Rural collective asset intelligent identification method, device and equipment and storage medium

Abstract

The invention discloses an intelligent recognition method for rural collective assets, and the method comprises the steps: obtaining multi-source data related to the assets of a target region, and constructing an asset base map; inputting the assets with the spatial graph into the semantic segmentation model, outputting a first asset pattern spot, and optimizing the first asset pattern spot; for the assets without the spatial graph, obtaining spatial point locations corresponding to the assets by using a neural network model, and generating second asset graph spots; obtaining an asset space vector layer of the target area according to the first asset pattern spot and the second asset pattern spot; generating a non-overlapping asset ownership graph layer according to ownership conditions of all assets; overlapping the asset space vector layer and the non-overlapping asset ownership layer, screening out all collective assets, encoding the collective assets, and generating a collective asset overview map; according to the invention, the accuracy and working efficiency of checking and identifying rural collective assets can be effectively improved, and centralized display and systematic management of the collective assets are realized through the collective asset overview map.

Description

Technical Field

[0001] This invention relates to the field of asset identification technology, and in particular to a method, apparatus, equipment and storage medium for intelligent identification of rural collective assets. Background Technology

[0002] Rural collective assets refer to all types of assets owned by rural collective organizations, encompassing resource assets, operating assets, and non-operating assets. Currently, rural collective assets suffer from unclear inventory, incomplete registration of collective land and buildings, frequent off-book assets, and the failure to obtain property ownership certificates for some assets on schedule, resulting in a significant risk of asset loss. In actual management, villagers voluntarily report assets and related attributes and manage them in the form of ledgers, but the assets lack accurate geographical location and coordinate information, resulting in low visualization levels.

[0003] Traditional asset identification mainly relies on manual measurement, reviewing paper records, and verbal inquiries, which is not only inefficient and prone to errors, but also presents problems such as being unintuitive and easily causing disputes. There is a lack of an accurate and efficient intelligent identification method for rural collective assets. Summary of the Invention

[0004] This invention provides a method for intelligent identification of rural collective assets, which can effectively improve the accuracy and efficiency of rural collective asset inventory and identification, and realize centralized display and systematic management of collective assets through a collective asset overview map.

[0005] In a first aspect, embodiments of the present invention provide a method for intelligent identification of rural collective assets, comprising: Acquire multi-source data related to assets in the target area, and construct an asset base map based on the multi-source data; wherein, the multi-source data includes remote sensing data, ownership data, current status data, and asset information; Assets with spatial graphics are input into a pre-trained semantic segmentation model, and the first asset patch identified by the model is output. The first asset patch is then optimized based on standard land use patches and preset optimization rules. For assets that do not have spatial graphics, a neural network model is used to obtain the spatial points corresponding to the assets, and a second asset patch is generated based on the spatial points. Based on the optimized first asset patch and the second asset patch, the asset spatial vector layer of the target area is obtained; Generate a non-overlapping asset ownership layer based on the ownership status of all assets in the target area; The asset space vector layer is overlaid with the non-overlapping asset ownership layer. All collective assets are selected from the overlaid layer, and the collective assets are coded using a pre-set coding rule to generate a collective asset overview map.

[0006] Furthermore, the construction of the asset base map based on the multi-source data includes: The preprocessing of the multi-source data includes: converting all vector and raster data in the multi-source data into the 2000 National Geodetic Coordinate System; scanning and georegistering all paper materials in the multi-source data to generate spatial graphics in shapefile format; extracting key information from unstructured text using natural language processing technology; associating the extracted key information with the spatial graphics using a unique identifier; and storing the information in a structured database. A standardized asset base map is constructed based on the preprocessed multi-source data.

[0007] Furthermore, the optimization rules include boundary optimization rules, land parcel merging rules, and type arbitration rules. Therefore, the optimization of the first asset parcel based on standard land parcel types and preset optimization rules includes: Calculate the intersection-union ratio of each plot in the first asset patch with the standard land use patch. When the intersection-union ratio is greater than a preset intersection-union ratio threshold, perform weighted fusion of the plot at the boundary between the first asset patch and the standard land use patch to obtain the final boundary. By comparing the standard land parcels, asset parcels belonging to the same category are merged to obtain the complete asset parcels corresponding to each category; When the category of a plot in the first asset patch is inconsistent with the category in the standard land use patch, the spectral characteristics of the plot and the confidence scores of the two different categories are calculated, and the category with the larger confidence score is taken as the final category of the plot.

[0008] Furthermore, for assets without spatial graphics, the method of using a neural network model to obtain the spatial points corresponding to the assets and generating a second asset patch based on the spatial points includes: For each asset that does not have a spatial graphic, keywords for the location description information of the asset are obtained from the asset ledger. The location description information includes the boundaries and address description. The keywords were modified using a Hidden Markov Model; A Bi-LSTM neural network model based on an attention mechanism is constructed and trained. This model is used to compare the corrected keywords with existing information in the asset database and calculate the attention weight. The asset with the largest attention weight is selected for matching so that the asset without spatial graphics can obtain the corresponding spatial location. A buffer zone is determined centered on the spatial point, and one or more planar patches that intersect with the buffer zone are found from the optimized first asset patch. The asset patch is determined from the intersecting area patches, and the second asset patch is obtained by integrating all assets that do not have spatial graphics.

[0009] Furthermore, generating a non-overlapping asset ownership layer based on the ownership status of all assets in the target area includes: Obtain ownership certificates for all assets in the target area, spatially overlay the ownership certificates, and filter out assets whose ownership certificates overlap. One by one, all overlapping assets are traversed, and the effective time of the ownership certificates of the assets is queried using SQL tools. The ownership certificate with the latest effective time is retained, and the asset range corresponding to the remaining ownership certificates is erased to generate a non-overlapping ownership layer.

[0010] Furthermore, the step of filtering out all collective assets from the overlaid layers and encoding the collective assets using pre-defined encoding rules includes: Assets owned or used collectively are selected from the overlaid layers and designated as collective assets. A unique identifier is generated for each collective asset. The identifier consists of an administrative division code, a village code, an economic organization code, an asset code, and a serial number arranged in sequence.

[0011] Furthermore, the method also includes: Acquire the latest remote sensing images of the target area in real time; When the latest remote sensing image changes, the Siamese neural network is used to calculate the similarity between the changed patches and the original patches of all collective assets in the collective asset overview map. When the calculated similarity is greater than or equal to the preset similarity threshold, the asset corresponding to the changed patch is determined to be the original collective asset, and the overview map of the collective asset is updated. When the calculated similarity is less than the similarity threshold, the asset corresponding to the changed patch is determined to be a newly added collective asset, and the newly added collective asset is added to the collective asset overview map.

[0012] Secondly, embodiments of the present invention provide a smart identification device for rural collective assets, comprising: The asset base map construction module is used to acquire multi-source data related to assets in the target area and construct an asset base map based on the multi-source data; wherein, the multi-source data includes remote sensing data, ownership data, current status data and asset information; The first asset identification module is used to input assets with spatial graphics into a pre-trained semantic segmentation model, output the first asset patch identified by the model, and optimize the first asset patch based on standard land use patches and preset optimization rules. The second asset identification module is used to obtain the spatial points corresponding to assets that do not have spatial graphics using a neural network model, and generate a second asset patch based on the spatial points. The spatial layer generation module is used to obtain the asset spatial vector layer of the target area based on the optimized first asset patch and the second asset patch; The ownership layer generation module is used to generate a non-overlapping asset ownership layer based on the ownership status of all assets in the target area. The asset overview map generation module is used to overlay the asset spatial vector map layer with the non-overlapping asset ownership layer, filter out all collective assets from the overlaid layer, encode the collective assets using a pre-set encoding rule, and generate a collective asset overview map.

[0013] Thirdly, embodiments of the present invention provide an electronic device, comprising: Memory, used to store computer programs; A processor for executing the computer program; Wherein, when the processor executes the computer program, it implements the intelligent identification method for rural collective assets as described in any of the first aspects above.

[0014] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program that, when executed, implements the intelligent identification method for rural collective assets as described in any of the first aspects.

[0015] Compared with existing technologies, the intelligent identification method for rural collective assets provided in this invention has the following advantages: It acquires multi-source data related to assets in a target area and constructs an asset base map based on the multi-source data; wherein the multi-source data includes remote sensing data, ownership data, current status data, and asset information; it inputs assets with spatial graphics into a pre-trained semantic segmentation model, outputs the first asset patch identified by the model, and optimizes the first asset patch based on standard land use patches and preset optimization rules; for assets without spatial graphics, it uses a neural network model to obtain the spatial points corresponding to the assets, and optimizes the spatial points... The first asset patch is generated; based on the optimized first and second asset patches, an asset spatial vector layer for the target area is obtained; based on the ownership status of all assets in the target area, a non-overlapping asset ownership layer is generated; the asset spatial vector layer is superimposed on the non-overlapping asset ownership layer, all collective assets are selected from the superimposed layer, the collective assets are coded using a pre-set coding rule, and a collective asset overview map is generated; this invention can effectively improve the accuracy and efficiency of rural collective asset inventory and identification, and realize the centralized display and systematic management of collective assets through the collective asset overview map. Attached Figure Description

[0016] To more clearly illustrate the technical features of the embodiments of the present invention, the drawings used in the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating a method for intelligent identification of rural collective assets provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a smart identification device for rural collective assets provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein is for the purpose of describing embodiments of the invention only and is not intended to limit the invention.

[0021] In a first aspect, embodiments of the present invention provide a method for intelligent identification of rural collective assets, see [link to relevant documentation]. Figure 1 This is a flowchart illustrating an embodiment of an intelligent identification method for rural collective assets provided by the present invention.

[0022] like Figure 1 As shown, the method includes the following steps: S1: Acquire multi-source data related to assets in the target area, and construct an asset base map based on the multi-source data; wherein, the multi-source data includes remote sensing data, ownership data, current status data, and asset information; S2: Input the asset with spatial graphics into the pre-trained semantic segmentation model, output the first asset patch identified by the model, and optimize the first asset patch based on the standard land use patch and the preset optimization rules; S3: For assets that do not have spatial graphics, use a neural network model to obtain the spatial points corresponding to the assets, and generate a second asset patch based on the spatial points; S4: Based on the optimized first asset patch and the second asset patch, obtain the asset space vector layer of the target area; S5: Generate a non-overlapping asset ownership layer based on the ownership status of all assets in the target area; S6: Overlay the asset space vector layer with the non-overlapping asset ownership layer, filter out all collective assets from the overlaid layer, encode the collective assets using a pre-set encoding rule, and generate a collective asset overview map.

[0023] In practice, multi-source data related to assets in the target area is acquired, including remote sensing data, ownership data, current status data, and asset information. Remote sensing data is raster data, which can be obtained through high-resolution satellite imagery or drone aerial imagery of the target area. Different resolution base maps are used for different types of assets; for example, sub-meter resolution is used for resource assets (such as large areas of forest land and fishponds), while centimeter-level drone imagery is used for property assets (such as individual buildings). Ownership data is vector data, which can be obtained from planning and land resources departments, including data on the confirmation of collective land ownership, confirmation of rural land contracting and management rights, forest / tree ownership, homestead use rights, and unified real estate registration data. According to the data, the current status data is vector data, and the latest land change survey vector data (including land type, area, spatial location, and changes) can be obtained from the planning and land resources department. Asset information can be obtained from the agricultural and rural affairs department, including existing management ledgers and related archives (such as historical contracts, survey maps, transaction announcements, etc.). The ledgers contain core attribute information such as asset name, type, area, affiliated organization, quantity, operating status, and boundaries. Historical / current status contracts contain key information such as the location, scope, use, and term of the assets. Survey maps and planning drawings are important bases for the spatial positioning of assets. Although most assets do not have such materials, if they do, they should be included as the primary basis for identifying asset boundaries.

[0024] Based on the collected multi-source data, an asset base map is constructed. Asset data with spatial graphics in the asset base map are input into a pre-trained semantic segmentation model. The model automatically identifies asset areas and delineates boundaries (such as identifying the scope of fish ponds, factories, and woodlands). Finally, the first identified asset patch is output. The first identified asset patch can reflect the general distribution of assets, but there may be problems such as inaccurate boundaries, misjudgment of land types, and fragmented shapes, which need to be further optimized. The first asset patch and the standard land type patch are processed collaboratively according to the pre-set optimization rules to output the optimized first asset patch.

[0025] For assets that do not have spatial graphics, such as those with only textual information like ledgers, boundaries, and address descriptions, a neural network model is used to parse the textual information of the asset, accurately obtain the spatial location corresponding to the asset, and generate a complete second asset map based on the spatial location.

[0026] The optimized first asset patch and the generated second asset patch are integrated into the same layer to obtain the asset spatial vector layer of the target area. All assets in this layer have complete and accurate spatial boundaries and land use information.

[0027] Based on the previously collected ownership data, the ownership status of all assets in the target area is determined, and a non-overlapping asset ownership layer with no ownership conflicts is generated. The asset space vector layer is overlaid with the non-overlapping asset ownership layer, and only collective assets are retained. Each collective asset is coded according to preset rules, and all coded collective assets are integrated into a single visual map to generate a collective asset overview map.

[0028] In summary, this invention, in terms of data integration, constructs an asset base map by acquiring multi-source data covering remote sensing, ownership, current status, and asset information of the target area. This breaks down barriers between data sources and comprehensively and systematically integrates various types of information related to rural collective assets, providing a rich and reliable data foundation for subsequent accurate identification and management. Regarding asset patch processing, for assets with spatial graphics, a pre-trained semantic segmentation model is used to identify the first asset patch, which is then optimized based on standard land use patches and preset optimization rules, improving the accuracy and rationality of asset patch identification. For assets without spatial graphics, a neural network model is used to obtain spatial locations and generate a second asset patch, effectively solving the positioning problem of assets without spatial graphics, ensuring that all types of assets can be accurately presented spatially. Through the integration and optimization of the first asset... The asset spatial vector layer of the target area, obtained from the map features and the second asset map features, clearly and intuitively displays the spatial distribution of assets, providing strong spatial support for asset management and planning. A non-overlapping asset ownership layer is generated based on asset ownership, and assets are sorted according to ownership status, eliminating ownership overlap issues and clarifying the ownership of each asset, providing a clear basis for the management and protection of collective assets. Finally, the asset spatial vector layer is overlaid with the non-overlapping asset ownership layer to filter out collective assets and generate an overview map, achieving centralized display and systematic management of collective assets. This allows managers to intuitively understand detailed information such as the distribution, quantity, and ownership of collective assets, providing strong technical support for the planning, decision-making, and supervision of rural collective assets, and helping to improve the management level of rural collective assets.

[0029] In one optional implementation, constructing the asset base map based on the multi-source data includes: The preprocessing of the multi-source data includes: converting all vector and raster data in the multi-source data into the 2000 National Geodetic Coordinate System; scanning and georegistering all paper materials in the multi-source data to generate spatial graphics in shapefile format; extracting key information from unstructured text using natural language processing technology; associating the extracted key information with the spatial graphics using a unique identifier; and storing the information in a structured database. A standardized asset base map is constructed based on the preprocessed multi-source data.

[0030] Specifically, preprocessing is performed on multi-source data of different types and formats. Preprocessing includes coordinate unification, digitization of paper materials, and extraction of unstructured text.

[0031] Coordinate unification: All vector data (such as ownership boundaries) and raster data (such as remote sensing images) are converted into the 2000 national geodetic coordinate system to ensure that all data are on the same spatial reference and there will be no positioning deviation; Digitization of paper materials: Scan historical survey maps, land rights maps and other paper archives, perform georegistration on the scanned maps to match the locations on the maps with the actual geographic coordinates, and then vectorize the asset boundaries to generate spatial graphics in shapefile format; Unstructured text extraction: Using natural language processing technology, key information such as the location, boundaries, area, and ownership of assets is extracted from textual materials such as contracts and transaction announcements. Unique identifiers are assigned to the extracted key information and the corresponding spatial graphics, so that the textual information and boundary graphics are bound one by one. Finally, they are stored together in a structured database for easy subsequent querying and retrieval.

[0032] A standardized asset base map is formed based on the preprocessed multi-source data. This map is a basic map containing unified coordinates and complete graphic and textual information. All existing assets have corresponding boundaries and attribute information on this map, which is the core data carrier for all subsequent operations.

[0033] This embodiment effectively integrates various heterogeneous data, eliminates differences in data format and coordinates, and improves data consistency and usability. Based on this, a standardized asset base map is constructed, providing a unified, comprehensive, and reliable data foundation for subsequent precise asset management, analysis, and decision-making, which helps to improve asset management efficiency and quality.

[0034] In one optional implementation, the optimization rules include boundary optimization rules, land parcel merging rules, and type arbitration rules. Then, the optimization of the first asset parcel based on standard land use parcels and preset optimization rules includes: Calculate the intersection-union ratio of each plot in the first asset patch with the standard land use patch. When the intersection-union ratio is greater than a preset intersection-union ratio threshold, perform weighted fusion of the plot at the boundary between the first asset patch and the standard land use patch to obtain the final boundary. By comparing the standard land parcels, asset parcels belonging to the same category are merged to obtain the complete asset parcels corresponding to each category; When the category of a plot in the first asset patch is inconsistent with the category in the standard land use patch, the spectral characteristics of the plot and the confidence scores of the two different categories are calculated, and the category with the larger confidence score is taken as the final category of the plot.

[0035] Specifically, the initial asset patches may have problems such as inaccurate boundaries, fragmented shapes, and incorrect land use classification. This embodiment uses collaborative rules to process the initial asset patches and standard land use patches collaboratively. Among them, the boundary optimization rule is used to make the asset boundary accurate and in line with the current situation. For each piece of land in the initial patch, the intersection-over-union ratio (IoU) with the corresponding area of ​​the standard patch is calculated, which is the proportion of the overlapping part of the two pieces of land. An IoU threshold is set in advance. When the IoU is greater than the IoU threshold, it indicates that the two are mostly overlapping and the direction is correct. The boundary of the plot is weighted and fused at the boundary between the initial asset patch and the standard land use patch to obtain the final boundary. The final boundary after fusion ensures both accuracy and conforms to the current reality.

[0036] Land parcel merging rules are used to make asset boundaries more complete. Authoritative DLTB land parcels may break down a complete asset into very small parts. Merging these asset parcels belonging to the same category will result in complete asset parcels corresponding to each category, avoiding the fragmentation of assets and meeting actual management needs.

[0037] The type arbitration rule is used to handle disputes over land categories. When the category of a land parcel in the initial asset patch is inconsistent with the category in the standard land category patch, the spectral features of the land parcel (such as color and texture in the image) and the confidence scores of the two land categories are calculated. The category with the higher confidence score is given priority as the final category of the land parcel to avoid subjective judgment, ensure the accuracy of land category determination, and add the disputed sample (image features + correct classification) to the sample library so that the model can be more accurate when it encounters similar situations in the future.

[0038] This embodiment effectively solves the boundary misalignment problem caused by differences in data sources through boundary optimization, thereby improving boundary accuracy. By merging asset patches of the same category through land parcel fusion, complete asset patches are generated, reducing data fragmentation and enhancing the integrity and coherence of the data. By scientifically determining the final category of land parcels with inconsistent categories through type arbitration, the accuracy and consistency of asset category classification are ensured. The synergistic effect of these three aspects significantly improves the collaborative processing quality of initial asset patches and standard land parcels, providing a reliable data foundation for subsequent asset management and analysis.

[0039] In one optional implementation, the step of using a neural network model to obtain the spatial points corresponding to the assets that do not have spatial graphics, and generating a second asset patch based on the spatial points, includes: For each asset that does not have a spatial graphic, keywords for the location description information of the asset are obtained from the asset ledger. The location description information includes the boundaries and address description. The keywords were modified using a Hidden Markov Model; A Bi-LSTM neural network model based on an attention mechanism is constructed and trained. This model is used to compare the corrected keywords with existing information in the asset database and calculate the attention weight. The asset with the largest attention weight is selected for matching so that the asset without spatial graphics can obtain the corresponding spatial location. A buffer zone is determined centered on the spatial point, and one or more planar patches that intersect with the buffer zone are found from the optimized first asset patch. The asset patch is determined from the intersecting area patches, and the second asset patch is obtained by integrating all assets that do not have spatial graphics.

[0040] Specifically, for the challenge of locating assets that only have textual descriptions (such as the boundaries and address of the ledger) and no spatial graphics, standardized processing of textual information and AI-powered intelligent matching can enable these assets to be accurately mapped onto the asset base map, thus filling the data gap.

[0041] First, the location description information of the assets is obtained from the asset ledger, including the boundaries and address description. Keywords are extracted from this information. For example, the keywords for the address description "fishpond in Group 3 of Shidong Village" are "Shidong Village", "Group 3" and "fishpond". The keywords for the boundaries "east to the small river and south to the village road" are "small river" and "village road". By extracting keywords and removing redundant text, the core information that can define the location is focused on, which facilitates subsequent AI processing.

[0042] Hidden Markov Models (HMMs) are used to validate and correct the extracted keywords, such as correcting typos, standardizing abbreviations, and completing ambiguous expressions, to avoid errors in subsequent matching and positioning due to textual errors or non-standard expressions, thus ensuring that the information input into the AI ​​is accurate.

[0043] First, construct and train a Bi-LSTM neural network model based on an attention mechanism to familiarize the model with the existing information in the asset database (such as POIs, waterways, administrative divisions, ownership boundaries, etc. of registered assets). Input the corrected keywords into the trained model and compare them one by one with the existing information in the database. At the same time, calculate the attention weight. The higher the relevance between the keyword and the existing information, the greater the weight. Select the existing information with the largest attention weight for matching, which is equivalent to finding a reference benchmark for the unmapped assets and obtaining the corresponding spatial location.

[0044] A buffer zone is determined centered on the spatial location of the asset. In the first asset patch optimized in the above embodiment, one or more planar patches intersect with this buffer zone. The intersecting patches are optimized based on the actual situation. Finally, the asset patch of the asset is selected from multiple intersecting patches according to the standard that the patch area is closest to the registered area of ​​the asset. The asset patches of all assets without spatial graphics are integrated to obtain the second asset patch.

[0045] This embodiment utilizes a Hidden Markov Model and an attention-based Bi-LSTM neural network model to accurately locate asset positions. It determines a buffer zone centered on spatial points and finds intersecting planar patches, then identifies asset patches from these patches and integrates them to generate a second asset patch. This effectively solves the problem of how to spatially represent assets without spatial graphics, improves the accuracy and reliability of asset spatial positioning, and makes the spatial distribution information of assets more complete and accurate.

[0046] In one optional implementation, generating a non-overlapping asset ownership layer based on the ownership status of all assets in the target area includes: Obtain ownership certificates for all assets in the target area, spatially overlay the ownership certificates, and filter out assets whose ownership certificates overlap. One by one, all overlapping assets are traversed, and the effective time of the ownership certificates of the assets is queried using SQL tools. The ownership certificate with the latest effective time is retained, and the asset range corresponding to the remaining ownership certificates is erased to generate a non-overlapping ownership layer.

[0047] Specifically, the system obtains ownership certificates for all assets within the target area (such as collective land ownership certificates, state-owned construction land use right certificates, and contracted management right confirmation certificates, all of which are vector data with spatial boundaries). The ownership certificates are then spatially overlaid, and assets with overlapping boundaries are automatically filtered out (for example, the land area of ​​certificate A and the land area of ​​certificate B overlap and completely cover each other).

[0048] For each overlapping asset, an SQL tool is used to retrieve the effective dates of all ownership certificates related to it (e.g., Certificate A became effective in 2010, Certificate B became effective in 2020). Based on the principle of the latest effective date being the only valid one, only the boundary of the ownership certificate with the latest effective date is retained. At the same time, a GIS tool is used to erase the overlapping areas corresponding to other old ownership certificates, which is equivalent to canceling the spatial validity of the old certificates. The new ownership certificate is the latest confirmation of the asset ownership and has priority, avoiding ownership disputes caused by multiple certificates in one location. Finally, a non-overlapping asset ownership layer is generated. Each asset in this layer corresponds to only one currently valid ownership certificate, with no overlapping boundaries and no ownership conflicts, which is the authoritative basis for subsequent determination of asset ownership.

[0049] This embodiment effectively resolves spatial conflicts caused by overlapping ownership certificates by clearing up spatial overlaps, ensuring the accuracy and uniqueness of the ownership layer, and providing reliable spatial data support for the clear definition, reasonable planning, and efficient management of asset ownership.

[0050] In one optional implementation, the step of filtering all collective assets from the overlaid layers and encoding the collective assets using pre-defined encoding rules includes: Assets owned or used collectively are selected from the overlaid layers and designated as collective assets. A unique identifier is generated for each collective asset. The identifier consists of an administrative division code, a village code, an economic organization code, an asset code, and a serial number arranged in sequence.

[0051] Specifically, collective assets are filtered from the overlaid layers. The filtering rules are that the ownership of the asset belongs to the collective (such as land and forest land under the name of the village collective) or the right to use the asset belongs to the collective (such as factory buildings and office buildings rented or managed by the village collective). Collective assets are extracted from all assets, and a unique identification code is generated for each collective asset. The identification code consists of the administrative division code (6 digits), the village code (2 digits), the economic organization code (2 digits), the asset code (4 digits), and the serial number (4 digits) arranged in sequence. This code embeds the core ownership and type attributes of the asset, which facilitates rapid retrieval, classification, and traceability, and provides a data structure foundation for dynamic monitoring and change tracking of assets.

[0052] This embodiment ensures the comprehensiveness and accuracy of collective asset identification by accurately filtering assets whose ownership or usage rights belong to the collective from the overlay layer. At the same time, it adopts a unique identifier code that is an orderly combination of multiple levels of codes such as administrative division, village area, economic organization, asset and serial number, to give each collective asset a clear and unique identity, which greatly improves the standardization and refinement of collective asset management, facilitates subsequent rapid retrieval, statistics and dynamic tracking, and lays a solid foundation for the informatization and intelligent management of collective assets.

[0053] In one optional implementation, the method further includes: Acquire the latest remote sensing images of the target area in real time; When the latest remote sensing image changes, the Siamese neural network is used to calculate the similarity between the changed patches and the original patches of all collective assets in the collective asset overview map. When the calculated similarity is greater than or equal to the preset similarity threshold, the asset corresponding to the changed patch is determined to be the original collective asset, and the overview map of the collective asset is updated. When the calculated similarity is less than the similarity threshold, the asset corresponding to the changed patch is determined to be a newly added collective asset, and the newly added collective asset is added to the collective asset overview map.

[0054] Specifically, the system acquires the latest remote sensing images of the target area in real time, automatically compares them with historical images, and identifies areas that have changed, i.e., changed patches. For example, a plot of land may have added buildings or the boundaries of an existing fishpond may have expanded. The system also periodically checks the dynamic ownership database to see if any new ownership certificates have been issued, old certificates have been cancelled, or usage rights have been adjusted. The system uses a Siamese neural network to perform similarity calculations between the identified changed patches and the original patches of all registered collective assets in the collective asset overview map in the feature space (such as shape, color, location, and surrounding environment) to determine whether they are the same asset.

[0055] When the calculated similarity is greater than or equal to the preset similarity threshold, it indicates that the changed patch is highly similar to a certain historical collective asset, which means that "the original collective asset has changed" (boundary adjustment, demolition and reconstruction, etc.). The system will automatically prompt the staff to confirm. After confirmation, the patch of the asset in the collective asset overview map will be updated directly (such as modifying the boundary or updating the status), without having to go through the complete ownership identification process again.

[0056] When the calculated similarity is less than the similarity threshold, it indicates that the changed patch is not similar to any of the historical collective assets, which means it is a "newly added collective asset" (such as a newly built factory or newly confirmed forest land of the village collective). The patch and information of the newly added collective asset are added to the collective asset overview map and the database is updated synchronously.

[0057] This embodiment can accurately and efficiently identify whether the changed map features are boundary changes or reconstructions of existing collective assets or newly added collective assets, realizing dynamic monitoring and intelligent updating of collective asset map features, greatly improving the timeliness and accuracy of collective asset management, effectively avoiding the lag and errors of manual monitoring, and reducing management costs.

[0058] Secondly, embodiments of the present invention provide a smart identification device for rural collective assets, see [link to relevant documentation]. Figure 2 This is a schematic diagram of one embodiment of a smart identification device for rural collective assets provided by the present invention.

[0059] like Figure 2 As shown, the device includes: The asset base map construction module 21 is used to acquire multi-source data related to assets in the target area and construct an asset base map based on the multi-source data; wherein, the multi-source data includes remote sensing data, ownership data, current status data and asset information; The first asset identification module 22 is used to input assets with spatial graphics into a pre-trained semantic segmentation model, output the first asset patch identified by the model, and optimize the first asset patch based on standard land use patches and preset optimization rules. The second asset identification module 23 is used to obtain the spatial points corresponding to the assets without spatial graphics using a neural network model, and generate a second asset patch based on the spatial points. The spatial layer generation module 24 is used to obtain the asset spatial vector layer of the target area based on the optimized first asset patch and the second asset patch; The ownership layer generation module 25 is used to generate a non-overlapping asset ownership layer based on the ownership status of all assets in the target area. The asset overview map generation module 26 is used to overlay the asset space vector map layer with the non-overlapping asset ownership layer, filter out all collective assets from the overlaid layer, encode the collective assets using a pre-set encoding rule, and generate a collective asset overview map.

[0060] In one optional implementation, constructing the asset base map based on the multi-source data includes: The preprocessing of the multi-source data includes: converting all vector and raster data in the multi-source data into the 2000 National Geodetic Coordinate System; scanning and georegistering all paper materials in the multi-source data to generate spatial graphics in shapefile format; extracting key information from unstructured text using natural language processing technology; associating the extracted key information with the spatial graphics using a unique identifier; and storing the information in a structured database. A standardized asset base map is constructed based on the preprocessed multi-source data.

[0061] In one optional implementation, the optimization rules include boundary optimization rules, land parcel merging rules, and type arbitration rules. Then, the optimization of the first asset parcel based on standard land use parcels and preset optimization rules includes: Calculate the intersection-union ratio of each plot in the first asset patch with the standard land use patch. When the intersection-union ratio is greater than a preset intersection-union ratio threshold, perform weighted fusion of the plot at the boundary between the first asset patch and the standard land use patch to obtain the final boundary. By comparing the standard land parcels, asset parcels belonging to the same category are merged to obtain the complete asset parcels corresponding to each category; When the category of a plot in the first asset patch is inconsistent with the category in the standard land use patch, the spectral characteristics of the plot and the confidence scores of the two different categories are calculated, and the category with the larger confidence score is taken as the final category of the plot.

[0062] In one optional implementation, the step of using a neural network model to obtain the spatial points corresponding to the assets that do not have spatial graphics, and generating a second asset patch based on the spatial points, includes: For each asset that does not have a spatial graphic, keywords for the location description information of the asset are obtained from the asset ledger. The location description information includes the boundaries and address description. The keywords were modified using a Hidden Markov Model; A Bi-LSTM neural network model based on an attention mechanism is constructed and trained. This model is used to compare the corrected keywords with existing information in the asset database and calculate the attention weight. The asset with the largest attention weight is selected for matching so that the asset without spatial graphics can obtain the corresponding spatial location. A buffer zone is determined centered on the spatial point, and one or more planar patches that intersect with the buffer zone are found from the optimized first asset patch. The asset patch is determined from the intersecting area patches, and the second asset patch is obtained by integrating all assets that do not have spatial graphics.

[0063] In one optional implementation, generating a non-overlapping asset ownership layer based on the ownership status of all assets in the target area includes: Obtain ownership certificates for all assets in the target area, spatially overlay the ownership certificates, and filter out assets whose ownership certificates overlap. One by one, all overlapping assets are traversed, and the effective time of the ownership certificates of the assets is queried using SQL tools. The ownership certificate with the latest effective time is retained, and the asset range corresponding to the remaining ownership certificates is erased to generate a non-overlapping ownership layer.

[0064] In one optional implementation, the step of filtering all collective assets from the overlaid layers and encoding the collective assets using pre-defined encoding rules includes: Assets owned or used collectively are selected from the overlaid layers and designated as collective assets. A unique identifier is generated for each collective asset. The identifier consists of an administrative division code, a village code, an economic organization code, an asset code, and a serial number arranged in sequence.

[0065] In an optional embodiment, the device is further configured to: Acquire the latest remote sensing images of the target area in real time; When the latest remote sensing image changes, the Siamese neural network is used to calculate the similarity between the changed patches and the original patches of all collective assets in the collective asset overview map. When the calculated similarity is greater than or equal to the preset similarity threshold, the asset corresponding to the changed patch is determined to be the original collective asset, and the overview map of the collective asset is updated. When the calculated similarity is less than the similarity threshold, the asset corresponding to the changed patch is determined to be a newly added collective asset, and the newly added collective asset is added to the collective asset overview map.

[0066] Thirdly, embodiments of the present invention provide an electronic device, see [link to previous document]. Figure 3 The diagram shown is a structural schematic of an electronic device provided in an embodiment of the present invention.

[0067] like Figure 3 As shown, the device includes: Memory 31 is used to store computer programs; Processor 32 is used to execute the computer program; When the processor 32 executes the computer program, it implements the intelligent identification method for rural collective assets as described in any of the above embodiments.

[0068] For example, the computer program may be divided into one or more modules / units, which are stored in the memory 31 and executed by the processor 32 to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0069] The processor 32 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0070] The memory 31 can be used to store the computer programs and / or modules. The processor 32 implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory 31 and calling the data stored in the memory 31. The memory 31 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory 31 may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital card (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0071] It should be noted that the aforementioned electronic devices include, but are not limited to, processors and memory, as will be understood by those skilled in the art. Figure 3 The structural diagram is merely an example of the electronic device described above and does not constitute a limitation on the electronic device. It may include more components than shown in the diagram, or combine certain components, or use different components.

[0072] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a computer program that, when executed, implements the intelligent identification method for rural collective assets described in any of the above embodiments.

[0073] It should be understood that the present invention can implement all or part of the processes in the above-described intelligent identification method for rural collective assets, or it can be accomplished by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the above-described intelligent identification method for rural collective assets. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0074] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. It should be noted that, for those skilled in the art, several equivalent obvious modifications and / or equivalent substitutions can be made without departing from the technical principles of the present invention, and these obvious modifications and / or equivalent substitutions should also be considered within the scope of protection of the present invention.

Claims

1. A method for intelligent identification of rural collective assets, characterized in that, include: Acquire multi-source data related to assets in the target area, and construct an asset base map based on the multi-source data; wherein, the multi-source data includes remote sensing data, ownership data, current status data, and asset information; Assets with spatial graphics are input into a pre-trained semantic segmentation model, and the first asset patch identified by the model is output. The first asset patch is then optimized based on standard land use patches and preset optimization rules. For assets that do not have spatial graphics, a neural network model is used to obtain the spatial points corresponding to the assets, and a second asset patch is generated based on the spatial points. Based on the optimized first asset patch and the second asset patch, the asset spatial vector layer of the target area is obtained; Generate a non-overlapping asset ownership layer based on the ownership status of all assets in the target area; The asset space vector layer is overlaid with the non-overlapping asset ownership layer. All collective assets are selected from the overlaid layer, and the collective assets are coded using a pre-set coding rule to generate a collective asset overview map.

2. The intelligent identification method for rural collective assets as described in claim 1, characterized in that, The construction of the asset base map based on the multi-source data includes: The preprocessing of the multi-source data includes: converting all vector and raster data in the multi-source data into the 2000 National Geodetic Coordinate System; scanning and georegistering all paper materials in the multi-source data to generate spatial graphics in shapefile format; extracting key information from unstructured text using natural language processing technology; associating the extracted key information with the spatial graphics using a unique identifier; and storing the information in a structured database. A standardized asset base map is constructed based on the preprocessed multi-source data.

3. The intelligent identification method for rural collective assets as described in claim 1, characterized in that, The optimization rules include boundary optimization rules, land parcel merging rules, and type arbitration rules. Therefore, the optimization of the first asset parcel based on standard land parcel types and preset optimization rules includes: Calculate the intersection-union ratio of each plot in the first asset patch with the standard land use patch. When the intersection-union ratio is greater than a preset intersection-union ratio threshold, perform weighted fusion of the plot at the boundary between the first asset patch and the standard land use patch to obtain the final boundary. By comparing the standard land parcels, asset parcels belonging to the same category are merged to obtain the complete asset parcels corresponding to each category; When the category of a plot in the first asset patch is inconsistent with the category in the standard land use patch, the spectral characteristics of the plot and the confidence scores of the two different categories are calculated, and the category with the larger confidence score is taken as the final category of the plot.

4. The intelligent identification method for rural collective assets as described in claim 1, characterized in that, For assets that do not have spatial graphics, a neural network model is used to obtain the spatial points corresponding to the assets, and a second asset patch is generated based on the spatial points, including: For each asset that does not have a spatial graphic, keywords for the location description information of the asset are obtained from the asset ledger. The location description information includes the boundaries and address description. The keywords were modified using a Hidden Markov Model; A Bi-LSTM neural network model based on an attention mechanism is constructed and trained. This model is used to compare the corrected keywords with existing information in the asset database and calculate the attention weight. The asset with the largest attention weight is selected for matching so that the asset without spatial graphics can obtain the corresponding spatial location. A buffer zone is determined centered on the spatial point, and one or more planar patches that intersect with the buffer zone are found from the optimized first asset patch. The asset patch is determined from the intersecting area patches, and the second asset patch is obtained by integrating all assets that do not have spatial graphics.

5. The intelligent identification method for rural collective assets as described in claim 1, characterized in that, The step of generating a non-overlapping asset ownership layer based on the ownership status of all assets in the target area includes: Obtain ownership certificates for all assets in the target area, spatially overlay the ownership certificates, and filter out assets whose ownership certificates overlap. One by one, all overlapping assets are traversed, and the effective time of the ownership certificates of the assets is queried using SQL tools. The ownership certificate with the latest effective time is retained, and the asset range corresponding to the remaining ownership certificates is erased to generate a non-overlapping ownership layer.

6. The intelligent identification method for rural collective assets as described in claim 1, characterized in that, The step of filtering all collective assets from the overlaid layers and encoding the collective assets using pre-defined encoding rules includes: Assets owned or used collectively are selected from the overlaid layers and designated as collective assets. A unique identifier is generated for each collective asset. The identifier consists of an administrative division code, a village code, an economic organization code, an asset code, and a serial number arranged in sequence.

7. The intelligent identification method for rural collective assets as described in claim 1, characterized in that, The method further includes: Acquire the latest remote sensing images of the target area in real time; When the latest remote sensing image changes, the Siamese neural network is used to calculate the similarity between the changed patches and the original patches of all collective assets in the collective asset overview map. When the calculated similarity is greater than or equal to the preset similarity threshold, the asset corresponding to the changed patch is determined to be the original collective asset, and the overview map of the collective asset is updated. When the calculated similarity is less than the similarity threshold, the asset corresponding to the changed patch is determined to be a newly added collective asset, and the newly added collective asset is added to the collective asset overview map.

8. A smart identification device for rural collective assets, characterized in that, include: The asset base map construction module is used to acquire multi-source data related to assets in the target area and construct an asset base map based on the multi-source data; wherein, the multi-source data includes remote sensing data, ownership data, current status data and asset information; The first asset identification module is used to input assets with spatial graphics into a pre-trained semantic segmentation model, output the first asset patch identified by the model, and optimize the first asset patch based on standard land use patches and preset optimization rules. The second asset identification module is used to obtain the spatial points corresponding to assets that do not have spatial graphics using a neural network model, and generate a second asset patch based on the spatial points. The spatial layer generation module is used to obtain the asset spatial vector layer of the target area based on the optimized first asset patch and the second asset patch; The ownership layer generation module is used to generate a non-overlapping asset ownership layer based on the ownership status of all assets in the target area. The asset overview map generation module is used to overlay the asset spatial vector map layer with the non-overlapping asset ownership layer, filter out all collective assets from the overlaid layer, encode the collective assets using a pre-set encoding rule, and generate a collective asset overview map.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program; The processor executes the computer program to implement the intelligent identification method for rural collective assets as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed, implements the intelligent identification method for rural collective assets as described in any one of claims 1 to 7.

Images