the greater the occlusion rate, the greater the penalty.

[0036] Set weights , , , and If the condition is non-negative and the sum is 1, then the initial quality score is defined as: ; in, It indicates that fewer clouds are better. Higher visibility is better. The weaker the rainfall, the better. This indicates that the higher the solar altitude, the better. This means that less occlusion is better. The larger the value, the better the image quality.

[0037] Assume that multiple source images from the same time period form a group. Calculate the overall quality score based on the initial quality score. : ; in, Effective pixel ratio of the available image .

[0038] Step S202: Set a time-fitting factor based on time proximity. The logic for setting the time-fitting factor is that time proximity and the corresponding time-fitting factor are positively correlated.

[0039] Temporal proximity is used to characterize the closeness between the image timestamp and the target reconstruction time. Let the target time be... , No. The timestamp of each image is Then the time difference Time fit factor for: ; in, This is a time scale parameter, set by the business logic.

[0040] An angular complementarity factor is set based on angular coverage complementarity. The logic for setting the angular complementarity factor is that angular coverage complementarity and angular complementarity factor are positively correlated.

[0041] Angular coverage complementarity is used to measure whether multi-source images are geometrically complementary, thereby reducing occlusion and information loss caused by a single viewpoint. For images... Extract its observation angle parameters: incident angle and azimuth angle.

[0042] The angular space is discretized into several angular bins. This invention uses 5° increments for the incident angle and 10° increments for the azimuth angle, resulting in [image / text / image]. The corresponding bucket set is denoted as .

[0043] The current set to be merged is It has covered the bucket and been collected into .

[0044] Then the relative set of images Angle Complementary Factor Defined as: ; The value ranges from 0 to 1, and the more new angles are covered, the closer it gets to 1.

[0045] The image quality factor is set based on image sharpness. The logic for setting the image quality factor is that image sharpness and image quality factor are positively correlated.

[0046] Image sharpness reflects the strength of texture detail and edge information. For images... Calculate the Laplace operator response Define the sharpness index : ; in, For variance, the sharpness index Perform normalization and output image quality factor : ; Step S202 sets a time-fitting factor, an angle-complementary factor, and an image quality factor based on time proximity, angle coverage complementarity, and image sharpness to construct a key parameter system for measuring the priority of image fusion.

[0047] Step S203: Construct a fusion model based on the comprehensive quality score, event fitting factor, angle complementarity factor, and image quality factor, and perform fusion calculation on the initial spatial raster remote sensing image set to generate a multi-dimensional remote sensing fusion image.

[0048] The fusion model of this invention adopts a feasible structure of multi-factor weight-pixel level weighted fusion, which includes: a weight generation sub-model and a fusion calculation sub-model.

[0049] Candidate image set corresponding to the same spatial raster location For each image Calculate the raw score with overall weight: ; in, , , and The coefficient can be set.

[0050] Then perform softmax to obtain the weights. : ; in, The computational logic and same.

[0051] Define the effective pixel mask Then the effective weight at the meta level : ; in, To prevent extremely small values ​​with a denominator of 0, when a pixel is completely occluded, neighborhood interpolation can be used as a fallback.

[0052] When optical radiation values ​​are used as the fusion object, the fused image is at the pixel level. The location is: ; If it is a multi-dimensional remote sensing fusion image that combines optical, radar and infrared, then the fusion channel is obtained for each modality in the manner described above, and then the channels are stitched together to form a multi-dimensional fusion result (RGB, multispectral channels, SAR backscatter channels and thermal infrared channels can be stitched together).

[0053] Step S203 constructs a fusion model based on the comprehensive quality score, time-fitting factor, angle complementarity factor, and image quality factor, and performs fusion processing on the initial spatial raster remote sensing image set to generate a multi-dimensional remote sensing fusion image with temporal continuity and spatial integrity.

[0054] Step S2 establishes a remote sensing image fusion mechanism. By evaluating image quality and constructing a fusion weight model, the initial spatial raster remote sensing image set is weighted to generate a multi-dimensional remote sensing fused image, providing a high-quality, information-rich image input for subsequent modeling.

[0055] Step S3 includes the following sub-steps: Step S301: Select image sequences from historical remote sensing image data that contain the same observation area at different times, and the area has known surface state change annotations at different time nodes. Construct a sample set based on the image time series and annotation information. The sample set includes image sequences, change correspondences and surface classification information.

[0056] Step S301 involves selecting data with temporal continuity and labeled surface conditions from historical remote sensing images to construct a sample set containing image sequences, change labels, and classification information, providing a supervised data foundation for training the evolutionary model.

[0057] Step S302: Establish a land surface change evolution model based on the sample set. The land surface change evolution model consists of a spatial feature extraction module, a time series modeling module, and a land surface state identification module.

[0058] Land surface change evolution models can employ an end-to-end structure of spatial encoders, temporal units, and state decoders: Spatial feature extraction module (spatial encoder): fusing images at each time point. Extracting spatial features . It is a convolutional network (several convolutional layers and downsampling).

[0059] Time series modeling module (temporal unit): performs time series modeling on feature sequences. Get the hidden state ConvLSTM is available: ; Surface condition recognition module (condition decoder): Outputs a probability map and a change rate map of surface types. ; ; in, The probability distribution of each pixel belonging to each local surface type. The degree of change.

[0060] Step S302 establishes a land surface change evolution model based on the sample set. This model consists of spatial feature extraction, time series modeling and land surface state identification modules, which are used to learn the spatial characteristics and temporal evolution patterns of land surface changes.

[0061] Step S303: Input the multidimensional remote sensing fusion image and target time into the land surface change evolution model. The output of the land surface change evolution model is a land surface state prediction layer corresponding to multiple time nodes. The land surface state prediction layer contains land surface type and change degree identifiers. Output the land surface state prediction layer as a virtual land surface evolution sequence.

[0062] The surface condition prediction layer is defined as follows: Surface type layer ; Change level layer .

[0063] Multiple time points Organized chronologically, the output is a virtually performative sequence.

[0064] For the missing time frame layer, assume the target time axis to be generated is an equally spaced sequence. Interpolation of the observed hidden states is performed using temporal attention weights: ; ; Then output by the decoder and Generate a layer for the missing measurement time.

[0065] Step S303 inputs the multidimensional remote sensing fusion image and target time into the land surface change evolution model, and outputs land surface state prediction layers at multiple time points. The layers contain land surface type and change degree identifiers, forming a virtual land surface evolution sequence.

[0066] Step S304: Perform continuity and consistency detection on the virtual land performance sequence, correct non-contiguous areas between land state prediction layers, fill in boundary anomalies and undefined areas according to the land state of the nearest time node, and output the corrected land performance result.

[0067] Constraining category transitions between adjacent layers: If a pixel experiences an isolated transition within a short time window (i.e., it only changes to a certain category in a single frame, and remains consistent across the preceding and following frames), then the category consistency backfilling is dominated by the nearest time node, i.e.: pixels ,like: ; Then let .

[0068] The surface state filling in step S304 includes: selecting the surface state of the same area in the time node adjacent to the area to be filled as the filling basis, using the dominant category consistency strategy in the area to specify the missing state, and generating an updated virtual performance layer.

[0069] Step S304 performs a continuity and consistency check on the virtual evolution sequence, performs surface state filling operation on non-continuous, boundary anomaly and undefined areas, and outputs the corrected evolution result. The surface state filling is based on the surface state of the same area in the adjacent time nodes, and the dominant category consistency strategy is used to determine the state of missing areas, complete the virtual evolution layer content, and ensure the logical consistency of the evolution sequence in time and space.

[0070] Step S3 constructs a land surface change evolution model, performs temporal reasoning on the input multidimensional remote sensing fusion imagery to generate a complete virtual land surface evolution sequence, fills in the state of discontinuous areas, and outputs land surface evolution results with continuous spatiotemporal attributes, providing a predictive basis for subsequent change analysis and label generation.

[0071] Step S4 includes the following sub-steps: Step S401: Calculate the change intensity map based on each adjacent layer in the virtual performance sequence.

[0072] The intensity change map is used to quantify the magnitude of changes in land surface condition at adjacent time points. (For each pixel...) At adjacent times The intensity of change is defined as: ; in, To measure changes in the distribution of land surface types, To measure the degree of change This is a weighting factor.

[0073] The maximum value can be taken for the intensity spectrum of the entire sequence: ; Output This is the intensity variation map (raster layer, value 0~1).

[0074] Step S401 calculates the magnitude of state change in each region over time based on adjacent time layers in the virtual performance sequence, and generates a change intensity map that reflects the trend of change.

[0075] Step S402: Identify regions in the intensity change map that exceed a set threshold and mark their spatial range.

[0076] This invention uses thresholding of intensity change maps to obtain candidate change masks. : ; Among them, threshold A fixed value can be taken (0.6 is taken in this invention).

[0077] Morphological opening and closing operations are performed on the binary mask for denoising and hole filling, and connected component labeling yields a set of changed regions. For each Boundary extraction (using marching squares in this invention) is performed to generate polygonal outlines. .

[0078] The output should include at least: region ID, raster mask, vector outline (GeoJSON), bounding rectangle, and centroid coordinates.

[0079] Step S402 involves filtering out areas in the change intensity map where the change intensity exceeds a preset threshold, and then performing spatial contour recognition and labeling on these areas to clarify the geographical extent of the change.

[0080] Step S403: Call the historical change type database, and determine the driving type of the change based on the morphological characteristics, duration and spatial location of the labeled change area through a preset classification model, and output the corresponding intent recognition label.

[0081] For each Define the eigenvector: Morphological characteristics: area ,perimeter Tightness Aspect ratio and principal orientation angle (PCA principal axis direction).

[0082] Duration: In a virtual performance sequence, the region is tracked by IoU matching at the same spatial location to obtain the set of existing frames and the duration.

[0083] Spatial location: centroid and administrative division code (obtained by spatial overlay).

[0084] Driver type is defined as a classification label for the cause of change, and can be divided into at least the following categories: Natural evolution: floods, landslides, seasonal vegetation changes, river channel shifts; Human intervention: construction, mining, deforestation, land reclamation, etc.

[0085] The classification model is a random forest. Its input is the aforementioned feature vector and environmental context (such as rainfall, slope, and land use base). The output is the driving type ID, which is mapped to the intent recognition label (such as "natural flooding change" or "construction expansion").

[0086] Step S403 calls the historical change type database, combines the morphological features of the marked area, the duration of the change and the spatial location, and uses a classification model to determine the type of change driving, and outputs an intent recognition label representing the change attribute to distinguish between natural evolution and human intervention changes.

[0087] Step S4 analyzes the intensity of changes in the land surface state in the virtual performance sequence, identifies spatial areas with significant changes, and combines historical change type databases and classification models to determine the type of change driver, generating corresponding intent recognition labels, providing a basis for the generation of subsequent geographic semantic labels.

[0088] Step S5 includes the following sub-steps: Step S501: Extract change features based on the intent recognition label and the corresponding change area location. The change features include change cycle, change direction, change range and change frequency.

[0089] Polygons of the same changing region at adjacent time points and Matching by IoU: ; like If the value is greater than or equal to the threshold (0.3 in this invention), then the same object is considered to evolve continuously, thus obtaining the time trajectory of the region.

[0090] The range of variation is defined as the union of all time-regions in the trajectory. : ; It outputs its bounding rectangle, area, and administrative region coverage.

[0091] With regional centroid Construct the displacement vector: ; ; To change the main direction, if there are multiple directional changes, a directional histogram can be generated to output the directional range.

[0092] Define sequence indicators (This invention takes) ).

[0093] Count the number of events that change significantly within a time window: If If we denote it as one event, then the frequency of change is... : ; right Perform autocorrelation: ; Take the first significant peak other than 0. It serves as a period and can be converted into an actual time period.

[0094] Map the driving type and spatiotemporal index to structured semantic label fields (geographic feature type, change type, spatiotemporal attribute, multilingual description), that is: Geographic feature types: cultivated land, forest land, water bodies, built-up areas... (from surface type layer); Types of change: expansion, contraction, abrupt change, periodic fluctuations… (from…) Trends and Freq); Spatiotemporal attributes: start and end times, center position, and range polygon; Multilingual description field: generated according to template.

[0095] Step S501 identifies the location of the label and the changed area based on the intent, and extracts feature parameters such as change cycle, change direction, change range and change frequency to describe the specific manifestation of surface changes in time and space.

[0096] Step S502: Generate structured semantic tags based on change characteristics. The structured semantic tags include geographic feature type, change type, spatiotemporal attributes, and multilingual description fields.

[0097] Step S502 transforms the change features into structured semantic tags. The tag content includes geographic feature type, change type, spatiotemporal attributes, and multilingual description fields, realizing the semantic expression of change information, which is convenient for system parsing and user understanding.

[0098] Step S5 extracts multidimensional change features based on intent recognition labels and spatial change areas, and generates structured semantic labels containing geographic information structure, attribute classification and language information, providing semantic support for the data expression of geographic information systems.

[0099] Step S6 includes the following sub-steps: Step S601: Enter the structured semantic tag data into the geographic information system database according to the spatial range and time index method.

[0100] Step S601 involves inputting the structured semantic tags into the geographic information system database according to their spatial range and temporal index, establishing a spatiotemporal index relationship, and ensuring that the tag data is queryable and traceable.

[0101] Step S602: Compare the geographic information data before and after the update, identify the changes in added, adjusted or eliminated geographic elements and update the database records.

[0102] Step S602 compares the content of the old and new geographic information data, identifies the addition, adjustment or elimination of geographic elements, and completes the corresponding record update in the database to maintain data version consistency.

[0103] Step S603: Synchronize the updated data to the map platform and data service platform through the geographic information system interface.

[0104] Step S603 synchronizes the updated geographic information data to the map platform and data service platform through the geographic information system interface, realizing information linkage and data synchronization and sharing with external systems.

[0105] Step S6 enables the organization, updating, and publishing of structured semantic tag data in the geographic information system. Through dynamic data update mechanisms and system interfaces, it achieves timely management and multi-platform sharing of geographic information.

[0106] Example 2, refer to Figure 2 This paper provides a geographic information acquisition system, which includes an image tiling module, an image fusion module, a geographic performance module, an intent recognition module, a feature extraction module, and a data update module.

[0107] The image tiling module is used to perform image projection consistency processing on the initial spatial raster remote sensing image set and output an image tile set; The image fusion module is used to calculate the overall quality score, build a fusion model, and generate multi-dimensional remote sensing fused images.

[0108] The land performance module is used to construct a sample set and establish a land surface change evolution model, and output a virtual land performance sequence.

[0109] The intent recognition module is used to calculate the intensity change map and identify the change area, and output the intent recognition label.

[0110] The feature extraction module is used to extract changing features and generate structured semantic labels.

[0111] The data update module is used to input structured semantic label data and update the geographic information system database.

[0112] This system utilizes a unified spatial reference system based on multi-source remote sensing images to improve the comparability and processing consistency between remote sensing images. It integrates meteorological data, solar altitude angle, and image occlusion rate to comprehensively evaluate image quality, constructs a multi-factor weighting model to enhance the clarity and information integrity of the fused images, establishes a land surface change evolution model based on labeled image sequences, and enables reasonable prediction and continuous reconstruction of land surface conditions in occluded areas or at time discontinuities. It introduces change intensity maps and preset classification models to identify the driving types of land surface changes, possessing the ability to distinguish between natural and human-induced changes. It automatically generates structured semantic labels and realizes real-time updates of the geographic information system database, enhancing the timeliness and applicability of geographic information data. It has the ability to recover land surface evolution processes in complex environments and can be widely applied in scenarios such as disaster assessment, engineering monitoring, and spatial planning.

[0113] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0114] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A geographic information acquisition method, characterized in that, Includes the following steps: Step S1: Perform image projection consistency processing on the initial spatial raster remote sensing image set and output an image tile set; Step S2: Calculate the overall quality score, construct the fusion model, and generate multi-dimensional remote sensing fusion image; Step S3: Construct a sample set and establish a land surface change evolution model, outputting a virtual land surface evolution sequence; Step S4: Calculate the change intensity map and identify the change area, and output the intent recognition label; Step S5: Extract change features and generate structured semantic labels; Step S6: Input structured semantic tag data and update the geographic information system database.

2. The geographic information acquisition method as described in claim 1, characterized in that, Step S1 includes the following sub-steps: Step S101: Call the satellite remote sensing image database to obtain optical remote sensing images, radar remote sensing images and infrared remote sensing images within the set area; Step S102: Perform image projection consistency processing on optical remote sensing images, radar remote sensing images and infrared remote sensing images to unify spatial resolution and coordinate system, and output the initial spatial raster remote sensing image set. Step S103: Classify and index the images according to timestamps and image types, and output the image slice set.

3. The geographic information acquisition method as described in claim 2, characterized in that, Step S2 includes the following sub-steps: Step S201: Evaluate the quality score of remote sensing images based on meteorological data, solar altitude angle, and image occlusion rate. The meteorological data includes cloud cover data, visibility data, and rainfall intensity data. Calculate the initial quality score according to a preset weather impact factor calculation strategy. In the weather impact factor calculation strategy, solar altitude angle and image occlusion rate are negatively correlated with the corresponding scores. Normalize the meteorological data, solar altitude angle, and image occlusion rate. Calculate the comprehensive quality score corresponding to the initial spatial raster remote sensing image set according to the weather impact factor calculation strategy. Step S202: Set a time-fitting factor based on time proximity. The logic for setting the time-fitting factor is that time proximity is positively correlated with the corresponding time-fitting factor. An angle complementarity factor is set based on the angle coverage complementarity. The setting logic of the angle complementarity factor is that the angle coverage complementarity and the angle complementarity factor are positively correlated. An image quality factor is set based on image sharpness, and the setting logic of the image quality factor is that image sharpness and image quality factor are positively correlated. Step S203: Construct a fusion model based on the comprehensive quality score, event fitting factor, angle complementarity factor, and image quality factor, and perform fusion calculation on the initial spatial raster remote sensing image set to generate a multi-dimensional remote sensing fusion image.

4. The geographic information acquisition method as described in claim 3, characterized in that, Step S3 includes the following sub-steps: Step S301: Select image sequences containing the same observation area at different times from historical remote sensing image data, and the area has known surface state change annotations at different time nodes. Construct a sample set based on the image time series and annotation information. The sample set includes image sequences, change correspondences and surface classification information. Step S302: Establish a land surface change evolution model based on the sample set. The land surface change evolution model consists of a spatial feature extraction module, a time series modeling module, and a land surface state identification module. Step S303: Input the multi-dimensional remote sensing fusion image and target time into the land surface change evolution model. The output of the land surface change evolution model is a land surface state prediction layer corresponding to multiple time nodes. The land surface state prediction layer contains land surface type and change degree identifiers. Output the land surface state prediction layer as a virtual land surface evolution sequence. Step S304: Perform continuity and consistency detection on the virtual land performance sequence, correct non-contiguous areas between land state prediction layers, fill in boundary anomalies and undefined areas according to the land state of the nearest time node, and output the corrected land performance result.

5. A geographic information acquisition method as described in claim 4, characterized in that, Step S4 includes the following sub-steps: Step S401: Calculate the change intensity map based on each adjacent layer in the virtual performance sequence; Step S402: Identify regions in the intensity change map that exceed a set threshold and mark their spatial range; Step S403: Call the historical change type database, and determine the driving type of the change based on the morphological characteristics, duration and spatial location of the labeled change area through a preset classification model, and output the corresponding intent recognition label.

6. The geographic information acquisition method as described in claim 5, characterized in that, Step S5 includes the following sub-steps: Step S501: Extract change features based on the intent recognition label and the corresponding change area location. The change features include change period, change direction, change range and change frequency. Step S502: Generate structured semantic tags based on change characteristics. The structured semantic tags include geographic feature type, change type, spatiotemporal attributes, and multilingual description fields.

7. A geographic information acquisition method as described in claim 6, characterized in that, Step S6 includes the following sub-steps: Step S601: Input the structured semantic tag data into the geographic information system database according to the spatial range and time index method; Step S602: Compare the geographic information data before and after the update, identify the changes in added, adjusted or removed geographic elements and update the database records. Step S603: Synchronize the updated data to the map platform and data service platform through the geographic information system interface.

8. A geographic information acquisition method as described in claim 7, characterized in that, The surface state filling in step S304 includes: selecting the surface state of the same area in the time node adjacent to the area to be filled as the filling basis, using the dominant category consistency strategy in the area to specify the missing state, and generating an updated virtual performance layer.

9. A geographic information acquisition method as described in claim 8, characterized in that, The driving types mentioned in step S403 include natural evolution and human intervention, which are classified by historical map matching and deep model judgment. The deep model is based on the change contour, change rate and environmental context for identification.

10. A geographic information acquisition system, applied in a geographic information acquisition method as described in any one of claims 1-9, characterized in that, It includes an image slicing module, an image fusion module, a geospatialization module, an intent recognition module, a feature extraction module, and a data update module; The image slicing module is used to perform image projection consistency processing on the initial spatial raster remote sensing image set and output an image slice set. The image fusion module is used to calculate the comprehensive quality score, construct the fusion model, and generate multi-dimensional remote sensing fusion images. The land evolution module is used to construct a sample set and establish a land surface change evolution model, and output a virtual land evolution sequence; The intent recognition module is used to calculate the change intensity map and identify the change area, and output the intent recognition label; The feature extraction module is used to extract change features and generate structured semantic labels; The data update module is used to input structured semantic tag data and update the geographic information system database.

Images