Urban renewal plot boundary division method and system based on cadastral data

Abstract

The application relates to the technical field of urban cadastral data analysis, and discloses a city renewal plot area boundary division method and system based on cadastral data. The method comprises the following steps: collecting division associated cadastral data of a target city renewal plot area; screening out present situation boundary effective points and newly added candidate points with historical continuity from present situation measured data by using an improved DTW algorithm; extracting spatial features and right ownership features for fusion; constructing a feature matrix by using a fused feature vector; constructing a spatial right ownership double-constraint graph neural network; predicting a final boundary point set of the target city renewal plot area; and generating a final plot boundary. Therefore, by introducing the improved DTW algorithm and the spatial right ownership double-constraint graph neural network, the application can establish accurate association between historical and present situation boundary data, realize joint optimization of spatial rationality and right ownership rationality by using double-constraint feature fusion, and improve the division precision and efficiency of the city renewal plot area boundary.

Description

Technical Field

[0001] This invention relates to the field of urban cadastral data analysis technology, and in particular to a method and system for delineating the boundaries of urban renewal areas based on cadastral data. Background Technology

[0002] Boundary demarcation within urban renewal areas is a crucial foundational task throughout the entire urban renewal process. Its results directly determine the legality of ownership and spatial rationality of the integrated land parcels, serving as the core basis for subsequent planning approvals, property registration, and construction. In practice, urban renewal areas often involve complex situations such as demolition and redevelopment of old urban areas, consolidation of scattered parcels, and integration of historically inherited parcels. Boundary demarcation faces multiple challenges: Firstly, historical cadastral data has inherent deficiencies, with significant differences in the measurement accuracy of boundary points across different periods (approximately 5 years apart) (coordinate deviations can reach 3-8 meters). Some key boundary points are missing due to damaged archives, making it difficult to trace historical ownership boundaries. Secondly, the current topography and features have undergone significant changes; road widening, new wall construction, and temporary building erection have caused a disconnect between actual boundaries and historical records, increasing the difficulty of boundary adaptation.

[0003] Traditional boundary demarcation methods mainly rely on manual comparison of historical drawings and current measured data, and subjective judgment of boundary locations based on experience. This has three major problems: First, there is a lack of quantitative standards, and manual judgment is easily influenced by subjective factors, resulting in large deviations in boundary demarcation that do not match the actual ownership scope or current topography. Second, the temporal correlation of historical boundaries and the rigid requirements of ownership constraints are not fully considered, often leading to ownership disputes such as boundaries exceeding the original total area of ​​the plot or overlapping with adjacent plots. Third, it is inefficient, and the basis for boundary demarcation is not traceable, making it difficult to meet the needs of efficient urban renewal projects.

[0004] Therefore, how to improve the accuracy and efficiency of delineation of boundaries in urban renewal areas, while solving the adaptation problems caused by the time sequence deviation of historical boundary points, the lack of key information, and changes in the current terrain, and meeting the constraints of undisputed ownership and no boundary conflicts, is a technical problem that urgently needs to be solved. Summary of the Invention

[0005] This invention provides a method and system for delineating the boundaries of urban renewal areas based on cadastral data, aiming to solve at least one of the above-mentioned technical problems.

[0006] To achieve the above objectives, this invention provides a method for delineating the boundaries of urban renewal areas based on cadastral data, comprising the following steps:

[0007] S1: Collect cadastral data related to the division of the target urban renewal area; wherein, the cadastral data related to the division includes historical boundary data, ownership constraint data and current measured data;

[0008] S2: Based on the historical boundary data and the current measured data, the improved DTW algorithm is used to establish the temporal correlation between historical boundary points and current candidate points, and to filter out valid current boundary points with historical continuity and new candidate points from the current measured data.

[0009] S3: Based on the ownership constraint data, the current land feature data in the current measured data, and the selected current boundary valid points and newly added candidate points, extract spatial features and ownership features and fuse them, and use the fused feature vector to construct a feature matrix;

[0010] S4: Using the feature matrix and the ownership constraint rules of the ownership constraint data, construct a spatial ownership dual-constraint graph neural network to predict the final set of boundary points of the target urban renewal area, and use polygon boundary fitting to generate the final land parcel boundary.

[0011] Optionally, in step S1, the historical boundary data is configured to represent the target urban renewal area over the past three time periods. The corresponding set of historical boundary points , , Boundary points are numbered, and the attributes of each historical boundary point are configured as follows: , Indicates the coordinate system position of the historical boundary point. Indicates measurement accuracy.

[0012] Optionally, step S2: Based on the historical boundary data and the current measured data, an improved DTW algorithm is used to establish a temporal correlation between historical boundary points and current candidate points, and to filter out valid current boundary points with historical continuity and new candidate points from the current measured data, specifically including:

[0013] S21: Based on the historical boundary point data and the current measured data, construct a time-series subsequence of the historical boundary data and a current sequence of the current measured data;

[0014] S22: Using an improved DTW algorithm that incorporates a precision weighting factor, the temporal similarity between the temporal subsequence and the current sequence is calculated. Based on the temporal similarity, valid current boundary points with historical continuity and new candidate points are selected from the current measured data.

[0015] Optionally, step S21: Based on the historical boundary data and the current measured data, construct a time-series subsequence of the historical boundary data and a current-state sequence of the current measured data, specifically including:

[0016] S211: Based on the set of historical boundary points in the historical boundary data According to the boundary point number The process involves splitting the sequence and constructing a time-series subsequence corresponding to three time periods for each boundary point number. ;

[0017] S212: Based on the current situation candidate point set according to the measured data. Candidate point number Each current candidate point Construct a repeating sequence that repeats for three time periods. As the current status sequence.

[0018] Optionally, step S22: Using an improved DTW algorithm incorporating a precision weighting factor, the temporal similarity between the time-series subsequence and the current status sequence is calculated. Based on the temporal similarity, valid current boundary points with historical continuity and new candidate points are selected from the current measured data. Specifically, this includes:

[0019] S221: The improved DTW algorithm, which introduces a precision weighting factor, is used to calculate the temporal distance between the time series subsequence and the current sequence. The specific expression is as follows:

[0020]

[0021]

[0022] In the formula, Time-series subsequences representing historical boundary data Current Status Sequence with Current Measured Data The weighted temporal distance is such that the smaller the value, the higher the similarity. Indicates the first

[0023] Historical boundary points in a time period The accuracy weighting factor, and the measurement accuracy Inversely proportional, the higher the precision, the greater the weight; express and The Euclidean distance is used to measure the similarity of spatial locations. This represents the gap penalty coefficient, with a value of 0.1. The interval length in the time sequence; Indicates the first Historical boundary points in a time period The coordinate system position; Indicates current candidate points Coordinate system position; Indicates historical boundary points Measurement accuracy; S222: Based on the time distance between the time series subsequence and the current sequence. The temporal similarity between the time-series subsequence and the current sequence is calculated using the following expression:

[0024]

[0025] In the formula, for and Temporal similarity, For time-series distance, For all The maximum time distance between pairs;

[0026] S223: For current sequences whose temporal similarity exceeds a similarity threshold... The corresponding candidate points are used as valid points for the current boundary. The current sequences with temporal similarity not exceeding a similarity threshold are... The corresponding candidate points are used as new candidate points. .

[0027] Optionally, in step S1, the ownership constraint data is configured to include a set of ownership scopes for several land parcels to be integrated. and ownership constraint rules , The total ownership scope of all land parcels to be integrated , The total ownership scope of all land parcels to be integrated With adjacent land parcels to be integrated No overlap.

[0028] Optionally, step S3: Based on the ownership constraint data, the existing land feature data in the current measured data, and the selected existing boundary valid points and newly added candidate points, extract spatial features and ownership features and fuse them, and construct a feature matrix using the fused feature vector, specifically including:

[0029] S31: For each point among the selected existing boundary valid points and newly added candidate points... Calculation points The first distance between the coordinates of the existing features and the coordinates of each existing feature in the current measured data. and points The coordinates and the range of adjacent plots to be integrated The second distance of the boundary ;

[0030] S32: [The first distance] The second distance and temporal similarity Determined historical reliability score As a spatial feature;

[0031] S33: For each point among the selected existing boundary valid points and newly added candidate points... According to the point Is it within the total ownership scope of all land parcels to be integrated in the ownership constraint data? Internally determined binary characteristics and points To the total ownership scope of all land parcels to be integrated Distance of the boundary Determined ownership constraint fit ;

[0032] S34: Transfer the binary features The distance and distance Determined ownership constraint fit As a characteristic of ownership constraints;

[0033] S35: Standardize the spatial features and the ownership constraint features, and concatenate the standardized feature vectors to generate each point. fused feature vector Construct a feature matrix containing m points. .

[0034] Optionally, step S4: Using the feature matrix and the ownership constraint rules of the ownership constraint data, a spatial ownership dual-constraint graph neural network is constructed to predict the final boundary point set of the target urban renewal area, and polygon boundary fitting is used to generate the final land parcel boundary, specifically including:

[0035] S41: Use all points between the existing valid boundary points and the newly added candidate points as nodes. , using the fused feature vector of each point The node features are defined by the spatial distance between nodes as the edge E, and the reciprocal of the spatial distance between nodes is used as the edge weight. , construct graph ;

[0036] S42: Fit based on ownership constraints A loss function based on real boundary point labels and backpropagation optimization model parameters is established as the node constraint weights. A spatial weight dual-constraint graph neural network is constructed, and the graph is updated by aggregating the three graph convolutional layers of the spatial weight dual-constraint graph neural network. The neighbor node features are aggregated and updated, and then input into the Sigmoid output layer to predict the probability of each node being the final boundary point. ;

[0037] S43: Based on the probability of each node becoming the final boundary point. Select the final set of boundary points that exceed the probability threshold. The final land parcel boundary is generated by polygon boundary fitting, and ownership constraint rules are used. Verify the final site boundary.

[0038] Optionally, in step S42, the loss function, the aggregated update of neighbor node features, and the predicted probability of each node being the final boundary point are calculated. The expressions are as follows:

[0039]

[0040]

[0041]

[0042] In the formula, For the first Nodes after layer graph convolution eigenvectors, For the first Layer weight matrix and bias terms, For nodes The neighborhood group, For edge weights, For the embedding coefficients of the ownership constraint features, For nodes binary feature vectors, It is the ReLU activation function. For vector concatenation, For the output layer weight matrix and bias terms, For the final layer feature vector, For cross-entropy loss, For real labels, For ownership constraint loss weight, For nodes The degree of adaptation of ownership constraints.

[0043] Furthermore, to achieve the above objectives, the present invention also provides an urban renewal area boundary delineation system based on cadastral data, comprising:

[0044] The data acquisition module is used to collect cadastral data related to the division of the target urban renewal area; wherein, the cadastral data related to the division includes historical boundary data, ownership constraint data, and current measured data;

[0045] The filtering module is used to establish a temporal correlation between historical boundary points and current candidate points based on the historical boundary data and the current measured data, using an improved DTW algorithm, and to filter out valid current boundary points with historical continuity and new candidate points from the current measured data.

[0046] The construction module is used to extract spatial features and ownership features based on the ownership constraint data, the current land feature data in the current measured data, and the selected current boundary valid points and newly added candidate points, and to construct a feature matrix using the fused feature vector;

[0047] The prediction module is used to construct a spatial ownership dual-constraint graph neural network using the feature matrix and the ownership constraint rules of the ownership constraint data, predict the final set of boundary points of the target urban renewal area, and generate the final land parcel boundary using polygon boundary fitting.

[0048] The beneficial effects of this invention are as follows: It proposes a method and system for delineating urban renewal area boundaries based on cadastral data. By collecting cadastral data related to the delineation of the target urban renewal area, and using an improved DTW algorithm, it selects valid existing boundary points with historical continuity and new candidate points from the current measured data. Spatial features and ownership features are extracted and fused, and a feature matrix is ​​constructed using the fused feature vector. A spatial ownership dual-constraint graph neural network is then built to predict the final set of boundary points for the target urban renewal area. The final land parcel boundary is generated using polygon boundary fitting. Therefore, by introducing the improved DTW algorithm and the spatial ownership dual-constraint graph neural network, this invention establishes a precise correlation between historical and current boundary data. Simultaneously, it utilizes dual-constraint feature fusion to achieve joint optimization of spatial rationality and ownership rationality, improving the accuracy and efficiency of urban renewal area boundary delineation. While solving the adaptation problems caused by historical boundary point temporal deviations, missing key information, and changes in current terrain, it can meet the actual needs of subsequent urban renewal planning and construction, real estate registration, and other related businesses. Attached Figure Description

[0049] Figure 1 This is a flowchart illustrating the urban renewal area boundary delineation method based on cadastral data according to an embodiment of the present invention.

[0050] Figure 2 This is a schematic diagram of the urban renewal area boundary delineation system based on cadastral data, according to an embodiment of the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0052] This invention provides a method for delineating the boundaries of urban renewal areas based on cadastral data, referring to... Figure 1 , Figure 1This is a flowchart illustrating the urban renewal area boundary delineation method based on cadastral data according to an embodiment of the present invention.

[0053] In this embodiment, a method for delineating the boundary of an urban renewal area based on cadastral data includes the following steps:

[0054] S1: Collect cadastral data related to the division of the target urban renewal area; wherein, the cadastral data related to the division includes historical boundary data, ownership constraint data and current measured data;

[0055] S2: Based on the historical boundary data and the current measured data, the improved DTW algorithm is used to establish the temporal correlation between historical boundary points and current candidate points, and to filter out valid current boundary points with historical continuity and new candidate points from the current measured data.

[0056] S3: Based on the ownership constraint data, the current land feature data in the current measured data, and the selected current boundary valid points and newly added candidate points, extract spatial features and ownership features and fuse them, and use the fused feature vector to construct a feature matrix;

[0057] S4: Using the feature matrix and the ownership constraint rules of the ownership constraint data, construct a spatial ownership dual-constraint graph neural network to predict the final set of boundary points of the target urban renewal area, and use polygon boundary fitting to generate the final land parcel boundary.

[0058] Traditional boundary demarcation methods mainly rely on manual comparison of historical drawings and current measured data, and subjective judgment of boundary locations based on experience. This has three major problems: First, there is a lack of quantitative standards, and manual judgment is easily influenced by subjective factors, resulting in large deviations in boundary demarcation that do not match the actual ownership scope or current topography. Second, the temporal correlation of historical boundaries and the rigid requirements of ownership constraints are not fully considered, often leading to ownership disputes such as boundaries exceeding the original total area of ​​the plot or overlapping with adjacent plots. Third, it is inefficient, and the basis for boundary demarcation is not traceable, making it difficult to meet the needs of efficient urban renewal projects.

[0059] To address the aforementioned issues, this embodiment collects cadastral data related to the division of the target urban renewal area, employs an improved DTW algorithm, selects valid existing boundary points with historical continuity and new candidate points from the current measured data, extracts spatial features and ownership features for fusion, constructs a feature matrix using the fused feature vector, builds a spatial ownership dual-constraint graph neural network, predicts the final boundary point set of the target urban renewal area, and fits to generate the final land parcel boundary.

[0060] Specifically, the process first identifies three key data categories: historical boundary points, ownership constraints, and current measured data. Then, an improved DTW algorithm is used to establish a temporal correlation between historical boundary points and current candidate points, eliminating false boundary points and selecting valid points with historical continuity to provide a reliable point basis for boundary delineation. Next, spatial features (such as terrain adaptability and conflict risk) and ownership features (such as ownership inclusion relationships and constraint adaptability) are extracted. Standardization is then used to eliminate dimensional differences and form a unified feature matrix. Finally, a spatial ownership dual-constraint graph neural network is constructed to learn the spatial distribution patterns of boundary points and ownership constraint requirements, select highly reliable boundary points, and fit the boundary. The validity of the results is ensured through ownership constraint verification. Therefore, by introducing an improved DTW algorithm and a spatial ownership dual-constraint graph neural network, this invention establishes a precise correlation between historical and current boundary data, and utilizes dual-constraint feature fusion to achieve joint optimization of spatial rationality and ownership rationality. This improves the accuracy and efficiency of boundary delineation in urban renewal areas. While solving the adaptation problems caused by historical boundary point temporal deviations, missing key information, and changes in current terrain, it can also meet the actual needs of subsequent urban renewal planning and construction, real estate registration, and other businesses.

[0061] In a preferred embodiment, in step S1, the historical boundary data is configured to represent the target urban renewal area over the past three time periods. The corresponding set of historical boundary points , , Boundary points are numbered, and the attributes of each historical boundary point are configured as follows: , Indicates the coordinate system position of the historical boundary point. Indicates measurement accuracy.

[0062] Regarding historical boundary data, this embodiment extracts boundary point data from three key time periods with an interval of approximately 5 years, based on the historical cadastral archives of the urban renewal area. This ensures that the time series reflects the changing trends of historical boundaries. Each historical boundary point is uniquely numbered (i), and its coordinates for the corresponding time period (t) are recorded. ) and measurement accuracy ( The data is then organized by time period and boundary point number to form a structured set of historical boundary points, ensuring that subsequent time-series subsequences can be split by number. It should be noted that the measurement accuracy here is primarily determined through the measurement instrument grade and error specifications in the historical archives.

[0063] In a preferred embodiment, step S2: Based on the historical boundary data and the current measured data, an improved DTW algorithm is used to establish a temporal correlation between historical boundary points and current candidate points, and to filter out valid current boundary points with historical continuity and new candidate points from the current measured data, specifically including:

[0064] S21: Based on the historical boundary point data and the current measured data, construct a time-series subsequence of the historical boundary data and a current sequence of the current measured data;

[0065] In a preferred embodiment, step S21: Based on the historical boundary data and the current measured data, constructing a time-series subsequence of the historical boundary data and a current-state sequence of the current measured data, specifically including:

[0066] S211: Based on the set of historical boundary points in the historical boundary data According to the boundary point number The process involves splitting the sequence and constructing a time-series subsequence corresponding to three time periods for each boundary point number. ;

[0067] S212: Based on the current situation candidate point set according to the measured data. Candidate point number Each current candidate point Construct a repeating sequence that repeats for three time periods. As the current status sequence.

[0068] When constructing the temporal subsequence of historical boundary data, the boundary point number (i) is used as a unique identifier. The coordinates of the corresponding boundary point are extracted from the historical boundary point set of three time periods and arranged in chronological order to form a temporal subsequence of length 3. Each subsequence corresponds to the historical changes of the same physical location. When constructing the current status sequence of measured data, for each current status candidate point (k), its coordinates are repeated three times to construct a repeating sequence with the same length as the historical temporal subsequence, ensuring that the DTW algorithm can achieve temporal alignment through dynamic programming. Therefore, this invention proposes a method of splitting historical temporal subsequences by boundary point number, aggregating scattered historical boundary points into a sequence with temporal correlation, which can reflect the historical change trajectory of the same boundary point. In this way, the current status candidate points are repeatedly expanded into a 3-period sequence, solving the problem of mismatch between the length of the historical temporal sequence (3 periods) and the current single-point sequence, ensuring that the current status candidate points can be compared with the temporal change patterns of historical boundary points, and improving the rationality and accuracy of subsequent similarity calculations.

[0069] S22: Using an improved DTW algorithm that incorporates a precision weighting factor, the temporal similarity between the temporal subsequence and the current sequence is calculated. Based on the temporal similarity, valid current boundary points with historical continuity and new candidate points are selected from the current measured data.

[0070] In a preferred embodiment, step S22: using an improved DTW algorithm incorporating a precision weighting factor, the temporal similarity between the time-series subsequence and the current status sequence is calculated. Based on the temporal similarity, valid current boundary points with historical continuity and new candidate points are selected from the current status measured data. Specifically, this includes:

[0071] S221: The improved DTW algorithm, which introduces a precision weighting factor, is used to calculate the temporal distance between the time series subsequence and the current sequence. The specific expression is as follows:

[0072]

[0073]

[0074]

[0075] In the formula, Time-series subsequences representing historical boundary data Current Status Sequence with Current Measured Data The weighted temporal distance is such that the smaller the value, the higher the similarity. Indicates the first Historical boundary points in a time period The accuracy weighting factor, and the measurement accuracy Inversely proportional, the higher the precision, the greater the weight; express and The Euclidean distance is used to measure the similarity of spatial locations. This represents the gap penalty coefficient, with a value of 0.1. The interval length in the time sequence; Indicates the first Historical boundary points in a time period The coordinate system position; Indicates current candidate points Coordinate system position; Indicates historical boundary points Measurement accuracy;

[0076] S222: Based on the temporal distance between the time series subsequence and the current sequence The temporal similarity between the time-series subsequence and the current sequence is calculated using the following expression:

[0077]

[0078] In the formula, for and Temporal similarity, For time-series distance, For all The maximum time distance between pairs;

[0079] S223: For current sequences whose temporal similarity exceeds a similarity threshold... The corresponding candidate points are used as valid points for the current boundary. The current sequences with temporal similarity not exceeding a similarity threshold are... The corresponding candidate points are used as new candidate points. .

[0080] In this embodiment, the measurement accuracy of historical boundary points is first determined. The system calculates the weights for each period (higher precision means greater weight, ensuring that high-precision historical data dominates the matching process). Then, it combines the precision weighting factor and Euclidean distance to calculate the minimum weighted distance between the historical time series subsequence and the current sequence through dynamic programming (the smaller the distance, the stronger the association). The time series distance is normalized to similarity, eliminating the difference in distance dimensions between different matching pairs. Finally, by setting a similarity threshold (e.g., 0.85), the system selects current candidate points with high similarity as valid points and those with low similarity as new candidate points.

[0081] It should be noted that this embodiment introduces a precision weighting factor into the traditional DTW algorithm. This enables the reliability of time-series distance calculation to distinguish historical boundary points in different periods, breaking through the limitation of the traditional DTW algorithm that treats all time-series data equally. At the same time, it transforms the absolute distance into a similarity index in the [0,1] interval, which makes it easier to set a screening threshold and improves the interpretability of the results.

[0082] In practical applications, the measurement accuracy of historical boundary points in urban renewal scenarios varies significantly (e.g., early measurements were 5 meters, while later measurements were only 1 meter). Introducing a precision weighting factor can highlight the reference value of high-precision data and prevent low-precision data from interfering with matching results. Furthermore, for the application of the improved DTW distance formula, the introduction of a precision weighting factor... Then, a dynamic programming algorithm is used to find the optimal alignment path between the historical subsequence and the current sequence, utilizing the gap penalty coefficient. To avoid matching bias caused by sequence gaps, the minimum temporal distance is calculated through weighted summation. This improved DTW distance formula can enhance matching stability in urban renewal scenarios when there are slight temporal offsets in historical boundary points by employing dynamic programming for flexible alignment and weighted summation. Finally, the DTW distance (difference index) is converted into a similarity index, and normalization is achieved by dividing by the maximum DTW distance, so that the similarity value ranges from [0,1]. The larger the value, the stronger the association, which can intuitively reflect the association strength between the current point and the historical boundary.

[0083] Therefore, this embodiment realizes intelligent screening of current candidate points, eliminates pseudo boundary points that are not related to historical boundaries, ensures that valid points have historical ownership basis, reduces ownership disputes in subsequent boundary demarcation, solves the problem that the traditional DTW algorithm does not consider the difference in cadastral data measurement accuracy, resulting in large matching deviations, and improves the accuracy of association matching between historical boundary points and current points.

[0084] In a preferred embodiment, in step S1, the ownership constraint data is configured to include a set of ownership scopes for several land parcels to be integrated. and ownership constraint rules , The total ownership scope of all land parcels to be integrated , The total ownership scope of all land parcels to be integrated With adjacent land parcels to be integrated No overlap.

[0085] In this embodiment, the ownership constraint data includes: (1) a set of ownership scopes, which means: collecting the ownership scope data of n historical land parcels to be integrated. The ownership scope of each land parcel is represented by a polygon formed by its boundary points, forming a set of ownership scopes. (2) Ownership constraint rules, which mean: based on the requirements of urban renewal business, two types of core rules are defined: Ensure that the integrated land parcels do not infringe upon external rights. Ensure there are no boundary conflicts with adjacent plots, forming a rigid constraint standard.

[0086] For example, a set of ownership scopes :

[0087] The rectangle formed by boundary points A(116.4000, 39.9000), B(116.4005, 39.9000), C(116.4005, 39.9005), and D(116.4000, 39.9005).

[0088] The rectangle formed by the boundary points D(116.4000,39.9005), E(116.4005,39.9005), F(116.4005,39.9010), and G(116.4000,39.9010);

[0089] The rectangle formed by the boundary points G(116.4000,39.9010), H(116.4005,39.9010), I(116.4005,39.9015), and J(116.4000,39.9015);

[0090] For example, ownership constraint rules : Scope of the integrated land parcel Must include (That is, the large rectangle covering AJ).

[0091] : Adjacent plots (The plots formed by boundary points K(116.4005,39.9000), L(116.4010,39.9000), etc.) do not overlap.

[0092] In a preferred embodiment, step S3: Based on the ownership constraint data, the existing land feature data in the current measured data, and the selected existing boundary valid points and newly added candidate points, spatial features and ownership features are extracted and fused, and a feature matrix is ​​constructed using the fused feature vector, specifically including:

[0093] S31: For each point among the selected existing boundary valid points and newly added candidate points... Calculation points The first distance between the coordinates of the existing features and the coordinates of each existing feature in the current measured data. and points The coordinates and the range of adjacent plots to be integrated The second distance of the boundary Among them, the first distance With the second distance The specific expression is:

[0094]

[0095]

[0096] In the formula, For point coordinates Current location of features coordinates The Euclidean distance from the point to the boundary of the polygon;

[0097] S32: [The first distance] The second distance and temporal similarity Determined historical reliability score As a spatial feature;

[0098] S33: For each point among the selected existing boundary valid points and newly added candidate points... According to the point Is it within the total ownership scope of all land parcels to be integrated in the ownership constraint data? Internally determined binary characteristics and points To the total ownership scope of all land parcels to be integrated Distance of the boundary Determined ownership constraint fit ;

[0099] Among them, binary features The value is configured as follows: when point Within the total ownership scope of all land parcels to be integrated in the ownership constraint data Take 1 when the time is inside, when the point is... The total ownership scope of all land parcels to be integrated that are not included in the ownership constraint data The internal value is set to 0; the fitness degree of the ownership constraint is... The specific expression is:

[0100]

[0101] In the formula, Indicates the degree of fit of ownership constraints. Point To the total ownership scope of all land parcels to be integrated Distance to the boundary For all points The maximum value;

[0102] S34: Transfer the binary features The distance and distance Determined ownership constraint fit As a characteristic of ownership constraints;

[0103] S35: Standardize the spatial features and the ownership constraint features, and concatenate the standardized feature vectors to generate each point. fused feature vector Construct a feature matrix containing m points. .

[0104] In this embodiment, the distance between each candidate point and existing features (terrain adaptability) and the distance to adjacent plots (conflict risk) are first calculated. The historical reliability score is determined by combining temporal similarity to form a spatial feature vector. Then, it is determined whether the candidate point is within the original ownership range (binary feature). The distance from the point to the ownership boundary is calculated, the ownership constraint adaptability is quantified, and an ownership feature vector is formed. Then, Min-Max standardization is used to transform the spatial feature and ownership feature to the [0,1] interval to eliminate the difference in dimensions. Finally, the standardized spatial feature and ownership feature are concatenated to form the fused feature vector of each candidate point and construct a feature matrix.

[0105] Target The first distance between the coordinates of the existing features and the coordinates of each existing feature in the current measured data. The main consideration is that the urban renewal boundary needs to conform to the existing terrain and features. The minimum distance can directly reflect the degree of fit between the point and the feature, and meet the requirements of the boundary's current adaptation. Therefore, the minimum Euclidean distance between the candidate point and the existing features (roads, walls, etc.) is calculated. The smaller the distance, the higher the terrain adaptation.

[0106] Target The coordinates and the range of adjacent plots to be integrated The second distance of the boundary The main consideration is that boundary conflicts between adjacent plots are the core risk point of urban renewal. This distance can quantify the conflict risk and provide a basis for the model to screen low-risk points. Therefore, the minimum distance between the candidate point and the boundary of the adjacent non-integrated plot is calculated. The larger the distance, the lower the conflict risk.

[0107] Fit for ownership constraints First, by generating points Is it within the total ownership scope of all land parcels to be integrated in the ownership constraint data? Internally determined binary characteristics (1 if within the range, 0 otherwise), then based on when the candidate point is within the original total ownership range ( When the distance to the ownership boundary decreases, the fit increases (the closer the distance, the better the fit to the ownership range). When the point is outside the range ( When the fit is 0 (completely does not meet the ownership requirements), the ownership constraint fit is calculated. By quantifying the degree to which the points satisfy the ownership rules, the model prioritizes the selection of points that fit the ownership boundary, thus ensuring the legality of the boundary.

[0108] For example, step S3: Based on the ownership constraint data, the existing land feature data in the current measured data, and the selected existing boundary valid points and newly added candidate points, spatial features and ownership features are extracted and fused, and a feature matrix is ​​constructed using the fused feature vector. The following execution process is adopted:

[0109] Based on the selected candidate points For example, (coordinates X°, Y°):

[0110] (1) Computational spatial characteristics:

[0111] The minimum distance to the existing road centerline is 8 meters;

[0112] The minimum distance to the boundary of the adjacent unconformable land parcel is 15 meters;

[0113] The temporal similarity score is 0.92 (historical reliability score).

[0114] (2) The spatial feature vector is [8, 15, 0.92];

[0115] (3) Calculate ownership characteristics:

[0116] :exist Within the range, the value is 1;

[0117] :arrive The distance to the boundary is 3 meters;

[0118] : 1−3 / 20=0.85(max(d right =20 meters);

[0119] (4) The weighted feature vector is [1,3,0.85];

[0120] (5) Standardization processing (assuming the maximum values ​​of spatial features are 20, 30, and 1, and the maximum values ​​of weight features are 1, 20, and 1):

[0121] Standardized spatial characteristics: [8 / 20=0.4, 15 / 30=0.5, 0.92 / 1=0.92];

[0122] Standardized ownership characteristics: [1 / 1=1, 3 / 20=0.15, 0.85 / 1=0.85];

[0123] Fusion feature vector: [0.4, 0.5, 0.92, 1, 0.15, 0.85];

[0124] (6) Construct the feature matrix: Arrange the fused feature vectors of all candidate points in rows to form an m×6 feature matrix (m is the number of candidate points).

[0125] Therefore, this embodiment proposes a two-dimensional feature system of spatial features and ownership constraint features. Spatial features are used to focus on the adaptability of the current terrain and the risk of conflict, while ownership features are used to focus on the legality of ownership. At the same time, the concept of ownership constraint fit degree is introduced to transform the binary ownership inclusion relationship into a continuous value, so that the model can accurately perceive the degree to which a point satisfies the ownership rules. Then, standardization and feature splicing are used to eliminate the dimensional differences of different types of features and form a unified fusion feature matrix. This solves the problem of the separation of spatial features and ownership constraints in traditional methods, so that the model can simultaneously take into account spatial rationality and ownership legality, and avoid boundary deviation caused by a single dimension.

[0126] In a preferred embodiment, step S4: using the feature matrix and the ownership constraint rules of the ownership constraint data, a spatial ownership dual-constraint graph neural network is constructed to predict the final boundary point set of the target urban renewal area, and polygon boundary fitting is used to generate the final land parcel boundary, specifically including:

[0127] S41: Use all points between the existing valid boundary points and the newly added candidate points as nodes. , using the fused feature vector of each point The node features are defined by the spatial distance between nodes as the edge E, and the reciprocal of the spatial distance between nodes is used as the edge weight. , construct graph ;

[0128] S42: Fit based on ownership constraints A loss function based on real boundary point labels and backpropagation optimization model parameters is established as the node constraint weights. A spatial weight dual-constraint graph neural network is constructed, and the graph is updated by aggregating the three graph convolutional layers of the spatial weight dual-constraint graph neural network. The neighbor node features are aggregated and updated, and then input into the Sigmoid output layer to predict the probability of each node being the final boundary point. ;

[0129] S43: Based on the probability of each node becoming the final boundary point. Select the final set of boundary points that exceed the probability threshold. The final land parcel boundary is generated by polygon boundary fitting, and ownership constraint rules are used. Verify the final site boundary.

[0130] First, using existing valid boundary points and newly added candidate points as nodes, and node features as fused feature vectors, a graph structure is constructed with the spatial distance between nodes as edges and the edge weight as the reciprocal of the distance. This model the spatial relationships between boundary points. Then, the fitness of the ownership constraints is used as the node constraint weights and incorporated into the loss function. The features of neighboring nodes are aggregated through three layers of graph convolutional layers to update the node feature vectors. Next, the final layer feature vectors are input into the Sigmoid output layer to predict the probability of each node as a final boundary point. High-probability nodes are selected to form the final boundary point set. Finally, a polygon boundary is fitted, and the boundary is verified to meet the ownership constraint rules through IOU calculation and spatial overlay analysis. If it does not meet the rules, the model parameters are adjusted and the prediction is repeated.

[0131] In this embodiment, by constructing a spatial ownership dual-constraint graph neural network, the fitness of ownership constraints is incorporated into the loss function, and ownership features are embedded in the graph convolutional layer. This overcomes the limitation of traditional GNNs that only focus on spatial features, and solves the core problem that traditional methods cannot balance spatial rationality and ownership legitimacy, achieving the dual goals of spatial adaptation and ownership legitimacy in boundary partitioning. Furthermore, by automatically learning the spatial distribution patterns and ownership constraint requirements of boundary points through the graph neural network, it can replace manual subjective judgment, improving the efficiency and standardization of boundary partitioning.

[0132] In a feasible embodiment, the convex hull algorithm can be used for fitting polygon boundaries, and then the weight constraint rules can be applied after fitting. The final land parcel boundary is verified, and the specific execution logic is as follows: based on The coordinates are used to fit the polygon boundary using the convex hull algorithm. Verification through spatial overlay analysis Does it meet the ownership constraint rules? ( )and (No overlap with adjacent unintegrated plots); if the constraints are not met, adjust the weights of the loss function of the spatial ownership dual-constraint graph neural network. or Then, retrain the model until the boundary satisfies all constraints.

[0133] In a preferred embodiment, in step S42, the loss function, the aggregated update of neighbor node features, and the predicted probability of each node as the final boundary point are calculated. The expressions are as follows:

[0134]

[0135]

[0136]

[0137] In the formula, For the first Nodes after layer graph convolution eigenvectors, For the first Layer weight matrix and bias terms, For nodes The neighborhood group, For edge weights, For the embedding coefficients of the ownership constraint features, For nodes binary feature vectors, It is the ReLU activation function. For vector concatenation, For the output layer weight matrix and bias terms, For the final layer feature vector, For cross-entropy loss, For real labels, For ownership constraint loss weight, For nodes The degree of adaptation of ownership constraints.

[0138] The design for graph convolutional layer feature update primarily involves aggregating the weighted features of neighboring nodes (edge ​​weights are the inverse of distance), combining them with the node's own features and ownership constraint features to update the node feature vector, capturing spatial correlation and ownership requirements. For the loss function design, cross-entropy loss is mainly used to optimize node prediction accuracy, and an ownership constraint penalty term is introduced to penalize nodes with low ownership fit but high prediction probability, forcing the model to prioritize legitimate nodes. For node probability prediction, a sigmoid output layer maps the node feature vector to a probability in the [0,1] interval, quantifying the reliability of a node as a boundary point and providing a basis for selection.

[0139] In practical applications, the model training process of a spatial weighted dual-constraint graph neural network specifically involves: adjusting the weighted constraint fit... As node constraint weights, the loss function of the spatial weight dual-constraint graph neural network is embedded; neighbor node features are aggregated through three graph convolutional layers to update the node feature vector; the feature vector output from the final layer is input into the sigmoid output layer to predict the probability of each node as the final boundary point. Real boundary point labels based on historical verification data The model parameters (weight matrix) are optimized through backpropagation. Bias terms (etc.) to minimize the loss function. In this way, graph convolutional layers can capture spatial correlations by aggregating neighbor node features, ownership constraint feature embedding and loss function penalty terms force the model to prioritize nodes that satisfy ownership rules, and the sigmoid output layer quantifies the reliability of nodes as boundary points. In urban renewal scenarios, this sub-step enables the model to simultaneously learn the spatial distribution patterns of boundary points and ownership constraint requirements, avoiding unreasonable boundaries or disputed ownership caused by considering only a single dimension.

[0140] Reference Figure 2 , Figure 2 This is a schematic diagram of the urban renewal area boundary delineation system based on cadastral data, according to an embodiment of the present invention.

[0141] like Figure 2 As shown, the urban renewal area boundary delineation system based on cadastral data proposed in this embodiment of the invention includes:

[0142] The data acquisition module 10 is used to acquire cadastral data related to the division of the target urban renewal area; wherein, the cadastral data related to the division includes historical boundary data, ownership constraint data and current measured data;

[0143] The filtering module 20 is used to establish a temporal correlation between historical boundary points and current candidate points based on the historical boundary data and the current measured data, using an improved DTW algorithm, and to filter out valid current boundary points with historical continuity and new candidate points from the current measured data.

[0144] The construction module 30 is used to extract spatial features and ownership features based on the ownership constraint data, the current land feature data in the current measured data, and the selected current boundary valid points and newly added candidate points, and to construct a feature matrix using the fused feature vector;

[0145] The prediction module 40 is used to construct a spatial ownership dual-constraint graph neural network by utilizing the feature matrix and the ownership constraint rules of the ownership constraint data, predict the final set of boundary points of the target urban renewal area, and generate the final land parcel boundary by using polygon boundary fitting.

[0146] Other embodiments or specific implementations of the urban renewal area boundary delineation system based on cadastral data of the present invention can be referred to the above-described method embodiments, and will not be repeated here.

[0147] It is understood that in the description of this specification, references to terms such as "one embodiment," "another embodiment," "other embodiments," or "first embodiment to Nth embodiment," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0148] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system 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, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0149] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for delineating the boundaries of urban renewal areas based on cadastral data, characterized in that, Includes the following steps: S1: Collect cadastral data related to the delineation of the target urban renewal area; wherein, the cadastral data related to the delineation includes historical boundary data, ownership constraint data, and current measured data; the historical boundary data is configured to reflect the target urban renewal area over the past three time periods. The corresponding set of historical boundary points , , Boundary points are numbered, and the attributes of each historical boundary point are configured as follows: , Indicates the coordinate system position of the historical boundary point. Indicates measurement accuracy; S2: Based on the historical boundary data and the current measured data, an improved DTW algorithm is used to establish a temporal correlation between historical boundary points and current candidate points, and to filter out valid current boundary points with historical continuity and new candidate points from the current measured data; specifically including: S211: Based on the set of historical boundary points in the historical boundary data According to the boundary point number The process involves splitting the sequence and constructing a time-series subsequence corresponding to three time periods for each boundary point number. ; S212: Based on the current situation candidate point set according to the measured data. Candidate point number Each current candidate point Construct a repeating sequence that repeats for three time periods. As the current sequence; S221: The improved DTW algorithm, which introduces a precision weighting factor, is used to calculate the temporal distance between the time series subsequence and the current sequence. The specific expression is as follows: In the formula, Time-series subsequences representing historical boundary data Current Status Sequence with Current Measured Data The weighted temporal distance is such that the smaller the value, the higher the similarity. Indicates the first Historical boundary points in a time period The accuracy weighting factor, and the measurement accuracy Inversely proportional, the higher the precision, the greater the weight; express and The Euclidean distance is used to measure the similarity of spatial locations. This represents the gap penalty coefficient, with a value of 0.

1. The interval length in the time sequence; Indicates the first Historical boundary points in a time period The coordinate system position; Indicates current candidate points Coordinate system position; Indicates historical boundary points Measurement accuracy; S222: Based on the temporal distance between the time series subsequence and the current sequence The temporal similarity between the time-series subsequence and the current sequence is calculated using the following expression: In the formula, for and Temporal similarity, For time-series distance, For all The maximum time distance between pairs; S223: For current sequences whose temporal similarity exceeds a similarity threshold... The corresponding candidate points are used as valid points for the current boundary. The current sequences with temporal similarity not exceeding a similarity threshold are... The corresponding candidate points are used as new candidate points. ; S3: Based on the ownership constraint data, the current land feature data in the current measured data, and the selected current boundary valid points and newly added candidate points, extract spatial features and ownership features and fuse them, and use the fused feature vector to construct a feature matrix; S4: Using the feature matrix and the ownership constraint rules of the ownership constraint data, construct a spatial ownership dual-constraint graph neural network to predict the final set of boundary points of the target urban renewal area, and use polygon boundary fitting to generate the final land parcel boundary.

2. The method for delineating urban renewal area boundaries based on cadastral data as described in claim 1, characterized in that, In step S1, the ownership constraint data is configured to include a set of ownership scopes for several land parcels to be integrated. and ownership constraint rules , The total ownership scope of all land parcels to be integrated , The total ownership scope of all land parcels to be integrated With adjacent land parcels to be integrated No overlap.

3. The method for delineating urban renewal area boundaries based on cadastral data as described in claim 2, characterized in that, Step S3: Based on the ownership constraint data, the existing land cover data in the current measured data, and the selected existing boundary valid points and newly added candidate points, extract spatial features and ownership features and fuse them. Construct a feature matrix using the fused feature vector, specifically including: S31: For each point among the selected existing boundary valid points and newly added candidate points... Calculation points The first distance between the coordinates of the existing features and the coordinates of each existing feature in the current measured data. and points The coordinates and the range of adjacent plots to be integrated The second distance of the boundary ; S32: [The first distance] The second distance and temporal similarity Determined historical reliability score As a spatial feature; S33: For each point among the selected existing boundary valid points and newly added candidate points... According to the point Is it within the total ownership scope of all land parcels to be integrated in the ownership constraint data? Internally determined binary characteristics and points To the total ownership scope of all land parcels to be integrated Distance of the boundary Determined ownership constraint fit ; S34: Transfer the binary features The distance and distance Determined ownership constraint fit As a characteristic of ownership constraints; S35: Standardize the spatial features and the ownership constraint features, and concatenate the standardized feature vectors to generate each point. fusion feature vector Construct a feature matrix containing m points. .

4. The method for delineating urban renewal area boundaries based on cadastral data as described in claim 3, characterized in that, Step S4: Using the feature matrix and the ownership constraint rules of the ownership constraint data, construct a spatial ownership dual-constraint graph neural network to predict the final boundary point set of the target urban renewal area. Use polygon boundary fitting to generate the final land parcel boundary, specifically including: S41: Use all points between the existing valid boundary points and the newly added candidate points as nodes. , using the fused feature vector of each point The node features are defined by the spatial distance between nodes as the edge E, and the reciprocal of the spatial distance between nodes is used as the edge weight. , construct graph ; S42: Fit based on ownership constraints A loss function based on real boundary point labels and backpropagation optimization model parameters is established as the node constraint weights. A spatial weight dual-constraint graph neural network is constructed, and the graph is aggregated and updated using the three-layer graph convolutional layer of the spatial weight dual-constraint graph neural network. The neighbor node features are aggregated and updated, and then input into the Sigmoid output layer to predict the probability of each node being the final boundary point. ; S43: Based on the probability of each node becoming the final boundary point. Select the final set of boundary points that exceed the probability threshold. The final land parcel boundary is generated by polygon boundary fitting, and ownership constraint rules are used. Verify the final site boundary.

5. The method for delineating urban renewal area boundaries based on cadastral data as described in claim 4, characterized in that, In step S42, the loss function is calculated, the neighbor node features are aggregated and updated, and the probability of each node being the final boundary point is predicted. The expressions are as follows: In the formula, For the first Nodes after layer graph convolution eigenvectors, For the first Layer weight matrix and bias terms, For nodes The neighborhood group, For edge weights, For the embedding coefficients of the ownership constraint features, For nodes binary feature vectors, It is the ReLU activation function. For vector concatenation, For the output layer weight matrix and bias terms, For the final layer feature vector, For cross-entropy loss, For real labels, For ownership constraint loss weight, For nodes The degree of adaptation of ownership constraints.

6. A boundary delineation system for urban renewal areas based on cadastral data, characterized in that, The method for delineating urban renewal area boundaries based on cadastral data as described in any one of claims 1-5 includes: The data acquisition module is used to collect cadastral data related to the division of the target urban renewal area; wherein, the cadastral data related to the division includes historical boundary data, ownership constraint data, and current measured data; The filtering module is used to establish a temporal correlation between historical boundary points and current candidate points based on the historical boundary data and the current measured data, using an improved DTW algorithm, and to filter out valid current boundary points with historical continuity and new candidate points from the current measured data. The construction module is used to extract spatial features and ownership features based on the ownership constraint data, the current land feature data in the current measured data, and the selected current boundary valid points and newly added candidate points, and to construct a feature matrix using the fused feature vector; The prediction module is used to construct a spatial ownership dual-constraint graph neural network using the feature matrix and the ownership constraint rules of the ownership constraint data, predict the final set of boundary points of the target urban renewal area, and generate the final land parcel boundary using polygon boundary fitting.

Images