Parcel state data generation method and apparatus, device, and medium

Abstract

The present application relates to the technical field of intelligent decision-making, and discloses a plot state data generation method, device, equipment and medium, comprising: acquiring supplementary images based on quality evaluation of basic remote sensing images, generating a target multispectral image set through registration and fusion; identifying plot boundaries and attributes, and performing change comparison and checking and confirmation in combination with historical plot state data; generating early warning events based on multi-period regional observation images, historical state baseline data and plot multi-source observation data, and forming plot panoramic state data. The present application can be applied to business scenarios such as financial technology, improves the data basis of plot recognition through supplementary fusion, forms updateable plot state data through historical comparison and checking and confirmation, generates early warning events through time series observation and multi-source observation, and uniformly organizes change information, state trends and early warning information, thereby improving the accuracy and integrity of plot dynamic state data.

Description

Technical Field

[0001] This invention relates to the field of intelligent decision-making technology, and in particular to a method, apparatus, device, and medium for generating land parcel status data. Background Technology

[0002] In data management scenarios where land parcels are used as risk targets, parcel boundaries, attributes, and status change over time. Data generated from a single collection or static registration is insufficient to continuously reflect the actual state of the parcel. Especially when image quality is unstable, parcel shapes are complex, and status changes are continuous, existing processing methods struggle to correlate spatial changes with abnormal status changes, resulting in incomplete dynamic status data for specific parcels.

[0003] In the fintech sector, crop insurance typically requires underwriting review, risk identification during the insurance period, loss assessment, and claims review based on the boundaries, area, use, and growth status of the insured plot. Existing business data largely comes from insurance application information or single-period remote sensing imagery, making it difficult to update in a timely manner as the boundaries, use, and growth status of the insured plot change. When the actual state of the plot deviates from existing records during the insurance period, insurance institutions struggle to obtain continuous status data corresponding to specific insured plots, thus affecting the accuracy of risk assessment.

[0004] Current crop insurance monitoring and processing typically lacks historical change records, time-series status records, and abnormal event records for the same insured plot. It is difficult to continuously express plot change information and abnormal growth information on a unified data foundation. Therefore, when insured plots experience spatial changes or abnormal conditions, existing technologies struggle to promptly generate dynamic plot status data capable of supporting risk identification and claims verification. Summary of the Invention

[0005] The main objective of this invention is to provide a method, apparatus, device, and storage medium for generating land parcel status data, aiming to solve the technical problem that existing agricultural insurance land parcel monitoring technologies struggle to generate dynamic land parcel status data that is synchronously updated with the actual status of the insured land parcels when image quality is unstable and the actual status of the insured land parcels is constantly changing.

[0006] To achieve the above objectives, the present invention provides a method for generating land parcel status data, comprising: Acquire basic remote sensing images of the target area, trigger supplementary acquisition based on the image quality evaluation of the basic remote sensing images to obtain supplementary images, and register and fuse the basic remote sensing images and the supplementary images into a target multispectral image set. The target multispectral image set is processed based on the joint recognition model to generate current land parcel status data containing land parcel boundary data and land parcel attribute data. The historical land parcel status data corresponding to the current land parcel status data is obtained from the spatiotemporal version data storage area. The current land parcel status data and the historical land parcel status data are compared and verified to generate confirmed land parcel status data and land parcel change events. The confirmed land parcel status data is written into the spatiotemporal version data storage area. Based on the land parcel boundary data in the confirmed land parcel status data, multi-period regional observation images are obtained; land parcel time-series observation data is generated based on the multi-period regional observation images; and status trend data is generated based on the land parcel time-series observation data. Obtain historical status baseline data and multi-source observation data of the land parcel corresponding to the confirmed land parcel status data, and generate early warning events based on the status trend data, the historical status baseline data and the multi-source observation data of the land parcel; Based on the confirmed land parcel status data, the land parcel change events, the status trend data, the land parcel multi-source observation data, and the early warning events, panoramic land parcel status data is generated.

[0007] Furthermore, to achieve the above objectives, the present invention provides a land parcel status data generation apparatus, comprising: The image acquisition and fusion module is used to acquire basic remote sensing images of the target area, trigger acquisition based on the image quality evaluation of the basic remote sensing images to acquire acquisition images, and register and fuse the basic remote sensing images and the acquisition images into a target multispectral image set. The land parcel status identification module is used to process the target multispectral image set based on the joint identification model to generate current land parcel status data containing land parcel boundary data and land parcel attribute data; The land parcel change confirmation module is used to obtain historical land parcel status data corresponding to the current land parcel status data from the spatiotemporal version data storage area, compare and verify the current land parcel status data with the historical land parcel status data, generate confirmed land parcel status data and land parcel change event, and write the confirmed land parcel status data into the spatiotemporal version data storage area. The status trend analysis module is used to obtain multi-period regional observation images based on the land parcel boundary data in the confirmed land parcel status data, generate land parcel time-series observation data based on the multi-period regional observation images, and generate status trend data based on the land parcel time-series observation data. The status risk warning module is used to acquire historical status baseline data and multi-source observation data of the land parcel corresponding to the confirmed land parcel status data, and generate warning events based on the status trend data, the historical status baseline data and the multi-source observation data of the land parcel. The panoramic status construction module is used to generate panoramic status data of the land parcels based on the confirmed land parcel status data, the land parcel change events, the status trend data, the land parcel multi-source observation data, and the early warning events.

[0008] Furthermore, to achieve the above objectives, the present invention also provides a computer device, the computer device including a memory, a processor, and a land parcel status data generation program stored in the memory and executable on the processor, wherein when the land parcel status data generation program is executed by the processor, it implements the steps of the land parcel status data generation method as described above.

[0009] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing a land parcel status data generation program, wherein when the land parcel status data generation program is executed by a processor, it implements the steps of the land parcel status data generation method as described above.

[0010] Beneficial Effects: This invention relates to the field of intelligent decision-making technology, and discloses a method, apparatus, device, and medium for generating land parcel status data. The method includes: acquiring supplementary images based on quality assessment of basic remote sensing images, and generating a target multispectral image set through registration and fusion; identifying land parcel boundaries and attributes, and performing change comparison and verification in conjunction with historical land parcel status data; generating early warning events based on multi-period regional observation images, historical status baseline data, and multi-source observation data of land parcels, and forming panoramic land parcel status data. This invention can be applied to business scenarios such as fintech, improving the data foundation for land parcel identification through supplementary image fusion, forming updatable land parcel status data through historical comparison and verification, generating early warning events through time-series observation and multi-source observation, and unifying change information, status trends, and early warning information, thereby improving the accuracy and completeness of dynamic land parcel status data. Attached Figure Description

[0011] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings: Figure 1 This is a schematic diagram of an application environment for a land parcel status data generation method according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating an embodiment of the land parcel status data generation method of the present invention; Figure 3 This is a schematic diagram of an application architecture for the land parcel status data generation method of the present invention; Figure 4 This is a schematic diagram of a remote sensing image re-capture and fusion and land parcel status update process for the land parcel status data generation method of the present invention. Figure 5 This is a schematic diagram of a land parcel status trend analysis and early warning event generation process according to the land parcel status data generation method of the present invention. Figure 6 This is a schematic diagram illustrating a multi-source data fusion and panoramic land parcel status display method for generating land parcel status data according to the present invention. Figure 7 This is a schematic diagram of the functional modules of a preferred embodiment of the land parcel status data generation device of the present invention; Figure 8 This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention; Figure 9 This is another structural schematic diagram of a computer device according to one embodiment of the present invention. Detailed Implementation

[0012] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0013] The land parcel status data generation method provided in this embodiment of the invention can be applied to, for example... Figure 1 In this application environment, the client communicates with the server via a network. The server can acquire supplementary images based on the quality assessment of basic remote sensing images through the client, and generate a target multispectral image set through registration and fusion; identify land parcel boundaries and attributes, and perform change comparison and verification based on historical land parcel status data; generate early warning events based on multi-period regional observation images, historical status baseline data, and multi-source observation data of land parcels, and form a panoramic status data of land parcels. This invention can be applied to business scenarios such as fintech, improving the data foundation for land parcel identification through supplementary image fusion, forming updatable land parcel status data through historical comparison and verification, generating early warning events through time-series observation and multi-source observation, and unifying change information, status trends, and early warning information, thereby improving the accuracy and completeness of dynamic status data of land parcels. The client can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server can be implemented using a standalone server or a server cluster consisting of multiple servers. The invention will be described in detail below through specific embodiments.

[0014] Please see Figure 2 , Figure 2 This is a flowchart illustrating an embodiment of the land parcel status data generation method provided by the present invention. It should be noted that although the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than that shown here.

[0015] like Figure 2 As shown, the land parcel status data generation method proposed in this invention includes the following steps: S10, acquire basic remote sensing images of the target area, trigger supplementary acquisition based on the image quality evaluation of the basic remote sensing images to acquire supplementary images, and register and fuse the basic remote sensing images and the supplementary images into a target multispectral image set. In this embodiment, the target area is used to determine the spatial extent of image acquisition and can consist of multiple continuous or discrete surface areas. The base remote sensing image is a multi-band image covering the target area and carrying spatial reference information and acquisition time information, which can be obtained from satellite imaging equipment or aerial imaging equipment. Image quality assessment is used to identify areas with occlusion, blurring, missing bands, or excessively long observation time intervals, and to determine the areas that cannot meet the current image usage requirements as the re-acquisition range.

[0016] The supplementary imagery is a newly acquired imagery for the supplementary area, and its spatial coverage corresponds to the corresponding area in the base remote sensing imagery. The image bands can be configured according to the surface condition identification requirements. During registration and fusion, spatial offsets are corrected based on spatial reference information or stable surface locations in both types of images, and the content of the supplementary imagery within the supplementary area is combined with the effective content in the base remote sensing imagery to form a target multispectral image set covering the target area.

[0017] In one implementation, the base remote sensing imagery uses satellite multispectral imagery covering the insured land parcels. When cloud shadows obscure the imagery or the edges of the land parcels are discontinuous, a supplementary imagery range is established based on the affected area, and low-altitude multispectral imaging equipment is used to acquire supplementary images. During registration, spatial correction is performed at road intersections, ditch edges, or field ridge turning points. During fusion, the supplementary images are used to supplement the image content of the obscured areas, forming a target multispectral image set.

[0018] In another implementation, the base remote sensing imagery uses aerial images generated in the previous observation period. To address situations where images in the post-disaster survey area are outdated or where local surface changes are difficult to identify, supplementary images are acquired by partitioning the surveyed area, adjusting the acquisition path and spatial sampling interval, and then registering and fusing these supplementary images with the base remote sensing imagery to form a target multispectral image set corresponding to the current survey batch.

[0019] In the scenario of verifying the underwriting of crop insurance, when the basic remote sensing image covering the insured plot is obscured by cloud shadows, supplementary images are acquired for the obscured plot area, and the supplementary images are registered and fused with the basic remote sensing image to form a target multispectral image set for verifying the spatial range of the insured plot.

[0020] This embodiment determines the areas that need to be supplemented by image quality assessment and obtains the corresponding supplemented images, which can supplement the image content in the basic remote sensing image that is affected by occlusion, blurring or observation lag; by registration and fusion, the two types of images are kept in a consistent spatial position relationship, forming a target multispectral image set that covers the target area and contains multi-band information, thereby improving the usability of land parcel image data in underwriting verification and survey review.

[0021] S20, Based on the joint recognition model, process the target multispectral image set to generate current land parcel status data containing land parcel boundary data and land parcel attribute data; In this embodiment, the joint recognition model receives multi-band image data and spatial reference information from the target multispectral image set, and uses texture edges, surface contours, and band responses within the same image coverage area as the recognition basis. The multi-band image data can be input according to fixed-size image blocks or according to the target area coverage area, and the pixel scale, band order, and coordinate reference are unified before input to ensure that the spatial positions expressed by different bands are consistent.

[0022] Plot boundary data is used to record the spatial extent of identified plots and can be represented as boundary point sequences, closed contours, or areal graphic data. The model identifies plot boundaries from field ridges, ditches, road edges, locations of changes in cover texture, and locations of abrupt changes in band response in the imagery, and continuously connects the perimeter of the same plot to form a closed boundary. For closely adjacent areas or areas with locally blurred boundaries, the boundary results that rely solely on texture can be corrected by incorporating multi-band response differences.

[0023] Land parcel attribute data records the identified content corresponding to the spatial extent of a land parcel, which may include land use category, crop cover category, area information, and identification confidence level. The model aggregates the band response and texture distribution within the same land parcel boundary in the imagery into land parcel-level data, and then outputs the corresponding attributes. Land parcel boundary data and land parcel attribute data are associated through the same land parcel marker, combined to form the current land parcel status data, ensuring that the spatial extent of a land parcel corresponds to its attribute content.

[0024] In one implementation, the joint identification model includes a shared image coding component, a boundary output component, and an attribute output component. The target multispectral image set is cropped into image blocks with overlapping areas. The image coding component extracts contour texture and band response from each image block. The boundary output component forms closed contours of the land parcels, and the attribute output component forms land use categories and crop cover categories. Repeating contours in adjacent image blocks are spatially merged, and the closed contours are converted into land parcel boundary data based on image spatial reference information. The identified categories are then written into the land parcel attribute data to form the current land parcel status data.

[0025] In another implementation, the joint recognition model takes the entire multispectral image covering the target area as input and outputs the spatial zoning results of the land parcels and the attribute results corresponding to each spatial zoning. For contiguous planting areas in the verification of crop insurance coverage, areal boundaries can be generated based on the spatial zoning results of the land parcels, and land parcel attribute data can be formed based on the band responses within the boundaries. For areas with small and adjacent land parcels, the spatial sampling precision of the input image can be improved, and overlapping image content can be retained near the boundaries to reduce the merging and identification of adjacent land parcels.

[0026] In the scenario of crop insurance underwriting verification, the target multispectral image set covers the planting area declared by the insured entity. The joint recognition model identifies the boundary contour of each insured plot and outputs the corresponding planting coverage category and usage category, forming the current plot status data, which is used to record the spatial range and attribute content of the insured plot at the time of verification.

[0027] This embodiment uses a joint recognition model to jointly process the spatial texture and multi-band response of the target multispectral image set, which can generate land parcel boundary data and land parcel attribute data corresponding to the spatial range of the land parcel. By associating and combining the boundary results and attribute results into the current land parcel status data, it can reduce the data deviation caused by the mismatch between the land parcel range and attribute content, and improve the accuracy and completeness of the land parcel status record of the target area.

[0028] S30, retrieve historical land parcel status data corresponding to the current land parcel status data from the spatiotemporal version data storage area, compare and verify the changes between the current land parcel status data and the historical land parcel status data, generate confirmed land parcel status data and land parcel change event, and write the confirmed land parcel status data into the spatiotemporal version data storage area; In this embodiment, the spatiotemporal version data storage area stores the land parcel status records formed by each observation batch according to spatial range and time information. The historical land parcel status data is used to reflect the boundary and attribute status of the target area at previous observation points. The current land parcel status data carries the spatial range and attribute content of the land parcel formed by the current observation. Historical records corresponding to the current observation range can be retrieved from the spatiotemporal version data storage area based on the land parcel identifier, spatial coverage relationship, and observation time.

[0029] Change comparison compares the boundary locations and attribute content of current and historical land parcel status data under a unified spatial reference, identifying discrepancies to be confirmed. Verification and confirmation overlays current, historical, and discrepancy content onto an image or spatial layer, receiving corrections to boundary ranges or attribute content. The corrected land parcel data is then used to create confirmed land parcel status data, and land parcel change events are generated based on the confirmation results.

[0030] When confirming that the parcel status data is written to the spatiotemporal version data storage area, the observation time, spatial range and version association information can be written, and existing historical records can be retained, so that the same target area can form distinguishable parcel status versions at different observation times.

[0031] In one implementation, the spatiotemporal version data storage area uses a joint index of plot identifier and observation time. For the target area in the crop insurance underwriting verification, the earlier version is retrieved based on the spatial range and observation time in the current plot status data, and the boundary graphics and attribute content of the two periods are overlaid. After the reviewers confirm the boundary changes or attribute changes, confirmed plot status data is generated, and the current confirmed version is written to the storage area.

[0032] In another implementation, for areas with significant changes in land parcel boundaries that are difficult to directly use land parcel identifiers, historical land parcel status data can be retrieved based on spatial coverage, and a correspondence between current and historical records can be established according to spatial overlap. For the central verification area, differing land parcel images, current status records, and historical status records are displayed synchronously. After receiving the verification results, new confirmed land parcel status data and land parcel change events are generated and saved to the storage area with a new observation time.

[0033] In the context of crop insurance underwriting review, after the current status data of the insured plot is generated in this observation, the historical plot status data of the previous observation batch is retrieved from the spatiotemporal version data storage area. The changes in boundary range and planting attributes are compared, and the reviewers confirm the differences. The confirmed plot status is then saved as the current underwriting verification record.

[0034] This embodiment extracts historical records corresponding to the current land parcel status data from the spatiotemporal version data storage area, and compares and verifies the changes between the current status and the historical status. This enables the land parcel changes formed by image recognition to form the current valid status after confirmation. By writing the confirmed land parcel status data into the spatiotemporal version data storage area, the land parcel status records corresponding to different observation points can be retained, improving the accuracy and traceability of the dynamic change records of land parcels.

[0035] S40, based on the land parcel boundary data in the confirmed land parcel status data, obtain multi-period regional observation images, generate land parcel time-series observation data based on the multi-period regional observation images, and generate status trend data based on the land parcel time-series observation data; In this embodiment, the parcel boundary data in the confirmed parcel status data is used to determine the spatial extent of the same parcel at different observation points. Multi-period regional observation images can be composed of remote sensing images acquired at different times, each image covering the area defined by the parcel boundary data and carrying acquisition time and spatial reference information.

[0036] Based on the parcel boundary data, observation content within the parcel area is extracted from imagery from various periods. The extracted content is then spatially standardized and the image data is organized. Observations for the same parcel are organized according to the acquisition time to form parcel time-series observation data. Based on the image response that changes over time in the parcel time-series observation data, the direction and magnitude of continuous change are extracted to form state trend data.

[0037] In one implementation, multiple remote sensing images covering the area are retrieved according to a fixed observation cycle. The image content of the area is extracted period by period based on the boundary data of the area. Image data with the same spatial location are arranged according to the acquisition time. Status trend data is formed by the changes in image response between adjacent observation cycles.

[0038] In another implementation, to address situations where image acquisition intervals are inconsistent or images are missing for certain periods, regional observation images covering the area and of usable quality are selected. The spatial locations of the images from each period are unified, and the observation content is temporally organized according to the acquisition time interval. Then, state trend data is formed based on the organized temporal observation data of the plots.

[0039] In the context of agricultural insurance inspections, for insured plots with confirmed spatial boundaries, multiple images covering the plot area are obtained according to the inspection cycle. Based on the plot boundary data, the observation content of each period is extracted, and plot time-series observation data is formed according to the collection time to record the changing trend of the plot surface state during the insurance period.

[0040] This embodiment constrains the spatial range of multi-period regional observation images by using plot boundary data, enabling the image content at different observation points to correspond to the same plot. By organizing the observation content of each period into plot time-series observation data and forming state trend data from it, the problem of single-period images being unable to reflect continuous state changes can be reduced, and the continuity of plot dynamic state records can be improved.

[0041] S50, acquire historical status baseline data and multi-source observation data of the plot corresponding to the confirmed plot status data, and generate an early warning event based on the status trend data, the historical status baseline data and the multi-source observation data of the plot; In this embodiment, confirmed land parcel status data is used to determine the land parcel range and attributes corresponding to the early warning analysis. Historical status baseline data records the range of status changes of the same or similar land parcels during the corresponding observation period, and can be formed by filtering existing status records according to land parcel location, crop type, and observation period. After comparing the status trend data with the historical status baseline data, it can be identified whether the current land parcel status deviates from the status range of the corresponding period.

[0042] Multi-source observation data for a plot is used to supplement the observation basis for state deviations and can include data corresponding to the plot area and observation time, such as precipitation, temperature, soil moisture, soil nutrients, or irrigation records. When generating an early warning event, the deviation reflected by the state trend data is combined with the environmental state reflected by the multi-source observation data of the plot for judgment, and the corresponding plot area, anomaly category, early warning time, and early warning level are written into the early warning event.

[0043] In one implementation, for the monitoring records of the insured plots, historical baseline data is retrieved according to crop type and current observation period, and the state trend data is compared with the historical baseline data. When the state trend shows a continuous decline, and the multi-source observation data of the plot contains information on declining soil moisture and insufficient precipitation, a drought early warning event corresponding to the insured plot is generated, and the time and severity of the early warning are recorded.

[0044] In another implementation, for inspection records of contiguous planting areas, historical baseline data for each plot is organized based on confirmed plot status data. When an abnormal change occurs in the plot status trend, and the multi-source observation data for the plot includes information on rising temperature, changes in soil nutrients, or abnormal irrigation, a status anomaly warning event is generated for the corresponding plot. For adjacent plots exhibiting different observation states, the warning events for each plot are saved separately to prevent local anomalies from spreading to the entire area.

[0045] In the risk management scenario of crop insurance, after the insured plots generate status trend data during the insurance period, they are combined with historical status baseline data of the corresponding crop period and multi-source observation data of the plots such as precipitation and soil moisture to form early warning events corresponding to the insured plots, providing data records for the arrangement of risk inspection tasks.

[0046] This embodiment compares the current state trend data with historical state baseline data to identify deviations in the current state of a land parcel relative to the corresponding observation period. By combining multi-source observation data of the land parcel to determine the deviation, early warning events corresponding to specific land parcels and observation times can be generated, thereby improving the accuracy and timeliness of land parcel state anomaly identification.

[0047] S60, generate panoramic status data of the land parcel based on the confirmed land parcel status data, the land parcel change event, the status trend data, the land parcel multi-source observation data, and the early warning event.

[0048] In this embodiment, the confirmed land parcel status data provides verified land parcel boundaries and attribute content. Land parcel change event records show changes in the spatial range or attribute content of the land parcel over observation time. Status trend data reflects the direction of status changes of the same land parcel at multiple observation points. Multi-source land parcel observation data associates observation content such as meteorology, soil, temperature, or irrigation with the spatial range and observation time of the land parcel. Early warning event records the anomaly type, warning level, and time information of the corresponding land parcel.

[0049] A spatial graphic record is established for each plot based on its boundaries, and attribute content, change content, trend content, observation content, and early warning content are associated with the same plot marker. Change events and early warning events generated at different observation times can be organized into event records according to time information; for status trend data and multi-source observation data of plots, they can be organized into status records according to plot markers and observation times. The panoramic status data of a plot is formed by combining spatial graphic records, status records, and event records, enabling the spatial location, attribute status, changes, observation status, and early warning status of the same plot to be expressed in a unified data structure.

[0050] In one implementation, a planar graphical object is used to store the confirmed land parcel boundaries, and land parcel tags are used to associate land parcel attributes, status trends, and early warning information. Land parcel change events are written into an event sequence according to their occurrence time, and multi-source observation data of land parcels are written into a status sequence according to their observation time. The planar graphical object, event sequence, and status sequence are then combined to form panoramic status data of the land parcels, which is suitable for monitoring areas with a small number of land parcels and relatively dispersed boundaries.

[0051] In another implementation, for a large number of contiguous land parcels, the confirmed land parcel status data is organized into a land parcel layer, the status trend data and multi-source observation data of the land parcels are organized into time-dimensional data, and land parcel change events and early warning events are organized into event-dimensional data. The data in each dimension are linked through land parcel tags, and corresponding records are selected and combined according to the display time period to form panoramic land parcel status data. This approach is suitable for business environments that require viewing changes in multiple insured land parcels in batches.

[0052] In the context of risk management in crop insurance, after the boundary of the insured plot is adjusted or the planting status changes, the confirmed plot range, plot change events, status trends and early warning information are organized into a panoramic status data of the plot, so that the insurance institution can view the status change records during the insurance period by insured plot.

[0053] This embodiment can retain the current spatial status of a plot and its changes by associating confirmed plot status data with plot change events; by associating status trend data, multi-source plot observation data and early warning events with the same plot object, it can uniformly organize the status information and abnormal information of the plot formed over time; the resulting panoramic plot status data can centrally express the plot location, attributes, changes, observation and early warning content, improving the completeness and accessibility of the dynamic status information of the plot.

[0054] In one embodiment, step S10 above includes: S101, acquire basic remote sensing images of the target area, divide the basic remote sensing images into multiple evaluation image areas, extract occlusion status data, spatial resolution data, radiometric consistency data and temporal continuity data from each evaluation image area, and combine the occlusion status data, spatial resolution data, radiometric consistency data and temporal continuity data into image quality evaluation data. S102, obtain the supplementary acquisition judgment conditions, compare the image quality evaluation data corresponding to each evaluation image area with the supplementary acquisition judgment conditions, and determine the evaluation image area that meets the supplementary acquisition judgment conditions as the supplementary acquisition image area. S103, acquire the acquisition time period data corresponding to the target area, generate acquisition path and acquisition time data based on the acquired image area and the acquisition time period data, generate multispectral acquisition configuration data based on the image quality evaluation data corresponding to the acquired image area, and combine the acquisition path, the acquisition time data and the multispectral acquisition configuration data into acquisition task data. S104: Collect images of the supplementary image area according to the acquisition path, supplementary acquisition time data and multispectral acquisition configuration data in the supplementary acquisition task data, and generate supplementary images; S105, extract registration control points covering the area of ​​the supplementary image from the basic remote sensing image and the supplementary image respectively, combine the registration control points corresponding to the spatial location into registration control point pairs, and register the supplementary image to the basic remote sensing image based on the registration control point pairs to generate a registration supplementary image. S106, extract the supplementary multispectral image data corresponding to the supplementary image region from the registered supplementary image, extract the basic multispectral image data corresponding to the spatial location of the supplementary multispectral image data from the basic remote sensing image, generate fusion weight data based on the image quality evaluation data corresponding to the supplementary image region, fuse the supplementary multispectral image data and the basic multispectral image data based on the fusion weight data, and generate the target multispectral image set.

[0055] In this embodiment, the target area is determined by the boundary coordinates of the surface area to be observed, and can be represented by a closed areal area or multiple adjacent areal areas. Based on the target area, basic remote sensing images with spatial reference information, acquisition time information, and band markers are retrieved, ensuring that the coverage of the basic remote sensing images includes the target area, and that the corresponding image content can be extracted according to spatial location.

[0056] The basic remote sensing imagery is divided into evaluation image regions according to a spatial grid or surface extent. For each evaluation image region, occlusion data can be formed by the spatial extent occupied by clouds, shadows, or missing pixels; spatial resolution data can be formed by the ground size corresponding to the pixels; radiometric consistency data can be formed by the differences in band response within adjacent or overlapping areas; and temporal continuity data can be formed by the time interval between acquisition time and observation period. These data are correlated with the spatial markers of the evaluation image regions to form image quality evaluation data.

[0057] The criteria for supplementary image acquisition record the permissible occlusion range, spatial resolution, radiation difference range, and observation time interval when the image is used for surface observation. The image quality evaluation data of each evaluation image area are compared with the criteria for supplementary image acquisition. Evaluation image areas with occlusion exceeding the permissible range, insufficient spatial resolution, radiation difference exceeding the permissible range, or excessive observation time interval are identified as supplementary image areas, and the spatial boundaries of the supplementary image areas are preserved.

[0058] The acquisition time period data records the time interval within the target area that can be used for imaging. Based on the distribution range of the supplementary image area and the acquisition time period data, an acquisition path and supplementary acquisition time data covering the supplementary image area are formed. Multispectral acquisition configuration data is formed based on the image quality evaluation data corresponding to the supplementary image area. For areas with insufficient spatial resolution, finer spatial sampling intervals can be configured; for areas with insufficient band response or significant radiometric differences, corresponding band combinations and imaging parameters can be configured. The acquisition path, supplementary acquisition time data, and multispectral acquisition configuration data are combined into supplementary acquisition task data. Images of the supplementary image area are acquired according to the supplementary acquisition task data to generate supplementary images.

[0059] Stable and corresponding registration control points are extracted from the overlapping spatial area of ​​the base remote sensing image and the supplementary image. These spatially corresponding registration control points are then paired. Based on the registration control point pairs, spatial transformation parameters are determined, and the supplementary image is transformed to the spatial coordinate reference used by the base remote sensing image to generate a registered supplementary image, ensuring that the same surface location in both types of images has a corresponding pixel location.

[0060] The re-acquired multispectral image data corresponding to the re-acquired image region is extracted from the registered and re-acquired image. Basic multispectral image data corresponding to the spatial location of the re-acquired multispectral image data is extracted from the basic remote sensing image. Fusion weight data is formed based on the image quality assessment data corresponding to the re-acquired image region. Fusion weights biased towards the re-acquired multispectral image data are assigned to locations with occlusion, defects, or large observation time intervals, while fusion weights biased towards the basic multispectral image data are assigned to locations with complete content in the basic remote sensing image. The two types of multispectral image data corresponding to the spatial location are fused according to the fusion weight data and written into an image data structure with spatial reference information and band labels to generate the target multispectral image set.

[0061] This embodiment reduces image content loss caused by occlusion, image defects, insufficient spatial resolution, and excessively large observation intervals by performing zonal quality assessment on basic remote sensing images and determining the areas for supplementary image acquisition. By organizing acquisition paths, acquisition time data, and multispectral acquisition configuration data according to supplementary acquisition needs, supplementary images corresponding to the defective areas can be obtained. Through spatial registration and pixel fusion based on image quality assessment data, a target multispectral image set with consistent spatial location and continuous multi-band content can be formed, improving the integrity and usability of image data for the target area.

[0062] In one embodiment, step S20 above includes: S201, Obtain the joint identification model including the boundary identification branch, attribute identification branch, confidence fusion module and land parcel separation branch; S202, input the target multispectral image set into the boundary recognition branch, extract boundary texture features and spectral distribution features from the target multispectral image set, and generate candidate land parcel boundary masks and boundary confidence data based on the boundary texture features and the spectral distribution features; S203, Based on the candidate plot boundary mask, extract candidate plot image blocks from the target multispectral image set, input the candidate plot image blocks into the attribute recognition branch, and generate candidate plot use data, candidate crop type data and attribute confidence data; S204, input the boundary confidence data and the attribute confidence data into the confidence fusion module to generate fused confidence data; input the candidate plot boundary mask, the candidate plot use data, the candidate crop type data, the fused confidence data and the target multispectral image set into the plot separation branch to generate mutually separated plot instance masks, as well as plot use data, crop type data and plot confidence data corresponding to the plot instance masks; S205, generate plot instance identifiers for each plot instance mask, and associate each plot instance identifier with the corresponding plot use data, crop type data and plot confidence data; S206, acquire spatial reference data corresponding to the target multispectral image set, convert each plot instance mask into plot boundary data associated with the plot instance identifier based on the spatial reference data, generate plot area data based on the plot boundary data, combine the plot instance identifier, the plot use data, the crop type data, the plot area data and the plot confidence data into plot attribute data, and combine the plot boundary data and the plot attribute data into current plot status data.

[0063] In this embodiment, the joint identification model includes a boundary identification branch, an attribute identification branch, a confidence fusion module, and a land parcel separation branch. The target multispectral image set carries multi-band pixel information and spatial reference data within the same spatial range. Before being fed into the joint identification model, the spatial resolution scale, band arrangement order, and pixel position relationship of each band can be unified, so that the same land surface location remains corresponding in different bands.

[0064] The boundary recognition branch reads image content from the target multispectral image set, forming boundary texture features from texture variations between adjacent pixels and spectral distribution features from response variations between different bands. Boundary texture features reflect the contour continuity at the edge of the field, while spectral distribution features reflect the coverage differences between adjacent areas. Organizing these two types of features according to spatial location creates candidate field boundary masks, and boundary confidence data is generated for each candidate boundary location to record the recognition reliability of each candidate boundary location.

[0065] Candidate plot boundary masks are used to extract candidate plot image blocks within the coverage area of ​​each candidate plot from the target multispectral image set. The attribute recognition branch reads the texture distribution, color response, and multiband response from the candidate plot image blocks to form candidate plot use data and candidate crop type data, and generates attribute confidence data. Candidate plot use data records the plot coverage category, candidate crop type data records the crop coverage category within the image range, and attribute confidence data records the degree of matching between the candidate attribute content and the image response.

[0066] The confidence fusion module receives boundary confidence data and attribute confidence data, and correlates the confidence level of spatial boundary identification with the confidence level of attribute identification within the land parcel to generate fused confidence data. When the edge texture of the land parcel is weak but the band response differences within the land parcel are clear, the fused confidence data can use the attribute identification results to supplement the boundary determination basis; when the responses within the land parcel are similar and the edge texture is continuous, the fused confidence data can use the boundary identification results to maintain the integrity of the land parcel outline.

[0067] The parcel separation branch, based on candidate parcel boundary masks, candidate parcel use data, candidate crop type data, fused confidence data, and target multispectral image set, distinguishes the spatial affiliation of adjacent parcels, forming mutually separated parcel instance masks. Each parcel instance mask represents the coverage area of ​​an independent parcel within the image pixel range and is associated with corresponding parcel use data, crop type data, and parcel confidence data, enabling adjacent parcels with similar attributes to be recorded separately.

[0068] Generating plot instance identifiers for each plot instance mask allows for the association of spatial extent, usage, crop category, and confidence content for the same plot. Plot instance identifiers can be unique values ​​within the current observation batch and written to the data record corresponding to the plot instance mask. When multiple plots exist within an image area, the data content for each plot can be organized separately using plot instance identifiers, reducing the overlap between boundary data and attribute data.

[0069] Spatial reference data records the transformation relationship between pixel locations in the target multispectral image set and the spatial coordinates of the target area. Based on spatial reference data, the pixel boundaries corresponding to the plot instance mask can be converted into plot boundary data, giving the plot boundary data coordinate content that can be used for spatial recording. Plot area data is generated based on the spatial range enclosed by the plot boundary data. Plot instance identifiers, plot use data, crop type data, plot area data, and plot confidence data are combined into plot attribute data. Then, the plot boundary data and plot attribute data are combined into current plot status data, giving each plot a corresponding spatial range and attribute content.

[0070] This embodiment extracts plot outline information and multi-band response information within the plot through boundary recognition and attribute recognition branches, respectively. It then uses a confidence fusion module and a plot separation branch to form the spatial range and attribute content corresponding to an independent plot, which can reduce data confusion caused by unclear boundaries between adjacent plots or similar plot attributes. By associating plot boundary data and plot attribute data through plot instance identifiers, and using spatial reference data to form current plot status data with spatial coordinates and area content, it can improve the accuracy of the correspondence between plot spatial range and attribute records and the data integrity.

[0071] In one embodiment, step S30 above includes: S301, extract land area data, land use data and crop type data from the land attribute data in the current land status data, and obtain historical land status data corresponding to the target area from the spatiotemporal version data storage area; S302, extract historical land parcel boundary data and historical land parcel attribute data from the historical land parcel status data, and extract historical land parcel area data, historical land parcel use data and historical crop type data from the historical land parcel attribute data; S303, the land parcel boundary data is spatially overlaid with the historical land parcel boundary data, and land parcel spatial correspondence data is formed based on the spatial overlay result. Land parcel boundary data that does not form a spatial correspondence with the historical land parcel boundary data is determined as newly added land parcel data, and historical land parcel boundary data that does not form a spatial correspondence with the land parcel boundary data is determined as disappeared land parcel data. S304, based on the spatial corresponding data of the land parcels, compare the spatially corresponding land parcel boundary data with the historical land parcel boundary data to generate boundary change data, compare the spatially corresponding land parcel area data with the historical land parcel area data to generate area change data, compare the spatially corresponding land parcel use data and the crop type data with the historical land parcel use data and the historical crop type data to generate attribute change data, and combine the boundary change data, the area change data, the attribute change data, the newly added land parcel data, and the disappeared land parcel data into change data to be verified; S305, superimpose the current land parcel status data, the historical land parcel status data, and the change data to be verified onto the target multispectral image set for display, and receive verification operation data for the land parcel boundary data, the land parcel use data, or the crop type data; S306, Based on the current land parcel status data and the verification operation data, the land parcel boundary data, land parcel use data, and crop type data are verified and updated. The land parcel area data is updated based on the verified and updated land parcel boundary data. The land parcel attribute data is updated based on the verified and updated land parcel use data, the verified and updated crop type data, and the updated land parcel area data. The verified and updated land parcel boundary data and the updated land parcel attribute data are combined to confirm the land parcel status data. S307, the confirmed land parcel status data is compared with the historical land parcel status data, the change area data and change category data are determined based on the comparison results, the change time data is obtained, the change area data, the change category data and the change time data are combined into a land parcel change event, the verification record data is generated based on the verification operation data, and the confirmed land parcel status data, the land parcel change event and the verification record data are associated and written into the spatiotemporal version data storage area.

[0072] In this embodiment, the spatiotemporal version data storage area is used to store land parcel status records formed at different observation points. The stored content may include land parcel boundary data, land parcel attribute data, observation time, spatial range markers, and version markers. After the current land parcel status data is formed, historical land parcel status data formed earlier in the same target area can be read from the spatiotemporal version data storage area based on the spatial range of the target area, observation time, and land parcel spatial location. This allows the current observation results to be compared with existing land parcel status records within the same spatial range.

[0073] The land parcel attribute data includes parcel area data, land use data, and crop type data. Parcel area data reflects the spatial size of the area enclosed by the parcel boundary; land use data reflects the current use of the parcel for planting, non-planting, or other surface uses; and crop type data reflects the main crop types covering the parcel. Historical parcel boundary data and historical parcel attribute data extracted from historical parcel status data are used as comparison objects for the current parcel boundary and current parcel attributes, respectively.

[0074] Spatial overlay is used to place land parcel boundary data and historical land parcel boundary data within the same spatial coordinate range and determine whether there is an overlap, inclusion, intersection, or separation relationship between them. Land parcels with spatial overlap or spatial proximity can form spatial correspondence data; land parcel boundary data that currently exists but does not form a spatial correspondence in historical records can form newly added land parcel data; historical land parcel boundary data that exists in historical records but does not currently form a spatial correspondence can form disappeared land parcel data. Through this processing, the three categories of land parcel addition, land parcel disappearance, and land parcel continuation can be distinguished.

[0075] Based on spatial parcel correspondence data, differences in boundaries, area, and attributes can be identified among spatially corresponding parcels. Boundary change data can be formed by boundary offset, boundary expansion, boundary contraction, or changes in boundary morphology; area change data can be formed by the difference or proportional change between the current parcel area data and historical parcel area data; attribute change data can be formed by category changes between parcel use data, crop type data, and historical parcel use data, and historical crop type data. Boundary change data, area change data, attribute change data, newly added parcel data, and disappeared parcel data are combined into change data to be verified, which carries the automatically identified changes.

[0076] The target multispectral image set serves as the image basis for verification display and can be overlaid with current plot status data, historical plot status data, and data to be verified for changes. Verification operation data can be formed from boundary adjustment data, plot merging data, plot segmentation data, and attribute modification data. Boundary adjustment data is used to correct boundary positions, plot merging data is used to combine multiple adjacent plots into one plot, plot segmentation data is used to split the same boundary area into multiple plots, and attribute modification data is used to correct land use or crop type. Verification operation data participates in updating plot boundary data, plot use data, and crop type data, enabling the automatic identification results to undergo spatial and attribute-level verification.

[0077] The updated land parcel boundary data is used to reconstruct the land parcel area data. The updated land use data, crop type data, and land parcel area data are used to update the land parcel attribute data. The updated land parcel boundary data and updated land parcel attribute data are combined to form confirmed land parcel status data. This confirmed land parcel status data is then compared with historical land parcel status data to generate change area data, change category data, and change time data. The change area data records the spatial extent of the change; the change category data records categories such as additions, disappearances, boundary changes, area changes, or attribute changes; and the change time data records the time when the change was confirmed. This data combination constitutes a land parcel change event and is written to the spatiotemporal version data storage area in conjunction with the updated data and the confirmed land parcel status data.

[0078] This embodiment reads historical land parcel status data from the spatiotemporal version data storage area and compares the current land parcel status data with the historical land parcel status data within the same spatial range in terms of boundaries, area, and attributes. This enables the identification of land parcel additions, disappearances, boundary changes, area changes, and attribute changes. By overlaying the change data to be verified with the target multispectral image set and receiving verification operation data, it can correct boundary and attribute deviations generated during the automatic identification process. By associating confirmed land parcel status data, land parcel change events, and verification record data and writing them into the spatiotemporal version data storage area, it can form a continuous land parcel status version record, improving the accuracy and traceability of land parcel dynamic change records.

[0079] In one embodiment, step S40 above includes: S401, extract land parcel boundary data from the confirmed land parcel status data, obtain regional observation images of land parcels with observation time identifiers and spatial ranges covering the land parcel boundary data, associate each observation time identifier with the corresponding regional observation image, and generate multi-period regional observation images. S402, Based on the land parcel boundary data, extract the image data within the land parcel range defined by the land parcel boundary data corresponding to each observation time marker from the multi-period regional observation images, and generate land parcel observation image segments. S403, extract red band data and near-infrared band data from the observation image segments of each plot, perform normalization difference processing on the red band data and the near-infrared band data, and generate spectral state index data corresponding to each observation time marker. S404, Arrange the spectral state index data corresponding to the boundary data of each plot in time sequence according to the observation time identifier to generate plot time sequence observation data; S405, Obtain smoothed window data, extract spectral state index data within adjacent observation time ranges from the plot time series observation data based on the smoothed window data, and smooth the plot time series observation data based on the spectral state index data within adjacent observation time ranges to generate smoothed time series observation data. S406, Based on the spectral state index data arranged according to the observation time identifier in the smooth time series observation data, trend fitting is performed to generate trend sequence data corresponding to the boundary data of each plot, and the trend sequence data and the observation time identifier are combined into state trend data.

[0080] In this embodiment, the verified spatial range of the land parcels is stored in the land parcel status data. The land parcel boundary data can be recorded using closed boundary point sequences or areal spatial data. Regional observation images covering the corresponding land parcel range are retrieved based on the land parcel boundary data, and the observation time identifier and spatial reference information of each regional observation image are read. The regional observation images can originate from different observation batches, and the coverage area of ​​each image should include the surface area determined by the same land parcel boundary data. After associating the observation time identifier with the corresponding regional observation image, multi-period regional observation images that can distinguish observation time points are formed.

[0081] Based on the spatial location of the land parcel boundary data within each regional observation image, image content within the land parcel area is extracted to generate land parcel observation image segments. Before extraction, the coordinate reference and pixel positions of each regional observation image can be unified based on spatial reference information, ensuring that image segments extracted at different times cover the same surface area. For cases where some pixels at the edge of a land parcel cross the boundary, pixel content falling within the areal range formed by the land parcel boundary data can be selected, reducing the mixing of image content from adjacent land parcels into the same land parcel observation image segment.

[0082] Red light and near-infrared data were extracted from image segments of each plot. Red light data reflects the absorption of red light by vegetation, while near-infrared data reflects the reflection of the vegetation canopy. Normalization was performed on the two types of data within the same plot and at the same observation time to generate spectral state index data. Normalization was used to generate index values ​​at the pixel level, and then plot-level index records were generated based on the index values ​​within the coverage area of ​​the plot boundary data, ensuring that the data generated at each observation time point have the same spatial objects.

[0083] The spectral status index data corresponding to each plot are arranged in time series according to the observation time identifier to generate plot time series observation data. The plot time series observation data retains the correlation between plot boundary data, observation time identifier, and spectral status index data at each observation time point. When multiple plots exist simultaneously, each plot forms its own time series to avoid mixing observation results from different plots.

[0084] Smoothing window data is used to determine the adjacent observation time ranges involved in the smoothing process. Based on the smoothing window data, spectral state index data from adjacent observation time points are read, and data affected by local imaging noise, short-term occlusion, or single observation bias are smoothed to generate smoothed time-series observation data. The smoothing window data can be configured as a continuous observation time range based on the observation interval, or the range of image records involved in the processing can be determined based on the actual time span to maintain the temporal continuity of state changes.

[0085] Trend fitting is performed on spectral state index data and observation time markers from smoothed time-series observation data to generate trend sequence data for each plot. The trend sequence data records the direction and degree of change of spectral state indices over observation time, and is combined with the observation time markers to form state trend data. Thus, the state trend data preserves the correspondence between plot spatial objects, observation time, and state change content.

[0086] This embodiment constrains the coverage and image capture range of multi-period regional observation images by using plot boundary data, enabling observations at different times to correspond to the same plot. It generates spectral state index data using red and near-infrared band data and organizes it into plot time-series observation data according to the observation time markers, enabling continuous recording of plot state changes. By smoothing the data, it reduces the impact of local observation biases on the time series and generates state trend data based on the smoothed time-series observation data, thereby improving the continuity and accuracy of plot state change recording.

[0087] In one embodiment, step S50 above includes: S501, extract plot boundary data and plot attribute data from the confirmed plot status data, extract crop type data from the plot attribute data, extract observation time identifier from the status trend data, obtain growth stage matching data corresponding to the crop type data, and determine growth stage data based on the observation time identifier and the growth stage matching data. S502, based on the plot boundary data, the crop type data and the growth stage data, filter the historical spectral state index sequence from the historical state baseline data storage area, and align the historical spectral state index sequence according to the growth stage data to generate historical state baseline data; S503, based on the plot boundary data and the observation time identifier, obtain meteorological observation data, soil observation data, thermal infrared observation data and irrigation observation data, align the meteorological observation data, soil observation data, thermal infrared observation data and irrigation observation data according to the observation time identifier, and spatially associate them according to the plot boundary data to generate multi-source observation data of the plot; S504, compare the state trend data with the historical state baseline data according to the growth stage data, extract the state trend segments that deviate from the historical state baseline data from the state trend data, and generate state deviation data; S505, extract the moisture state data, temperature state data and soil nutrient state data corresponding to the state trend section from the multi-source observation data of the plot, and associate the moisture state data, temperature state data and soil nutrient state data with the state deviation data to generate risk evidence data; S506, determine the state deviation type based on the state deviation data, determine the observation state combination based on the risk evidence data, obtain the early warning discrimination condition containing the correspondence between the state deviation type, the observation state combination, the risk category identifier, and the early warning level identifier, generate risk category data and early warning level data based on the state deviation type, the observation state combination, and the early warning discrimination condition, generate early warning area data based on the land parcel boundary data, obtain early warning time data, and combine the confirmed land parcel status data, the risk category data, the early warning level data, the risk evidence data, the early warning area data, and the early warning time data into an early warning event.

[0088] In this embodiment, the land parcel status data is confirmed to include verified land parcel boundary data and land parcel attribute data. The land parcel boundary data can record the land parcel coverage area using closed coordinate ranges or areal spatial data. The crop type data in the land parcel attribute data is used to distinguish the status change characteristics corresponding to different crops. The status trend data carries an observation time identifier to determine the time and location of the current status change.

[0089] Growth stage matching data can be stored using a corresponding data structure between crop type, time interval, and growth stage markers. Matching content is retrieved based on crop type data, and the observation time marker is placed within the corresponding time interval to form growth stage data, enabling subsequent state comparisons to be conducted under the same crop type and corresponding growth stage.

[0090] The historical state baseline data storage area organizes historical spectral state index sequences according to spatial extent, crop type, and growth stage. Index sequences covering the corresponding plot area are selected based on plot boundary data. Index sequences with similar state change characteristics are selected based on crop type and growth stage data. Then, index records from different observation times are aligned according to growth stage data to generate historical state baseline data. The historical state baseline data can record the range and direction of index changes corresponding to each aligned time position.

[0091] Meteorological, soil, thermal infrared, and irrigation observation data are retrieved based on plot boundary data and observation time identifiers. Meteorological data may include precipitation and air temperature data; soil data may include soil moisture and nutrient status; thermal infrared data may include surface temperature response; and irrigation data may include irrigation time and amount. All types of observation data are time-aligned according to their observation time identifiers and correlated according to the spatial range defined by the plot boundary data to generate multi-source plot observation data.

[0092] The trend data corresponding to the growth stage in the state trend data is compared with the historical state baseline data to identify state trend segments that exceed the range of change corresponding to the historical state baseline data, generating state deviation data. State deviation data can record the direction, duration, and degree of deviation, making the differences between state changes and historical references readable data content.

[0093] Observational records corresponding to the state trend segments in time and space are extracted from multi-source observational data of the plot. Moisture state data is generated based on soil moisture content, precipitation, and irrigation status; temperature state data is generated based on air temperature and surface temperature response; and soil nutrient state data is generated based on soil nutrient status. The moisture state data, temperature state data, and soil nutrient state data are correlated with state deviation data to generate risk evidence data, ensuring that state deviations are supported by corresponding observational data.

[0094] State deviation types are generated based on state deviation data, and observation state combinations are generated based on risk evidence data. Early warning discrimination conditions store the correspondence between state deviation types, observation state combinations, risk category identifiers, and early warning level identifiers, and generate risk category data and early warning level data based on the matching results. Early warning area data is generated based on plot boundary data, and early warning time data is generated based on the time range corresponding to the state trend segment or the early warning generation time. Confirmed plot state data, risk category data, early warning level data, risk evidence data, early warning area data, and early warning time data are combined to form an early warning event.

[0095] This embodiment forms historical state baseline data based on plot boundary data, crop type data, and growth stage data, enabling comparison of state trend data with historical states of the corresponding spatial range and growth stage. By temporally aligning and spatially correlating meteorological observation data, soil observation data, thermal infrared observation data, and irrigation observation data, it provides multi-source observational basis for state deviation data. By combining state deviation types and observation states to convert them into risk category data, warning level data, warning area data, and warning time data, it can generate warning events corresponding to specific plots and observation periods, improving the accuracy of plot state anomaly identification and the completeness of warning information.

[0096] In one embodiment, step S60 above includes: S601, extract land parcel boundary data and land parcel attribute data from the confirmed land parcel status data, convert the land parcel boundary data into land parcel spatial graphic data, write land parcel association identifiers for each land parcel spatial graphic data, and combine the land parcel spatial graphic data, the land parcel attribute data and the land parcel association identifiers into a land parcel status record; S602, extract change area data, change category data and change time data from the land parcel change event; based on the spatial correspondence between the change area data and the land parcel spatial graphic data, associate the change area data that forms a spatial correspondence with the land parcel spatial graphic data with the corresponding land parcel association identifier; generate land parcel association identifiers for change area data that does not form a spatial correspondence with the land parcel spatial graphic data; and combine the land parcel association identifier, the change area data, the change category data and the change time data into a change event record; S603, based on the land parcel boundary data, extract trend sequence data and observation time identifier from the state trend data, extract observation state data corresponding to the land parcel boundary data and the observation time identifier from the multi-source observation data of the land parcel, determine the land parcel association identifier corresponding to the trend sequence data based on the spatial correspondence between the land parcel boundary data and the land parcel spatial graphic data, and combine the land parcel association identifier, the trend sequence data, the observation time identifier and the observation state data into a state observation record; S604, extract warning area data, risk category data, warning level data and warning time data from the warning event, determine the land parcel association identifier corresponding to the warning event based on the spatial correspondence between the warning area data and the land parcel spatial graphic data, and combine the land parcel association identifier, the risk category data, the warning level data and the warning time data into a warning event record; S605, generate land parcel distribution layer data based on the land parcel status record, generate status trend display data based on the status observation record, and arrange the change event record and the early warning event record in chronological order according to the change time data and the early warning time data to generate land parcel event time series data; S606, combine the land parcel status record, the status observation record, the change event record, the early warning event record, the land parcel distribution layer data, the status trend display data, and the land parcel event time series data into a panoramic land parcel status data.

[0097] In this embodiment, the land parcel status data is confirmed to include verified land parcel boundary data and land parcel attribute data. Land parcel boundary data can be stored as a closed set of boundary points, a planar coordinate object, or a vector range with spatial reference information. When converting the land parcel boundary data into land parcel spatial graphic data, planar graphic objects suitable for layer organization can be formed based on coordinate reference relationships, and the boundary position corresponding to each graphic object is retained. Land parcel attribute data can be saved along with the corresponding planar graphic objects, ensuring that spatial location, area, use, and crop type correspond to the same land parcel.

[0098] The land parcel association identifier is used to establish data associations between various records corresponding to the same land parcel. After writing the land parcel association identifier into the spatial graphic data of each land parcel, the spatial graphic data, land parcel attribute data, and land parcel association identifier are combined into a land parcel status record. The land parcel status record can be stored as a data object indexed by the land parcel association identifier, so that subsequent changes, observations, and warnings can be located at the specific land parcel.

[0099] Land parcel change events include change area data, change category data, and change time data. Change area data records the spatial extent of boundary changes, parcel additions, or parcel disappearances; change category data records the corresponding change type; and change time data records the time the change was confirmed. For change area data that can form a spatial correspondence with the current parcel spatial graphic data, the change area data is associated with the corresponding parcel association identifier. For change area data that cannot form a spatial correspondence with the current parcel spatial graphic data, an independent parcel association identifier can be generated to preserve parcels that have disappeared or spatial objects that only exist in the change record. The resulting change event record simultaneously saves parcel ownership, change range, change type, and occurrence time.

[0100] The status trend data includes trend sequence data organized according to observation time. Based on the spatial correspondence between plot boundary data and plot spatial graphic data, the plot association identifier to which the trend sequence data belongs can be determined. The multi-source plot observation data stores observation status data corresponding to the plot range and observation time; the corresponding observation content is read based on the same plot boundary data and observation time identifier. The plot association identifier, trend sequence data, observation time identifier, and observation status data are combined into a status observation record, ensuring that the change trend of the same plot corresponds to the observation content at the corresponding time point.

[0101] The early warning event includes data on the warning area, risk category, warning level, and warning time. Based on the spatial correspondence between the warning area data and the spatial graphic data of the land parcels, the associated identifier of the land parcel to which the early warning event belongs is determined, and the risk category data, warning level data, and warning time data are combined with this identifier to form the early warning event record. Thus, the early warning content can be located at a specific land parcel and can be associated with the attribute records, change records, and status observation records of that land parcel.

[0102] When generating land parcel distribution layer data based on land parcel status records, land parcel spatial graphic data and land parcel attribute data can be written into the layer data structure. When generating status trend display data based on status observation records, trend sequence data and observation status data can be organized according to observation time identifiers. Change event records and warning event records are arranged according to change time data and warning time data to form land parcel event time series data. After combining land parcel status records, status observation records, change event records, warning event records, land parcel distribution layer data, status trend display data, and land parcel event time series data, a panoramic land parcel status data is formed.

[0103] This embodiment converts confirmed land parcel status data into land parcel status records with land parcel association identifiers, ensuring that the spatial range and attributes of land parcels remain consistent. By organizing land parcel change events, status trend data, multi-source observation data of land parcels, and early warning events into records corresponding to land parcel association identifiers, it can retain the changes in currently existing and disappeared land parcels and associate status changes, observation content, and early warning content with the corresponding land parcels. By forming land parcel distribution layer data, status trend display data, and land parcel event time series data, it can centrally record the spatial status, change process, and abnormal information of land parcels, improving the completeness and accessibility of the panoramic status data of land parcels.

[0104] In one embodiment, a schematic diagram of an application architecture for a method for generating land parcel status data is provided, such as... Figure 3 As shown, the application architecture includes a client, an application programming interface gateway, a microservice cluster, a data layer, and an external system integration layer.

[0105] The user interface, available via a web browser or application (Web / App), includes a digital farmland panoramic dashboard, a plot editing and confirmation interface, and a risk warning notification center. The digital farmland panoramic dashboard displays the spatial extent of plots, status changes, and warning information. The plot editing and confirmation interface receives requests for plot boundary adjustments or attribute corrections. The risk warning notification center displays warning information associated with specific plots.

[0106] The Application Programming Interface Gateway (API Gateway) uses Kong gateway components and Nginx network service components, sets up identity authentication based on the Open Authorization Protocol (OAuth2.0), and performs traffic control, circuit breaking, and log auditing. User-initiated requests for image queries, land parcel verification, status display, and alert queries are transmitted to the microservice cluster via the API Gateway.

[0107] The microservice cluster utilizes the Spring Cloud microservice framework and the Kubernetes container orchestration platform. The remote sensing mission scheduling service invokes satellites or drones to acquire imagery of target areas. The artificial intelligence (AI) inference service leverages a graphics processing unit (GPU) cluster for boundary identification, usage classification, and growth analysis. The change detection service performs differential comparison and version control. The drone scheduling engine plans flight missions. The risk warning service performs Normalized Difference Vegetation Index (NDVI) analysis, multi-source fusion, and early warning push notifications. The data fusion engine performs map construction and multi-source data fusion. The approval process service receives human-machine collaborative verification results. The data synchronization service uses Kafka event streaming to synchronize the generated data to the insurance system in an idempotent manner.

[0108] In the data layer, PostgreSQL relational database and PostGIS spatial data extension are used to store plot spatial data, Neo4j graph data storage component is used to store digital farmland map data, MinIO object storage component is used to store remote sensing imagery, message queue (MQ) or Kafka is used to deliver task messages and update messages, Redis cache component is used to store frequently accessed data, Model Registry is used to store artificial intelligence model versions, and Sensor DB is used to store real-time meteorological data and irrigation data.

[0109] In the external system integration layer, the satellite remote sensing interface is used to receive remote sensing images provided by Gaofen, Sentinel, or Planet image services; the drone platform is used to connect to DJI or XAG flight equipment; the meteorological interface is used to receive national or local meteorological data; the soil survey platform is used to provide soil observation content; the intelligent irrigation system is used to provide irrigation observation content; the planting insurance core system is used to receive data that needs to be queried for underwriting, surveying, and claims business; and the notification platform is used to send early warning information via DingTalk or WeChat Work.

[0110] In one embodiment, a schematic diagram of a remote sensing image re-capture and fusion and land parcel status update process is provided as a method for generating land parcel status data, such as... Figure 4 As shown, image quality is assessed after satellite imagery is acquired. The image uses an image quality score below 60 or an image not updated for more than 15 days as a trigger for UAV re-capture, representing a parameter configuration method to indicate when the image quality assessment results meet the requirements for supplementary acquisition and thus initiate a supplementary acquisition task.

[0111] After the drone re-capture is initiated, the drone mission scheduling engine generates flight paths and parameter configurations for the areas to be acquired, controls the drone to fly and acquire high-resolution multispectral images. The acquired images undergo registration, fusion, and cloud removal processing. Registration is used to establish the spatial relationship between satellite and drone images, fusion is used to combine the available multispectral content from the two types of images, and cloud removal is used to process image areas affected by cloud cover.

[0112] The processed multispectral imagery is input into an artificial intelligence (AI) model for land parcel identification, outputting updated land parcel patches. The updated land parcel patches in the image represent the spatial graphic results of the land parcels identified in the current batch of observations, and can include parcel boundary information. The land parcel attribute information identified by the model can be associated and saved with the spatial graphic results of these parcels, forming the current parcel status data. The update here indicates an update to the current image identification result relative to the existing displayed content, and does not indicate that change comparison and verification have been completed.

[0113] The data generated from the current identification is synchronized to the main database and used to update the digital farmland panorama. In this embodiment, the main database is used to store plot spatial data and corresponding status information, while the digital farmland panorama is used to display the spatial distribution of plots and the status information formed after data organization.

[0114] In one embodiment, a schematic diagram of a land parcel status trend analysis and early warning event generation process is provided as a method for generating land parcel status data, such as... Figure 5 As shown, the Normalized Difference Vegetation Index (NDVI) sequence was obtained according to the weekly observation cycle. The NDVI sequence is used to record the spectral status of vegetation in the same plot at multiple observation times and can be used as a status indicator in the plot's time-series observation data.

[0115] The normalized vegetation index (NVI) sequence was denoised and trend-fitted to generate the state change results. The Savitzky-Golay filter was used for sequence smoothing, and a Long Short-Term Memory (LSTM) network was used to extract the sequence change trend. Smoothing was used to reduce sequence fluctuations caused by local observation biases, and trend fitting was used to generate the state trend content over time.

[0116] The status trend is compared with the historical baseline. In the figure, a deviation greater than twice the standard deviation is used as a parameter condition for judging anomalies. The historical baseline is used to record the range of status changes for the corresponding plot or crop stage. When the current status trend exceeds this range, an anomaly judgment result is formed.

[0117] Anomaly detection results are combined with multi-source data to identify risk categories. Low soil moisture and little rainfall lead to a drought risk assessment. Abnormally high temperatures lead to a pest and disease risk assessment. Abnormal nutrient indicators lead to a nutrient deficiency risk assessment. The results of all risk assessments are combined to form an early warning event.

[0118] The diagram also shows a three-tiered early warning system. Level 1 warnings are for agricultural technicians to view; Level 2 warnings trigger inspection tasks; and Level 3 warnings freeze compensation eligibility. Level 3 warnings indicate a high-level abnormal situation triggering business control measures; related business processing can still continue based on inspection results and review records.

[0119] In one embodiment, a schematic diagram of multi-source data fusion and panoramic land parcel status display is provided as a method for generating land parcel status data, such as... Figure 6 As shown, the data fusion engine integrates remote sensing imagery, meteorological data, soil data, irrigation data, and farmer behavior data. Remote sensing imagery provides the spatial extent and surface imagery of the land parcels; meteorological, soil, and irrigation data provide environmental status information related to the land parcels and observation time; and farmer behavior data can be used to store records of land parcel declaration, confirmation, inspection, or management operations.

[0120] The data fusion engine organizes various types of data into a digital farmland map. Nodes in the digital farmland map include plots, crops, sensors, and events. Plot nodes store the spatial extent and attribute information of the plot; crop nodes store crop category information; sensor nodes store the sources of meteorological, soil, or irrigation observations; and event nodes store event information generated by plot changes, early warnings, or management operations.

[0121] Edges in digital farmland maps include spatial, temporal, and causal relationships. Spatial relationships record the spatial association between plots of land and the image area, sensor observation range, or event occurrence area. Temporal relationships record the chronological order of observed content, changes, and warnings. Causal relationships record the correlation information formed between changes in observation status, anomaly identification results, and event content.

[0122] Digital farmland maps are used to generate digital farmland panoramic dashboards. These dashboards can display plot distribution, crop attributes, status trends, multi-source observation data, and event time records, and provide query content for underwriting, surveying, and claims processing. The digital farmland maps and digital farmland panoramic dashboards in the diagrams serve as a structured organization and display format for panoramic plot status data.

[0123] In one embodiment, a land parcel status data generation device is provided, which corresponds one-to-one with the land parcel status data generation method described in the above embodiments. (Refer to...) Figure 7 , Figure 7 This is a schematic diagram of the functional modules of a preferred embodiment of the land parcel status data generation device of the present invention. The modules include: image acquisition and fusion module 10, land parcel status identification module 20, land parcel change confirmation module 30, status trend analysis module 40, status risk early warning module 50, and panoramic status construction module 60. Detailed descriptions of each functional module are as follows: The image acquisition and fusion module 10 is used to acquire basic remote sensing images of the target area, trigger acquisition based on the image quality evaluation of the basic remote sensing images to acquire acquisition images, and register and fuse the basic remote sensing images and the acquisition images into a target multispectral image set. The land parcel status identification module 20 is used to process the target multispectral image set based on the joint identification model to generate current land parcel status data containing land parcel boundary data and land parcel attribute data; The land parcel change confirmation module 30 is used to obtain historical land parcel status data corresponding to the current land parcel status data from the spatiotemporal version data storage area, compare and verify the current land parcel status data with the historical land parcel status data, generate confirmed land parcel status data and land parcel change event, and write the confirmed land parcel status data into the spatiotemporal version data storage area. The status trend analysis module 40 is used to obtain multi-period regional observation images based on the land parcel boundary data in the confirmed land parcel status data, generate land parcel time-series observation data based on the multi-period regional observation images, and generate status trend data based on the land parcel time-series observation data. The status risk warning module 50 is used to acquire historical status baseline data and multi-source observation data of the land parcel corresponding to the confirmed land parcel status data, and generate warning events based on the status trend data, the historical status baseline data and the multi-source observation data of the land parcel. The panoramic status construction module 60 is used to generate panoramic status data of the land parcel based on the confirmed land parcel status data, the land parcel change event, the status trend data, the land parcel multi-source observation data, and the early warning event.

[0124] In one embodiment, the image acquisition and fusion module 10 is specifically used for: Acquire basic remote sensing images of the target area, divide the basic remote sensing images into multiple evaluation image regions, extract occlusion status data, spatial resolution data, radiometric consistency data and temporal continuity data from each evaluation image region, and combine the occlusion status data, spatial resolution data, radiometric consistency data and temporal continuity data into image quality evaluation data. Obtain the supplementary acquisition criteria, compare the image quality evaluation data corresponding to each evaluation image area with the supplementary acquisition criteria, and determine the evaluation image area that meets the supplementary acquisition criteria as the supplementary acquisition image area. Acquire the acquisition time period data corresponding to the target area, generate acquisition path and acquisition time data based on the acquired image area and the acquisition time period data, generate multispectral acquisition configuration data based on the image quality evaluation data corresponding to the acquired image area, and combine the acquisition path, the acquisition time data and the multispectral acquisition configuration data into acquisition task data. Collect images of the supplementary image area according to the acquisition path, supplementary acquisition time data and multispectral acquisition configuration data in the supplementary acquisition task data, and generate supplementary images; Registration control points covering the area of ​​the supplementary image are extracted from the base remote sensing image and the supplementary image respectively. The registration control points corresponding to the spatial location are combined into registration control point pairs. Based on the registration control point pairs, the supplementary image is registered to the base remote sensing image to generate a registration supplementary image. Extract the multispectral image data corresponding to the area of ​​the supplemented image from the registered supplemented image, extract the basic multispectral image data corresponding to the spatial location of the supplemented multispectral image data from the basic remote sensing image, generate fusion weight data based on the image quality evaluation data corresponding to the area of ​​the supplemented image, and fuse the supplemented multispectral image data and the basic multispectral image data based on the fusion weight data to generate the target multispectral image set.

[0125] In one embodiment, the land parcel status identification module 20 is specifically used for: Obtain a joint identification model including boundary recognition branch, attribute recognition branch, confidence fusion module and land parcel separation branch; The target multispectral image set is input into the boundary recognition branch, and boundary texture features and spectral distribution features are extracted from the target multispectral image set. Based on the boundary texture features and the spectral distribution features, candidate land parcel boundary masks and boundary confidence data are generated. Based on the candidate plot boundary mask, candidate plot image blocks are extracted from the target multispectral image set. The candidate plot image blocks are then input into the attribute recognition branch to generate candidate plot use data, candidate crop type data, and attribute confidence data. The boundary confidence data and the attribute confidence data are input into the confidence fusion module to generate fused confidence data. The candidate plot boundary mask, the candidate plot use data, the candidate crop type data, the fused confidence data, and the target multispectral image set are input into the plot separation branch to generate mutually separated plot instance masks, as well as plot use data, crop type data, and plot confidence data corresponding to the plot instance masks. Generate plot instance identifiers for each plot instance mask, and associate each plot instance identifier with the corresponding plot use data, crop type data, and plot confidence data; Acquire spatial reference data corresponding to the target multispectral image set; convert each plot instance mask into plot boundary data associated with the plot instance identifier based on the spatial reference data; generate plot area data based on the plot boundary data; combine the plot instance identifier, the plot use data, the crop type data, the plot area data, and the plot confidence data into plot attribute data; and combine the plot boundary data and the plot attribute data into current plot status data.

[0126] In one embodiment, the land parcel change confirmation module 30 is specifically used for: Extract land area data, land use data, and crop type data from the land attribute data in the current land status data, and obtain historical land status data corresponding to the target area from the spatiotemporal version data storage area; Historical land parcel boundary data and historical land parcel attribute data are extracted from the historical land parcel status data. Historical land parcel area data, historical land parcel use data, and historical crop type data are extracted from the historical land parcel attribute data. The land parcel boundary data is spatially overlaid with the historical land parcel boundary data. Based on the spatial overlay result, spatial correspondence data of land parcels is formed. Land parcel boundary data that does not form a spatial correspondence with the historical land parcel boundary data is identified as newly added land parcel data, and historical land parcel boundary data that does not form a spatial correspondence with the land parcel boundary data is identified as disappeared land parcel data. Based on the spatial corresponding data of the land parcels, the spatially corresponding land parcel boundary data is compared with the historical land parcel boundary data to generate boundary change data; the spatially corresponding land parcel area data is compared with the historical land parcel area data to generate area change data; the spatially corresponding land parcel use data and the crop type data are compared with the historical land parcel use data and the historical crop type data to generate attribute change data; the boundary change data, the area change data, the attribute change data, the newly added land parcel data, and the disappeared land parcel data are combined into change data to be verified. The current land parcel status data, the historical land parcel status data, and the change data to be verified are overlaid on the target multispectral image set for display, and verification operation data for the land parcel boundary data, the land parcel use data, or the crop type data are received; Based on the current land parcel status data and the verification operation data, the land parcel boundary data, land parcel use data, and crop type data are verified and updated. The land parcel area data is updated based on the verified and updated land parcel boundary data. The land parcel attribute data is updated based on the verified and updated land parcel use data, the verified and updated crop type data, and the updated land parcel area data. The verified and updated land parcel boundary data and the updated land parcel attribute data are combined to confirm the land parcel status data. The confirmed land parcel status data is compared with the historical land parcel status data. Based on the comparison results, the change area data and change category data are determined, and the change time data is obtained. The change area data, the change category data, and the change time data are combined into a land parcel change event. Based on the verification operation data, verification record data is generated. The confirmed land parcel status data, the land parcel change event, and the verification record data are associated and written into the spatiotemporal version data storage area.

[0127] In one embodiment, the state trend analysis module 40 is specifically used for: Extract land parcel boundary data from the confirmed land parcel status data, obtain regional observation images of land parcels with observation time identifiers and spatial ranges covering the land parcel boundary data, associate each observation time identifier with the corresponding regional observation image, and generate multi-period regional observation images. Based on the land parcel boundary data, image data within the land parcel range defined by the land parcel boundary data corresponding to each observation time marker is extracted from the multi-period regional observation images to generate land parcel observation image segments. Red band data and near-infrared band data are extracted from the observation image segments of each plot. Based on the red band data and the near-infrared band data, normalization difference processing is performed to generate spectral state index data corresponding to each observation time marker. The spectral state index data corresponding to the boundary data of each plot are arranged in time sequence according to the observation time identifier to generate plot time-series observation data. Obtain smoothed window data, extract spectral state index data within adjacent observation time ranges from the plot time series observation data based on the smoothed window data, and smooth the plot time series observation data based on the spectral state index data within adjacent observation time ranges to generate smoothed time series observation data. Based on the spectral state index data arranged according to the observation time identifier in the smooth time series observation data, trend fitting is performed to generate trend sequence data corresponding to the boundary data of each plot, and the trend sequence data and the observation time identifier are combined into state trend data.

[0128] In one embodiment, the state risk warning module 50 is specifically used for: Extract plot boundary data and plot attribute data from the confirmed plot status data, extract crop type data from the plot attribute data, extract observation time identifier from the status trend data, obtain growth stage matching data corresponding to the crop type data, and determine growth stage data based on the observation time identifier and the growth stage matching data. Based on the plot boundary data, the crop type data, and the growth stage data, historical spectral state index sequences are selected from the historical state baseline data storage area, and the historical spectral state index sequences are aligned according to the growth stage data to generate historical state baseline data. Meteorological observation data, soil observation data, thermal infrared observation data, and irrigation observation data are obtained based on the plot boundary data and the observation time identifier. The meteorological observation data, soil observation data, thermal infrared observation data, and irrigation observation data are time-aligned according to the observation time identifier and spatially correlated according to the plot boundary data to generate multi-source observation data of the plot. The state trend data is compared with the historical state baseline data according to the growth stage data, and the state trend segments that deviate from the historical state baseline data are extracted from the state trend data to generate state deviation data. Extract moisture state data, temperature state data, and soil nutrient state data corresponding to the state trend segment from the multi-source observation data of the plot, and associate the moisture state data, temperature state data, and soil nutrient state data with the state deviation data to generate risk evidence data; Based on the state deviation data, the state deviation type is determined; based on the risk evidence data, the observation state combination is determined; and a warning discrimination condition containing the correspondence between the state deviation type, the observation state combination, the risk category identifier, and the warning level identifier is obtained. Based on the state deviation type, the observation state combination, and the warning discrimination condition, risk category data and warning level data are generated; based on the land parcel boundary data, warning area data is generated; and warning time data is obtained. The confirmed land parcel status data, the risk category data, the warning level data, the risk evidence data, the warning area data, and the warning time data are combined into a warning event.

[0129] In one embodiment, the panoramic state construction module 60 is specifically used for: Extract land parcel boundary data and land parcel attribute data from the confirmed land parcel status data, convert the land parcel boundary data into land parcel spatial graphic data, write land parcel association identifiers for each land parcel spatial graphic data, and combine the land parcel spatial graphic data, the land parcel attribute data, and the land parcel association identifiers into a land parcel status record; Extract change area data, change category data, and change time data from the land parcel change event. Based on the spatial correspondence between the change area data and the land parcel spatial graphic data, associate the change area data that forms a spatial correspondence with the land parcel spatial graphic data with the corresponding land parcel association identifier. Generate land parcel association identifiers for change area data that does not form a spatial correspondence with the land parcel spatial graphic data. Combine the land parcel association identifier, the change area data, the change category data, and the change time data into a change event record. Based on the land parcel boundary data, trend sequence data and observation time identifiers are extracted from the state trend data. Observation state data corresponding to the land parcel boundary data and observation time identifiers are extracted from the multi-source observation data of the land parcel. Based on the spatial correspondence between the land parcel boundary data and the land parcel spatial graphic data, a land parcel association identifier corresponding to the trend sequence data is determined. The land parcel association identifier, the trend sequence data, the observation time identifier, and the observation state data are combined into a state observation record. Data on the warning area, risk category, warning level, and warning time are extracted from the warning event. Based on the spatial correspondence between the warning area data and the land parcel spatial graphic data, a land parcel association identifier corresponding to the warning event is determined. The land parcel association identifier, the risk category data, the warning level data, and the warning time data are combined into a warning event record. Based on the land parcel status records, land parcel distribution layer data is generated; based on the status observation records, status trend display data is generated; and the change event records and warning event records are arranged chronologically according to the change time data and the warning time data to generate land parcel event time series data. The land parcel status records, status observation records, change event records, early warning event records, land parcel distribution layer data, status trend display data, and land parcel event time series data are combined into a panoramic land parcel status data.

[0130] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 8 As shown. The computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides determination and control capabilities. The memory includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with external clients via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a land parcel state data generation method on the server side.

[0131] In one embodiment, a computer device is provided, which may be a client, and its internal structure diagram may be as follows: Figure 9 As shown, the computer device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides determination and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with an external server via a network connection. When executed by the processor, the computer program implements the functions or steps of a land parcel state data generation method on the client side.

[0132] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps: Acquire basic remote sensing images of the target area, trigger supplementary acquisition based on the image quality evaluation of the basic remote sensing images to obtain supplementary images, and register and fuse the basic remote sensing images and the supplementary images into a target multispectral image set. The target multispectral image set is processed based on the joint recognition model to generate current land parcel status data containing land parcel boundary data and land parcel attribute data. The historical land parcel status data corresponding to the current land parcel status data is obtained from the spatiotemporal version data storage area. The current land parcel status data and the historical land parcel status data are compared and verified to generate confirmed land parcel status data and land parcel change events. The confirmed land parcel status data is written into the spatiotemporal version data storage area. Based on the land parcel boundary data in the confirmed land parcel status data, multi-period regional observation images are obtained; land parcel time-series observation data is generated based on the multi-period regional observation images; and status trend data is generated based on the land parcel time-series observation data. Obtain historical status baseline data and multi-source observation data of the land parcel corresponding to the confirmed land parcel status data, and generate early warning events based on the status trend data, the historical status baseline data and the multi-source observation data of the land parcel; Based on the confirmed land parcel status data, the land parcel change events, the status trend data, the land parcel multi-source observation data, and the early warning events, panoramic land parcel status data is generated.

[0133] In one embodiment, a computer-readable storage medium is provided, which may be non-volatile or volatile, and a computer program is stored thereon, which, when executed by a processor, performs the following steps: Acquire basic remote sensing images of the target area, trigger supplementary acquisition based on the image quality evaluation of the basic remote sensing images to obtain supplementary images, and register and fuse the basic remote sensing images and the supplementary images into a target multispectral image set. The target multispectral image set is processed based on the joint recognition model to generate current land parcel status data containing land parcel boundary data and land parcel attribute data. The historical land parcel status data corresponding to the current land parcel status data is obtained from the spatiotemporal version data storage area. The current land parcel status data and the historical land parcel status data are compared and verified to generate confirmed land parcel status data and land parcel change events. The confirmed land parcel status data is written into the spatiotemporal version data storage area. Based on the land parcel boundary data in the confirmed land parcel status data, multi-period regional observation images are obtained; land parcel time-series observation data is generated based on the multi-period regional observation images; and status trend data is generated based on the land parcel time-series observation data. Obtain historical status baseline data and multi-source observation data of the land parcel corresponding to the confirmed land parcel status data, and generate early warning events based on the status trend data, the historical status baseline data and the multi-source observation data of the land parcel; Based on the confirmed land parcel status data, the land parcel change events, the status trend data, the land parcel multi-source observation data, and the early warning events, panoramic land parcel status data is generated.

[0134] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions on the server side and client side in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0135] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0136] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0137] It should be noted that any AI models, software tools, or components not belonging to this company appearing in the embodiments of this application are merely illustrative examples and do not represent actual use. All user personal information involved in the embodiments of this application has been authorized (with the knowledge and consent) by the relevant parties or has been fully authorized by all parties, and the executing entity may obtain it through various legal and compliant means. The collection, storage, use, processing, transmission, provision, and disclosure of the information, data, and signals involved all comply with relevant laws and regulations and do not violate public order and good morals.

[0138] The above-described 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 the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for generating land parcel status data, characterized in that, Includes the following steps: Acquire basic remote sensing images of the target area, trigger supplementary acquisition based on the image quality evaluation of the basic remote sensing images to obtain supplementary images, and register and fuse the basic remote sensing images and the supplementary images into a target multispectral image set. The target multispectral image set is processed based on the joint recognition model to generate current land parcel status data containing land parcel boundary data and land parcel attribute data. The historical land parcel status data corresponding to the current land parcel status data is obtained from the spatiotemporal version data storage area. The current land parcel status data and the historical land parcel status data are compared and verified to generate confirmed land parcel status data and land parcel change events. The confirmed land parcel status data is written into the spatiotemporal version data storage area. Based on the land parcel boundary data in the confirmed land parcel status data, multi-period regional observation images are obtained; land parcel time-series observation data is generated based on the multi-period regional observation images; and status trend data is generated based on the land parcel time-series observation data. Obtain historical status baseline data and multi-source observation data of the land parcel corresponding to the confirmed land parcel status data, and generate early warning events based on the status trend data, the historical status baseline data and the multi-source observation data of the land parcel; Based on the confirmed land parcel status data, the land parcel change events, the status trend data, the land parcel multi-source observation data, and the early warning events, panoramic land parcel status data is generated.

2. The method for generating land parcel status data as described in claim 1, characterized in that, Acquire basic remote sensing imagery of the target area; trigger supplementary acquisition based on image quality evaluation of the basic remote sensing imagery to obtain supplementary images; register and fuse the basic remote sensing imagery and the supplementary images into a target multispectral image set, including: Acquire basic remote sensing images of the target area, divide the basic remote sensing images into multiple evaluation image regions, extract occlusion status data, spatial resolution data, radiometric consistency data and temporal continuity data from each evaluation image region, and combine the occlusion status data, spatial resolution data, radiometric consistency data and temporal continuity data into image quality evaluation data. Obtain the supplementary acquisition criteria, compare the image quality evaluation data corresponding to each evaluation image area with the supplementary acquisition criteria, and determine the evaluation image area that meets the supplementary acquisition criteria as the supplementary acquisition image area. Acquire the acquisition time period data corresponding to the target area, generate acquisition path and acquisition time data based on the acquired image area and the acquisition time period data, generate multispectral acquisition configuration data based on the image quality evaluation data corresponding to the acquired image area, and combine the acquisition path, the acquisition time data and the multispectral acquisition configuration data into acquisition task data. Collect images of the supplementary image area according to the acquisition path, supplementary acquisition time data and multispectral acquisition configuration data in the supplementary acquisition task data, and generate supplementary images; Registration control points covering the area of ​​the supplementary image are extracted from the base remote sensing image and the supplementary image respectively. The registration control points corresponding to the spatial location are combined into registration control point pairs. Based on the registration control point pairs, the supplementary image is registered to the base remote sensing image to generate a registration supplementary image. Extract the multispectral image data corresponding to the area of ​​the supplemented image from the registered supplemented image, extract the basic multispectral image data corresponding to the spatial location of the supplemented multispectral image data from the basic remote sensing image, generate fusion weight data based on the image quality evaluation data corresponding to the area of ​​the supplemented image, and fuse the supplemented multispectral image data and the basic multispectral image data based on the fusion weight data to generate the target multispectral image set.

3. The method for generating land parcel status data as described in claim 1, characterized in that, Based on the joint recognition model, the target multispectral image set is processed to generate current land parcel status data containing land parcel boundary data and land parcel attribute data, including: Obtain a joint identification model including boundary recognition branch, attribute recognition branch, confidence fusion module and land parcel separation branch; The target multispectral image set is input into the boundary recognition branch, and boundary texture features and spectral distribution features are extracted from the target multispectral image set. Based on the boundary texture features and the spectral distribution features, candidate land parcel boundary masks and boundary confidence data are generated. Based on the candidate plot boundary mask, candidate plot image blocks are extracted from the target multispectral image set. The candidate plot image blocks are then input into the attribute recognition branch to generate candidate plot use data, candidate crop type data, and attribute confidence data. The boundary confidence data and the attribute confidence data are input into the confidence fusion module to generate fused confidence data. The candidate plot boundary mask, the candidate plot use data, the candidate crop type data, the fused confidence data, and the target multispectral image set are input into the plot separation branch to generate mutually separated plot instance masks, as well as plot use data, crop type data, and plot confidence data corresponding to the plot instance masks. Generate plot instance identifiers for each plot instance mask, and associate each plot instance identifier with the corresponding plot use data, crop type data, and plot confidence data; Acquire spatial reference data corresponding to the target multispectral image set; convert each plot instance mask into plot boundary data associated with the plot instance identifier based on the spatial reference data; generate plot area data based on the plot boundary data; combine the plot instance identifier, the plot use data, the crop type data, the plot area data, and the plot confidence data into plot attribute data; and combine the plot boundary data and the plot attribute data into current plot status data.

4. The method for generating land parcel status data as described in claim 1, characterized in that, Retrieve historical land parcel status data corresponding to the current land parcel status data from the spatiotemporal version data storage area, compare and verify the changes between the current land parcel status data and the historical land parcel status data, generate confirmed land parcel status data and land parcel change events, and write the confirmed land parcel status data into the spatiotemporal version data storage area, including: Extract land area data, land use data, and crop type data from the land attribute data in the current land status data, and obtain historical land status data corresponding to the target area from the spatiotemporal version data storage area; Historical land parcel boundary data and historical land parcel attribute data are extracted from the historical land parcel status data. Historical land parcel area data, historical land parcel use data, and historical crop type data are extracted from the historical land parcel attribute data. The land parcel boundary data is spatially overlaid with the historical land parcel boundary data. Based on the spatial overlay result, spatial correspondence data of land parcels is formed. Land parcel boundary data that does not form a spatial correspondence with the historical land parcel boundary data is identified as newly added land parcel data, and historical land parcel boundary data that does not form a spatial correspondence with the land parcel boundary data is identified as disappeared land parcel data. Based on the spatial corresponding data of the land parcels, the spatially corresponding land parcel boundary data is compared with the historical land parcel boundary data to generate boundary change data; the spatially corresponding land parcel area data is compared with the historical land parcel area data to generate area change data; the spatially corresponding land parcel use data and the crop type data are compared with the historical land parcel use data and the historical crop type data to generate attribute change data; the boundary change data, the area change data, the attribute change data, the newly added land parcel data, and the disappeared land parcel data are combined into change data to be verified. The current land parcel status data, the historical land parcel status data, and the change data to be verified are overlaid on the target multispectral image set for display, and verification operation data for the land parcel boundary data, the land parcel use data, or the crop type data are received; Based on the current land parcel status data and the verification operation data, the land parcel boundary data, land parcel use data, and crop type data are verified and updated. The land parcel area data is updated based on the verified and updated land parcel boundary data. The land parcel attribute data is updated based on the verified and updated land parcel use data, the verified and updated crop type data, and the updated land parcel area data. The verified and updated land parcel boundary data and the updated land parcel attribute data are combined to confirm the land parcel status data. The confirmed land parcel status data is compared with the historical land parcel status data. Based on the comparison results, the change area data and change category data are determined, and the change time data is obtained. The change area data, the change category data, and the change time data are combined into a land parcel change event. Based on the verification operation data, verification record data is generated. The confirmed land parcel status data, the land parcel change event, and the verification record data are associated and written into the spatiotemporal version data storage area.

5. The method for generating land parcel status data as described in claim 1, characterized in that, Based on the land parcel boundary data in the confirmed land parcel status data, multi-period regional observation images are obtained; based on the multi-period regional observation images, time-series observation data of the land parcels is generated; and based on the time-series observation data of the land parcels, status trend data is generated, including: Extract land parcel boundary data from the confirmed land parcel status data, obtain regional observation images of land parcels with observation time identifiers and spatial ranges covering the land parcel boundary data, associate each observation time identifier with the corresponding regional observation image, and generate multi-period regional observation images. Based on the land parcel boundary data, image data within the land parcel range defined by the land parcel boundary data corresponding to each observation time marker is extracted from the multi-period regional observation images to generate land parcel observation image segments. Red band data and near-infrared band data are extracted from the observation image segments of each plot. Based on the red band data and the near-infrared band data, normalization difference processing is performed to generate spectral state index data corresponding to each observation time marker. The spectral state index data corresponding to the boundary data of each plot are arranged in time sequence according to the observation time identifier to generate plot time-series observation data. Obtain smoothed window data, extract spectral state index data within adjacent observation time ranges from the plot time series observation data based on the smoothed window data, and smooth the plot time series observation data based on the spectral state index data within adjacent observation time ranges to generate smoothed time series observation data. Based on the spectral state index data arranged according to the observation time identifier in the smooth time series observation data, trend fitting is performed to generate trend sequence data corresponding to the boundary data of each plot, and the trend sequence data and the observation time identifier are combined into state trend data.

6. The method for generating land parcel status data as described in claim 1, characterized in that, Acquire historical status baseline data and multi-source observation data of the land parcel corresponding to the confirmed land parcel status data, and generate early warning events based on the status trend data, the historical status baseline data, and the multi-source observation data of the land parcel, including: Extract plot boundary data and plot attribute data from the confirmed plot status data, extract crop type data from the plot attribute data, extract observation time identifier from the status trend data, obtain growth stage matching data corresponding to the crop type data, and determine growth stage data based on the observation time identifier and the growth stage matching data. Based on the plot boundary data, the crop type data, and the growth stage data, historical spectral state index sequences are selected from the historical state baseline data storage area, and the historical spectral state index sequences are aligned according to the growth stage data to generate historical state baseline data. Meteorological observation data, soil observation data, thermal infrared observation data, and irrigation observation data are obtained based on the plot boundary data and the observation time identifier. The meteorological observation data, soil observation data, thermal infrared observation data, and irrigation observation data are time-aligned according to the observation time identifier and spatially correlated according to the plot boundary data to generate multi-source observation data of the plot. The state trend data is compared with the historical state baseline data according to the growth stage data, and the state trend segments that deviate from the historical state baseline data are extracted from the state trend data to generate state deviation data. Extract moisture state data, temperature state data, and soil nutrient state data corresponding to the state trend segment from the multi-source observation data of the plot, and associate the moisture state data, temperature state data, and soil nutrient state data with the state deviation data to generate risk evidence data; Based on the state deviation data, the state deviation type is determined; based on the risk evidence data, the observation state combination is determined; and a warning discrimination condition containing the correspondence between the state deviation type, the observation state combination, the risk category identifier, and the warning level identifier is obtained. Based on the state deviation type, the observation state combination, and the warning discrimination condition, risk category data and warning level data are generated; based on the land parcel boundary data, warning area data is generated; and warning time data is obtained. The confirmed land parcel status data, the risk category data, the warning level data, the risk evidence data, the warning area data, and the warning time data are combined into a warning event.

7. The method for generating land parcel status data as described in claim 1, characterized in that, Based on the confirmed land parcel status data, the land parcel change events, the status trend data, the land parcel multi-source observation data, and the early warning events, panoramic land parcel status data is generated, including: Extract land parcel boundary data and land parcel attribute data from the confirmed land parcel status data, convert the land parcel boundary data into land parcel spatial graphic data, write land parcel association identifiers for each land parcel spatial graphic data, and combine the land parcel spatial graphic data, the land parcel attribute data, and the land parcel association identifiers into a land parcel status record; Extract change area data, change category data, and change time data from the land parcel change event. Based on the spatial correspondence between the change area data and the land parcel spatial graphic data, associate the change area data that forms a spatial correspondence with the land parcel spatial graphic data with the corresponding land parcel association identifier. Generate land parcel association identifiers for change area data that does not form a spatial correspondence with the land parcel spatial graphic data. Combine the land parcel association identifier, the change area data, the change category data, and the change time data into a change event record. Based on the land parcel boundary data, trend sequence data and observation time identifiers are extracted from the state trend data. Observation state data corresponding to the land parcel boundary data and observation time identifiers are extracted from the multi-source observation data of the land parcel. Based on the spatial correspondence between the land parcel boundary data and the land parcel spatial graphic data, a land parcel association identifier corresponding to the trend sequence data is determined. The land parcel association identifier, the trend sequence data, the observation time identifier, and the observation state data are combined into a state observation record. Data on the warning area, risk category, warning level, and warning time are extracted from the warning event. Based on the spatial correspondence between the warning area data and the land parcel spatial graphic data, a land parcel association identifier corresponding to the warning event is determined. The land parcel association identifier, the risk category data, the warning level data, and the warning time data are combined into a warning event record. Based on the land parcel status records, land parcel distribution layer data is generated; based on the status observation records, status trend display data is generated; and the change event records and warning event records are arranged chronologically according to the change time data and the warning time data to generate land parcel event time series data. The land parcel status records, status observation records, change event records, early warning event records, land parcel distribution layer data, status trend display data, and land parcel event time series data are combined into a panoramic land parcel status data.

8. A land parcel status data generation device, characterized in that, The land parcel status data generation device includes: The image acquisition and fusion module is used to acquire basic remote sensing images of the target area, trigger acquisition based on the image quality evaluation of the basic remote sensing images to acquire acquisition images, and register and fuse the basic remote sensing images and the acquisition images into a target multispectral image set. The land parcel status identification module is used to process the target multispectral image set based on the joint identification model to generate current land parcel status data containing land parcel boundary data and land parcel attribute data; The land parcel change confirmation module is used to obtain historical land parcel status data corresponding to the current land parcel status data from the spatiotemporal version data storage area, compare and verify the current land parcel status data with the historical land parcel status data, generate confirmed land parcel status data and land parcel change event, and write the confirmed land parcel status data into the spatiotemporal version data storage area. The status trend analysis module is used to obtain multi-period regional observation images based on the land parcel boundary data in the confirmed land parcel status data, generate land parcel time-series observation data based on the multi-period regional observation images, and generate status trend data based on the land parcel time-series observation data. The status risk warning module is used to acquire historical status baseline data and multi-source observation data of the land parcel corresponding to the confirmed land parcel status data, and generate warning events based on the status trend data, the historical status baseline data and the multi-source observation data of the land parcel. The panoramic status construction module is used to generate panoramic status data of the land parcels based on the confirmed land parcel status data, the land parcel change events, the status trend data, the land parcel multi-source observation data, and the early warning events.

9. A computer device, characterized in that, The computer device includes a memory, a processor, and a land parcel status data generation program stored in the memory and executable on the processor. When executed by the processor, the land parcel status data generation program implements the steps of the land parcel status data generation method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The storage medium stores a land parcel status data generation program, which, when executed by a processor, implements the steps of the land parcel status data generation method as described in any one of claims 1-7.

Images