Investigation data sharing processing method

Abstract

The invention provides a survey data sharing processing method, which belongs to the technical field of sharing management, and specifically comprises the following steps of: taking a data packet without a manual processing data packet as a to-be-processed target, and identifying the to-be-processed target on the basis of an identification processing method of the to-be-processed target and a similar condition of boundary association data between the to-be-processed target and other to-be-processed targets; and determining an identification processing sequence of the to-be-processed target, and performing identification processing on a cultivated land layer of the to-be-processed target based on the identification processing sequence and an identification processing method, and a sharing processing method of the to-be-processed target is determined by utilizing a recognition processing method of the to-be-processed target in the later period of the to-be-processed target and the similarity condition of the boundary association data between the to-be-processed target and the recognized to-be-processed target, so that the reliability degree of boundary recognition processing is improved.

Description

Technical Field

[0001] This invention belongs to the field of shared management technology, and in particular relates to a method for sharing and processing survey data. Background Technology

[0002] Land survey data (such as data from the Third National Land Survey) is characterized by high precision and confidentiality; direct release could potentially involve national security and public interests. Currently, simple methods such as geographic coordinate offsetting and local fuzzification are commonly used in government data sharing.

[0003] According to encryption requirements, except for the cultivated map layer, the other layers are processed using nationally recognized nonlinear confidentiality processing technology for geographic information data to reduce their positional accuracy. The planar positional accuracy is controlled to meet the technical requirements. Existing technical solutions often rely on manual optimization, which has technical problems such as high processing difficulty and low accuracy.

[0004] To address the aforementioned technical problems, the invention patent application CN202311661619.2, "A Rubber Sheet Grid Decryption Algorithm Based on Controllable Vector Data Precision," determines the layer data requiring optimization based on image recognition results, thereby improving the efficiency and reliability of the decryption process. However, the above technical solution has the following technical problems: When dividing the cultivated map layer into layers and other layers, relying solely on image recognition models may result in low reliability. Therefore, combining image features of different layer connection locations can help improve the reliability of the recognition process. This makes it an urgent technical problem to determine the parsing order and shared processing methods for data packets with different layer types, thereby improving the reliability of parsing and data reuse.

[0005] Therefore, there is an urgent need for a method for sharing and processing survey data. Summary of the Invention

[0006] To achieve the objectives of this invention, the following technical solution is adopted: Specifically, this application provides a method for sharing and processing survey data, which includes: S1 uses the recognition and processing results of the cultivated map layer to determine the boundary association data between the cultivated map layer and different layers in the data packet. Based on the deviation of the boundary association data of the cultivated map layer, it determines the manually processed data packet in the data packet. Based on the similarity of the boundary association data between the manually processed data packet and the manually processed data packet, it determines the recognition and processing method to remove the manually processed data packet. S2 takes the data packets excluding manually processed data packets as the targets to be processed, and determines the order of identification and processing of the targets to be processed based on the identification and processing method of the targets to be processed and the similarity of the boundary association data between the targets to be processed and other targets to be processed. S3 performs the identification processing of the cultivated map layer of the target to be processed based on the identification processing order and identification processing method, and determines the shared processing method of the target to be processed by utilizing the identification processing method of the target to be processed in the later stage and the similarity of the boundary association data between the target to be processed and the target to be processed that has been identified.

[0007] The beneficial effects of this invention are as follows: Based on the deviation of the boundary association data of the cultivated map layers, manually processed data packets are determined within the data packets. This enables the determination of a method for filtering manually processed data packets based on the dispersion of layer distribution in different data packets. This not only ensures the reliability of manual verification but also, while maintaining a high degree of dispersion in layer distribution, increases the number and proportion of manually processed data packets, further enhancing the reliability of distinguishing between layers. By utilizing the identification and processing methods of the target in the later stages and the similarity of the boundary association data between the target and the target that has already been identified, a shared processing method for the target is determined. This realizes the reuse processing requirement of the image features of the target in the later stages of boundary segmentation, as well as the degree of identification deviation of the target boundary. It also determines whether further manual identification processing is needed before shared processing, thereby further improving the reliability of the overall boundary segmentation identification and processing.

[0008] Furthermore, the different layers include woodland, garden land, roads, ditches, and construction land.

[0009] Furthermore, the boundary association data is determined based on the layer type of the cultivated map layer connection of the image in the data packet.

[0010] Furthermore, the deviation of the boundary association data of the cultivated map layer is determined based on the deviation of the layer type of the cultivated map layer connection between different data packets.

[0011] Furthermore, the method for determining the manually processed data packets in the data packets is as follows: Based on the boundary association data of the cultivated map layer, determine the layer type connected to the cultivated map layer in the data packet and use it as the connection layer type; Group data packets with the same connection layer type into a single connection layer group; Based on the combined data of the connection layers and the connection layer type of the data packet, determine whether the data packet is a manually processed data packet.

[0012] Furthermore, the method for determining the shared processing method of the target to be processed is as follows: Using the later-stage targets to be processed of the target to be processed, identify the matching targets that belong to the target to be processed among the later-stage targets to be processed, and treat them as unidentified targets; According to the identification and processing method for unidentified targets, the identification data of the unidentified targets that belong to the data packets that meet the requirements is determined, and the unidentified targets that belong to the data packets that meet the requirements are identified as the identified targets. Based on the unidentified targets and the identified targets of the unidentified targets to be processed, a shared processing method for the target to be processed is determined.

[0013] Other features and advantages will be set forth in the following description, and the objects and other advantages of the invention are realized and obtained through the structures particularly pointed out in the description and the drawings.

[0014] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0015] The above and other features and advantages of the present invention will become more apparent from a detailed description of exemplary embodiments thereof with reference to the accompanying drawings.

[0016] Figure 1 This is a flowchart of a survey data sharing and processing method; Figure 2 It is a flowchart of the method for determining the manual processing of data packets in the data packet; Figure 3 This is a flowchart illustrating the method for determining the identification and processing method of data packets, excluding manual processing. Detailed Implementation

[0017] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0018] Example 1 like Figure 1As shown, this application provides a method for sharing and processing survey data, specifically including: S1 uses the recognition and processing results of the cultivated map layer to determine the boundary association data between the cultivated map layer and different layers in the data packet. Based on the deviation of the boundary association data of the cultivated map layer, it determines the manually processed data packet in the data packet. Based on the similarity of the boundary association data between the manually processed data packet and the manually processed data packet, it determines the recognition and processing method to remove the manually processed data packet. S2 takes the data packets excluding manually processed data packets as the targets to be processed, and determines the order of identification and processing of the targets to be processed based on the identification and processing method of the targets to be processed and the similarity of the boundary association data between the targets to be processed and other targets to be processed. S3 performs the identification processing of the cultivated map layer of the target to be processed based on the identification processing order and identification processing method, and uses the target data of the target to be processed in the later stage to determine the shared processing method of the target to be processed.

[0019] Furthermore, the different layers include woodland, garden land, roads, ditches, and construction land.

[0020] Furthermore, the boundary association data is determined based on the layer type of the cultivated map layer connection of the image in the data packet.

[0021] Furthermore, the deviation of the boundary association data of the cultivated map layer is determined based on the deviation of the layer type of the cultivated map layer connection between different data packets.

[0022] Specifically, such as Figure 2 As shown, the method for determining the manually processed data packet in the data packet is as follows: The core objective of this method is to achieve optimal division of labor in human-machine collaborative processing through an intelligent data packet classification mechanism. Its decision-making logic is based on a key understanding: different types of boundary association patterns correspond to different processing difficulties and sample values. Through multi-level rule analysis, the system can automatically identify two types of data packets most suitable for manual processing: 1) ultra-high complexity data packets, whose multiple boundary types intersect and require comprehensive judgment by human experts; 2) typical representative data packets, whose processing results can generate high-quality training samples and feature benchmarks. The knowledge base formed by these manual processing results can be directly used to improve the accuracy of automated processing, especially to discover and correct omissions or errors in the classification of similar feature regions by automated models.

[0023] S11 uses the boundary association data of the cultivated map layer as a basis to determine the layer type connected to the cultivated map layer in the data packet and uses it as the connection layer type. Determine the type of connected layer: Keyword Explanation: "Connected Layer Type" refers to the specific category of other land cover layers that are spatially adjacent to cultivated land cover patches within a single data packet (usually a geographic map sheet or processing unit). For example, layers that may border cultivated land include forest land, orchard land, roads, ditches, and construction land.

[0024] This step transforms spatial topological relationships into quantifiable discrete attributes. Spatial adjacency relationships are inherently difficult to calculate and compare directly, but once transformed into type labels, set operations and statistical analysis can be performed. The significance lies in establishing a fundamental mapping from spatial relationships to attribute features, providing structured input for subsequent data-driven pattern recognition and classification. This is a prerequisite for achieving automated intelligent screening.

[0025] Example: In data packet P1, the cultivated map patch is bordered by woodland to the east and a rural road to the south. Therefore, the set of connection layer types for this data packet is {woodland, rural road}.

[0026] S12 groups data packets with the same connection layer type into a single connection layer group; Construct a group of connected layers: "Connection layer combination" refers to the unordered set of all "connection layer types" within a data packet, which fully describes the adjacency environment type composition of the cultivated map patches within the packet.

[0027] Grouping packets with exactly the same set of types together is essentially performing clustering based on the homogeneity of their adjacency environment. This helps to discover groups of packets that are processed in completely consistent ways.

[0028] Significance: It can identify "similar items" that can be processed in batches, making it easier to find the most representative samples or discover special cases in a group. This is the foundation for achieving process standardization and scientific sample selection.

[0029] Example: Data packet P1 is composed of {woodland, rural roads}. Although the cultivated land patches in data packet P2 have different shapes and areas, if they also only border woodland and rural roads, then P2 and P1 belong to the same connected layer combination.

[0030] S13 determines whether the data packet is a manually processed data packet based on the connection layer combination data and the connection layer type of the data packet.

[0031] It should be noted that if the number of connection layer types in the data packet is greater than the preset threshold for the number of connection layer types, then there are more image features between it and other connection layer types, and the data packet is determined to be a manually processed data packet.

[0032] Complexity-based filtering: The "Preset Threshold for the Number of Connection Layer Types" is a key parameter used to measure the complexity of the boundary environment. When the number of connection types contained in a data packet exceeds this threshold, its complexity is considered too high.

[0033] The more connection types there are, the higher the degree of feature mixing at the boundaries, the more complex the image features, and the exponentially increase the risk of feature confusion during model judgment. Such situations require advanced human spatial cognition and prior knowledge.

[0034] Significance: Prioritize assigning the most challenging and complex problems to human experts to ensure the quality of solutions to core problems, while preventing automated models from generating catastrophic errors in complex scenarios and spreading them downstream.

[0035] Example: Set the threshold to 5. Data packet A contains 6 connection types: {woodland, orchard, road, ditch, village, agricultural facility land}. Since 6 > 5, data packet A is directly determined to require manual processing.

[0036] Additionally, it can be understood that if the number of connection layer types in the data packet is not greater than the preset threshold for the number of connection layer types, then the number of connection layer combinations is further determined, and it is determined whether the number of connection layer combinations is greater than the preset threshold for the number of layer combinations. If so, the data packet with the number of connection layer types greater than the second preset threshold for the number of connection layer types is treated as a manually processed data packet; otherwise, the process proceeds to the next step. Pattern scarcity analysis: The "Second Preset Threshold for the Number of Connecting Layer Types" is a second parameter used to determine whether a combination of connecting layers "has a certain scale but is relatively scarce".

[0037] If there are many connection layer combinations, such as more than 8 groups, the deviation of connection layer types between different data packets is relatively high. In this case, for connection layer combinations with certain feature reuse value, and targeting those combinations with more connection layer types (such as more than 3 times), centralized manual processing is the most cost-effective. Moreover, the relevant image features have high value in the layer segmentation process, thereby improving the reliability of the segmentation process.

[0038] Example: Let the second threshold be 3. For packet B with no more than 5 connection types (e.g., only 4), check its combination {woodland, garden, road}. Therefore, packets containing connection layer combinations B multiple times need to proceed to the next step.

[0039] S112 determines whether the number of data packets in the connection layer combination where the data packet is located is greater than a preset data packet number threshold. If yes, proceed to the next step; otherwise, determine that the data packet does not belong to manually processed data packets. Group size verification, key explanation: "Preset data packet number threshold" is the threshold value for judging whether a combination of connected layers has a sufficient statistical sample size.

[0040] A "representative" selected from a very small group (such as only 2-3 samples) has low demand for reuse of its image features after extraction, and is not suitable as a basis for training models or formulating rules.

[0041] Significance: To ensure the statistical reliability of the selected samples. Only typical samples selected from a sufficiently large population have characteristics and experiences that have generalizable value and can effectively guide the processing of other data packets.

[0042] Example: Let the threshold be 20. The combination {woodland, ditch} has 50 data packets in the database (50>20), which meets the population size condition, and can proceed to the next step of screening.

[0043] S113 Based on the area of ​​the cultivated map layer of the data packet, from large to small, determine the sorting result of the data packet in the connection layer combination where the data packet is located, and determine whether the sorting result of the data packet is before the preset position. If yes, proceed to the next step; if no, determine that the data packet does not belong to the manually processed data packet. Individual significance ranking: Keyword Explanation: "Preset Position" refers to the ranking boundary (such as the top 5, top 10%, etc.) used to define "large area patches" in a list sorted by cultivated land area.

[0044] Area is an important attribute of geographical elements. Large cultivated land parcels typically have longer and more complete boundaries, a wider contact area with surrounding features, and their characteristics are more obvious and stable in images.

[0045] Significance: Prioritize the selection of samples with the clearest and most significant features. Manual processing of large-area patches can provide richer and less noisy boundary feature information, thereby generating higher-quality training samples or feature benchmarks.

[0046] Example: In a group of 50 packets containing {woodland, ditch}, sort them by cultivated land area. Assume the preset position is "top 5". Packet C, ranked 3rd in area, is selected for final evaluation.

[0047] S114 Determines the proportion of data packets with the connection layer type of the data packet in all data packets based on the connection layer type of the data packet, and determines whether the data packet is a manually processed data packet based on the proportion of the proportion.

[0048] It is understood that when the quantity ratio is greater than a preset quantity ratio threshold, the data packet is determined to be a manually processed data packet.

[0049] Element universality test: Keyword explanation: "Quantity ratio" and "Preset quantity ratio threshold": The former refers to the frequency of a certain connection layer type in the entire data packet; the latter is the threshold value for judging whether the type belongs to "common features".

[0050] If all connection types of a data packet are common features with extremely high frequency, then the boundary features obtained by processing this data packet are the standard features needed to deal with the most common and typical problems in that area.

[0051] Significance: Building a foundational feature "benchmark." The features obtained from processing these data packets are the "basic skills" that automated models should learn to correctly identify first. These features can serve as a benchmark for scanning the entire area and detecting whether the model makes mistakes when handling similar common feature boundaries in other locations.

[0052] Example: Data packet C has connection types of "rural road" (global occurrence rate 90%) and "ditch" (global occurrence rate 85%). Let the threshold for the proportion be 75%. Since both proportions exceed the threshold, data packet C is ultimately determined to be a typical sample requiring manual processing. After manually and precisely delineating its boundaries, standard image features for "farmland-road" and "farmland-ditch" can be extracted. Subsequently, when automatically processing other areas, the algorithm can proactively search for boundary segments with similar image features that have not been correctly classified by the model, providing hints or automatic corrections, thereby effectively solving the model's "missed classification" problem.

[0053] Specifically, such as Figure 3 As shown, the method for determining the identification and processing method for removing manually processed data packets is as follows: This embodiment provides a technical solution for dynamically selecting and constructing the most suitable automated identification and processing method for ordinary data packets (targets to be processed) remaining after excluding "priority manually processed data packets". Its core decision-making logic is: "Similar historical processing experience is the best basis for guiding the current automated processing". By analyzing the similarity of the target to be processed and existing manually processed data packets in boundary association features, the system can intelligently determine which processing strategy is more suitable for the current target: whether to use broader similar features for guidance (one type of method), or to rely on more strictly matched manual samples for precise guidance (two types of methods). The ultimate goal of this method is to achieve personalization and precision in the automated processing flow, avoid the accuracy loss caused by "one-size-fits-all" processing, and ensure that each data packet can be optimally segmented based on its most relevant "historical knowledge".

[0054] S21 takes the data packet excluding the manually processed data packet as the target to be processed, and determines the data packet with the same connection layer type as the target to be processed based on the similarity of the boundary association data between the target to be processed and the manually processed data packet, and takes it as the consistent data packet. Determine consistent data packets: Explanation of the term: "Consistent data packets" refer to all data packets (including those that have been manually processed and those that have not) whose "connection layer type" is exactly the same as the current "target to be processed". For example, if the connection type of the target T to be processed is {woodland, road}, then all data packets with the same connection type {woodland, road} are "consistent data packets" of T.

[0055] The fact that the connection layer types are completely identical means that these data packets and the target to be processed face the exact same adjacency environment type, which is the highest level of similarity. The challenges of boundary processing and the types of features to be identified are theoretically consistent.

[0056] Significance: Identifying the most directly relevant reference group. This is the most direct way to find similar cases to learn from. The size and quality of this group (including how many high-quality samples have been manually processed) directly determine the reliability of subsequent strategy selection.

[0057] Example: The type of target T1 to be processed is {woodland, road}. A query reveals 15 data packets of this type in the database, 5 of which have already been manually processed (denoted as A1-A5). Therefore, these 15 data packets (including A1-A5) are all "consistent data packets" for T1.

[0058] S22 determines the manually processed data packets that meet the requirements based on the similarity situation and the connection layer type of the target to be processed, and uses them as manually associated data packets; Identify manually associated data packets: "Artificially associated data packets" specifically refer to data packets whose connection layer types match (i.e., there is an intersection) with the target area to be processed. The tightness of the match is defined by preset rules (such as a deviation threshold). For example, if the target area to be processed is {woodland, road, ditch}, then artificially processed packets of {woodland, road} or {woodland, ditch, village} (with a deviation of 1 type) may become its "artificially associated data packets".

[0059] When a completely identical artificial sample cannot be found, it is necessary to find a "similar one". Partially matched artificial processing packages have at least some of their boundary features (such as the boundary features of cultivated land and forest land) that can be directly referenced, providing a local reference for processing complex targets.

[0060] Significance: Expanding the scope of available knowledge. When perfectly matching samples are lacking, by aggregating features from multiple partially matching artificial samples, most of the knowledge needed to process the current complex goal can be "pieced together," which is an effective knowledge transfer and combination strategy.

[0061] Example: For the target T1 {woodland, road}, in addition to the completely identical A1-A5, an artificial processing package B1 of type {woodland, road, ditch} is also its "artificial association package" because B1 contains all the types {woodland, road} required by T1, except for the additional "ditch".

[0062] S23 determines the identification and processing method for the target to be processed based on the consistent data packet data, the manually associated data packet data, and the manually processed data packet data in the consistent data packet data.

[0063] It should be noted that if the number of consistent data packets is not less than a preset data packet number threshold, then the identification and processing method of the target to be processed is determined to be a type of identification method.

[0064] Level 1: Decision-making based on the "same style" group: Logical judgment: Check whether the total number of "consistent data packets" is greater than or equal to the preset data packet number threshold (e.g., 10).

[0065] Decision and Interpretation: If the conditions are met, it indicates the existence of a sufficiently large group that is completely homogeneous with the target being processed. In this case, a single identification method is used.

[0066] Method Construction (Class I Method): The core of this method is broad feature learning and guidance. It extracts image features at the connection points between the cultivated map layer and various layers from all "qualified data packets" selected by the first strategy.

[0067] The first strategy is defined as follows: Select data packets whose entirety is contained within the set of target types to be processed. This means that the adjacency of these data packets is a subset or the entirety of the target to be processed, and their features are pure and can be directly referenced. When there are a large number of homogeneous samples, the requirement for extraction accuracy is high, so a single identification method is chosen.

[0068] Example: There are 15 consistent packets in T1 (>threshold 10). A single approach is used. The system collects image features of farmland-forestland and farmland-road boundaries from all packets with connection types {woodland}, {road}, or {woodland, road} (including A1-A5 and other unprocessed packets). Using these features, a model specifically designed to identify {woodland, road} combined boundaries is trained to partition T1.

[0069] It should be noted that if it belongs to a certain type of identification method, then it is necessary to consider the image features of the connection positions between the cultivated map layer and other layers in all the data packets that meet the requirements determined by the first strategy before the identification processing of the target to be processed, so as to assist in the division processing between the cultivated map layer and other layers based on the above image features. It should be noted that the first strategy determines the eligible data packets as those whose connection layer types are all within the connection layer types of the target to be processed, or whose deviation from the connection layer types of the target to be processed is less than a preset deviation threshold.

[0070] Additionally, it can be understood that if the number of consistent data packets is less than a preset data packet number threshold, the proportion of manually processed data packets in the consistent data packets is determined, and it is judged whether the proportion of manually processed data packets in the consistent data packets is greater than a preset proportion threshold. If so, the identification and processing method of the target to be processed is determined to be a second-class identification method, that is, it is necessary to consider the image features of the connection positions between the cultivated map layer and other layers in all the data packets that meet the requirements determined by the second strategy before the identification and processing of the target to be processed, so as to assist in the division processing between the cultivated map layer and other layers based on the above image features. If not, proceed to the next step. Decisions based on the quality of artificial samples within the "same product" group: Logical judgment: If the number of "consistent data packets" is less than the first-level threshold, calculate the proportion of data packets that have completed manual processing. Determine whether this proportion is greater than a preset proportion threshold (e.g., 60%).

[0071] Decision and Interpretation: If the proportion is high (>60%), it indicates that although the same group is small, its manual processing coverage is very high, meaning that the "knowledge level" of this small group is very high. In this case, a binary identification method is used.

[0072] Method Construction (Type II Method): The core of this method is accurate sample matching and feature application. It extracts features from the "qualified data packets" selected by the second strategy.

[0073] The second strategy is defined (same as the first strategy when triggered at this level): it also selects data packets whose connection types are all included in the set of target types to be processed.

[0074] When the population is small but the sample quality is extremely high, the goal is precise reuse. These high-precision manually processed results are directly used as "templates" or "standard answers," and their boundary features are used as the gold standard. Template matching, feature comparison, and other more direct methods are employed to guide the segmentation of the current target. At this point, it is already quite reliable. By controlling the range, the impact of recognition biases in other layer connection types can be further reduced.

[0075] Example: T2 has only 8 consistent data packets (<10), but 7 of them are manually processed packets (87.5% >60%). A binary approach is used. The system will directly use the precise boundary features of farmland-forest land and farmland-road land in these 7 manually processed packets as templates, and search for the most similar regions on the T2 image for division, similar to "copying a gourd".

[0076] It should be noted that the first strategy determines the eligible data packets as those whose connection layer types are all within the connection layer types of the target to be processed.

[0077] Determine whether the number of consistent data packets is within a preset data packet range. If yes, then determine that the identification and processing method for the target to be processed is a type of identification method; otherwise, proceed to the next step. The artificially associated data packets of the target to be processed are obtained, and the identification and processing method of the target to be processed is determined based on the artificially associated data packets of the target to be processed in different connection layer types.

[0078] It should be noted that when there is a connection layer type with a number of manually associated data packets less than the preset threshold for the number of associated data packets, the identification and processing method for the target to be processed is determined to be a Class I identification method. When there is no connection layer type with a number of manually associated data packets less than the preset threshold for the number of associated data packets, the identification and processing method for the target to be processed is determined to be a Class II identification method.

[0079] Composite decision-making based on population size and coverage of associated samples: Secondary verification of population size: If neither of the above two levels is satisfied, check whether the number of "consistent data packets" falls within a preset data packet range (e.g., [6, 9]). If it is within the range, it means that there is still a relatively high homogeneous population for reference, and the proportion of manually processed data packets is small, so the first-class method (broad learning) is still used.

[0080] Related sample coverage analysis: If the population size is also outside the range (i.e., very small, such as <3), then proceed to the final decision. Analyze the number of "manually associated data packets" corresponding to each connection layer type.

[0081] Decision rule: If there are certain connection layer types with a very small number of manually associated samples (less than the preset threshold for the number of associated data packets, such as 10), then there is a serious lack of knowledge to draw upon for that type.

[0082] Final decision: If knowledge of any one type of connection is insufficient, use method one; otherwise, if knowledge of all types is sufficient, use method two.

[0083] This is a trade-off between risk control and resource utilization. When reference knowledge of a certain type is severely lacking, it is safer to adopt a more conservative, first-class method that relies more on learning general features, avoiding the overall failure of accurate matching based on the second-class method due to a single "weak link". Only when all components have sufficient high-quality references can we dare to adopt the more accurate but also more fragile second-class method.

[0084] Example: The target T3 to be processed is of type {woodland, road, salt field}. There is only one consistent data packet, which does not meet any of the aforementioned conditions. Upon investigation, there are 8 manually associated packets for "woodland", 10 for "road", but only 1 manually associated packet for "salt field" (<2). Due to insufficient knowledge of the "salt field" type, the system determines to use a single identification method for T3. It will search for all data packets containing {woodland}, {road}, {salt field} and their combinations (such as {woodland}, {road}, {woodland, road}, {salt field}, etc.), learn the common boundary features of each type, and then combine them to process T3.

[0085] The method described in this embodiment implements an intelligent processing pipeline that "tailors its approach to individual needs." It completely changes the rigid model of fixed algorithms and parameters in traditional automated processing, instead dynamically tailoring the optimal processing strategy to the specific context of each target (the prevalence of its connection patterns, the quantity and quality of reference samples). Its core value lies in: Maximize knowledge utilization: Ensure that every high-quality manually processed result can be accurately reused in the most relevant and needed scenarios.

[0086] Enhanced robustness: Through a hierarchical decision-making mechanism, when faced with difficult situations such as scarce samples and rare types, it can automatically select more robust strategies, effectively reducing the risk of automated processing failure.

[0087] This embodiment aims to address how to determine the optimal execution order for numerous "tasks to be processed" in a multi-task parallel automated processing environment, in order to maximize overall processing efficiency and quality. Its core logic is to construct a multi-level, multi-dimensional dynamic priority evaluation system, comprehensively considering: 1) the timeliness requirements of the target's own processing strategy (identifying processing priorities); 2) the knowledge dependencies between targets (matching processing targets); and 3) the quality and completeness of knowledge preparation (deviations in connection layer types). Through this comprehensive evaluation, the system can intelligently prioritize the execution of those "most well-prepared, highest-value results, and those that can benefit other tasks to the greatest extent," thereby forming an efficient task scheduling mechanism, optimizing the utilization of computing resources, and promoting knowledge sharing and reuse of processing results.

[0088] Specifically, the method for determining the identification and processing order of the targets to be processed is as follows: S31 Based on the identification and processing method of the target to be processed, determine the identification and processing priority of the target to be processed; Determine the priority of basic identification processing: Explanation of keywords: "Identification and processing priority" is the basic priority level directly assigned based on the "identification and processing method" selected for the target to be processed (see the previous embodiment). It reflects the target's inherent requirement for processing timeliness.

[0089] Different processing methods correspond to different levels of knowledge dependence and processing logic. The second type of processing method heavily relies on existing high-quality human-made samples for precise matching; therefore, it is given a higher first priority, aiming to complete these "easy-to-succeed, high-yield" tasks as quickly as possible to rapidly expand the high-quality sample library. The first type of processing method relies more on learning general features from a broad range of samples, requiring a larger amount of reference data; therefore, it is given a lower second priority.

[0090] Significance: It establishes a first baseline for sorting, ensuring that tasks with more efficient processing strategies and higher value are prioritized, which aligns with the basic optimization principle of "easy first, difficult later, and high-yield first".

[0091] Example: Target Ta, using a binary method (based on precise template matching), is assigned "first priority"; Tb, using a single method (based on general feature learning), is assigned "second priority". In the initial ranking, Ta should be placed before Tb.

[0092] S32 determines the target to be processed as a target to be processed belonging to a data packet that meets the requirements based on the similarity of the boundary association data between the target to be processed and other targets to be processed, and uses it as a matching processing target; "Matching targets" refers to other targets that are highly similar to the current target in terms of "boundary association data" (i.e., connection layer type). Because these targets deal with similar problems, their processing methods or results may influence or reference each other.

[0093] Geoprocessing tasks are not isolated. Experience gained from processing one objective (such as model parameter optimization and feature extraction results) may be directly applicable to similar objectives. Identifying these groups of objectives with potential "knowledge-sharing" relationships is the foundation for collaborative scheduling.

[0094] Significance: Identifying the network of relationships between tasks. This allows the system to plan from the perspective of "task groups" rather than individual tasks, prioritizing those goals that pave the way or provide key references for a large number of other tasks within their "group," achieving a synergistic effect of "handling one, driving a whole group."

[0095] Example: The connection type of the target Tc to be processed is {woodland, road}. Upon searching, it was found that there are 8 other targets to be processed (Td1-Td8) that are also of this type or highly similar type (such as {woodland}, {road}, {woodland, road, ditch}). Therefore, these 8 targets are the "matching targets" of Tc.

[0096] S33 determines the recognition and processing order of the target to be processed based on the recognition and processing priority of the target to be processed, the matching target of the target to be processed, and the deviation of the connection layer type between the target to be processed and the matching target.

[0097] It is understood that the recognition processing priority is determined according to the recognition processing method. When the recognition processing method is a type I processing method, its recognition processing priority is the second priority. When the recognition processing method is a type II processing method, its recognition processing priority is the first priority.

[0098] Specifically, the first priority is greater than the second priority.

[0099] Specifically, the identification and processing order of the target to be processed is determined based on the recognition and processing priority of the target to be processed, the matching target of the target to be processed, and the deviation of the connection layer type between the target to be processed and the matching target. This includes: S331 first processes the target to be processed with high recognition processing priority. If the recognition processing priorities of the target to be processed are consistent, the number of matching targets of the target to be processed is determined. It is then determined whether the number of matching targets of the target to be processed is greater than a preset threshold for the number of matching targets. If yes, the sub-priority of the target to be processed is determined to be the first level. If no, proceed to the next step. Preliminary screening based on the number of associated tasks: Logical judgment: Among tasks with the same basic priority, prioritize the task with more "matching processing targets".

[0100] A task with a large number of matching targets implies a wide potential impact and high shared value in its processing results. Prioritizing its completion allows the knowledge generated (such as trained specialized models and optimized features) to benefit more subsequent tasks as quickly as possible, thereby maximizing the knowledge spillover effect of a single processing operation and improving overall process efficiency. This embodies the scheduling concept of global optimization.

[0101] Example: Both T1 and T2 are "first priority". T1 has 15 matching targets, and T2 has 5 matching targets. Since 15 > the preset threshold for the number of matching targets (e.g., 10), T1 is assigned a "first priority" sub-priority and should be executed before T2.

[0102] S332 determines whether the identification processing methods of the matching processing target of the target to be processed are all of the same type of processing method. If yes, the sub-priority of the target to be processed is determined to be the third level. If no, proceed to the next step. Negative filtering based on the maturity of related tasks: Logical judgment: If all matching targets of a task adopt a lower priority processing method, then the sub-priority of the task itself will be downgraded.

[0103] The matching targets employ a class of methods, indicating that they have low dependence on high-precision samples and require longer processing paths. Even if the current task is completed, its high-precision results may offer limited direct and immediate assistance to these matching targets, as they are not urgently awaiting accurate templates. Therefore, the urgency and synergistic value of the current task are relatively reduced, and it is downgraded. This avoids prematurely advancing a task for "potential beneficiaries" with weak relevance.

[0104] Example: T3 is of "first priority", and all 5 matching targets of it are processed using "one type of method". According to this rule, the sub-priority of T3 is determined to be the lowest, "third level".

[0105] S333 identifies matching processing targets with a processing priority less than or equal to the target to be processed as available processing targets, and determines whether the proportion of available processing targets in the number of matching processing targets is greater than a preset threshold for the proportion of processing targets. If yes, proceed to the next step; otherwise, determine the sub-priority of the target to be processed as the second level. Secondary evaluation based on the proportion of reusable tasks: Among the matching targets, those with a basic priority no higher than the current task are selected and called "available processing targets". Their proportion among the matching targets is calculated.

[0106] "Available processing targets" refer to tasks whose results can be reused immediately after the current task is completed (because they have the same or lower priority, meaning they have not yet been processed or are about to be processed). A high proportion indicates a high probability that "knowledge can be used immediately" after the current task is completed. This is a key indicator for measuring the "immediate value" of task processing.

[0107] Example: T4 is "Second Priority" and has 10 matching targets. Among them, 7 matching targets also have a priority of "Second Priority" or "First Priority" (those with a higher priority than T4 are unavailable), and 3 are "Third Priority" (those with a lower priority than T4 are available). The availability rate is only 30%. If the preset threshold for the percentage of targets processed is 50%, then 30% < 50%, and T4 is assigned a "Second Level" sub-priority.

[0108] S334 determines the matching target whose connection layer types do not all fall into the connection layer types of the matching target based on the deviation between different matching processing targets and the target to be processed, and takes it as the matching deviation target. Based on the proportion of the matching deviation targets in the matching processing targets, the sub-priority of the target to be processed is determined.

[0109] It is understood that if the proportion of the number of the matching deviation targets in the matching processing targets is greater than a preset deviation proportion threshold, then the sub-priority of the target to be processed is determined to be the second priority; if the proportion of the number of the matching deviation targets in the matching processing targets is not greater than the preset deviation proportion threshold, then the sub-priority of the target to be processed is determined to be the first priority.

[0110] Final ruling based on completeness of knowledge coverage: Logical Judgment: This is the finest layer. Analyze the deviation of each matching target from the current task in terms of "Connection Layer Type". Identify those matching targets that do not fully cover all connection types of the current task; these are called "Match Deviation Targets". Calculate their proportion.

[0111] If the proportion of biased targets is high, it indicates that most similar tasks are not entirely consistent with the current task in terms of problem structure. There may be noise affecting the accuracy of extracting biased targets due to recognition bias in the layer segmentation. In this case, the current task should be prioritized, as the boundary feature knowledge of these biased tasks has a greater impact when bias exists, thus its value is lower. Conversely, if the proportion of biased targets is low, it indicates that similar tasks are highly homogeneous with the current task. The processing of the current task is more of a "repetitive verification," and the impact of extraction bias is smaller, resulting in relatively higher value.

[0112] Example: T5 is of "second priority" and has 8 matching targets. T5 itself is of type {A, B, C}. Upon comparison, 6 matching targets are of type {A, B}, {A, C}, or {B}, none of which completely cover {A, B, C}, and are considered "mismatched targets", accounting for 75%. If the preset mismatch percentage threshold is 60%, since 75% > 60%, T5 is given a higher "second-level" sub-priority. This is because processing T5 can simultaneously provide these 6 tasks with the boundary processing references for the missing C, B, or A types.

[0113] It should be noted that the first priority is greater than the second priority, the second priority is greater than the third priority, and when the sub-priorities are the same, the identification and processing order is determined from high to low based on the number of matching targets.

[0114] It should be noted that the target to be processed after the target to be processed refers to the target to be processed whose processing order is after the target to be processed.

[0115] Specifically, the method for determining the shared processing method for the target to be processed is as follows: This embodiment aims to address a key issue: after an objective (referred to as the "current task") completes automated identification and processing, can its output (such as the delineated farmland patches and their boundary features) be safely and reliably shared directly with subsequent tasks ("later objectives")? Or is it necessary to manually review and correct the current task's processing results before sharing? The core decision-making logic is to proactively assess the "external dependency risk" of the current task's output. By analyzing which later tasks will rely on the current task's output and assessing the processing status and preparation of these dependents, the system can determine whether potential defects in the current task's output will have a widespread impact on the external world, thus intelligently deciding whether a manual quality control step needs to be inserted before the output is released. The goal is to minimize unnecessary human intervention while ensuring the quality of downstream task processing, achieving a balance between efficiency and reliability in results sharing.

[0116] S41 uses the later-stage targets to be processed of the target to be processed to determine the matching processing targets that belong to the target to be processed among the later-stage targets to be processed, and treats them as unidentified targets. Identify unidentified targets (dependency identification): Keyword Explanation: "Unidentified Targets" specifically refers to those late-stage tasks that have not yet begun processing and have been determined by the system to be highly similar to the "Current Task" in terms of boundary association characteristics (i.e., belonging to the "matching processing targets" of the current task). They are potential direct users or beneficiaries of the results of the current task.

[0117] Not all subsequent tasks depend equally on the current task. Only those tasks that address similar problems (matching processing objectives) are most likely to reuse the processing logic, model parameters, or feature templates of the current task. This step aims to accurately identify the true group of dependents, avoid including irrelevant tasks in the quality impact assessment, and clarify the boundaries of responsibility for sharing results. This forms the basis for subsequent risk assessments and decisions, ensuring that the analysis focuses on the most likely affected stakeholders.

[0118] Example: Task C1 (connection type {woodland, road}) is about to be completed. In the queue of tasks to be processed, there are tasks D1, D2, D3...D10. After comparison, D2, D5, D7, and D9 are matching targets of C1 (similar types). Therefore, {D2, D5, D7, D9} are the "unidentified targets" of C1, and they are very likely to use the results of C1 in the future.

[0119] S42 determines the identification data of the unidentified target that belongs to the data packet that meets the requirements, according to the identification processing method of the unidentified target, and takes the unidentified target that belongs to the data packet that meets the requirements as the identified target; Obtain the identification completion target (dependency readiness assessment): Explanation of keywords: "Identifying completed targets" refers to other tasks that are similar to (meet) a certain "unidentified target" and have already been completed. These are existing, referable knowledge sources for that unidentified target.

[0120] Assessing the risk of a dependent party (unidentified objective) requires more than just looking at its dependence on the current task (C1); it also requires considering whether it has other alternative sources of knowledge. If it already has multiple high-quality objectives to refer to, then its dependence on the results of the current task C1 is low; otherwise, its dependence is high.

[0121] Significance: This quantifies the "dependency vulnerability" of each dependent party to the current task. This is a key dimension for assessing the potential impact of errors in the current task's outcome. The fewer reference samples a dependent party has, the weaker its ability to withstand errors in the current task's outcome.

[0122] Example: For an unidentified target D2, the system finds that, in addition to C1 which is about to be completed, three other tasks similar to D2 (A2, B2, E2) have already been completed with high quality. Therefore, {A2, B2, E2} are the "identified completed targets" of D2. D2 has multiple reference sources.

[0123] S43 determines the shared processing method for the target to be processed based on the unidentified targets and the identified target data of the unidentified targets.

[0124] Furthermore, based on the unidentified targets and the identified targets of the unidentified targets to be processed, a shared processing method for the target to be processed is determined, specifically including: S431 determines whether the target to be processed exists. If it belongs to the cultivated map layer among the already processed targets and has the image features of the dividing line between the connecting layer type of the target to be processed, but has not been divided, then it is determined that the target to be processed needs to be manually divided twice before sharing the processing. If not, proceed to the next step. Integrity check based on its own results (internal defect scan): Logical judgment: Check whether there are obvious missed areas in the processing results of the current task C1 itself. That is, on the processed cultivated map layer, are there some boundary segments whose image features are highly similar to other similar boundaries that C1 has successfully divided, but the model failed to divide them.

[0125] This is a baseline quality check. If the current task's output contains obvious, automatically detectable low-level errors (such as missing large areas of clearly defined features), then the output does not meet the shared basic quality standards and must be manually corrected. This prevents obviously defective outputs from being passed down downstream.

[0126] Example: After C1 processing, the system automatically scans the boundaries between cultivated land patches and roads. Through feature comparison, it is found that a continuous section of road boundary with typical "hardened edge" characteristics on the image was incorrectly retained as cultivated land by the model. The system determines that C1 has an internal defect, and its results must undergo manual secondary subdivision processing to correct this error before being shared.

[0127] S432 determines whether the number of unidentified targets of the target to be processed is greater than a preset value for the number of unidentified targets. If yes, it determines that the target to be processed needs to be manually divided into secondary segments before sharing processing. If no, it proceeds to the next step. Risk assessment based on the number of dependent parties (judging the breadth of impact): Logical judgment: If the current task itself has no obvious defects, assess the number of "unidentified targets". If the number exceeds the preset value for the number of unidentified targets (e.g., 5), then the potential impact of the current task's results is considered to be very broad.

[0128] Risk is directly proportional to the scope of its impact. Even if the current task's outcome has only a minor deviation, if it is reused in many subsequent tasks, this minor deviation can be amplified and spread, leading to widespread systemic errors. Therefore, for tasks involving "critical suppliers" with a wide impact, adopting a more cautious strategy (increasing manual verification) is reasonable and constitutes a risk mitigation measure.

[0129] Example: C1 itself is flawless, but it has 8 unidentified targets (>5). The system considers C1 a "critical knowledge node," and the quality of its output is crucial. To ensure that the downstream 8 tasks are not misled, it is decided to conduct a second manual review of C1's output before sharing it.

[0130] S433 uses the identified target data of different unidentified targets to determine the number of unidentified targets that have been identified. Based on the number of identified targets, it determines whether there are reference deviation targets among the unidentified targets. If so, proceed to the next step. If not, it determines that the target to be processed does not need to undergo manual secondary division processing before sharing processing. It should be noted that the reference deviation target is the number of uncompleted targets whose number of completed targets is less than a preset threshold for the number of completed targets.

[0131] S434 determines the reference deviation weight coefficient of the reference deviation target by the number of targets identified by the reference deviation target and the proportion of data packets that meet the requirements in the reference deviation target. It then determines whether the sum of the reference deviation weight coefficients of different reference deviation targets is greater than a preset deviation weight coefficient threshold. If so, it is determined that the target to be processed needs to undergo manual secondary division processing before sharing processing. If not, it is determined that the target to be processed does not need to undergo manual secondary division processing before sharing processing.

[0132] A detailed assessment based on the vulnerability of dependent parties (impact on depth assessment:) Logical judgment: When the impact is not significant, conduct a deep assessment. First (S433), identify those "reference deviation targets"—that is, unidentified targets with a very small number of "identified completed targets" (less than the preset threshold for the number of completed targets, such as 2). These targets are highly dependent on the current task C1, have weak knowledge reserves, and are the most vulnerable.

[0133] The fewer reference samples a dependent party has, the worse its ability to "correct errors" is, and the more likely it is to accept the current task results (including possible errors) in their entirety.

[0134] Then (S434), not all reference deviation targets are equally important. Calculate the "reference deviation weight coefficient" for each reference deviation target. This coefficient is determined by the ratio of its "number of identified targets" to the "total number of data packets that meet its requirements." The lower the ratio, the more "isolated" the target is, and the higher the weight coefficient.

[0135] The weighting coefficient quantifies the degree of vulnerability. A target with no other references (0%) has a very high weight; a target with one or two references (20%) has a slightly lower weight.

[0136] Final decision: Sum the weight coefficients of all reference deviation targets. If the sum exceeds a preset deviation weight coefficient threshold (e.g., 1.5), it indicates the existence of one or more extremely vulnerable downstream targets that are highly dependent on the current task. Any error in the current task's outcome could have a "fatal" impact on them. Therefore, the current task must be manually re-divided to protect these vulnerable downstream targets.

[0137] Significance: Achieving precise risk control. This level of assessment avoids a "one-size-fits-all" approach; human intervention is only triggered when an error in the current task could seriously harm downstream tasks that lack resilience. This minimizes unnecessary manual checks, allowing valuable human resources to be used on truly high-risk quality control checkpoints.

[0138] Example: C1 has 3 unidentified targets: D2, D5, and D7.

[0139] D2 has 3 targets that have been identified (with sufficient references, and are not fragile targets).

[0140] D5 has one target that has been identified (with few references, it is a reference deviation target). Assuming there are a total of 10 data packets that meet the D5 requirements, the reference ratio of D5 is 1 / 10 = 0.1, and the weighting coefficient is high (e.g., 0.9).

[0141] D7 has 0 targets that were successfully identified (extremely fragile, reference bias targets). Its reference ratio is 0, and its weighting coefficient is extremely high (e.g., 1.0).

[0142] The sum of the weights of the vulnerable targets = 0.9 + 1.0 = 1.9 > threshold 1.5.

[0143] Decision: Since D5 and D7 are highly dependent on C1 and lack their own error correction capabilities, in order to ensure the future processing quality of D5 and D7, the results of C1 must be manually divided and processed again to ensure their high reliability before sharing.

[0144] The method described in this embodiment introduces an intelligent "quality gating" mechanism into the automated processing pipeline. It changes the traditional, crude approach of either completely trusting automated results or manually sampling all results, and instead establishes a dynamic quality inspection strategy based on risk perception. Its core value lies in: Defending against systemic risks: By identifying "critical knowledge nodes" and "vulnerable dependencies," the system can proactively deploy defenses (manual verification) at "gaps" where errors may spread on a large scale, effectively preventing the cascading amplification of automated errors in the task chain and ensuring the overall output quality of the project.

[0145] Optimize human resource input: Direct the secondary manual processing (the most costly step) precisely to the results with the highest risk and greatest impact, avoiding the waste of manpower on low-risk results and achieving the optimal balance between quality control costs and benefits.

[0146] Enhancing System Credibility: This mechanism imbues the entire automated system with a sense of responsibility. It not only produces results but also assesses the potential external impact of its outputs and proactively takes safeguards. This significantly increases user trust in the entire automated process and promotes deeper and more harmonious human-machine collaboration.

[0147] Forming a quality feedback loop: The results of manual secondary processing (whether confirming correctness or correcting errors) themselves become high-quality new samples, which can be fed back into the knowledge base to improve future automated processing and risk assessment models, forming a continuous improvement quality enhancement loop.

[0148] This method is a prime example of successfully applying the "prevention-oriented" quality management philosophy to the intelligent production process of geographic information, and is a key technological guarantee for achieving large-scale, automated production without sacrificing accuracy.

[0149] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0150] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0151] The above description is merely one or more embodiments of this specification and is not intended to limit this specification. Various modifications and variations can be made to the one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of this specification.

Claims

1. A method for sharing and processing survey data, characterized in that, Specifically, it includes: Based on the identification and processing results of the cultivated map layer, the boundary association data between the cultivated map layer and different layers in the data packet is determined. Based on the deviation of the boundary association data of the cultivated map layer, the manually processed data packet in the data packet is determined. Based on the similarity of the boundary association data between the manually processed data packet and the manually processed data packet, the identification and processing method for removing the manually processed data packet is determined. The data packets excluding manually processed data packets are taken as the targets to be processed. Based on the identification and processing methods of the targets to be processed and the similarity of the boundary association data between them and other targets to be processed, the identification and processing order of the targets to be processed is determined. Based on the identification and processing order and the identification and processing method, the cultivated map layer of the target to be processed is identified and processed. Then, by using the identification and processing method of the target to be processed in the later stages and the similarity of the boundary association data between the target to be processed and the target to be processed that has already been identified, the shared processing method of the target to be processed is determined.

2. The survey data sharing and processing method as described in claim 1, characterized in that, The different layers include woodland, garden land, roads, ditches, and construction land.

3. The survey data sharing and processing method as described in claim 1, characterized in that, The boundary association data is determined based on the layer type of the cultivated map layer connection of the image in the data packet.

4. The survey data sharing and processing method as described in claim 1, characterized in that, The deviation of the boundary association data of the cultivated map layer is determined based on the deviation of the layer type of the cultivated map layer connection between different data packets.

5. The survey data sharing and processing method as described in claim 1, characterized in that, The method for determining the manually processed data packets in the data packets is as follows: Based on the boundary association data of the cultivated map layer, determine the layer type connected to the cultivated map layer in the data packet and use it as the connection layer type. Group data packets with the same connection layer type into a single connection layer group; Based on the combined data of the connection layers and the connection layer type of the data packet, determine whether the data packet is a manually processed data packet.

6. The survey data sharing and processing method as described in claim 5, characterized in that, If the number of connection layer types in the data packet is less than a preset threshold for the number of connection layer types, then the data packet is determined to be a manually processed data packet.

7. The survey data sharing and processing method as described in claim 1, characterized in that, The method for determining the identification and processing method for removing manually processed data packets is as follows: The data packets excluding manually processed data packets are taken as the target to be processed. Based on the similarity of the boundary association data between the target to be processed and the manually processed data packets, data packets with the same connection layer type as the target to be processed are determined and taken as the consistent data packets. Based on the similarity situation and the connection layer type of the target to be processed, determine the manually processed data packets that meet the requirements and use them as manually associated data packets; Based on the consistent data packet data, the manually associated data packet data, and the manually processed data packet data in the consistent data packet data, the identification and processing method for the target to be processed is determined.

8. The survey data sharing and processing method as described in claim 7, characterized in that, If the number of consistent data packets is not less than a preset data packet number threshold, then the identification and processing method of the target to be processed is determined to be a type of identification method.

9. The survey data sharing and processing method as described in claim 8, characterized in that, If it belongs to a certain type of recognition method, then the image features of the connection positions between the cultivated map layer and other layers in all eligible data packets before the recognition processing of the target to be processed need to be considered.

10. The survey data sharing and processing method as described in claim 1, characterized in that, The method for determining the shared processing method for the target to be processed is as follows: Using the later-stage targets to be processed of the target to be processed, identify the matching targets that belong to the target to be processed among the later-stage targets to be processed, and treat them as unidentified targets; According to the identification and processing method for unidentified targets, the identification data of the unidentified targets that belong to the data packets that meet the requirements is determined, and the unidentified targets that belong to the data packets that meet the requirements are identified as the identified targets. Based on the unidentified targets and the identified targets of the unidentified targets to be processed, a shared processing method for the target to be processed is determined.

Images