An agent-based method for land use determination and processing in territorial spatial planning

Abstract

This invention relates to the technical field of land planning and discloses a method for determining and processing land use planning purposes based on intelligent agents. The method includes: performing parcel-level spatial correlation on legally managed data and spatially perceived data to screen parcels with legally stipulated construction land use as candidate land parcels; extracting the pre-development disturbance characteristics and legal status time-series codes of the candidate land parcels; using a construction legality identification model to determine whether the candidate land parcels have the legality of construction land development; generating initial land use labels and surrounding construction environment neighborhood characteristics of the candidate land parcels; and using a construction use correction model to determine the planned use labels of the candidate land parcels and the land use planning determination result. This invention intelligently filters approved but undeveloped candidate land parcels and determines the legality of construction land development through parcel-level multi-source data correlation, and outputs the land use planning determination result after intelligent correction, thus achieving intelligent processing for determining land use planning purposes.

Description

Technical Field

[0001] This invention relates to multi-agent systems, particularly to the field of land planning, and specifically to a method for determining and processing the use of land space planning based on agents. Background Technology

[0002] Territorial spatial planning is a crucial foundation for the modernization of the national spatial governance system, undertaking the core task of coordinating the layout of various spatial functions such as ecological protection, agricultural production, and urban construction. The accurate determination of land use is a core prerequisite for planning, land use control, resource allocation, and subsequent supervision. Determining land use requires integrating and analyzing multi-source data, including land surveys, approval and registration, land supply management, and remote sensing monitoring, according to unified planning classification rules. This results in compliant, reasonable, and traceable land use labels, forming the core basis for supporting planning implementation and land use control. The accuracy of land use determination not only determines the effectiveness of urban and rural land use structure layout, construction land classification statistics, and development boundary control, but also directly affects the implementation of the arable land protection red line, the revitalization of existing land, the approval of new land use, and the scientific and legal nature of the management of natural resource assets across the entire region.

[0003] In existing technologies, some research and patents have attempted to automate and intelligently upgrade land use management. For example, patent CN118521182B discloses a land planning adjustment auxiliary system based on remote sensing data. This system includes an image marker recognition module, a data analysis module, a conflict prediction module, and a planning adjustment module. By extracting vegetation index, hydrological cycle data, and soil quality indicators and conducting time series analysis, it completes the assessment of land restoration cycle and quality change, dynamically monitors land use evolution patterns, predicts potential land use conflicts, and optimizes land use plans based on the analysis results, thus coordinating ecological protection and development needs. While this type of remote sensing data-based technical solution can achieve land use status monitoring and planning scheme optimization, it has significant shortcomings in the intelligent determination of construction land use. The scheme focuses on land change trend analysis, planning conflict early warning and land use plan optimization, but does not deeply integrate legal management data such as agricultural land conversion approval, state-owned construction land supply and real estate registration. It is difficult to take into account the coordinated verification of the legal approval status of the plot, the actual development form and the planned use, and cannot effectively solve the problem of accurately determining the use of plots with inconsistent legal attributes and surface forms, such as "approved but not built" and "approved for early development".

[0004] To address this issue, this invention proposes a method for determining and processing the use of land space planning based on intelligent agents. This method utilizes artificial intelligence technology to improve the automatic identification and consistency of difficult land use in planning and construction land use, reduce the burden of manual verification, and minimize statistical distortions in construction land use, deviations in planning land use structure, and conflicts in subsequent approval and supervision caused by misjudgments. This enhances the rule of law, precision, and intelligence of land and resources management. Summary of the Invention

[0005] This invention proposes a method for determining and processing the use of land space planning based on intelligent agents. Step S1 involves collecting legally managed land resources data and spatial sensing data, and performing parcel-level spatial correlation. This achieves unified matching and spatial alignment of multi-source heterogeneous data on a parcel-by-parcel basis. Based on legal approval conditions and building identification rules, candidate plots of approved but undeveloped land are screened, solving the technical problems of scattered and misaligned multi-source data, disconnect between legal and spatial information, and low efficiency of manual screening of approved but undeveloped land parcels. Step S2 involves extracting early development disturbance features and legal status temporal codes to construct a confirmation feature vector, and using a dual-branch gating fusion model to intelligently determine the use of land. The legality of land development is established by jointly determining surface disturbance and legal procedures. Step S3 generates initial land use labels through priority rules and extracts three types of neighborhood features: road access direction, functional compatibility, and spatial morphology adaptability, to achieve a quantitative correlation between the initial land use labels and the surrounding construction environment. Step S4 optimizes the planned use labels under the allowable label deviation constraint through the construction use correction model, and correlates the legal time sequence, legality, and final use to form the result of determining the construction land use, realizing intelligent use correction and standardized output of results, and solving the technical problems of rigid and abrupt use correction and difficulty in balancing legal constraints and spatial adaptability.

[0006] To achieve the above objectives, the present invention provides a method for determining and processing the land use in land spatial planning based on intelligent agents, comprising the following steps: S1: Collect legal management data and spatial perception data of land use, perform parcel-level spatial correlation on the legal management data and spatial perception data, and select parcels that meet the legal requirements for construction land as candidate map patches that have been approved but not yet built based on the legal management data and spatial perception data after parcel-level spatial correlation, and construct a set of candidate map patches. S2: Extract the early development disturbance features and legal status time sequence codes of the candidate patches in the candidate patch set, concatenate the early development disturbance features and legal status time sequence codes to obtain the confirmation feature vector of the candidate patch, and use the construction legality recognition model to generate whether the candidate patch associated with the confirmation feature vector has the legality of construction land development. S3: Select candidate land parcels with legality for construction land development, and generate the initial land use label and surrounding construction environment neighborhood characteristics of the selected candidate land parcels based on the legal management data and spatial perception data after the land parcel spatial association. S4: Using the construction use correction model, the initial land use label and surrounding construction environment neighborhood characteristics of the candidate map patch are received, the planned use label of the candidate map patch is generated, and the legal status time sequence code, planned use label and construction land development legality of the candidate map patch are associated as the result of determining the land space planning use of the candidate map patch.

[0007] As a further improvement of the present invention: Further, in step S1, legal management data and spatial sensing data of land use are collected, and parcel-level spatial correlation is performed on the legal management data and spatial sensing data, including: S11: Extract the legal management data of the land to be planned from the land and resources database. The legal management data includes land parcel boundary vector data, agricultural land conversion approval data, state-owned construction land supply data, land use right registration data, and control detailed planning plot map data. S12: Acquire remote sensing images of the location of the land to be planned for use as spatial perception data; S13: Convert the vector data and raster data in the legal management data and spatial perception data to a unified coordinate system, and divide the land to be planned into multiple non-overlapping parcels based on the parcel boundary in the parcel boundary vector data. S14: Extract the location data of the land parcel from the legal management data as the legal management information of the land parcel; trim the spatial sensing data according to the land parcel boundary to obtain remote sensing image sub-blocks of the land parcel as the spatial sensing information of the land parcel. S15: Construct the legal management information and spatial perception information of the land parcel into a set form, which serves as the legal management data and spatial perception data after spatial association at the land parcel level.

[0008] Furthermore, step S1, based on the legally mandated management data and spatial perception data after spatial correlation at the parcel level, selects parcels that meet the legal requirements for construction land as candidate plots that have been approved but not yet developed. This also includes: Based on the legal management information of the land parcel, the legal procedural legality of the land parcel is generated, wherein the value of legal procedural legality is 0 or 1, where 0 indicates that the legal procedural requirements are not met and 1 indicates that the legal procedural requirements are met. Based on the spatial perception information of the land parcel, the area and shape regularity of the land parcel are generated. Based on the area and shape regularity, the main building identification label of the land parcel is generated according to the preset main building identification rule. The value range of the main building identification label is 0 or 1, where 0 indicates that there is no main building in the land parcel and 1 indicates that there is a main building in the land parcel. Land parcels with a legal procedure validity value of 1 and a main building identification tag value of 0 are selected as candidate land parcels that have been approved but not yet built.

[0009] Further, in step S2, the early development disturbance features and legal state temporal codes of the candidate patches in the candidate patch set are extracted, and a confirmation feature vector of the candidate patches is constructed, including: S21: Obtain the spatial perception information of the candidate patch, perform grayscale processing on the obtained spatial perception information, calculate the peak intensity of the directional gradient of the spatial perception information in each gradient direction after grayscale processing, and calculate the directional consistency parameter of the candidate patch based on the peak intensity of the directional gradient. S22: Extract the image boundary of the spatial perception information of the candidate patch, and perform line fitting on the extracted image boundary to obtain multiple sets of line segments. Connect any two sets of line segments according to the endpoint connection rule, and use the connection result as the connection fence. Calculate the ratio between the total length of the connection fence and the minimum perimeter of the bounding rectangle of the spatial perception information of the candidate patch, and use it as the fence closure parameter of the candidate patch. S23: Calculate the mean gray value of the spatial perception information after grayscale processing of the candidate image patch, and obtain the mean gray value and standard deviation of the gray value of the grayscale ring area of ​​the candidate image patch. Based on the mean gray value, mean gray value and standard deviation of the gray value of the spatial perception information after grayscale processing, calculate the high contrast intensity parameter of the candidate image patch. S24: The orientation consistency parameter, fence closure parameter, and high contrast intensity parameter of the candidate patch are spliced ​​together as the early development disturbance feature of the candidate patch; S25: Obtain the legal management information of the candidate map patch, extract the time sequence information from the legal management information, perform time difference correction and stage coding processing on the time sequence information, and generate the legal status time sequence code of the candidate map patch; S26: The early development disturbance features and legal state time sequence codes of the candidate patches are concatenated to form the confirmation feature vector of the candidate patches.

[0010] Furthermore, step S2, which uses a construction legality identification model to generate a confirmation feature vector to determine whether candidate land parcels associated with the feature vector possess the legality for construction land development, also includes: The legality identification model adopts a dual-branch coding structure, including a temporal coding branch and a perturbation coding branch. The temporal coding branch and the perturbation coding branch encode the legal status temporal code and the early development perturbation feature in the confirmation feature vector, respectively. A consistency-gated fusion unit is used to perform weighted fusion of the coding results to obtain a weighted fusion feature. The weighted fusion features are converted into legality probabilities using a probabilistic activation function. If the legality probability is higher than a preset legality threshold, it indicates that the candidate map patch has the legality for construction land development; otherwise, it indicates that the candidate map patch does not have the legality for construction land development.

[0011] Furthermore, step S3 generates the initial land use label and surrounding construction environment neighborhood characteristics of candidate land parcels with legality for development, including: S31: Extract the legal management information and spatial perception information of the candidate land parcels that have the legality for construction land development; S32: The initial land use label of candidate plots is extracted from the statutory management information using priority determination rules; S33: Extract the road length and normal direction angle of the road in the candidate map from the spatial perception information, and calculate the road access direction feature of the candidate map; Specifically, the formula for calculating the road access direction feature of the candidate patch is as follows: ; in, This indicates the road access direction features of the candidate map features. This represents the length of the m-th road in the candidate polygon. Let represent the normal direction angle of the m-th road in the candidate polygon, and M represent the total number of roads in the candidate polygon. Represents the two-parameter arctangent function; S34: Using the candidate plot with legal construction land development as the center and R as the radius, construct a circular neighborhood area of ​​the candidate plot. Mark the plot with the main building identification tag of 1 in the circular neighborhood area as the reference plot of the candidate plot, and obtain the legal management information and spatial perception information of the reference plot to calculate the land function compatibility of the reference plot. Specifically, the formula for calculating the land use compatibility of the reference parcels for the candidate map features is as follows: ; in, Indicates the functional compatibility of the land parcel. This indicates the number of reference parcels for the candidate map features. This indicates the initial land use label for the e-th reference parcel. This indicates the actual built-up use label for the e-th reference parcel. Indicating the initial use label of the land parcel Labels with actual construction purpose The compatibility discrimination function between them, if the initial use label of the land parcel Labels with actual construction purpose Consistent, then =1, otherwise =0; S35: Calculate the average street-to-area ratio of the reference parcel and the street-to-area ratio of the candidate parcels, and then calculate the spatial morphological fit of the candidate parcels. S36: The road access direction features, plot functional compatibility and spatial morphological adaptability are spliced ​​together to form the surrounding construction environment neighborhood features.

[0012] Furthermore, the formula for calculating the spatial morphological fit of the candidate patches in step S35 is as follows: ; in, This indicates the spatial morphological fit of the candidate image patches. The area represents the spatial perception information of the candidate patch. This represents the total road length of all roads in the candidate patch. This indicates the face-to-street ratio of the candidate image patch. This represents the average street-facing ratio of the reference land parcel. This represents an exponential function with the natural constant as its base. Indicates spatial morphology control parameters.

[0013] Further, in step S4, the initial land use label and surrounding built environment neighborhood features of the candidate land parcels are received using the construction use correction model to generate the planned use label of the candidate land parcels, including: S41: Initialize and generate label deviation values ​​between labels for different purposes, and generate standard spatial morphology adaptation degree and standard road access direction characteristics for labels for different purposes; S42: Based on the land parcel functional compatibility of the candidate patches, generate an adaptive planning threshold; S43: Based on the initial land use label of the candidate map patch and the surrounding construction environment neighborhood characteristics, calculate the degree of fit between the candidate map patch and the use label, and calculate the label deviation value between the initial land use label and the use label of the candidate map patch. S44: Select the land parcel whose initial use label and use label have a label deviation value lower than the adaptive planning threshold and the highest degree of adaptation as the planned use label of the candidate land parcel.

[0014] Compared with existing technologies, this invention proposes a method for determining and processing the use of land space planning based on intelligent agents. This technology has the following beneficial effects: First, it should be noted that this invention generates initial land use labels based on usage priority, aligning with legal planning and control logic. It then extracts three neighborhood features: road access direction, plot functional compatibility, and spatial morphology adaptability. These features comprehensively reflect the candidate plot's traffic location, consistency with surrounding functions, and spatial morphology fit. Neighborhood features supplement the basis for use determination from a spatial environment perspective, making the initial land use labels more consistent with the actual construction pattern. This provides multi-dimensional constraints for use modification and improves the rationality of the planned use labels. Specifically, the road access direction feature characterizes the relationship between the candidate plot's main street-facing direction and road spatial layout, reflecting the plot's traffic access conditions and directly relating to the reasonable layout of commercial, residential, and industrial uses. Plot functional compatibility reflects the consistency between the initial land use label and the uses of surrounding existing plots; a higher value indicates a more unified regional function and less room for use adjustment, while a lower value indicates mixed functions and allows for reasonable minor adjustments. Spatial morphology adaptability reflects the similarity between the candidate plot's area, street-facing scale, and other morphological characteristics and surrounding existing plots; a higher value indicates that the candidate plot is more suitable for maintaining the same development intensity and planned use as its surroundings.

[0015] Meanwhile, this invention adaptively generates permissible label deviations based on the functional compatibility of land parcels, flexibly constrains the range of use corrections, and calculates the degree of adaptation by combining standard form and directional benchmarks. Within the permissible deviation range, it selects the optimal planned use label, respecting both the rigid constraints of the legal initial use and the rationality of the spatial environment, thus avoiding abrupt changes in use. The final output is a use determination result that is associated with the legal time sequence, planned use, and legality, with a complete, traceable, and verifiable structure, meeting the standardization requirements of territorial spatial planning. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a method for determining and processing the use of land space planning based on intelligent agents, provided in an embodiment of the present invention. Figure 2 This is a diagram of an intelligent agent system architecture provided in an embodiment of the present invention. Detailed Implementation

[0017] The realization of the objectives, functional characteristics, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0018] This invention provides a method for determining and processing land use in land spatial planning based on intelligent agents. The executing entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this invention: a server, a terminal, etc. In other words, this method can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0019] Reference Figure 1 Embodiment 1 of the present invention is as follows: A method for determining and processing land use in land spatial planning based on intelligent agents, the method comprising: S1: Collect legal management data and spatial perception data of land use, perform parcel-level spatial correlation on the legal management data and spatial perception data, and select parcels that meet the legal requirements for construction land as candidate map patches that have been approved but not yet built based on the legal management data and spatial perception data after parcel-level spatial correlation, and construct a set of candidate map patches.

[0020] Specifically, step S1 involves collecting legally mandated land use management data and spatial sensing data, and performing parcel-level spatial correlation on the legally mandated land use management data and spatial sensing data, including: S11: Extract the legal management data of the land to be planned from the land and resources database. The legal management data includes land parcel boundary vector data, agricultural land conversion approval data, state-owned construction land supply data, land use right registration data, and control detailed planning plot map data. As an embodiment of the present invention, the land and resources database includes a cadastral survey database, a construction land approval management system, a land supply management system, a real estate registration database, and a land and space detailed planning results database, and extracts parcel boundary vector data, agricultural land conversion approval data, state-owned construction land supply data, land use right registration data, and control detailed planning plot map data from the above databases in sequence. The parcel boundary vector data is in vector data form and includes the boundary location information, parcel number, and area field of the parcel boundary. The agricultural land conversion approval data is in the form of vector points + attribute table, where the vector points are land location information, and the attribute table of the agricultural land conversion approval data includes the agricultural land conversion approval document number, approval time and approved purpose of the land corresponding to the vector points; The state-owned construction land supply data is in the form of vector points + attribute table, wherein the attribute table of the state-owned construction land supply data includes the land supply contract number, signing time, land supply purpose, and term of use of the land corresponding to the vector point; The land use right registration data is in the form of vector points + attribute table, wherein the attribute table of the land use right registration data includes the real estate unit number, registration time, registration purpose, and right holder of the land corresponding to the vector point; The control detailed planning plot map data is in the form of vector points + attribute table. The attribute table of the control detailed planning plot map data includes the planned primary use of the land corresponding to the vector points and the plot control indicators. The planned primary use includes residential land, commercial service land, industrial land, logistics and warehousing land, public management and public service land, and urban community service facility land. The plot control indicators are mandatory control parameters for the plot's construction intensity, spatial form, and use function, including plot ratio, building density, green space ratio, and building height limit. S12: Acquire remote sensing images of the location of the land to be planned for use as spatial perception data; Specifically, by acquiring high-resolution satellite or aerial orthophotos of the location of the land to be planned, and then performing radiometric correction, orthorectification, and cloud removal processing, remote sensing images of the location of the land to be planned are formed. S13: Convert the vector data and raster data in the legal management data and spatial perception data to a unified coordinate system, and divide the land to be planned into multiple non-overlapping parcels based on the parcel boundary in the parcel boundary vector data. S14: Extract the location data of the land parcel from the legal management data as the legal management information of the land parcel; trim the spatial sensing data according to the land parcel boundary to obtain remote sensing image sub-blocks of the land parcel as the spatial sensing information of the land parcel. Specifically, the legal management information of the land parcel includes the land parcel boundary vector data of the location of the land parcel, agricultural land conversion approval data, state-owned construction land supply data, land use right registration data, and control detailed planning plot map data; S15: Construct the legal management information and spatial perception information of the land parcel into a set form, which serves as the legal management data and spatial perception data after spatial association at the land parcel level.

[0021] Specifically, the data formats of the legally managed data and spatial sensing data after spatial association at the parcel level are as follows: ; in, This represents legally managed data and spatial perception data after spatial association at the parcel level. Includes statutory management information for the nth land parcel. and spatial perception information N represents the total number of land parcels.

[0022] Specifically, the land parcel mentioned above is the smallest unit of cadastral land registration. It generally refers to a closed plot of land composed of ownership boundaries, which is a piece of land with definite boundaries and definite ownership.

[0023] Step S1, based on the legally mandated management data and spatial perception data after spatial correlation at the parcel level, selects parcels that meet the legal requirements for construction land as candidate plots that have been approved but not yet developed. This also includes: Based on the legal management information of the land parcel, the legal procedural legality of the land parcel is generated, wherein the value of legal procedural legality is 0 or 1, where 0 indicates that the legal procedural requirements are not met and 1 indicates that the legal procedural requirements are met. Specifically, the rule for generating the legality of the land parcel's legal procedures is as follows: if the land parcel has an approval number for agricultural land conversion and a land supply contract number, then the legality of the land parcel's legal procedures is 1; otherwise, the legality of the land parcel's legal procedures is 0. Based on the spatial perception information of the land parcel, the area and shape regularity of the land parcel are generated. Based on the area and shape regularity, the main building identification label of the land parcel is generated according to the preset main building identification rule. The value range of the main building identification label is 0 or 1, where 0 indicates that there is no main building in the land parcel and 1 indicates that there is a main building in the land parcel. Specifically, the area and perimeter of the land parcel are calculated based on the spatial perception information, and then the shape regularity of the land parcel is calculated. The formula for calculating the shape regularity is as follows: ; in, Indicates shape regularity. Len represents the area of ​​the spatially perceived information, and Len represents the perimeter of the spatially perceived information. The shape regularity is higher, which means that the spatial perception information of the land parcel is closer to a rectangle, that is, there is a main building within the land parcel after development. The preset main building identification rule is as follows: if the land parcel meets either the rule of area greater than 36 square meters or shape regularity greater than 0.7, then the main building identification label of the land parcel is 1; otherwise, the main building identification label of the land parcel is 0. Optionally, by extracting the area field of the land parcel in the land parcel boundary vector data, if the absolute difference between the calculated area of ​​the spatial perception information and the area field is higher than the preset area threshold (the default setting is 8 square meters), the legality of the land parcel's legal procedures is marked as 0, and the land parcel boundary is calibrated to regenerate the land parcel boundary vector data or spatial perception information. Land parcels with a legal procedure validity value of 1 and a main building identification tag value of 0 are selected as candidate land parcels that have been approved but not yet built.

[0024] It should be noted that this invention uses a dual-condition screening process—legal procedural compliance and main building identification—to quickly locate approved but undeveloped land parcels. Specifically, it employs shape regularity and area thresholds to determine the main building, objectively reflecting the construction status of the land parcel and avoiding subjective misjudgments. Simultaneously, it performs boundary calibration on parcels with abnormal areas to improve data quality. This dual screening mechanism ensures that candidate land parcels possess both legal approval procedures and an undeveloped status, eliminating constructed, illegal, and incompletely documented parcels, narrowing the scope of subsequent analysis, improving processing efficiency, and guaranteeing the legality and compliance of construction land objects awaiting use determination.

[0025] S2: Extract the early development disturbance features and legal status time sequence codes of the candidate map patches in the candidate map patch set, concatenate the early development disturbance features and legal status time sequence codes as the confirmation feature vector of the candidate map patches, and use the construction legality recognition model to generate whether the candidate map patches associated with the confirmation feature vector have the legality of construction land development.

[0026] Specifically, step S2 extracts the early development disturbance features and legal state temporal codes of the candidate patches in the candidate patch set, and constructs a confirmation feature vector for the candidate patches, including: S21: Obtain the spatial perception information of the candidate patch, perform grayscale processing on the obtained spatial perception information, calculate the peak intensity of the directional gradient of the spatial perception information in each gradient direction after grayscale processing, and calculate the directional consistency parameter of the candidate patch based on the peak intensity of the directional gradient. Specifically, the gradient direction angle of any pixel in the spatially perceived information after grayscale processing is calculated, and eight equally sized direction intervals are obtained. Based on the gradient direction angle, the pixels are divided into the corresponding direction intervals. The pixel gradient of any pixel in the direction interval is calculated, and the largest pixel gradient is selected as the peak intensity of the directional gradient of that direction interval. The gradient direction angle and pixel gradient calculation formulas are as follows: ; ; ; in, This represents the gradient direction angle of the pixel in the x-th row and y-th column of the spatially perceived information after grayscale processing. These represent the horizontal and vertical gradients of the pixel in the x-th row and y-th column of the spatially perceived information after grayscale processing, respectively. This represents the pixel gradient of the pixel in the x-th row and y-th column of the spatially perceived information after grayscale processing. This represents the grayscale value of the pixel in the (x+1)th row and yth column of the spatially perceived information after grayscale processing. This represents the grayscale value of the pixel in the (x-1)th row and yth column of the spatially perceived information after grayscale processing. This represents the grayscale value of the pixel in the x-th row and y+1-th column of the spatially perceived information after grayscale processing. This represents the grayscale value of the pixel in the x-th row and y-1-th column of the spatially perceived information after grayscale processing. Represents the two-parameter arctangent function; The formula for calculating the directional consistency parameter is as follows: ; in, Indicates the direction consistency parameter. This represents the peak intensity of the directional gradient in the k-th directional interval. This represents the center angle of the k-th direction interval; for example, the center angle between 0 and 90 degrees is 45 degrees. Furthermore, since pixel gradients have directional duality, this invention divides 0 to 180 degrees into 8 equally sized directional intervals; S22: Extract the image boundary of the spatial perception information of the candidate patch, and perform line fitting on the extracted image boundary to obtain multiple sets of line segments. Connect any two sets of line segments according to the endpoint connection rule, and use the connection result as the connection fence. Calculate the ratio between the total length of the connection fence and the minimum perimeter of the bounding rectangle of the spatial perception information of the candidate patch, and use it as the fence closure parameter of the candidate patch. Specifically, the endpoint connection rule is as follows: if the distance between the endpoints of two sets of straight line segments is less than 3 meters, then the endpoints of the two sets of straight line segments are connected. It should be noted that after connecting the straight segments, the connection results with a length less than the preset length threshold (the default setting is 2 meters) are removed, and the remaining connection results are used as the connection fence. S23: Calculate the mean gray value of the spatial perception information after grayscale processing of the candidate image patch, and obtain the mean gray value and standard deviation of the gray value of the grayscale ring area of ​​the candidate image patch. Based on the mean gray value, mean gray value and standard deviation of the gray value of the spatial perception information after grayscale processing, calculate the high contrast intensity parameter of the candidate image patch. Specifically, a 3-meter ring around the spatially perceived information is selected as the ring area of ​​the candidate patch, and the ring area is grayscaled to form a grayscale ring area. The formula for calculating the high contrast intensity parameter of the candidate patch is as follows: ; in, Indicates a high contrast intensity parameter. This represents the average grayscale value of the region. This represents the standard deviation of the gray values ​​in the region. This represents the mean grayscale value of the spatial perception information after grayscale processing. This represents the grayscale control parameters; the default settings are as follows. It is 0.00001; S24: The orientation consistency parameter, fence closure parameter, and high contrast intensity parameter of the candidate patch are spliced ​​together as the early development disturbance feature of the candidate patch; S25: Obtain the legal management information of the candidate map patch, extract the time sequence information from the legal management information, perform time difference correction and stage coding processing on the time sequence information, and generate the legal status time sequence code of the candidate map patch; Specifically, the time-series information in the legally mandated management information includes the approval time in the agricultural land conversion approval data. The signing time in the data on the supply of state-owned construction land And the registration time in the land use right registration data The formulas for time-series information time difference correction and stage coding processing are as follows: ; ; ; ; in, Approval time, in order Signing time and registration time Time difference correction results This indicates a unit of time difference (default setting is 10 seconds). Represents the encoded value of the j-th segment, where The approval times are represented in order. Signing time and registration time The stage encoding value, Indicates the stage determination parameter, if satisfied ,but =1, otherwise If it is 0, then The sequence is spliced ​​together to form the legal state timing code of the candidate patch; S26: The early development disturbance features and legal state time sequence codes of the candidate patches are concatenated to form the confirmation feature vector of the candidate patches.

[0027] It should be noted that this invention extracts three early-stage development disturbance features from remote sensing images: directional consistency, fence closure, and high contrast intensity. This accurately quantifies typical disturbance phenomena such as site leveling, construction fencing, and spoil heaping, objectively reflecting the early-stage development intensity of undeveloped land. Furthermore, by applying subtle corrections and stage coding to the approval, land supply, and registration times, this invention forms a legal status time-series code, clearly reflecting the completeness of procedures and the processing sequence. The disturbance features and time-series code are concatenated to form a confirmation feature vector, taking into account both the actual surface conditions and legal procedures, providing high-quality input for legality identification and improving the model's judgment accuracy.

[0028] The S2 step, which uses a construction legality identification model to generate a confirmation feature vector to determine whether candidate land parcels associated with the feature vector possess the legality for construction land development, also includes: The legality identification model adopts a dual-branch coding structure, including a temporal coding branch and a perturbation coding branch. The temporal coding branch and the perturbation coding branch encode the legal status temporal code and the early development perturbation feature in the confirmation feature vector, respectively. A consistency-gated fusion unit is used to perform weighted fusion of the coding results to obtain a weighted fusion feature. The weighted fusion features are converted into legality probabilities using a probabilistic activation function. If the legality probability is higher than a preset legality threshold, it indicates that the candidate map patch has the legality for construction land development; otherwise, it indicates that the candidate map patch does not have the legality for construction land development.

[0029] In one embodiment of the present invention, the preset legal threshold is set to 0.62 by default, and the process of the construction legality identification model generating a confirmation feature vector to determine whether the construction land development is legal is as follows: Timing coding branch receives statutory state timing coding The legal state timing encoding Perform embedding encoding: ; Where T represents transpose. Indicates confirmation feature vector The embedding encoding result, This represents the parameters of the trainable embedding matrix. This represents the trainable embedding bias parameters. This indicates the first activation function, which is set to ReLU by default. Perturbation coding branch receives early-stage perturbation features ,in This represents the fence closure parameter, which relates to the previously developed disturbance characteristics. Perform nonlinear mapping encoding: ; in, This indicates the characteristics of the early development disturbance. The nonlinear mapping encoding result, Denotes the parameters of the first trainable nonlinear mapping matrix. Denotes the parameters of the trainable second nonlinear mapping matrix. Represents trainable nonlinear bias parameters; The encoding results are processed using a consistency-gated fusion unit. , Weighted fusion is performed, and the weighted fusion features are converted into legality probabilities using a probabilistic activation function. : ; ; ; in, Indicates the result of encoding processing , The concatenated vector, This represents vector concatenation. This represents the element-wise multiplication operator. Represents concatenated vectors The corresponding gating weights, This represents the probabilistic activation function. The default probabilistic activation function is the Sigmoid function. Indicates the result of encoding processing , The weighted fusion result, Indicates and A unit vector of uniform length with all elements equal to 1. Represents the parameters of the trainable gate matrix. This represents the trainable gating bias parameters. This represents the parameters of the trainable fusion matrix. Denotes the trainable fusion bias parameters, where The value range is between 0 and 1; It should be noted that the dual-branch coding structure processes legal time-series and perturbation features separately, preserving the independent representation capabilities of the two types of information. It also employs a consistency-gated fusion unit to achieve adaptive weighted fusion, strengthening the suppression of redundant noise in key information. A probabilistic activation function is used to output the legality probability, quantifying the degree of legality in land development. Compared to traditional rule-based methods, this approach more accurately identifies legal pre-development, reduces false positives and false negatives, and provides a reliable legal basis for land use planning.

[0030] In another embodiment of the present invention, by collecting the confirmation feature vectors of multiple candidate patches and the real labels of whether they have the legality of construction land development (where 1 indicates that the construction land development is legal and 0 indicates that the construction land development is not legal), a training loss function is constructed with the goal of minimizing the absolute value of the difference between the real label and the legality probability output by the construction legality recognition model. Based on the training loss function, the trainable parameters in the construction legality recognition model are optimized and trained using the gradient descent algorithm.

[0031] S3: Select candidate land parcels with legality for construction land development, and generate the initial land use label and surrounding construction environment neighborhood characteristics of the selected candidate land parcels based on the legal management data and spatial perception data after the land parcel spatial association.

[0032] Specifically, step S3 generates the initial land use label and surrounding construction environment neighborhood characteristics of candidate land parcels with legality for development, including: S31: Extract the legal management information and spatial perception information of the candidate land parcels that have the legality for construction land development; S32: The initial land use label of candidate plots is extracted from the statutory management information using priority determination rules; As an embodiment of the present invention, the planned primary use, land supply use and registered use of candidate land parcels are obtained from legal management information, and the use with the highest priority is selected as the initial land use label of the candidate land parcel according to the priority order of the planned primary use, land supply use and registered use. S33: Extract the road length and normal direction angle of the road in the candidate map from the spatial perception information, and calculate the road access direction feature of the candidate map; Specifically, the formula for calculating the road access direction feature of the candidate patch is as follows: ; in, This indicates the road access direction features of the candidate map features. This represents the length of the m-th road in the candidate polygon. Let represent the normal direction angle of the m-th road in the candidate polygon, and M represent the total number of roads in the candidate polygon. Represents the two-parameter arctangent function; S34: Using the candidate plot with legal construction land development as the center and R as the radius (R is set to 100 meters by default), construct a circular neighborhood area of ​​the candidate plot. Mark the plot with the main building identification tag of 1 in the circular neighborhood area as the reference plot of the candidate plot, and obtain the legal management information and spatial perception information of the reference plot. Calculate the land function compatibility of the reference plot. Specifically, the formula for calculating the land use compatibility of the reference parcels for the candidate map features is as follows: ; in, Indicates the functional compatibility of the land parcel. This indicates the number of reference parcels for the candidate map features. This indicates the initial land use label for the e-th reference parcel. This indicates the actual built-up use label for the e-th reference parcel. Indicating the initial use label of the land parcel Labels with actual construction purpose The compatibility discrimination function between them, if the initial use label of the land parcel Labels with actual construction purpose Consistent, then =1, otherwise =0; S35: Calculate the average street-to-area ratio of the reference parcel and the street-to-area ratio of the candidate parcels, and then calculate the spatial morphological fit of the candidate parcels. S36: The road access direction features, plot functional compatibility and spatial morphological adaptability are spliced ​​together to form the surrounding construction environment neighborhood features.

[0033] The formula for calculating the spatial morphological fit of the candidate image patches in step S35 is as follows: ; in, This indicates the spatial morphological fit of the candidate image patches. The area represents the spatial perception information of the candidate patch. This represents the total road length of all roads in the candidate patch. This indicates the face-to-street ratio of the candidate image patch. This represents the average street-facing ratio of the reference land parcel. This represents an exponential function with the natural constant as its base. Indicates spatial morphology control parameters, default settings. It is 0.2.

[0034] It should be noted that this invention calculates the street-facing ratio based on area and street-front length, and combines this with the average value of surrounding reference parcels to obtain the spatial morphological fit. This quantifies the degree to which the parcel's morphology matches the surrounding built-up area, and uses an exponential function to ensure a smooth and continuous fit, resulting in stable and easily comparable values. This feature effectively reflects whether candidate parcels and reference parcels belong to the same development unit, providing a morphological basis for the rationality of land use, avoiding contradictions between land use and spatial morphology, and improving the scientific nature of planning land use labeling.

[0035] S4: Using the construction use correction model, the initial land use label and surrounding construction environment neighborhood characteristics of the candidate map patch are received, the planned use label of the candidate map patch is generated, and the legal status time sequence code, planned use label and construction land development legality of the candidate map patch are associated as the result of determining the land space planning use of the candidate map patch.

[0036] Specifically, in step S4, the initial land use label and surrounding built environment neighborhood features of the candidate land parcels are received using the construction use correction model, and the planned use label of the candidate land parcels is generated, including: S41: Initialize and generate label deviation values ​​between labels for different purposes, and generate standard spatial morphology adaptation degree and standard road access direction characteristics for labels for different purposes; As an embodiment of the present invention, the average spatial morphology fit of land parcels under the use label and the average road access direction feature are collected as the standard spatial morphology fit and standard road access direction feature of the use label, respectively. The use label includes all planned dominant uses, land supply uses and registered uses. The value of the label deviation value ranges from 0 to 3. S42: Based on the land parcel functional compatibility of the candidate patches, generate an adaptive planning threshold; Specifically, the formula for calculating the adaptive planning threshold is: ; in, Indicates the adaptive planning threshold. This indicates the maximum possible value for the label deviation (default is 3). S43: Based on the initial land use label of the candidate map patch and the surrounding construction environment neighborhood characteristics, calculate the degree of fit between the candidate map patch and the use label, and calculate the label deviation value between the initial land use label and the use label of the candidate map patch. Specifically, the formula for calculating the degree of fit is: ; in, Indicates candidate patches and usage labels The degree of compatibility between them Indicating purpose label Standard spatial form adaptability Indicating purpose label Standard road access direction characteristics, All represent control parameters, default settings. They are 0.6 and 0.4 respectively; S44: Select the land parcel whose initial use label and use label have a label deviation value lower than the adaptive planning threshold and the highest degree of adaptation as the planned use label of the candidate land parcel.

[0037] The legal status time sequence code, planned use label, and legality of construction land development of the candidate map patches are associated as the result of determining the land space planning use of the candidate map patches, according to the planned use label.

[0038] Example 2 This invention employs a multi-agent collaborative distributed intelligent processing architecture, as described above. Figure 2 The system architecture diagram shown illustrates a system of intelligent agents. This system comprises five types of agents: a data fusion agent, a candidate land parcel screening agent, a construction legality identification agent, a land use feature extraction agent, and a planned land use correction agent. Each agent operates independently yet collaboratively, enabling intelligent processing of the entire process from multi-source data access to the output of land use determination results. Specific functions are as follows: The data fusion intelligent agent is responsible for collecting legal management data and spatial perception data of land and resources, completing unified coordinate transformation, topology cleaning, parcel boundary segmentation and image cropping, realizing accurate spatial correlation between legal information and remote sensing information at the parcel level, and outputting legal management data and spatial perception data after parcel-level spatial correlation.

[0039] The candidate map patch screening intelligent agent connects with the data fusion intelligent agent output results, determines the legality based on the legal procedures of agricultural land conversion, land supply, and registration, identifies the main building based on the shape regularity and area index of image pixels, and automatically filters out approved but unbuilt land parcels to form a set of candidate map patches.

[0040] The legality recognition agent adopts a dual-branch coding reasoning structure, independently extracts the early development disturbance features and the legal state time sequence code, completes feature weighting through a consistency gating fusion unit, outputs the probability of legality of construction land development, and marks candidate land patches with legality of construction land development.

[0041] The intelligent agent for extracting land use features analyzes the surrounding environment of the labeled candidate land parcels, generates initial land use labels according to priority, and extracts three types of neighborhood features: road access direction, land parcel functional compatibility, and spatial morphological adaptability.

[0042] The planning use correction agent adaptively generates permissible label deviations based on functional compatibility, calculates the comprehensive fit between the target map patch and various use labels, selects the optimal planning use label within the permissible deviation range, and finally associates the legal time sequence, legality, and planning use to output standardized land space planning use determination results.

[0043] Example 3 This invention selects 2,362 land parcels from the total land use data of a city, covering scenarios such as approved but not yet developed, approved but not used, simple disturbances, legal pre-development, and illegal land use. It compares the traditional manual interpretation methods and conventional rule matching methods with the agent-based land use determination and processing method for land use planning and construction described in this invention, focusing on four indicators: accuracy of parcel screening, accuracy of legality identification, consistency of use correction, and overall processing efficiency. This invention, through multi-agent collaborative data fusion, candidate patch screening, legality identification, use feature extraction, and planned use correction mechanisms, achieves a 18.3% improvement in patch screening accuracy, a 22.5% improvement in legality identification accuracy, a 24.4% improvement in use correction consistency, and a 42.3% improvement in overall processing efficiency compared to traditional manual interpretation and conventional rule matching methods. All four core indicators are significantly optimized, and the overall performance is superior to both traditional manual interpretation and conventional rule matching. This invention can efficiently support the automated, standardized, and high-precision determination of construction land use based on large volumes of land use data, demonstrating significant technical advantages and practical value in land spatial planning and land use control scenarios. The traditional manual interpretation method relies on land and resources personnel manually loading remote sensing images, land parcel ledgers, and planning maps, visually identifying the boundaries of each parcel, verifying approval procedures, and manually correcting land use. The conventional rule matching method is based on fixed business thresholds and hard-coded logic rules to develop automated programs. It relies on single field matching and numerical threshold judgment to complete the screening of map patches and determination of their uses, which is a semi-automated solution. Furthermore, traditional manual interpretation methods rely entirely on human labor, resulting in low accuracy and efficiency; conventional rule matching methods only achieve shallow automation, have poor scenario adaptability, and still require a large amount of manual intervention.

[0044] It should be noted that the terms "comprising," "including," or any other variations thereof used herein are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0045] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0046] The above are merely preferred embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for determining and processing land use in land spatial planning based on intelligent agents, characterized in that, The method includes: S1: Collect legal management data and spatial perception data of land use, perform parcel-level spatial correlation on the legal management data and spatial perception data, and select parcels that meet the legal requirements for construction land as candidate map patches that have been approved but not yet built based on the legal management data and spatial perception data after parcel-level spatial correlation, and construct a set of candidate map patches. S2: Extract the early development disturbance features and legal status time sequence codes of the candidate patches in the candidate patch set, concatenate the early development disturbance features and legal status time sequence codes to obtain the confirmation feature vector of the candidate patch, and use the construction legality recognition model to generate whether the candidate patch associated with the confirmation feature vector has the legality of construction land development. S3: Select candidate land parcels with legality for construction land development, and generate the initial land use label and surrounding construction environment neighborhood characteristics of the selected candidate land parcels based on the legal management data and spatial perception data after the land parcel spatial association. S4: Using the construction use correction model, the initial land use label and surrounding construction environment neighborhood characteristics of the candidate map patch are received, the planned use label of the candidate map patch is generated, and the legal status time sequence code, planned use label and construction land development legality of the candidate map patch are associated as the result of determining the land space planning use of the candidate map patch.

2. The method for determining and processing land use in land spatial planning based on intelligent agents as described in claim 1, characterized in that, Step S1 involves collecting legally mandated land use management data and spatial sensing data, and performing parcel-level spatial correlation on the legally mandated land use management data and spatial sensing data, including: S11: Extract the legal management data of the land to be planned from the land and resources database. The legal management data includes land parcel boundary vector data, agricultural land conversion approval data, state-owned construction land supply data, land use right registration data, and control detailed planning plot map data. S12: Acquire remote sensing images of the location of the land to be planned for use as spatial perception data; S13: Convert the vector data and raster data in the legal management data and spatial perception data to a unified coordinate system, and divide the land to be planned into multiple non-overlapping parcels based on the parcel boundary in the parcel boundary vector data. S14: Extract the location data of the land parcel from the legal management data as the legal management information of the land parcel; trim the spatial sensing data according to the land parcel boundary to obtain remote sensing image sub-blocks of the land parcel as the spatial sensing information of the land parcel. S15: Construct the legal management information and spatial perception information of the land parcel into a set form, which serves as the legal management data and spatial perception data after spatial association at the land parcel level.

3. The method for determining and processing land use in land spatial planning based on intelligent agents as described in claim 2, characterized in that, Step S1, based on the legally mandated management data and spatial perception data after spatial correlation at the parcel level, selects parcels that meet the legal requirements for construction land as candidate plots that have been approved but not yet developed. This also includes: Based on the legal management information of the land parcel, the legal procedural legality of the land parcel is generated, wherein the value of legal procedural legality is 0 or 1, where 0 indicates that the legal procedural requirements are not met and 1 indicates that the legal procedural requirements are met. Based on the spatial perception information of the land parcel, the area and shape regularity of the land parcel are generated. Based on the area and shape regularity, the main building identification label of the land parcel is generated according to the preset main building identification rule. The value range of the main building identification label is 0 or 1, where 0 indicates that there is no main building in the land parcel and 1 indicates that there is a main building in the land parcel. Land parcels with a legal procedure validity value of 1 and a main building identification tag value of 0 are selected as candidate land parcels that have been approved but not yet built.

4. The method for determining and processing land use in land spatial planning based on intelligent agents as described in claim 1, characterized in that, In step S2, the early development perturbation features and legal state temporal codes of the candidate patches in the candidate patch set are extracted, and a confirmation feature vector of the candidate patches is constructed, including: S21: Obtain the spatial perception information of the candidate patch, perform grayscale processing on the obtained spatial perception information, calculate the peak intensity of the directional gradient of the spatial perception information in each gradient direction after grayscale processing, and calculate the directional consistency parameter of the candidate patch based on the peak intensity of the directional gradient. S22: Extract the image boundary of the spatial perception information of the candidate patch, and perform line fitting on the extracted image boundary to obtain multiple sets of line segments. Connect any two sets of line segments according to the endpoint connection rule, and use the connection result as the connection fence. Calculate the ratio between the total length of the connection fence and the minimum perimeter of the bounding rectangle of the spatial perception information of the candidate patch, and use it as the fence closure parameter of the candidate patch. S23: Calculate the mean gray value of the spatial perception information after grayscale processing of the candidate image patch, and obtain the mean gray value and standard deviation of the gray value of the grayscale ring area of ​​the candidate image patch. Based on the mean gray value, mean gray value and standard deviation of the gray value of the spatial perception information after grayscale processing, calculate the high contrast intensity parameter of the candidate image patch. S24: The orientation consistency parameter, fence closure parameter, and high contrast intensity parameter of the candidate patch are spliced ​​together as the early development disturbance feature of the candidate patch; S25: Obtain the legal management information of the candidate map patch, extract the time sequence information from the legal management information, perform time difference correction and stage coding processing on the time sequence information, and generate the legal status time sequence code of the candidate map patch; S26: The early development disturbance features and legal state time sequence codes of the candidate patches are concatenated to form the confirmation feature vector of the candidate patches.

5. The method for determining and processing land use in land spatial planning based on intelligent agents as described in claim 4, characterized in that, The S2 step, which uses a construction legality identification model to generate confirmation feature vectors to determine whether candidate land parcels associated with them possess the legality for construction land development, also includes: The legality identification model adopts a dual-branch coding structure, including a temporal coding branch and a perturbation coding branch. The temporal coding branch and the perturbation coding branch encode the legal status temporal code and the early development perturbation feature in the confirmation feature vector, respectively. A consistency-gated fusion unit is used to perform weighted fusion of the coding results to obtain a weighted fusion feature. The weighted fusion features are converted into legality probabilities using a probabilistic activation function. If the legality probability is higher than a preset legality threshold, it indicates that the candidate map patch has the legality for construction land development; otherwise, it indicates that the candidate map patch does not have the legality for construction land development.

6. The method for determining and processing land use in land spatial planning based on intelligent agents as described in claim 1, characterized in that, The S3 step generates the initial land use label and surrounding construction environment neighborhood characteristics of candidate land parcels with legal construction land development, including: S31: Extract the legal management information and spatial perception information of the candidate land parcels that have the legality for construction land development; S32: The initial land use label of candidate plots is extracted from the statutory management information using priority determination rules; S33: Extract the road length and normal direction angle of the road in the candidate map from the spatial perception information, and calculate the road access direction feature of the candidate map; S34: Using the candidate plot with legal construction land development as the center and R as the radius, construct a circular neighborhood area of ​​the candidate plot. Mark the plot with the main building identification tag of 1 in the circular neighborhood area as the reference plot of the candidate plot, and obtain the legal management information and spatial perception information of the reference plot to calculate the land function compatibility of the reference plot. S35: Calculate the average street-to-area ratio of the reference parcel and the street-to-area ratio of the candidate parcels, and then calculate the spatial morphological fit of the candidate parcels. S36: The road access direction features, plot functional compatibility and spatial morphological adaptability are spliced ​​together to form the surrounding construction environment neighborhood features.

7. The method for determining and processing land use in land spatial planning based on intelligent agents as described in claim 6, characterized in that, The formula for calculating the spatial morphological fit of the candidate image patches in step S35 is as follows: ； in, This indicates the spatial morphological fit of the candidate image patches. The area represents the spatial perception information of the candidate patch. This represents the total road length of all roads in the candidate patch. This indicates the face-to-street ratio of the candidate image patch. This represents the average street-facing ratio of the reference land parcel. This represents an exponential function with the natural constant as its base. Indicates spatial morphology control parameters.

8. The method for determining and processing land use in land spatial planning based on intelligent agents as described in claim 1, characterized in that, In step S4, the initial land use label and surrounding built environment neighborhood features of the candidate land parcels are received using the construction use correction model, and the planned use label of the candidate land parcels is generated, including: S41: Initialize and generate label deviation values ​​between labels for different purposes, and generate standard spatial morphology adaptation degree and standard road access direction characteristics for labels for different purposes; S42: Based on the land parcel functional compatibility of the candidate patches, generate an adaptive planning threshold; S43: Based on the initial land use label of the candidate map patch and the surrounding construction environment neighborhood characteristics, calculate the degree of fit between the candidate map patch and the use label, and calculate the label deviation value between the initial land use label and the use label of the candidate map patch. S44: Select the land parcel whose initial use label and use label have a label deviation value lower than the adaptive planning threshold and the highest degree of adaptation as the planned use label of the candidate land parcel.

Images