Map sample library customization construction method and device, equipment and storage medium

Abstract

The present application relates to the technical field of data engineering, and particularly relates to a method and device for customized construction of a sample library of map patches, equipment and a storage medium, the present application first analyzes user demand, matches the semantic category corresponding to the demand with an authoritative geographical classification system, and screens a standard geographical category in combination with a geographical distribution knowledge base; then calculates the distribution density and area proportion of vector plot data in a target region on the standard category, evaluates a sampling priority, and finally extracts and combines sample units from remote sensing images and vector data according to the evaluation result to form a sample library of map patches. The present application ensures that the semantic category of the sample is clear and consistent with the actual regional geography through semantic analysis and geographical knowledge verification, and improves the accuracy and practicality of the sample. Priority evaluation and hierarchical sampling are implemented based on spatial distribution characteristics, so that sample collection is concentrated in key regions with rich information, and the present application realizes the automated construction of a high-quality sample library from user demand.

Description

Technical Field

[0001] This invention relates to the field of data engineering technology, and in particular to a method, apparatus, equipment, and storage medium for the customized construction of a sample library of image patches. Background Technology

[0002] In today's information age, geospatial information is of paramount importance for many fields, including natural resource management, ecological environmental protection, agricultural production monitoring, and emergency disaster reduction. With the rapid development of Earth observation technology, massive amounts of high-resolution remote sensing imagery have been continuously accumulated, providing a solid data foundation for macro-dynamic monitoring.

[0003] In recent years, artificial intelligence technologies such as deep learning have demonstrated enormous potential in this field. However, the performance of these models fundamentally depends on the quality and scale of the input training sample database. Currently, the construction of such databases heavily relies on professional personnel to perform manual visual interpretation and pixel-level annotation of images. my country has established a nationwide, up-to-date, and highly accurate vector geospatial data database, which is an extremely valuable asset of current geospatial data. However, on the other hand, the classification system of this data is insufficient to directly meet the personalized and fine-grained semantic classification needs of different vertical industries such as agricultural yield estimation, biodiversity conservation, and disaster assessment.

[0004] Therefore, how to systematically and efficiently transform standardized land survey vector data into customized high-quality sample libraries adapted to different application scenarios is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] The main objective of this invention is to provide a method, apparatus, device, and storage medium for the customized construction of a map feature sample library, aiming to solve the technical problem in the prior art of how to systematically and efficiently transform standardized land survey vector data into a customized high-quality sample library adapted to different application scenarios.

[0006] To achieve the above objectives, the present invention provides a method for customizing the construction of a patch sample library, the method comprising the following steps: Obtain user-defined requirements for a sample library of image patches, including target regions and target semantic category information; Based on a preset geographic classification system, the target semantic category information is parsed to determine the standard geographic category; Based on the distribution characteristics of vector parcel data within the target area on the standard geographic category, a sampling priority assessment is performed on the target area to obtain the priority assessment results; Based on the priority evaluation results, data are extracted and combined from the remote sensing imagery and the vector parcel data to obtain sample units; By summarizing the sample units, a target patch sample library is obtained.

[0007] Optionally, the step of parsing the target semantic category information according to a preset geographic classification system to determine the standard geographic category includes: Each category contained in the target semantic category information is semantically similar to each standard category in the geographic classification system to obtain a matching score between each pair of categories. The standard categories with matching scores higher than a preset score threshold are filtered and summarized to obtain a set of candidate standard categories; Based on the geographical distribution knowledge base corresponding to the target area, the candidate standard category set is filtered to determine the standard geographical category.

[0008] Optionally, the step of filtering the candidate standard category set based on the geographic distribution knowledge base corresponding to the target region to determine the standard geographic category includes: Based on the geographical distribution knowledge base corresponding to the target area, geographical constraint rules are obtained; Each category in the candidate standard category set is matched and verified with the geographical constraint rule to obtain the compliance judgment result of each category; Based on the compliance determination result, categories that do not conform to the geographical constraint rules are removed from the candidate standard category set to obtain the standard geographical category.

[0009] Optionally, the step of evaluating the sampling priority of the target area based on the distribution characteristics of vector parcel data within the target area on the standard geographic category, and obtaining the priority evaluation result, includes: Based on the vector land parcel data, the number and total area of ​​land parcels of each standard geographical category within the target area are statistically obtained; Based on the number of plots and total area of ​​each standard geographical category, calculate the distribution density and area proportion of each standard geographical category in the entire target area; Based on the distribution density and area proportion, the sampling priority of the target area is evaluated to obtain the priority evaluation result.

[0010] Optionally, the step of prioritizing the sampling of the target area based on the distribution density and area proportion to obtain the priority assessment result includes: The target region is divided into multiple regular grids; Based on the distribution density and area ratio, the information richness of each regular grid is calculated; The rule grids are sorted in descending order of information richness, and a number of rule grids with the highest information richness are selected according to a preset sampling quantity to obtain the priority evaluation result.

[0011] Optionally, before sorting the rule grids in descending order of information richness and selecting a corresponding number of top-ranked rule grids based on a preset sampling quantity to obtain the priority evaluation result, the method further includes: The target area, the preset national vegetation zoning data, and the national administrative boundary data are overlaid to obtain the zoning data overlap results; Based on the overlapping results of the partition data, the target area is divided into multiple geographical partitions; Based on the total number of samples set by the user and the area ratio of each geographic partition, a sampling quota is allocated to each geographic partition. The sampling quota is used when evaluating the sampling priority within each geographic partition, and the preset sampling quantity is the sampling quota allocated to the corresponding geographic partition.

[0012] Optionally, the step of extracting and combining data from the remote sensing image and the vector parcel data to obtain sample units based on the priority evaluation results includes: Based on the priority evaluation results, the grid to be sampled is determined from the target region; Based on the geographical range of the grid to be sampled, the corresponding image patches are cropped from the remote sensing image, and the corresponding vector map patches are extracted from the vector patch data. Based on the composition and spatial distribution statistics of the various land cover categories in the vector image patch, a text description associated with the image patch and the vector image patch is generated; The image patch, the vector image patch, and the text description are associated and combined to generate the sample unit.

[0013] Furthermore, to achieve the above objectives, the present invention also proposes a customized construction device for a patch sample library, the customized construction device for the patch sample library comprising: The data acquisition module is used to acquire user-defined requirements for a sample library of image patches, including target regions and target semantic category information. The data processing module is used to parse the target semantic category information according to a preset geographic classification system and determine the standard geographic category; The data evaluation module is used to perform sampling priority evaluation on the target area based on the distribution characteristics of vector parcel data in the target area on the standard geographic category, and obtain the priority evaluation result. The data processing module is also used to extract and combine data from the remote sensing image and the vector land parcel data according to the priority evaluation results to obtain sample units; The data processing module is also used to summarize the sample units to obtain a target patch sample library.

[0014] Furthermore, to achieve the above objectives, the present invention also proposes a custom construction device for a patch sample library, the patch sample library custom construction device comprising: a memory, a processor, and a patch sample library custom construction program stored in the memory and executable on the processor, the patch sample library custom construction program being configured to implement the steps of the patch sample library custom construction method described above.

[0015] In addition, to achieve the above objectives, the present invention also proposes a storage medium storing a custom construction program for a patch sample library, wherein when the custom construction program for a patch sample library is executed by a processor, it implements the steps of the custom construction method for a patch sample library as described above.

[0016] The one or more technical solutions proposed in this application have at least the following technical effects: This application ensures that the semantics of the sample categories are clear and consistent with the actual regional geography through semantic parsing and geographical knowledge verification, thereby improving the accuracy and practicality of the samples. Furthermore, by prioritizing based on spatial distribution characteristics and implementing stratified sampling, sample collection is concentrated in information-rich key areas, significantly improving sampling efficiency and the representativeness of the sample set. Finally, this method integrates the screening, sampling, and extraction processes, effectively reducing the degree of manual intervention, ensuring the standardization and consistency of the sample library construction, and providing a reliable data foundation for the training of subsequent geographical analysis models. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the first embodiment of the method for customizing the patch sample library of the present invention. Figure 2 This is a flowchart illustrating the second embodiment of the method for customizing the patch sample library of the present invention. Figure 3This is a structural block diagram of the first embodiment of the custom construction device for the image sample library of the present invention; Figure 4 This is a schematic diagram of the structure of the custom construction device for the image sample library of the hardware operating environment involved in the embodiments of the present invention.

[0020] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0022] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0023] The main solution of this application embodiment is to obtain user-defined requirements information for a sample library of geographic features, including target area and target semantic category information; to parse the target semantic category information according to a preset geographic classification system to determine a standard geographic category; to evaluate the sampling priority of the target area based on the distribution characteristics of vector parcel data within the target area on the standard geographic category, and to obtain a priority evaluation result; to extract and combine data from remote sensing images and vector parcel data based on the priority evaluation result, and to obtain sample units; and to summarize the sample units to obtain a target geographic feature sample library.

[0024] Currently, through periodic land surveys, my country has established a nationwide, up-to-date, and highly accurate vector map database, which is an extremely valuable current geospatial data asset. However, the classification system of this data primarily serves the specific purposes of land surveys and cannot directly meet the personalized, fine-grained semantic classification needs of different vertical industries such as agricultural yield estimation, biodiversity conservation, and disaster assessment. Therefore, how to systematically and efficiently transform standardized land survey vector data into customized, high-quality sample libraries adapted to different application scenarios is a pressing technical problem that needs to be solved.

[0025] It should be noted that the executing entity of this invention can be a custom construction device for a patch sample library, or a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or a thermal management device capable of realizing the above functions of a custom construction device for a patch sample library. This embodiment does not specifically limit this. The following uses a custom construction device for a patch sample library as an example to describe this embodiment and the following embodiments.

[0026] Based on this, embodiments of this application provide a method for customizing the construction of a patch sample library, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the method for customizing the sample library of image patches in this application.

[0027] In this embodiment, the method for customizing the patch sample library includes steps S10 to S50: Step S10: Obtain user-defined image patch sample library requirements information, the requirements information including target region and target semantic category information.

[0028] It should be noted that the target area refers to the geographical scope within which the user wishes to create the sample database. Specifically, it can be a complete administrative division, such as a province, city, or county, or a user-defined rectangular or polygonal region. The target semantic category information refers to the types of land features that the user is interested in under their specific business scenario. For example, when the user is an agricultural researcher, the corresponding target semantic category information might focus on crop types such as rice, wheat, and greenhouses.

[0029] Understandably, these requirements stem from users' actual business objectives, and are typically expressed using natural language or industry terminology within the field, rather than strictly adhering to a standard geographic information classification system. Therefore, the core function of this step is to transform users' subjective and diverse application intentions into objective and structured input parameters that can be accurately identified and processed in subsequent processes.

[0030] Step S20: Based on the preset geographic classification system, the target semantic category information is parsed to determine the standard geographic category.

[0031] It should be noted that the core purpose of this step is to accurately map the category expressions used by users in their requirements, which may be colloquial or domain-specific, to the unambiguous standard categories defined by the system.

[0032] Understandably, parsing target semantic category information involves natural language processing and semantic similarity calculation techniques. The system compares each category term entered by the user with all candidate standard categories in the classification system. By analyzing the semantic correlation of words, it finds one or more standard categories that match best, laying the foundation for subsequent screening of corresponding image patches from standardized data sources and avoiding data omissions or misjudgments caused by name differences.

[0033] Step S30: Based on the distribution characteristics of vector parcel data within the target area on the standard geographic category, perform sampling priority assessment on the target area to obtain priority assessment results.

[0034] It should be noted that vector parcel data refers to polygonal data within the target area that has been surveyed and confirmed, possessing precise boundaries and clearly defined land use attributes. This step aims to address how to scientifically select relatively high-value sample collection areas within a vast geographical space. The basic concept is that, since the spatial distribution of different types of parcels is not uniform, sampling priority assessment is a process of quantitatively evaluating the information value contained in each geographical unit based on standard geographical categories.

[0035] Understandably, distribution characteristics primarily include the number and area of ​​each category of patches, as well as their spatial clustering throughout the region. The evaluation system uses spatial statistical methods to analyze dense areas of a particular category of patches, or complex areas with highly heterogeneous distributions of multiple target categories. For example, a grid containing farmland, orchards, rural roads, and ponds / water surfaces typically has far greater information richness and representativeness than a grid containing only large areas of homogeneous forest. The priority evaluation results are essentially a spatial guideline, indicating which sub-regions should be prioritized for sampling.

[0036] It should be understood that this step is crucial for achieving optimal resource allocation and ensuring the representativeness of the sample set. By analyzing distribution characteristics, this method effectively avoids invalid sampling in homogeneous areas with scarce information or in areas outside the target category, thus concentrating limited computational and human resources on the most valuable "information-rich areas." This data-driven evaluation strategy ensures that the final sample database efficiently covers all land types of interest to users and fully reflects their true spatial distribution patterns and combinations.

[0037] In one embodiment, the step of performing a sampling priority assessment of the target area based on the distribution characteristics of vector land parcel data within the target area on the standard geographic categories, and obtaining a priority assessment result, includes: statistically determining the number and total area of ​​land parcels of each standard geographic category within the target area based on the vector land parcel data; calculating the distribution density and area proportion of each standard geographic category in the entire target area based on the number and total area of ​​land parcels of each standard geographic category; and performing a sampling priority assessment of the target area based on the distribution density and area proportion, thereby obtaining a priority assessment result.

[0038] It should be noted that this embodiment specifically introduces a quantitative implementation method for sampling priority assessment. The process first performs basic statistics on vector plots belonging to each standard geographic category within the target area, obtaining the number of individual plots in each category and their total covered geographic area. Distribution density reflects the frequency or clustering of a certain type of plot within a unit space, and is usually positively correlated with the number of plots, but also considers the concentration of their spatial arrangement. Area percentage refers to the percentage of the total area of ​​a certain type of plot in the entire target area, reflecting the dominance of that type of land cover in the landscape pattern.

[0039] Understandably, the purpose of calculating distribution density and area proportion is to characterize the spatial performance of each geographic category from different dimensions. High distribution density means that the type of land cover may have formed a typical cluster pattern, and sampling in such areas makes it easier to capture representative samples of this type of land cover and its typical spatial relationship with the surrounding environment. On the other hand, a large area proportion indicates that this type of land cover is a major component of the region, and to ensure that the sample database can represent the overall appearance of the region, it is necessary to ensure that it has a sufficient sample size. The evaluation process takes into account both of these factors. For example, for categories with a small area proportion but extremely high distribution density (such as scattered industrial land), their sampling priority may be higher than that of categories with a large area proportion but extremely uniform and singular distribution (such as contiguous plain farmland), because the former contains more complex boundary information and spatial variation characteristics.

[0040] It should be understood that prioritizing based on distribution density and area proportion ultimately aims to generate a spatial decision map to guide the sampling process. This assessment is not a simple ranking of land use importance, but rather an explicit spatial guide that identifies hotspots that combine land use typicality with spatial complexity. Through this data-driven approach, the system can automatically prioritize sampling resources to the geographic units with the highest information value per unit area. This maximizes the statistical representativeness and spatial diversity of the final sample library within limited sample size constraints, providing optimal data for subsequent model training.

[0041] In one embodiment, before sorting the rule grids in descending order of information richness and selecting a corresponding number of rule grids with high sorting based on a preset sampling quantity to obtain the priority evaluation result, the method further includes: performing overlay analysis on the target area, preset national vegetation zoning data, and national administrative boundary data to obtain zoning data overlap results; dividing the target area into multiple geographical zoning based on the zoning data overlap results; and allocating sampling quotas to each geographical zoning based on the total number of samples set by the user and the area ratio of each geographical zoning, wherein the sampling quota is used when performing sampling priority evaluation within each geographical zoning, and the preset sampling quantity used is the sampling quota allocated to the corresponding geographical zoning.

[0042] It should be noted that this embodiment provides a specific preparatory step for implementing a stratified sampling strategy within a target area. The core of this step lies in generating overlapping zoning data by first utilizing two types of basic geographic information—national vegetation zoning data and national administrative boundary data—through spatial overlay analysis with the target area. Essentially, this involves doubly subdividing the target area according to both natural geographic conditions (such as climate and soil characteristics reflected by vegetation types) and administrative units, thus forming finer geographic zoning. These zonings are relatively homogeneous but differ from one another. For example, a cross-provincial region can be divided into an eastern plain agricultural area and a western mountain forest area.

[0043] It is understandable that the total number of samples set by the user is the total number of samples that need to be obtained in the end, while the sampling quota is the number of samples that should be collected in each partition after allocating the total number of samples based on the proportion of the area of ​​each partition to the total area of ​​the target area. This ensures that the spatial distribution of samples is proportional to the actual size of each partition and avoids over-concentration of sampling in a certain type of area.

[0044] It should be understood that once each geographic partition is assigned an independent sampling quota, subsequent sorting and selection operations based on the richness of information within that partition, performed within that partition, are subject to clear quantitative constraints. In other words, within each partition, the system only needs to select no more than the quota number of grids as samples, according to their priority. This design ensures that the final sample set not only represents the most information-rich area within each partition at the micro level, but also provides balanced coverage of all important geographic units within the target area at the macro level, significantly enhancing the statistical inference and generalization capabilities of the sample database for the entire region.

[0045] Step S40: Based on the priority evaluation results, extract and combine data from the remote sensing image and the vector land parcel data to obtain sample units.

[0046] It should be noted that the core operation of this step lies in the targeted collection and integration of multi-source data based on the high-priority spatial locations identified by the evaluation results. A sample unit refers to the basic unit that constitutes the final sample library, typically corresponding to a specific geographical area and the various types of information it carries. Specifically, this includes the original remote sensing imagery covering the area, precisely registered vector patch data, and automatically generated descriptive text descriptions of land cover categories and spatial relationships.

[0047] In one embodiment, the step of extracting and combining data from remote sensing imagery and vector patch data to obtain sample units based on the priority evaluation result includes: determining a grid to be sampled from the target area based on the priority evaluation result; cropping corresponding image patches from the remote sensing imagery based on the geographical extent of the grid to be sampled, and extracting vector patches corresponding to the geographical extent from the vector patch data; generating text descriptions associated with the image patches and vector patches based on the composition and spatial distribution statistics of various land cover categories in the vector patches; and associating and combining the image patches, the vector patches, and the text descriptions to generate the sample units.

[0048] It should be noted that this embodiment presents an execution flow for generating specific sample units based on priority evaluation results. This flow begins with the spatial positioning of the grid to be sampled, which refers to the regular grid units ultimately selected after being sorted according to information richness. Based on the precise geographic boundaries of these grids, the system performs two core data extraction operations: first, cropping image patches from the original remote sensing imagery that completely cover the grid area; and second, extracting all vector map patches from the vector patch data that spatially fall within this area. Image patches provide the spectral and textural features of the land features, while vector map patches provide the precise boundary geometry and category attributes of each land feature.

[0049] Understandably, generating text descriptions is a crucial step in enhancing sample interpretability and machine learning friendliness. These descriptions are based on automated statistical analysis of all vector plots extracted from the current grid. The system identifies and statistically analyzes the main land cover categories within the area, their respective area proportions, and typical spatial adjacency relationships between them. For example, for a single sample cell, the text description might summarize it as "This area is primarily composed of arable land, with small rural settlements mosaicked within, and linear roads running through it." This description encapsulates important scene semantic information in natural language.

[0050] It should be understood that the final generated sample unit is a structured multimodal data entity. It associates and encapsulates three types of heterogeneous information originating from the same spatial location: image patches, vector patches, and text descriptions, together forming a complete training sample. This "three-piece set" combination ensures that the sample data possesses a rich information hierarchy, ranging from pixel-level features and object-level geometry to scene-level semantics. This provides a high-quality data foundation for training intelligent models capable of deeply understanding complex geographical scenes and is a core element in achieving automated, high-quality sample production.

[0051] Step S50: Summarize the sample units to obtain the target patch sample library.

[0052] It should be noted that the aggregation process follows specific data entry procedures. Each sample unit, as an independent record in the database, contains image patches, vector patches, and text descriptions, which are stored in a predefined structure and typically supplemented with metadata information such as sample number, spatial extent, and acquisition time to facilitate indexing and retrieval. The resulting target patch sample database is a product with a clear application focus. It systematically encapsulates all the value of on-demand planning, intelligent optimization, and automatic generation achieved in the preceding steps, forming a training dataset that can directly serve specific tasks such as land parcel classification, change detection, and feature identification.

[0053] Understandably, the establishment of this sample database has transformed model training activities that previously relied on scattered manual annotations into a systematic work based on standardized, high-quality datasets, significantly improving the development efficiency of geospatial analysis applications and the reliability of model performance.

[0054] This embodiment intelligently maps user needs to a classification system and prioritizes data based on spatial distribution characteristics, selecting information-rich regions as sampling units. Within each unit, it automatically crops image patches, extracts vector image patches, and generates descriptive text, ultimately associating and combining these into structured sample units and summarizing them. This method achieves a systematic transformation from user intent to high-quality sample output.

[0055] In summary, this technical solution ensures the accuracy and standardization of sample semantics through demand analysis and standardized mapping; it utilizes data-driven spatial priority assessment to focus the sampling process on the most representative complex areas, thereby significantly improving the statistical representativeness and spatial heterogeneity of the sample set with a limited sample size; and its fully automated processing transforms the tedious task of relying on manual interpretation into an efficient system engineering process, greatly reducing manpower and time costs and providing a high-quality data foundation for training geospatial intelligent models.

[0056] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 In the method for customizing the image patch sample library, the step of parsing the target semantic category information and determining the standard geographic category according to the preset geographic classification system includes steps S201 to S203: Step S201: Perform semantic similarity matching between each category contained in the target semantic category information and each standard category in the geographic classification system to obtain the matching score between each pair of categories.

[0057] It should be noted that the core of this step lies in calculating the semantic association strength between user terms and standard system terms. The matching score is used to objectively measure the semantic conceptual closeness between a specific category term in the target semantic category information and a potential standard category term in the geographic classification system.

[0058] Understandably, semantic similarity matching relies on natural language processing techniques, particularly word vector models or knowledge graph-based semantic computation. This method is not a simple string comparison, but rather delves into the semantic connotations of words for comparison. For example, for the user-defined word "factory," the system calculates its semantic distance to multiple standard categories such as "industrial land," "warehouse land," and "construction land," thereby obtaining a series of quantified matching scores.

[0059] It should be understood that this step transforms potentially ambiguous or domain-specific user descriptions into a series of comparable objective scores, providing a data foundation for subsequent accurate screening, avoiding mapping failures due to different terminology habits, and enhancing the system's robustness in handling diverse user needs.

[0060] Step S202: Filter and summarize the standard categories whose matching scores are higher than the preset score threshold to obtain a set of candidate standard categories.

[0061] It should be noted that the core of this step lies in filtering the preliminary semantic matching results based on a predetermined metric. The preset score threshold is a key criterion, which defines the minimum acceptable level of semantic relevance, thereby excluding matches with weak relevance or potential ambiguity.

[0062] Understandably, after threshold filtering, each user-defined category term may correspond to one or more standard categories. These filtered categories are aggregated to form a candidate standard category set. This set, as an intermediate result, combines broad inclusiveness of user intent with adherence to the system. For example, the user-proposed term "artificial water body" may simultaneously match two standard categories: "reservoir / pond" and "canal / channel." As long as their matching degree both exceed the threshold, both are included in the candidate set.

[0063] It should be understood that the resulting set of candidate standard categories lays the foundation for subsequent geospatial rationality verification. This avoids potential information omissions caused by overly strict one-to-one mapping in the initial stage, retains a certain degree of flexibility, and ensures the quality of candidate categories through quantification thresholds.

[0064] Step S203: Based on the geographical distribution knowledge base corresponding to the target area, filter the candidate standard category set to determine the standard geographical category.

[0065] It should be noted that the geographic distribution knowledge base refers to a knowledge system containing the patterns of various geographic elements in a specific region, such as the climatic adaptability of specific vegetation types and the regional characteristics of geological structures within that region. Its core purpose is to perform a secondary verification of the spatial rationality of the candidate standard category set obtained through semantic matching, ensuring that the determined categories not only match the names but also conform to the objective geographical background of the target area, thereby avoiding logical errors that may arise from mere semantic association and contradict the actual situation.

[0066] Understandably, this screening process involves placing abstract candidate categories within specific geospatial contexts for feasibility assessment. For instance, semantically, "salt fields," as an artificial water surface, might have a high degree of matching with "ponds and pools." However, if the target area is inland mountainous terrain, based on geographical distribution knowledge, the area lacks the natural and human conditions for salt production. Therefore, the candidate category "salt fields" would be deemed to be geographically incompatible and excluded. This mechanism effectively integrates semantic understanding with geographical expertise, enhancing the rigor of category mapping.

[0067] It should be understood that the final standard geographical categories, determined through screening based on a geographical distribution knowledge base, possess both semantic accuracy and spatial plausibility. This step elevates data processing from a simple text matching level to a comprehensive reasoning level that aligns with real-world patterns, significantly enhancing the scientific rigor and credibility of subsequent spatial sampling and data extraction results based on these categories.

[0068] In one embodiment, the step of filtering the candidate standard category set based on the geographic distribution knowledge base corresponding to the target region to determine the standard geographic category includes: obtaining geographic constraint rules based on the geographic distribution knowledge base corresponding to the target region; matching and verifying each category in the candidate standard category set with the geographic constraint rules to obtain a compliance judgment result for each category; and removing categories that do not conform to the geographic constraint rules from the candidate standard category set based on the compliance judgment result to obtain the standard geographic category.

[0069] It should be noted that geographical constraint rules refer to a series of judgment conditions derived from the geographical distribution knowledge base. Their content reflects the objective regularity of the spatial distribution of specific geographical features, such as the altitude range typically observed for a certain landform type or the suitable climate zone for a certain land use type. The matching and verification process is a systematic compliance review. The system will judge each category in the candidate standard category set one by one whether it meets the geographical constraint rules derived for the target area, thereby generating a qualitative compliance judgment result. For example, if the target area is a high-latitude plain, a typical geographical constraint rule might be "evergreen broad-leaved forests are only distributed in areas south of specific climatic conditions." Then, for the candidate category "evergreen broad-leaved forest," the system will determine, based on this rule and combined with regional location information, whether it has a reasonable distribution possibility within the target area.

[0070] Understandably, this method effectively encodes experts' prior geographical knowledge into computational logic, enabling automated spatial context verification of candidate categories. This allows for the precise elimination of options that significantly conflict with the regional geographical context. This not only improves the accuracy of standard geographical category determination but also enhances the automation and reliability of the entire sample database construction process.

[0071] In this embodiment, the association between user terms and standard categories is established and quantified by semantic similarity calculation; then, a set of candidate standard categories with similar semantics is selected according to a preset threshold; finally, the spatial rationality of the candidate categories is verified by combining the geographical distribution knowledge base of the target area, and options that do not conform to the regional distribution pattern are eliminated, thereby accurately determining the final standard geographical category.

[0072] In summary, this embodiment achieves intelligent and precise mapping between user intent and standards by combining semantic matching and spatial verification. By introducing constraints based on geographical distribution patterns on top of semantic association, it effectively avoids spatial logic errors that may arise from relying solely on text similarity. This ensures that the final determined geographical category is not only semantically consistent but also realistically plausible, thereby significantly improving the accuracy and scientific value of geospatial data processing.

[0073] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the customized construction method of the image sample library of this application. Any simple transformations based on this technical concept are within the protection scope of this application.

[0074] This application also provides a device for customizing the construction of a patch sample library; please refer to... Figure 3 The customized construction device for the image patch sample library includes: Data acquisition module 10 is used to acquire user-defined requirements for a sample library of image patches, the requirements including target region and target semantic category information; The data processing module 20 is used to parse the target semantic category information according to a preset geographic classification system and determine the standard geographic category; Data evaluation module 30 is used to perform sampling priority evaluation on the target area based on the distribution characteristics of vector parcel data in the target area on the standard geographic category, and obtain priority evaluation results; The data processing module 20 is also used to extract and combine data from the remote sensing image and the vector land parcel data according to the priority evaluation result to obtain sample units; The data processing module 20 is also used to summarize the sample units to obtain a target patch sample library.

[0075] In one embodiment, the data processing module 20 is further configured to perform semantic similarity matching between each category contained in the target semantic category information and each standard category in the geographic classification system to obtain a matching score between each pair of categories; filter and summarize the standard categories whose matching scores are higher than a preset score threshold to obtain a set of candidate standard categories; and filter the set of candidate standard categories based on the geographic distribution knowledge base corresponding to the target region to determine the standard geographic category.

[0076] In one embodiment, the data processing module 20 is further configured to obtain geographical constraint rules based on the geographical distribution knowledge base corresponding to the target area; match and verify each category in the candidate standard category set with the geographical constraint rules to obtain compliance judgment results for each category; and remove categories that do not conform to the geographical constraint rules from the candidate standard category set based on the compliance judgment results to obtain the standard geographical category.

[0077] In one embodiment, the data evaluation module 30 is further configured to: statistically obtain the number and total area of ​​each standard geographical category within the target area based on the vector land parcel data; calculate the distribution density and area proportion of each standard geographical category in the entire target area based on the number and total area of ​​each standard geographical category; and perform sampling priority evaluation on the target area based on the distribution density and area proportion to obtain priority evaluation results.

[0078] In one embodiment, the data evaluation module 30 is further configured to divide the target area into multiple regular grids; calculate the information richness of each regular grid according to the distribution density and area ratio; sort the regular grids in descending order of information richness, and select the corresponding number of regular grids that are ranked first according to a preset sampling quantity to obtain the priority evaluation result.

[0079] In one embodiment, the data evaluation module 30 is further configured to perform overlay analysis based on the target area, preset national vegetation zoning data, and national administrative boundary data to obtain zoning data overlap results; divide the target area into multiple geographical zoning based on the zoning data overlap results; and allocate sampling quotas to each geographical zoning based on the total number of samples set by the user and the area ratio of each geographical zoning, wherein the sampling quota is used when performing sampling priority evaluation within each geographical zoning, and the preset sampling quantity used is the sampling quota allocated to the corresponding geographical zoning.

[0080] In one embodiment, the data processing module 20 is further configured to: determine the grid to be sampled from the target area based on the priority evaluation result; crop the corresponding image patch from the remote sensing image based on the geographical range of the grid to be sampled, and extract the vector map patch corresponding to the geographical range from the vector patch data; generate a text description associated with the image patch and the vector map patch based on the composition and spatial distribution statistics of the various land cover categories in the vector map patch; and associate and combine the image patch, the vector map patch, and the text description to generate the sample unit.

[0081] This embodiment ensures that the sample categories are semantically clear and consistent with the actual regional geography through semantic parsing and geographical knowledge verification, thereby improving the accuracy and practicality of the samples. Prioritization based on spatial distribution characteristics and stratified sampling concentrate sample collection in information-rich key areas. This method automates the construction of a high-quality sample library from user needs.

[0082] The customizable map feature library construction device provided in this application, employing the customizable map feature library construction method in the above embodiments, can solve the technical problem of how to systematically and efficiently transform standardized land survey vector data into a customized high-quality map feature library adapted to different application scenarios. Compared with the prior art, the beneficial effects of the customizable map feature library construction device provided in this application are the same as those of the customizable map feature library construction method provided in the above embodiments, and other technical features in the customizable map feature library construction device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0083] This application provides a custom construction device for a patch sample library, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the custom construction method for the patch sample library in the first embodiment described above.

[0084] The following is for reference. Figure 4 This document illustrates a structural schematic diagram of a customizable image sample library construction device suitable for implementing embodiments of this application. The customizable image sample library construction device in embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 4 The illustrated custom construction device for the sample library is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0085] like Figure 4As shown, the custom construction device for the patch sample library may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 1002 or a program loaded from storage device 1003 into random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the custom construction device for the patch sample library. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the custom construction device for the patch sample library to communicate wirelessly or wiredly with other devices to exchange data. Although a custom construction device for the patch sample library with various systems is shown in the figure, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented alternatively.

[0086] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0087] The customizable map feature library construction device provided in this application, employing the customizable map feature library construction method described in the above embodiments, can solve the technical problem of how to systematically and efficiently transform standardized land survey vector data into customized high-quality map feature libraries adapted to different application scenarios. Compared with the prior art, the beneficial effects of the customizable map feature library construction device provided in this application are the same as those of the customizable map feature library construction method provided in the above embodiments, and other technical features in this customizable map feature library construction device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0088] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0089] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0090] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the custom construction method of the patch sample library in the above embodiments.

[0091] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0092] The aforementioned computer-readable storage medium may be included in the custom construction device for the patch sample library; or it may exist independently and not be assembled into the custom construction device for the patch sample library.

[0093] The aforementioned computer-readable storage medium carries one or more programs. When these programs are executed by the custom construction device for the map patch sample library, the custom construction device for the map patch sample library causes the following: it acquires user-defined map patch sample library requirement information, including target area and target semantic category information; it parses the target semantic category information according to a preset geographic classification system to determine a standard geographic category; it performs a sampling priority assessment of the target area based on the distribution characteristics of vector parcel data within the target area on the standard geographic category, obtaining a priority assessment result; it extracts and combines data from remote sensing images and the vector parcel data according to the priority assessment result to obtain sample units; and it summarizes the sample units to obtain the target map patch sample library.

[0094] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0095] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0096] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0097] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described method for customizing the map feature sample library. This solves the technical problem of how to systematically and efficiently transform standardized land survey vector data into customized, high-quality sample libraries adapted to different application scenarios. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the map feature sample library customization method provided in the above embodiments, and will not be elaborated upon here.

[0098] The computer program product provided in this application can solve the technical problem of customized construction of a patch sample library. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as the beneficial effects of the customized construction method of the patch sample library provided in the above embodiments, and will not be repeated here.

[0099] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for customizing the construction of a patch sample library, characterized in that, The method for customizing the image patch sample library includes: Obtain user-defined requirements for a sample library of image patches, including target regions and target semantic category information; Based on a preset geographic classification system, the target semantic category information is parsed to determine the standard geographic category; Based on the distribution characteristics of vector parcel data within the target area on the standard geographic category, a sampling priority assessment is performed on the target area to obtain the priority assessment results; Based on the priority evaluation results, data are extracted and combined from the remote sensing imagery and the vector parcel data to obtain sample units; By summarizing the aforementioned sample units, a target patch sample library is obtained; The step involves evaluating the sampling priority of the target area based on the distribution characteristics of vector parcel data within the target area according to the standard geographic category, and obtaining the priority evaluation results, including: Based on the vector land parcel data, the number and total area of ​​land parcels of each standard geographical category within the target area are statistically obtained; Based on the number of plots and total area of ​​each standard geographical category, calculate the distribution density and area proportion of each standard geographical category in the entire target area; Based on the distribution density and area proportion, the sampling priority of the target area is evaluated to obtain the priority evaluation result; The sampling priority assessment of the target area based on the distribution density and area proportion is used to obtain the priority assessment results, including: The target region is divided into multiple regular grids; Based on the distribution density and area ratio, the information richness of each regular grid is calculated; The rule grids are sorted in descending order of information richness, and a number of rule grids with the highest information richness are selected according to a preset sampling quantity to obtain the priority evaluation result. Before sorting the rule grids in descending order of information richness and selecting a corresponding number of top-ranked rule grids based on a preset sampling quantity to obtain the priority evaluation result, the method further includes: The target area, the preset national vegetation zoning data, and the national administrative boundary data are overlaid to obtain the zoning data overlap results; Based on the overlapping results of the partition data, the target area is divided into multiple geographical partitions; Based on the total number of samples set by the user and the area ratio of each geographic partition, a sampling quota is allocated to each geographic partition. The sampling quota is used when evaluating the sampling priority within each geographic partition, and the preset sampling quantity is the sampling quota allocated to the corresponding geographic partition.

2. The method for customized construction of a patch sample library according to claim 1, characterized in that, The step of parsing the target semantic category information according to a preset geographic classification system to determine the standard geographic category includes: Each category contained in the target semantic category information is semantically similar to each standard category in the geographic classification system to obtain a matching score between each pair of categories. The standard categories with matching scores higher than a preset score threshold are filtered and summarized to obtain a set of candidate standard categories; Based on the geographical distribution knowledge base corresponding to the target area, the candidate standard category set is filtered to determine the standard geographical category.

3. The method for customized construction of a patch sample library according to claim 2, characterized in that, The process of filtering the candidate standard category set based on the geographical distribution knowledge base corresponding to the target region to determine the standard geographical category includes: Based on the geographical distribution knowledge base corresponding to the target area, geographical constraint rules are obtained; Each category in the candidate standard category set is matched and verified with the geographical constraint rule to obtain the compliance judgment result of each category; Based on the compliance determination result, categories that do not conform to the geographical constraint rules are removed from the candidate standard category set to obtain the standard geographical category.

4. The method for customized construction of a patch sample library according to claim 1, characterized in that, The step of extracting and combining data from remote sensing imagery and vector land parcel data based on the priority evaluation results to obtain sample units includes: Based on the priority evaluation results, the grid to be sampled is determined from the target region; Based on the geographical range of the grid to be sampled, the corresponding image patches are cropped from the remote sensing image, and the corresponding vector map patches are extracted from the vector patch data. Based on the composition and spatial distribution statistics of the various land cover categories in the vector image patch, a text description associated with the image patch and the vector image patch is generated; The image patch, the vector image patch, and the text description are associated and combined to generate the sample unit.

5. A device for customizing the construction of a patch sample library, characterized in that, The customized construction device for the image patch sample library includes: The data acquisition module is used to acquire user-defined requirements for a sample library of image patches, including target regions and target semantic category information. The data processing module is used to parse the target semantic category information according to a preset geographic classification system and determine the standard geographic category; The data evaluation module is used to perform sampling priority evaluation on the target area based on the distribution characteristics of vector parcel data in the target area on the standard geographic category, and obtain the priority evaluation result. The data processing module is also used to extract and combine data from the remote sensing image and the vector land parcel data according to the priority evaluation results to obtain sample units; The data processing module is also used to summarize the sample units to obtain a target patch sample library; The data evaluation module is further configured to: statistically determine the number and total area of ​​plots of each standard geographic category within the target area based on the vector plot data; calculate the distribution density and area proportion of each standard geographic category in the entire target area based on the number and total area of ​​plots of each standard geographic category; perform sampling priority evaluation on the target area based on the distribution density and area proportion to obtain priority evaluation results; divide the target area into multiple regular grids; calculate the information richness of each regular grid based on the distribution density and area proportion; sort the regular grids in descending order of information richness, and then... Based on a preset sampling quantity, a corresponding number of rule grids with high priority are selected to obtain the priority evaluation result; based on the target area, preset national vegetation zoning data, and national administrative boundary data, an overlay analysis is performed to obtain the zoning data overlap result; based on the zoning data overlap result, the target area is divided into multiple geographical zoning; based on the total number of samples set by the user and the area ratio of each geographical zoning, a sampling quota is allocated to each geographical zoning, wherein the sampling quota is used when performing sampling priority evaluation within each geographical zoning, and the preset sampling quantity used is the sampling quota allocated to the corresponding geographical zoning.

6. A device for customizing the construction of a patch sample library, characterized in that, The image patch sample library customization construction device includes: a memory, a processor, and an image patch sample library customization construction program stored on the memory and executable on the processor, wherein the image patch sample library customization construction program is configured to implement the steps of the image patch sample library customization construction method as described in any one of claims 1 to 4.

7. A storage medium, characterized in that, The storage medium stores a custom construction program for a patch sample library, which, when executed by a processor, implements the steps of the custom construction method for a patch sample library as described in any one of claims 1 to 4.

Images