Urban and rural marginal area identification method, device and equipment based on human-computer interaction cooperation, and medium

Abstract

The invention relates to an urban and rural marginal area identification method and device based on human-computer interaction cooperation, equipment and a medium, and relates to the technical field of urban planning. The method comprises the following steps: repairing a road and generating a buffer area according to multi-source geographic space data, removing fine land parcels to form a plurality of continuous land parcel units, and extracting urban and rural features; inputting the urban and rural features of each land block unit and the field land block unit thereof into a neighborhood feature fusion model, fusing to generate a regional correction feature of each land block unit, and constructing a data set; inputting the data set into an urban and rural marginal region recognition model to calculate a cumulative probability, and determining a statistical result; screening is carried out through uncertainty evaluation, and high-uncertainty plot units are obtained; and then clustering labeling is carried out, labeled plot units are used as incremental samples to update the data set, and the urban and rural marginal region identification model is iteratively optimized based on the updated data set. According to the method, the continuity, accuracy and reliability of the urban and rural marginal region identification result can be improved.

Description

Technical Field

[0001] This invention relates to the field of urban planning technology, and in particular to a method, apparatus, equipment and medium for identifying urban-rural fringe areas through human-computer interaction and collaboration. Background Technology

[0002] As a transitional zone between urban and rural areas, the accurate identification of urban fringe areas is a crucial foundation for supporting optimized spatial planning and promoting coordinated urban-rural development. While current methods for identifying urban fringe areas have established a basic framework involving multi-source data fusion, spatial feature quantification, and boundary extraction, they still face numerous technical limitations in practical applications.

[0003] Existing technologies mostly employ grid-based processing to quantify multi-source data, obtaining urbanization feature values ​​through weighted summation, and then combining methods such as abrupt change detection and constrained clustering to extract edge zone boundaries. While this improves recognition accuracy to some extent, grid-based segmentation easily severs the continuous semantics of urban-rural fringe areas, resulting in an insufficiently comprehensive characterization of the overall regional features. In the feature fusion stage, existing methods mostly focus on the independent quantification and weighting of single indicators, failing to fully consider the spatial correlation between neighboring plots and making it difficult to reflect the gradual characteristics and spatial dependencies of urban-rural transition zones.

[0004] Meanwhile, existing technologies lack effective quantification and control mechanisms for uncertainties in the identification process. They only address data noise by removing pseudo-mutation points and redundant edges, without addressing dynamic optimization of sample selection and model parameters. This makes the reliability of the identification results susceptible to data quality and model assumptions. Furthermore, in the boundary delineation stage, existing methods largely rely on geometric constraints or empirical thresholds, lacking statistical significance support. This makes them ill-suited to the dynamic changes in the boundaries of edge areas within complex urban structures and unable to effectively address the heterogeneous differences in urbanization processes across different regions. These issues collectively result in deficiencies in existing urban edge area identification methods regarding continuity semantics, uncertainty control, and boundary accuracy, making it difficult to meet the high-precision and dynamic requirements of refined urban governance for edge area identification. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, device, equipment, and medium for identifying urban-rural fringe areas that can take into account the semantic continuity of urban-rural fringe areas and handle the uncertainty of urban-rural fringe area samples and results through human-computer interactive collaboration, in order to address the above-mentioned technical problems.

[0006] A human-computer interactive collaborative method for identifying urban-rural fringe areas, the method comprising: Acquire multi-source geospatial data of the target area, repair roads and generate buffer zones based on the multi-source geospatial data, remove fragmented plots to form several continuous plot units, and extract the urban and rural characteristics of each plot unit; A neighborhood feature fusion model is constructed based on the kernel density field. The urban and rural features of each land parcel unit and its neighboring land parcel units are input into the neighborhood feature fusion model to generate regional correction features for each land parcel unit. A dataset is constructed based on the regional correction features and urban and rural labels. A model for identifying urban-rural fringe areas is constructed. The dataset is input into the model to calculate the cumulative probability of urbanization intensity of each plot unit in the zero distribution of urban and rural areas, and to determine the statistical results corresponding to each plot unit. Based on the statistical results, high-uncertainty land parcel units are obtained through uncertainty assessment. The high-uncertainty land parcel units are clustered and labeled, and the labeled land parcel units are used as incremental samples to update the dataset. The urban-rural fringe area identification model is iteratively optimized based on the updated dataset. The optimized urban-rural fringe area identification model outputs the spatial range of the urban-rural fringe area of ​​the target region.

[0007] On the other hand, a human-computer interactive collaborative identification device for urban-rural fringe areas is also provided, the device comprising: The urban-rural feature extraction module is used to acquire multi-source geospatial data of the target area, repair roads and generate buffer zones based on the multi-source geospatial data, remove fragmented plots, and form a series of continuous plot units; and extract the urban-rural features of each plot unit. The regional correction feature generation module is used to construct a neighborhood feature fusion model based on the kernel density field. The urban and rural features of each land parcel unit and its neighboring land parcel units are input into the neighborhood feature fusion model to generate regional correction features for each land parcel unit. A dataset is constructed based on the regional correction features and urban and rural labels. The statistical results calculation module is used to construct an urban-rural fringe area identification model. The dataset is input into the urban-rural fringe area identification model to calculate the cumulative probability of urbanization intensity of each plot unit in the zero distribution of urban and rural areas, and to determine the statistical results corresponding to each plot unit. The high-uncertainty land parcel unit determination module is used to screen high-uncertainty land parcel units based on the statistical results and through uncertainty assessment; The iterative optimization module is used to cluster and label the high-uncertainty land parcel units, update the dataset with the labeled land parcel units as incremental samples, and iteratively optimize the urban-rural fringe area identification model based on the updated dataset. The urban-rural fringe area identification module is used to output the spatial range of the urban-rural fringe area of ​​the target region through the optimized urban-rural fringe area identification model.

[0008] On the other hand, a computer device is also provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the above-mentioned human-computer interactive collaborative method for identifying urban and rural fringe areas.

[0009] Furthermore, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the steps of the aforementioned human-computer interactive collaborative method for identifying urban-rural fringe areas.

[0010] Compared with existing technologies, the human-computer interactive collaborative method, apparatus, equipment, and medium for identifying urban-rural fringe areas provided by this invention have the following beneficial effects: 1. By repairing roads, generating buffer zones, and removing fragmented plots, continuous plot units that conform to geospatial logic are constructed. This avoids the fragmentation of regional continuity semantics at the data foundation level and ensures the continuity semantics of urban-rural fringe areas. At the same time, core urban-rural characteristics are extracted. Compared with the existing single-indicator quantification model, this fully covers the key feature dimensions of urban-rural transition and effectively improves the integrity of the overall regional feature characterization.

[0011] 2. By constructing a neighborhood feature fusion model based on kernel density field, the features of each plot unit and its neighboring plots are weighted and fused to generate regional correction features, which can fully explore the spatial dependence of the neighborhood and accurately reflect the gradual transition characteristics between urban and rural areas.

[0012] 3. By constructing an urban-rural fringe identification model, the cumulative probability of urbanization intensity of each plot unit in the zero distribution of urban and rural areas is calculated. The attributes of the plot unit are judged by the cumulative probability statistics, so that the boundary definition of the fringe area has strict statistical significance support, thereby improving the reliability and universality of the boundary definition of the fringe area.

[0013] 4. By clustering analysis of high-uncertainty land parcel units, updating incremental samples, and iteratively training the model, a closed-loop mechanism of uncertainty identification, sample supplementation, and model optimization is formed. This enables the model to dynamically adapt to the urban and rural development characteristics of different regions, effectively meeting the high-precision and dynamic requirements of refined urban governance for edge area identification, and significantly improving the practicality and scalability of the technical solution. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention, and those skilled in the art can obtain other related drawings based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart illustrating a human-computer interaction collaboration method for identifying urban-rural fringe areas in one embodiment. Figure 2 This is a structural block diagram of a human-computer interactive collaborative urban-rural fringe area identification device in one embodiment; Figure 3 This is an internal structural diagram of a computer device in one embodiment.

[0016] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0018] It should be noted that in this invention, the use of terms such as "first," "second," etc., is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0019] It is understood that the technical solutions of the various embodiments of the present invention can be combined with each other, but only if they are based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0020] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0021] Example 1 like Figure 1 As shown, this embodiment provides a human-computer interactive collaborative method for identifying urban-rural fringe areas, including the following steps: Step 201: Obtain multi-source geospatial data of the target area, repair roads and generate buffer zones based on the multi-source geospatial data, remove fragmented plots to form several continuous plot units, and extract the urban and rural characteristics of each plot unit.

[0022] It is understandable that multi-source geospatial data is the fundamental data support for identifying urban-rural fringe areas. It covers multiple dimensions such as road network data, land use data, point of interest data, and remote sensing image data, and can comprehensively reflect the geospatial attributes of the target area. Road repair and buffer generation are carried out to construct a complete and continuous road space area, and then obtain land parcel units that conform to geospatial logic through differential operations, so as to avoid the interference of fragmented land parcels on subsequent feature extraction and identification results. The extraction of urban and rural features involves mining key indicators that can distinguish urban and rural attributes from core dimensions such as nature, built environment, and socio-economic activities, providing feature support for subsequent model training and identification.

[0023] Step 202: Construct a neighborhood feature fusion model based on the kernel density field. Input the urban and rural features of each land parcel unit and its neighboring land parcel units into the neighborhood feature fusion model to generate regional correction features for each land parcel unit. Construct a dataset based on the regional correction features and urban and rural labels.

[0024] It is understandable that the core of the neighborhood feature fusion model is to fully consider the spatial correlation between land parcel units. As a transitional zone, the urban-rural fringe area has obvious spatial dependence in its features, and the features of a single land parcel are difficult to fully reflect the regional attributes. By using the weighted fusion method of kernel density field, the features of the target land parcel and neighboring land parcels can be organically combined to generate more regional and representative corrected features. The dataset constructed based on the corrected features and urban-rural labels can provide high-quality training data for the urban-rural fringe area identification model, ensuring the effectiveness of model learning.

[0025] Step 203: Construct an urban-rural fringe area identification model. Input the dataset into the urban-rural fringe area identification model, calculate the cumulative probability of urbanization intensity of each plot unit in the zero distribution of urban and rural areas, and determine the statistical results corresponding to each plot unit.

[0026] It is understandable that the urban-rural fringe area identification model achieves the mapping from features to urban-rural attributes through the collaborative work of multiple modules: the urban-rural classification identification model completes the initial classification based on the random forest algorithm, the urban-rural intensity identification model quantifies the urbanization intensity through weighted voting, and the urban-rural fringe area statistical test model provides statistical significance support through cumulative probability calculation; the cumulative probability calculation of urbanization intensity can effectively distinguish the attribute boundaries of cities, rural areas and urban-rural fringe areas, avoid the subjectivity of traditional experience threshold judgment, and improve the reliability of the identification results.

[0027] Step 204: Based on the statistical results, the high-uncertainty land parcel units are obtained through uncertainty assessment.

[0028] It is understandable that uncertainty assessment is a key step in identifying land parcels with low reliability in the model results. Due to the complexity of the landscape in urban and rural fringe areas and the limitations of sample coverage, the identification results of some land parcels may be uncertain. By constructing a feature vector similarity function, uncertainty can be quantified from both single sample point and comprehensive multi-sample point levels. This can accurately locate land parcels that lack sample support or have ambiguous features, providing a clear direction for subsequent sample supplementation and model optimization.

[0029] Step 205: Cluster and label the high-uncertainty land parcel units, use the labeled land parcel units as incremental samples to update the dataset, and iteratively optimize the urban-rural fringe area identification model based on the updated dataset.

[0030] It is understandable that clustering labeling can classify high-uncertainty land parcel units according to feature similarity, reducing the workload of manual labeling while ensuring the representativeness of labeled samples; using labeled samples as incremental updates to the dataset can make up for the problem of insufficient feature coverage in the initial sample set; and the iterative optimization of the model is to continuously supplement high-quality samples to continuously improve the model's adaptability to complex urban and rural landscapes, forming a closed-loop mechanism of "identification, evaluation, supplementation, and optimization".

[0031] Step 206: Output the spatial range of the urban-rural fringe area of ​​the target region using the optimized urban-rural fringe area identification model.

[0032] It is understandable that the urban-rural fringe area identification model, after multiple rounds of iterative optimization, has high performance in feature extraction, spatial correlation fusion, and uncertainty control; the output spatial range of the urban-rural fringe area can accurately reflect the actual situation of urban-rural transition in the target area, providing high-precision spatial boundary data support for applications such as land spatial planning and coordinated urban-rural development.

[0033] The human-computer interactive collaborative method for identifying urban-rural fringe areas provided by this invention ensures the semantic continuity of the region by constructing continuous land parcel units, utilizes neighborhood feature fusion to mine spatial dependencies, improves boundary reliability based on statistical tests, and iteratively optimizes model performance through human-computer interaction. This method achieves a significant improvement in the continuity, accuracy, and reliability of urban-rural fringe area identification results, effectively meeting the needs of refined urban governance.

[0034] In the specific implementation of step 201, multi-source geospatial data of the target area is first acquired. The multi-source geospatial data includes road network data, land use data, point of interest data, and remote sensing image data.

[0035] The road network data is topologically repaired by extending unconnected road segments, pruning suspended roads and independent road segments to construct a closed road network system. Then, buffer zones are generated. The buffer zone length is preset according to the requirements. In this embodiment, a 40-meter buffer zone is generated for highways or main roads, and a 20-meter buffer zone is generated for secondary roads to obtain a complete and continuous road space area.

[0036] Then, a difference operation is performed on the geographic space of the complete continuous road space area and the target area. A threshold for the area of ​​fragmented plots is preset according to requirements, and fragmented plots with an area less than the threshold are deleted. In this embodiment, the threshold for the area of ​​fragmented plots is set to 0.1 km². After deleting fragmented plots with an area less than 0.1 km², several continuous plot units are obtained.

[0037] Finally, based on land use data, remote sensing imagery data, and point-of-interest data, the urban-rural characteristics of each continuous land parcel unit are extracted. This includes natural feature characteristics, built environment characteristics, and socioeconomic activity characteristics. Natural feature characteristics include mean elevation (DEM), mean slope (SLO), and mean surface roughness (ROU); built environment characteristics include vegetation cover (VEG), building density (BLD), and impervious surface cover (IMP); and socioeconomic activity characteristics include population density (POP), point of interest density (POI), and mean nighttime light index (NLI).

[0038] This step avoids the fragmentation of regional continuity semantics at the data preprocessing level. By accurately dividing land parcel units and comprehensively extracting multi-dimensional urban and rural features, it provides high-quality basic data for subsequent model training, effectively improving the completeness of the overall regional feature characterization.

[0039] In the specific implementation of step 202, for any plot of land to be processed... and plot units Any urban-rural characteristic A neighborhood feature fusion model is constructed based on the kernel density field. The expression for the neighborhood feature fusion model is as follows: ; In the formula, Representing land parcel units The first after fusing neighborhood features One regional correction feature; Indicates the first Individual plots The Urban and rural characteristics; Represents the Gaussian kernel function; Indicates the first Individual plots To plot unit The Euclidean distance; Indicates bandwidth parameter; Indicates the number of land parcel units.

[0040] Dynamically adjust bandwidth parameters based on the spatial distribution density of elements. The expression is: ; In the formula, Indicates sample size; Indicates the first The standard deviation of the urban-rural characteristics; IQR represents the standard deviation of the first urban-rural characteristic; Interquartile range of urban and rural characteristics.

[0041] The urban and rural characteristics of each land parcel unit and its surrounding land parcel units are input into the neighborhood feature fusion model to generate regional correction features.

[0042] Subsequently, urban and rural samples are extracted by fusing data from multiple sources to obtain corresponding urban and rural labels. Specifically, the grid side length parameter is first specified. Number of plot units This embodiment sets =10 km × 10 km =10. Collect existing urban built-up area distribution data products, overlay the plot units constructed in step 201 with the existing urban built-up area distribution, and divide the data into sampling grids of 10 km × 10 km to obtain several sampling grids S. Randomly sample within each sampling grid S. =10 plot units, among which the plot units located within the existing urban built-up area are initially determined as urban samples. Conversely, it is initially determined to be a rural sample. .

[0043] Then, existing urban functional area data products and land use / land cover data products are collected, and the reliability of the initial sample is verified using a consistency test formula. The consistency test formula is as follows: ; In the formula, Indicates the first The classification labels for each land parcel unit in the urban built-up area distribution product, urban functional area product, and land use / surface cover product are as follows: urban built-up area and urban functional area are labeled as 1, while non-urban built-up area and non-urban functional area are labeled as 0. This represents a consistency judgment function, which checks if and only if the labels from the three data sources are consistent. Otherwise, it is 0; This represents the initial proportion of samples in multi-source data where the urban / rural labels are completely consistent. A proportion threshold Rα is set; if... If ≥Rα, then the current sampling grid S region is accepted. If a plot of land is selected as a candidate sample, otherwise, sampling and testing are repeated within the sampling grid S area until the conditions are met. If the value is greater than or equal to Rα, then proceed with the extraction of candidate samples for the next sampling grid.

[0044] Finally, the extracted candidate city sample set is calculated. and candidate rural sample sets The number of samples, from and Randomly extract from each The candidate urban and rural samples constitute the urban and rural labels corresponding to the plot units.

[0045] Finally, the regional correction features of each land parcel are associated with the corresponding urban / rural labels to construct a dataset for training a model for identifying urban / rural fringe areas. .

[0046] This step, through weighted fusion of kernel density fields, fully explores the spatial dependence between neighboring plots, accurately reflects the gradual characteristics of urban-rural transition, and generates more representative regional correction features, providing higher-quality input data for subsequent model recognition.

[0047] In the specific implementation of step 203, the urban-rural fringe area identification model includes an urban-rural classification identification model, an urban-rural intensity identification model, and an urban-rural fringe area statistical test model.

[0048] Dataset Input the urban-rural classification and identification model, train it using a random forest algorithm based on classification decision trees to obtain a trained urban-rural fringe area identification model, and output the initial classification labels of each plot unit through the trained urban-rural fringe area identification model.

[0049] For each classification decision tree in the trained urban-rural fringe area identification model, the uncertainty value of each classification decision tree is calculated through the decision uncertainty function. Then, the weight of each classification decision tree is determined based on the uncertainty value. The initial classification label and weight are input into the urban-rural intensity identification model for weighted voting calculation to obtain the urbanization intensity of each plot unit.

[0050] The expression for the decision uncertainty function is as follows: ; In the formula, Indicates the first Classification decision tree When performing classification, the proportion of city classification results in its leaf nodes; Indicates the first The uncertainty value of a classification decision tree. ∈[0,1], The larger the value, the larger the decision tree. The more difficult it is to distinguish between urban landscapes and rural natural landscapes, the more... When the decision tree has a resolution of 0.5, it is completely unable to distinguish between urban and rural landscapes. It has reached its maximum value.

[0051] The weights of each classification decision tree are determined based on the uncertainty value. The expression is: .

[0052] The initial classification labels and weights are input into the urban-rural intensity identification model for weighted voting calculation, expressed as follows: ; In the formula, Representing land parcel units The intensity of urbanization; This represents the total number of classification decision trees in the trained urban-rural fringe area identification model; express The weights of each classification decision tree; This represents any plot of land to be processed. Indicates the first Classification decision trees for land parcel units The category labels are set as follows: city = 1, rural = 0. This represents a conditional function that outputs 1 when the condition is true.

[0053] Urbanization intensity and dataset The corresponding urban and rural labels are input into the statistical test model for urban-rural fringe areas. First, a zero-distribution model for urban intensity is constructed. and rural intensity zero distribution These two figures represent the empirical levels of urbanization intensity in highly confident urban or rural natural landscapes, respectively.

[0054] The one-dimensional kernel density estimation method based on [0,1] truncation is used to determine the discrete zero distribution of urban intensity. and rural intensity zero distribution Convert to continuous probability density distribution and Then calculate the cumulative probability of urbanization intensity of each plot unit in the two probability density distributions, and output the statistical results corresponding to each plot unit.

[0055] Specifically, the statistical test model for urban-rural fringe areas includes a module for calculating the cumulative probability of zero distribution of urban intensity and a module for calculating the cumulative probability of zero distribution of rural intensity, with the following expressions: ; ; In the formula, Representing land parcel units The cumulative probability in a zero distribution of urban intensity; Representing land parcel units The cumulative probability in the zero distribution of rural intensity; An interval indicator function representing the zero distribution of urban intensity; An interval indicator function representing the zero distribution of rural intensity; This represents the continuous probability density distribution after the zero-distribution transformation of urban intensity. This represents the continuous probability density distribution after the zero-distribution transformation of rural intensity; Represents probability operators; This represents a continuous value for urbanization intensity.

[0056] Then, the statistical results corresponding to each plot unit are obtained through judgment. Wherein, if If the value is greater than 0.05, then the null hypothesis that "the geographical unit belongs to the urban area" is accepted, that is, "the plot unit u0 belongs to the urban area" is considered with high confidence.

[0057] like If the value is greater than 0.05, then the null hypothesis that "the geographical unit belongs to the rural natural area" is accepted, that is, the plot unit u0 is considered to belong to the rural natural area with high confidence.

[0058] like <0.05 and If the value is less than 0.05, then both null hypotheses are rejected. In this case, it can be concluded with high confidence that "plot unit u0 belongs to the urban-rural fringe area".

[0059] like >0.05 and A value greater than 0.05 indicates that the dataset... There is a serious quality issue. The data needs to be re-examined and the process in step 202 needs to be repeated to rebuild the dataset.

[0060] This step ensures the reliability of the samples through multi-source data verification, and provides strict statistical significance support for the identification results through multi-module collaboration and statistical testing, effectively improving the reliability and universality of the boundary definition of the edge region, while also having the function of self-checking the reliability of the samples.

[0061] In the specific implementation of step 204, firstly, based on the statistical results, data that does not belong to the dataset is... The plot units to be evaluated will belong to the dataset. The plot units in the data are used as the benchmark plot units.

[0062] Then, a similarity function of urban and rural feature vectors is constructed. This function is used to calculate the uncertainty of the land parcel to be evaluated relative to a single benchmark land parcel, yielding a single-sample point uncertainty metric. The expression for the urban and rural feature vector similarity function is as follows: ; In the formula, D( ) represents the similarity function between urban and rural feature vectors; Indicates the land parcel unit to be evaluated; Indicates the benchmark parcel unit; Indicates the land parcel unit to be evaluated The One regional correction feature; Representing the benchmark parcel unit The One regional correction feature; , Represents the first unit among all land parcels. The maximum and minimum values ​​of each regional correction feature.

[0063] The uncertainty of the land parcel to be evaluated relative to a single benchmark land parcel is calculated using the urban-rural feature vector similarity function, expressed as follows: ; In the formula, This represents a measure of uncertainty for a single sample point. This represents the mean calculation function. In this embodiment, the urban and rural feature vectors include DEM, SLO, ROU, VEG, BLD, IMP, POP, POD, and NLI. It can be understood that... The larger the value, the more difficult it is for the model to be based on known benchmark parcel units. Inferring the land parcel unit to be evaluated The greater the uncertainty, the better.

[0064] Then, based on the single-sample point uncertainty metric... Calculate the similarity metric between the land parcel to be evaluated and all benchmark land parcels, and take the minimum value among the similarity metrics as the uncertainty of the outcome for the land parcel to be evaluated. The expression is as follows: ; In the formula, This indicates uncertainty about the outcome. This is understandable. The smaller the value, the better the evaluation unit. With dataset The greater the similarity, the higher the certainty of the model's inference; conversely, the less similar the assessed land parcels, the lower the certainty of the inference. The lack of representative samples to support model learning results in significant uncertainty in the current model.

[0065] Set an uncertainty tolerance threshold. The uncertainty of the result Greater than the tolerance threshold The land parcels to be evaluated were identified as high-uncertainty land parcels, forming a set of high-uncertainty land parcels. Set of highly uncertain land parcel units This indicates that, given an uncertainty tolerance threshold Below is the current sample dataset. The spatial region corresponding to the feature space that was not fully covered.

[0066] This step, through a quantitative uncertainty assessment method, accurately identifies land parcels with lower reliability in the identification results, providing a clear target for subsequent human-computer interaction optimization. At the same time, this assessment method provides a quantitative basis for sample optimization, which helps to achieve continuous improvement in model performance.

[0067] In the specific implementation of step 205, the set of high uncertainty land parcel units is first... Clustering and labeling were performed using Mahalanobis distance as the metric and the k-means clustering algorithm. Divided into Clusters, in this embodiment =20; each cluster's block units have similar urban and rural feature vectors, and lack sample support.

[0068] For the central plot unit of each cluster Through on-site investigation or in combination with other information, the urban-rural type is manually determined: if the labeling can be reliably done manually, then the central plot unit will be classified as such. After annotation, merge into the sample dataset In the middle; if it is difficult to reliably identify the central plot unit manually. For the urban-rural classification, the central plot unit will be temporarily ignored. The above methods are used to determine... =20 cluster center plot units, realizing the dataset Incremental updates.

[0069] Based on the incrementally updated dataset The urban-rural fringe area identification model in step 203 is reconstructed, and the uncertainty of the new urban-rural fringe area identification results is evaluated.

[0070] Iterate the above process until the uncertainty of all land parcel units is below the uncertainty tolerance threshold. Or, in a certain round, it may be impossible to continue updating samples incrementally based on manual methods, or the maximum number of labeled samples that the user can bear may be reached. (In this embodiment) =1000), stop iterating, and obtain the optimized urban-rural fringe area identification model.

[0071] This step reduces the workload of manual annotation through clustering and ensures the representativeness of incremental samples. Through iterative model training, a closed-loop mechanism of "uncertainty identification, sample supplementation, and model optimization" is formed, enabling the optimized urban-rural fringe area identification model to dynamically adapt to the urban-rural development characteristics of different regions, significantly improving the model's generalization ability and recognition accuracy.

[0072] In the specific implementation of step 206, the land parcel unit features of the target area are input into the urban-rural fringe area identification model after multiple rounds of iterative optimization. The model outputs the urban-rural attribute determination results of each land parcel unit through processes such as urban-rural classification identification, urbanization intensity calculation and statistical verification, and finally integrates them to obtain the spatial range of the urban-rural fringe area of ​​the target area.

[0073] The optimized urban-rural fringe area identification model in this step fully integrates the advantages of multi-source data, spatial correlation characteristics, and high-quality samples supplemented by human-computer interaction. The output spatial range of urban-rural fringe areas has high continuity, accuracy, and reliability, and can effectively meet the high-precision and dynamic requirements of urban-rural fringe area identification in scenarios such as refined urban governance and land spatial planning.

[0074] It should be understood that, although this embodiment Figure 1 The steps are shown sequentially as indicated by the arrows, but they are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order in which these steps are performed; they can be executed in other orders. Figure 1 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0075] Example 2 Based on the human-computer interaction collaborative method for identifying urban-rural fringe areas in Embodiment 1, this embodiment discloses a human-computer interaction collaborative device for identifying urban-rural fringe areas, such as... Figure 2 As shown, the human-computer interactive collaborative urban-rural fringe area identification device includes: an urban-rural feature extraction module 401, a regional correction feature generation module 402, a statistical result calculation module 403, a high-uncertainty land parcel unit determination module 404, an iterative optimization module 405, and an urban-rural fringe area identification module 406, wherein: The urban and rural feature extraction module 401 is used to acquire multi-source geospatial data of the target area, repair roads and generate buffers based on the multi-source geospatial data, remove fragmented plots, and form a number of continuous plot units; and extract the urban and rural features of each plot unit.

[0076] The regional correction feature generation module 402 is used to construct a neighborhood feature fusion model based on the kernel density field. It inputs the urban and rural features of each land parcel unit and its surrounding land parcel units into the neighborhood feature fusion model to generate regional correction features for each land parcel unit. The dataset is constructed based on the regional correction features and urban and rural labels.

[0077] The statistical results calculation module 403 is used to construct the urban-rural fringe area identification model. The dataset is input into the urban-rural fringe area identification model to calculate the cumulative probability of urbanization intensity of each plot unit in the zero distribution of urban and rural areas, and to determine the statistical results corresponding to each plot unit.

[0078] The high-uncertainty land parcel unit determination module 404 is used to screen high-uncertainty land parcel units based on statistical results and uncertainty assessment.

[0079] The iterative optimization module 405 is used to cluster and label high-uncertainty land parcel units, use the labeled land parcel units as incremental samples to update the dataset, and iteratively optimize the urban-rural fringe area identification model based on the updated dataset.

[0080] The urban-rural fringe area identification module 406 is used to output the spatial range of the urban-rural fringe area of ​​the target region through the optimized urban-rural fringe area identification model.

[0081] In this embodiment, the specific working processes and principles of the urban-rural feature extraction module 401, the regional correction feature generation module 402, the statistical result calculation module 403, the high-uncertainty land parcel unit determination module 404, the iterative optimization module 405, and the urban-rural fringe area identification module 406 are the same as those in Embodiment 1, and therefore will not be described again in this embodiment. Each unit module can be implemented entirely or partially through software, hardware, or a combination thereof. Each unit module can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each of the above unit modules.

[0082] Example 3 like Figure 3The diagram illustrates a terminal device disclosed in this embodiment, comprising a transmitter, a receiver, a memory, and a processor. The transmitter transmits instructions and data, the receiver receives instructions and data, the memory stores computer-executed instructions, and the processor executes the computer-executed instructions stored in the memory to implement the method described in Embodiment 1 above.

[0083] It is important to note that the aforementioned memory can be either standalone or integrated with the processor. When the memory is set up independently, the terminal device also includes a bus for connecting the memory and the processor.

[0084] Example 4 This embodiment discloses a computer-readable storage medium storing computer-executable instructions. When a processor executes the computer-executable instructions, it implements the method in Embodiment 1 above.

[0085] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0086] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0087] The embodiments described above are merely examples of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A human-computer interactive collaborative method for identifying urban-rural fringe areas, characterized in that, The method includes: Acquire multi-source geospatial data of the target area, repair roads and generate buffer zones based on the multi-source geospatial data, remove fragmented plots to form several continuous plot units, and extract the urban and rural characteristics of each plot unit; A neighborhood feature fusion model is constructed based on the kernel density field. The urban and rural features of each land parcel unit and its neighboring land parcel units are input into the neighborhood feature fusion model to generate regional correction features for each land parcel unit. A dataset is constructed based on the regional correction features and urban and rural labels. A model for identifying urban-rural fringe areas is constructed. The dataset is input into the model to calculate the cumulative probability of urbanization intensity of each plot unit in the zero distribution of urban and rural areas, and to determine the statistical results corresponding to each plot unit. Based on the statistical results, high-uncertainty land parcel units were obtained through uncertainty assessment. The high-uncertainty land parcel units are clustered and labeled, and the labeled land parcel units are used as incremental samples to update the dataset. The urban-rural fringe area identification model is iteratively optimized based on the updated dataset. The optimized urban-rural fringe area identification model outputs the spatial range of the urban-rural fringe area of ​​the target region.

2. The method for identifying urban-rural fringe areas through human-computer interaction collaboration according to claim 1, characterized in that, Acquire multi-source geospatial data of the target area, repair roads and generate buffer zones based on the multi-source geospatial data, and remove fragmented plots to form a series of continuous plot units. Extract the urban-rural characteristics of each parcel unit, including: Acquire multi-source geospatial data of the target area, including road network data, land use data, point of interest data, and remote sensing image data; The road network data is topology repaired and buffer generated to obtain a complete and continuous road space region; By performing a difference operation on the geographic space of the complete continuous road space area and the target area, and deleting fragmented plots, several continuous plot units are obtained. Based on the land use data, the remote sensing image data, and the point of interest data, urban and rural characteristics of each continuous land parcel unit are extracted. These urban and rural characteristics include natural base features, built environment features, and socio-economic activity features.

3. The method for identifying urban-rural fringe areas through human-computer interaction and collaboration according to claim 1, characterized in that, The expression for the neighborhood feature fusion model is: ； In the formula, Representing land parcel units The first after fusing neighborhood features One regional correction feature; Indicates the first Individual plots The Urban and rural characteristics; Represents the Gaussian kernel function; Indicates the first Individual plots To plot unit The Euclidean distance; Indicates bandwidth parameter; Indicates the number of land parcel units.

4. The method for identifying urban-rural fringe areas through human-computer interaction and collaboration according to claim 1, characterized in that, The urban-rural fringe area identification model includes an urban-rural classification identification model, an urban-rural intensity identification model, and an urban-rural fringe area statistical test model. The dataset is input into the urban-rural classification and identification model, and a trained urban-rural fringe area identification model is obtained by training through the random forest algorithm, and the initial classification labels of each plot unit are output. For each classification decision tree in the trained urban-rural fringe area identification model, the uncertainty value of each classification decision tree is calculated through the decision uncertainty function, and the weight of each classification decision tree is determined according to the uncertainty value. The initial classification label and weight are then input into the urban-rural intensity identification model for weighted voting calculation to obtain the urbanization intensity of each plot unit. The urbanization intensity and the corresponding urban and rural labels of the dataset are input into the statistical test model of the urban-rural fringe area. First, the zero distribution of urban intensity and the zero distribution of rural intensity are constructed and converted into continuous probability density distributions. Then, the cumulative probability of urbanization intensity of each plot unit in the two probability density distributions is calculated, and the statistical results corresponding to each plot unit are output.

5. The method for identifying urban-rural fringe areas through human-computer interaction collaboration according to any one of claims 1 to 4, characterized in that, The initial classification labels and weights are input into the urban-rural intensity identification model for weighted voting calculation, expressed as follows: ； In the formula, Representing land parcel units The intensity of urbanization; This represents the total number of classification decision trees in the trained urban-rural fringe area identification model; express The weights of each classification decision tree; This represents any plot of land to be processed. Indicates the first Classification decision trees for land parcel units Category tags; This represents a conditional function.

6. The method for identifying urban-rural fringe areas through human-computer interaction collaboration according to claim 5, characterized in that, The statistical test model for urban-rural fringe areas includes modules for calculating the cumulative probability of zero distribution of urban intensity and modules for calculating the cumulative probability of zero distribution of rural intensity, with the following expressions: ； ； In the formula, Representing land parcel units The cumulative probability in a zero distribution of urban intensity; Representing land parcel units The cumulative probability in the zero distribution of rural intensity; An interval indicator function representing the zero distribution of urban intensity; An interval indicator function representing the zero distribution of rural intensity; This represents the continuous probability density distribution after the zero-distribution transformation of urban intensity. This represents the continuous probability density distribution after the zero-distribution transformation of rural intensity; Represents probability operators; This represents a continuous value for urbanization intensity.

7. The method for identifying urban-rural fringe areas through human-computer interaction collaboration according to any one of claims 1 to 4, characterized in that, Based on the statistical results, high-uncertainty land parcel units were obtained through uncertainty assessment and screening, including: Based on the statistical results, land parcels that do not belong to the dataset are designated as land parcels to be evaluated, while land parcels that belong to the dataset are designated as benchmark land parcels. A similarity function of urban and rural feature vectors is constructed, and the uncertainty of the land parcel unit to be evaluated relative to a single benchmark land parcel unit is calculated through the similarity function of urban and rural feature vectors to obtain a single sample point uncertainty measure value; Based on the single-sample point uncertainty metric, the similarity metric between the land parcel unit to be evaluated and all benchmark land parcel units is calculated, and the minimum value among all similarity metric values ​​is taken as the result uncertainty of the land parcel unit to be evaluated. An uncertainty tolerance threshold is set, and land parcels whose uncertainty exceeds the tolerance threshold are identified as high-uncertainty land parcels, forming a set of high-uncertainty land parcels.

8. A human-computer interactive collaborative identification device for urban-rural fringe areas, characterized in that, The device includes: The urban-rural feature extraction module is used to acquire multi-source geospatial data of the target area, repair roads and generate buffer zones based on the multi-source geospatial data, remove fragmented plots, and form a series of continuous plot units; and extract the urban-rural features of each plot unit. The regional correction feature generation module is used to construct a neighborhood feature fusion model based on the kernel density field. The urban and rural features of each land parcel unit and its neighboring land parcel units are input into the neighborhood feature fusion model to generate regional correction features for each land parcel unit. A dataset is constructed based on the regional correction features and urban and rural labels. The statistical results calculation module is used to construct an urban-rural fringe area identification model. The dataset is input into the urban-rural fringe area identification model to calculate the cumulative probability of urbanization intensity of each plot unit in the zero distribution of urban and rural areas, and to determine the statistical results corresponding to each plot unit. The high-uncertainty land parcel unit determination module is used to screen high-uncertainty land parcel units based on the statistical results and through uncertainty assessment; The iterative optimization module is used to cluster and label the high-uncertainty land parcel units, update the dataset with the labeled land parcel units as incremental samples, and iteratively optimize the urban-rural fringe area identification model based on the updated dataset. The urban-rural fringe area identification module is used to output the spatial range of the urban-rural fringe area of ​​the target region through the optimized urban-rural fringe area identification model.

9. A computer device, comprising a memory and a processor, characterized in that, The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the human-computer interaction collaborative method for identifying urban-rural fringe areas as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores a computer program, which, when executed by a processor, implements the steps of the human-computer interaction collaborative method for identifying urban-rural fringe areas as described in any one of claims 1 to 7.

Images