Urban growth simulation form correction method and system based on grid-vector cooperation

Abstract

The invention discloses an urban growth simulation form correction method and system based on grid-vector collaboration, and the method comprises the steps: segmenting remote sensing image data to obtain segmented patches, and forming a fused space unit in combination with a block unit obtained through the processing of vector road data; inputting the land utilization and driving factor data into the grid CA model to obtain converted city cells, and obtaining a city growth simulation result map of the grid CA; and correcting the urban growth simulation result graph of the grid CA by using a form correction strategy to obtain a final corrected urban growth simulation result graph. According to the method, the fused composite space unit is obtained by combining the segmented plaque and the block unit, the fineness of the divided space unit is improved, a more accurate space base is provided for form correction, and effective balance among calculation efficiency, process rationality and form authenticity is realized by utilizing a grid simulation-vector correction cooperation mechanism.

Description

Technical Field

[0001] This invention relates to the field of urban spatial simulation and geographic information technology, and in particular to a method and system for morphological correction of urban growth simulation based on raster-vector collaboration. Background Technology

[0002] Currently, the technical solutions for CA (Cybernetic Architecture) modeling in the field of urban spatial simulation and geographic information technology mainly include two categories: traditional raster CA models and vector CA models. Traditional raster CA models use regular rasters as units. Although they have high computational efficiency, the simulation results are coarse in shape, exhibiting a significant "chessboard" effect, and cannot accurately reflect the continuous and reasonable urban outline shaped by real geographic elements. Vector CA models, although using irregular polygons as units to improve morphological realism, generally suffer from two major drawbacks: First, spatial units divided by relying on a single data source (such as only cadastral or road data) are insufficient in terms of data timeliness and precision, especially in areas with sparse data where the results are coarse; second, embedding complex vector operations into the iterative process significantly increases the computational load, and the design of morphological evolution rules during the iterative process is difficult, making it difficult to achieve an effective balance between computational efficiency, process rationality, and morphological realism. Summary of the Invention

[0003] To address the issues of reliance on a single data source and low computational efficiency in existing technologies, this invention provides a method and system for urban growth simulation morphology correction based on grid-vector collaboration, which improves the timeliness and accuracy of spatial unit division and achieves an effective balance between computational efficiency, process rationality, and morphological realism.

[0004] Therefore, the technical solution adopted by the present invention is as follows: A method for urban growth simulation morphology correction based on raster-vector coordination is provided, the method comprising: The remote sensing image data is segmented to obtain segmented patches, and combined with the street blocks obtained by processing vector road data to form a fused spatial unit; By inputting land use and driving factor data into the raster CA model, transformed urban cells are obtained, and the urban growth simulation results map of raster CA is obtained. The urban growth simulation results of the raster CA are corrected using a morphological correction strategy to obtain the final corrected urban growth simulation results. Specifically, the morphological correction strategy divides and codes the urban growth simulation results of the raster CA according to the fused spatial units to obtain each patch with a unique code. The proportion of urban land area of ​​each patch is calculated, and the cells in the corresponding patch are morphologically corrected according to the calculation results.

[0005] According to the above scheme, the remote sensing image data specifically refers to preprocessed remote sensing image data, and the preprocessing specifically includes: Acquire initial and final land use raster data and classify them into urban and non-urban categories, respectively; Select urban growth drivers and standardize the driver data. The acquired remote sensing image data is exported in RGB three-band TIFF format, and the projection coordinate system, geographic range and spatial resolution of all data are unified.

[0006] According to the above scheme, the integrated spatial units are obtained in the following ways: Seed points are distributed on remote sensing image data, and each pixel in the image data is assigned to the nearest seed point by optimizing the distance function to obtain segmented patches; The vector road data in the remote sensing image data is trimmed, extended and the hanging lines are cleaned up. Based on the processed vector road data, road buffers and spatial division operations are performed to obtain the street block units surrounded by the roads. By performing spatial overlay analysis on segmented patches and street units, and using the road network as a structural framework to optimize and reorganize the superpixel boundaries, a fused spatial unit is formed.

[0007] According to the above scheme, the raster CA model is specifically constructed based on preset transformation rules, as follows: Remote sensing image data is input into the raster CA model for iteration. In each iteration, the urban conversion probability of each non-urban cell is calculated. Non-urban cells with a probability greater than a certain threshold are converted into urban cells until the newly added urban land area reaches a predetermined threshold, and the simulation result map of raster CA is obtained. Specifically, the city conversion probability is calculated by multiplying the development suitability probability of each cell, the neighborhood effect value, and the development restriction state.

[0008] According to the above scheme, the development suitability probability is specifically calculated by the trained random forest model; the trained random forest model is obtained in the following way: Acquire and preprocess initial and final land use data, overlay and analyze the preprocessed initial and final land use data, and identify and divide them into urban growth areas and unchanged areas. Positive samples are drawn from urban growth areas at a certain sampling rate, and an equal number of negative samples are drawn from unchanged areas. The driving factor values ​​of each sample point are obtained to construct a sample set. The sample set is input into the random forest model for training to obtain the relationship between urban growth and driving factors, resulting in the trained random forest model.

[0009] According to the above scheme, the neighborhood effect value is specifically calculated based on the total number of cells in the neighborhood, the city status of the neighboring cells at that time, and the weight of the neighboring cells, where the weight of the neighboring cells is calculated by the distance decay function.

[0010] According to the above scheme, the morphological correction strategy includes a static correction strategy and a dynamic correction strategy. The static correction strategy corrects the morphology of cells in each patch based on the simulation results of the raster CA at this time after the newly added urban land area reaches a predetermined threshold and the iteration stops. The dynamic correction strategy corrects the morphology of cells in each patch in real time based on the simulation results of the raster CA in each iteration.

[0011] According to the above scheme, the static correction strategy is as follows: Spatial units are used to spatially divide the urban growth simulation results map of the raster CA after the iteration is completed, and each patch after division is coded. Traverse each patch and calculate the proportion of urban land area within each patch; Set a fill threshold and a trim threshold. If the proportion of urban land area is greater than or equal to the fill threshold, then all non-urban cells in the patch are converted to urban cells; if the proportion of urban land area is less than or equal to the trim threshold, then all urban cells in the patch are converted to non-urban cells; otherwise, maintain the original state.

[0012] According to the above scheme, the dynamic correction strategy is as follows: Spatial cells are used to spatially divide the simulation results of the raster CA in each iteration, and each patch after division is encoded. After each iteration, identify the newly added areas in this iteration, connect the newly added areas with the patch codes, traverse each patch containing the newly added areas, and calculate the proportion of urban land area in each patch; Set a fill threshold and a trim threshold. If the proportion of urban land area is greater than or equal to the fill threshold, then all non-urban cells in the patch are converted to urban cells; if the proportion of urban land area is less than or equal to the trim threshold, then all urban cells in the patch are converted to non-urban cells; otherwise, maintain the original state. The iteration stops when the maximum number of iterations is reached, and the final city growth simulation result map is output.

[0013] A grid-vector collaborative urban growth simulation morphology correction system is also provided, the system comprising: The spatial unit fusion module is used to segment remote sensing image data into segmented patches, and combine them with street blocks obtained by processing vector road data to form fused spatial units. The raster simulation module is used to input land use and driving factor data into the raster CA model to obtain transformed urban cells and to obtain the urban growth simulation results map of the raster CA. The correction module is used to correct the urban growth simulation results of the raster CA using a morphological correction strategy to obtain the final corrected urban growth simulation results. Specifically, the morphological correction strategy divides and codes the urban growth simulation results of the raster CA according to the fused spatial units to obtain each patch with a unique code, calculates the urban land area ratio of each patch, and performs morphological correction on the cells in the corresponding patch based on the calculation results.

[0014] The beneficial effects of this invention are as follows: By combining segmented patches and street units, this invention obtains a composite spatial unit with natural boundary recognition capabilities and planning semantic expression, which improves the precision of spatial unit division and provides a more accurate spatial basis for morphological correction. Furthermore, by utilizing the collaborative mechanism of "grid simulation-vector correction", spatial division is performed based on the urban growth simulation result map and the fused spatial unit, and the morphological correction strategy is used to perform hierarchical correction on the divided patches, thereby improving the reliability and accuracy of the final urban growth simulation result map. At the same time, a two-layer structure with a separation between the simulation calculation layer and the morphological control layer is formed, which improves the simulation and correction efficiency of the model and achieves an effective balance between computational efficiency, process rationality and morphological realism.

[0015] Furthermore, by designing static and dynamic correction strategies, this invention can correct the simulation results of the grid CA after iteration or the real-time simulation results during the iteration process based on the set fill and trim thresholds, thus achieving an effective balance between the realism of urban growth simulation morphology and simulation accuracy. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the method flow of the urban growth simulation morphology correction method based on grid-vector collaboration according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the data preprocessing steps in an embodiment of the present invention; Figure 3 This is a schematic flowchart of the steps of the fusion spatial unit according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the segmentation results of the segmented patches according to an embodiment of the present invention; Figure 5 This is a schematic diagram illustrating the generation result of a street block unit according to an embodiment of the present invention; Figure 6 This is a schematic diagram illustrating the fusion of segmented patches and street blocks according to an embodiment of the present invention; Figure 7This is a schematic diagram illustrating the development suitability probability calculation results of an embodiment of the present invention; Figure 8 This is a schematic diagram of morphological modification according to an embodiment of the present invention; Figure 9 This is a schematic diagram comparing simulation results of an embodiment of the present invention; Figure 10 This is a schematic diagram of the system structure of the urban growth simulation morphology correction system based on grid-vector collaboration according to an embodiment of the present invention. Detailed Implementation

[0017] 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.

[0018] To address the problems of existing technologies, such as reliance on a single data source, low computational efficiency, difficulty in designing morphological evolution rules during iteration, and the inability to achieve an effective balance between computational efficiency, process rationality, and morphological realism, this invention provides a grid-vector collaborative urban growth simulation morphological correction method, such as... Figure 1 As shown, the method includes: S1. The remote sensing image data is segmented to obtain segmented patches, and combined with the street block units obtained by processing the vector road data to form a fused spatial unit.

[0019] S2. Input land use and driving factor data into the raster CA model to obtain the transformed urban cells and obtain the urban growth simulation results map of the raster CA.

[0020] S3. Use the morphological correction strategy to correct the urban growth simulation result map of the raster CA to obtain the final corrected urban growth simulation result map. Specifically, the morphological correction strategy divides and codes the urban growth simulation result map of the raster CA according to the fused spatial units to obtain each patch with a unique code. The proportion of urban land area of ​​each patch is calculated, and the cells in the corresponding patch are morphologically corrected according to the calculation results.

[0021] Specifically, in this embodiment, the remote sensing image data refers to preprocessed remote sensing image data. The method also includes acquiring land use raster data and remote sensing image data and performing preprocessing, such as... Figure 2 As shown, data preprocessing specifically includes: S011. Obtain initial and final land use raster data and classify them into urban and non-urban categories, respectively.

[0022] S012. Select urban growth drivers and standardize the driver data.

[0023] S013. Export the acquired remote sensing image data in RGB three-band TIFF format, and unify the projection coordinate system, geographical range and spatial resolution of all data.

[0024] In this embodiment, 14 spatial variables as shown in Table 1 were selected as driving factors for urban growth. All driving factor data were standardized to eliminate the influence of dimensions. Table 1 Spatial Variables

[0025] Specifically, in step S1, the Simple Non-iterative Clustering (SNIC) algorithm is used to segment the preprocessed remote sensing image; such as Figure 3 As shown, the fused spatial units are obtained in the following ways: S11. Distribute seed points on the preprocessed remote sensing image, and assign each pixel in the image to the nearest seed point by optimizing the distance function to obtain segmented patches.

[0026] S12. Trim, extend, and clean up the hanging lines of the vector road data, and construct road buffers and perform spatial division operations based on the processed vector road data to obtain the block units surrounded by the roads.

[0027] S13. Perform spatial overlay analysis on the segmented patches and street units, and use the road network as a structural framework to optimize and reorganize the superpixel boundaries to form fused spatial units.

[0028] Specifically, the SNIC algorithm operates in a five-dimensional feature space combining the CIELAB color space and the XY coordinate space. It aggregates pixels into uniform superpixel objects with clear boundaries by measuring the spectral and spatial similarity between pixels. In step S11, initial, regularly spaced seed points are distributed on the preprocessed remote sensing image, and each pixel in the remote sensing image is assigned to the nearest seed point by optimizing the following distance function:

[0029] In the formula, Indicates the spectral characteristic distance, Represents the spatial Euclidean distance. It is a compactness parameter that balances spectral uniformity and object shape compactness. It is the target size parameter of the superpixel, and the calculation formula is: ,in It is the total number of pixels in the image. That is the required number of superpixels.

[0030] In a preferred embodiment, the following is provided: , Through iterative optimization, the SNIC algorithm divides the image into 25,134 units. After excluding invalid regions, 14,049 valid patches were obtained, with an average unit area of ​​approximately 1.17 square kilometers. The segmentation results are as follows: Figure 4 As shown.

[0031] In step S12, vector road data is processed through operations such as trimming, extending, and clearing overhead lines. Subsequently, street blocks surrounded by major roads are generated by constructing road buffer zones and performing spatial partitioning operations. This process produced 9937 street blocks with an average area of ​​approximately 1.65 square kilometers. The results are as follows: Figure 5 As shown.

[0032] Specifically, in step S13, to synergistically utilize image features and road network boundaries, spatial overlay analysis is used to optimize and reorganize superpixel boundaries using the road network as a structural framework, thereby forming a fused spatial unit system. This method preserves the spectral and textural features of the land surface captured by the SNIC algorithm, while combining the urban spatial structure defined by the road network, ultimately generating 39,665 spatial units with an average area of ​​0.41 square kilometers. The fusion process is as follows: Figure 6 As shown.

[0033] Preferably, the raster CA model is constructed based on preset transformation rules, specifically: Remote sensing image data is input into the raster CA model for iteration. In each iteration, the urban conversion probability of each non-urban cell is calculated. Non-urban cells with a probability greater than a certain threshold are converted into urban cells until the newly added urban land area reaches a predetermined threshold, and the simulation result map of raster CA is obtained. The city conversion probability is specifically calculated by multiplying the development suitability probability of each cell, the neighborhood effect value, and the development restriction state, and can be expressed as follows:

[0034] In the formula, Indicates the probability of city switching. Represents the probability of development suitability. Indicates the neighborhood effect value. This indicates cell assignment.

[0035] Among them, development suitability probability Specifically, it is calculated by the trained random forest model; the trained random forest model is obtained in the following way: Acquire initial and final land use data, overlay and analyze the initial and final land use data, and identify and divide them into urban growth areas and unchanged areas; Positive samples (labeled 1) are drawn from urban growth areas at a certain sampling rate, and an equal number of negative samples (labeled 0) are drawn from unchanged areas. The driving factor values ​​of each sample point are obtained to construct a sample set. The sample set is input into the random forest model for training to obtain the relationship between urban growth and driving factors, resulting in the trained random forest model.

[0036] In this embodiment, a sample set is constructed by sampling positive samples and an equal number of negative samples at a sampling rate of 1%. The random forest model is then trained based on this sample set, and the development suitability probability is calculated using the trained random forest model. The calculation results are as follows: Figure 7 As shown.

[0037] Among them, the domain effect value The specific calculation formula can be expressed as:

[0038] In the formula, It is the total number of cells in the neighborhood. Representing adjacent cells In time The status (urban land is 1, non-urban land is 0). These are the weights assigned to the corresponding neighboring cells. To more realistically capture the distance decay feature of spatial dependencies, a distance decay function is used, with the formula:

[0039] In the formula, It is the attenuation intensity index that controls the attenuation rate. From adjacent cells The Euclidean distance to the central cell. This function effectively expresses the spatial interaction mechanism: the influence from neighboring cells is strong, but weakens with increasing distance.

[0040] Specifically, cell assignment means that if there is a prohibited development area, cells within the prohibited development area are assigned a value of 0, and other cells are assigned a value of 1; in this embodiment, the water area is set as the prohibited development area.

[0041] Specifically, morphological correction strategies include static correction strategies and dynamic correction strategies. The core concept of both is to use spatial units with clear geographical significance as morphological constraint frameworks to geometrically optimize the spatial pattern output by raster simulation, thereby improving its morphological rationality and spatial accuracy.

[0042] The static correction strategy is a post-processing technique applied after the raster CA model has completed all simulation iterations and output the final urban land use layout. Specifically, the static correction strategy is as follows: Spatial units are used to spatially divide the simulation result map of the raster CA after the iteration is completed, and each patch after division is uniquely ID-coded. Traverse each patch and calculate the proportion of urban land area within each patch; Set a fill threshold and a trim threshold. If the proportion of urban land area is greater than or equal to the fill threshold, then all non-urban cells in the patch are converted to urban cells; if the proportion of urban land area is less than or equal to the trim threshold, then all urban cells in the patch are converted to non-urban cells; otherwise, maintain the original state.

[0043] Among them, the proportion of urban land area within each patch It can be represented as:

[0044] In the formula, It is the number of urban cells within the patch. This represents the total number of cells within the plaque.

[0045] Specifically, set the fill threshold. and pruning threshold .if If so, it is considered that the patch is developing rapidly, and all non-urban cells within it have been transformed into cities. Conversely, if If the area is considered underdeveloped, all urban cells within it are reverted to non-urban status, as detailed in the following process: Figure 8 As shown.

[0046] The dynamic correction strategy embeds the morphological optimization process into the iterative process of the raster CA model. After each iteration, it immediately corrects the newly urbanized areas for that period and uses the corrected results as input for the next iteration, thus providing real-time feedback for morphological control. Specifically, the dynamic correction strategy is as follows: Spatial cells are used to spatially divide the simulation results of the raster CA in each iteration, and each patch after division is encoded. After each iteration, identify the newly added areas in this iteration, connect the newly added areas with the patch codes, traverse each patch containing the newly added areas, and calculate the proportion of urban land area in each patch; Set a fill threshold and a pruning threshold. If the proportion of urban land area is greater than or equal to the fill threshold, all non-urban cells in the patch are converted to urban cells; if the proportion of urban land area is less than or equal to the pruning threshold, all urban cells in the patch are converted to non-urban cells; otherwise, the original state is maintained. After the correction, the total number of remaining cells available for development will be dynamically updated according to the actual number of cells developed, thus allowing for flexible adjustment of the urban growth scale in subsequent iterations. The iteration stops when the maximum number of iterations is reached, and the final city growth simulation result map is output.

[0047] Specifically, in each iteration, based on the selected morphological correction strategy, a corresponding CA model transformation rule is constructed, and the simulation is continuously iterated until the termination condition is met. The simulation results are then output and evaluated. The specific implementation includes the following sub-steps: Based on the simulated time step, the number of iterations is set to be once every six months. The actual urban land growth area or the predicted future urban land growth area during the initial to final period is taken as the target growth area and evenly distributed in each iteration process; When the maximum number of iterations is reached, the simulation is terminated, the simulation results are output, and an evaluation is performed.

[0048] Specifically, when the scale area of ​​practical application is the first range, the patch division accuracy is relatively coarse, and a static correction strategy is preferred for morphological correction. When the scale area of ​​practical application is the second range, the patch division accuracy is relatively fine, and a dynamic correction strategy is preferred for morphological correction. The first range specifically refers to a larger area, such as metropolitan areas, city clusters, and above; the second range specifically refers to a smaller area, such as cities and counties and below. Furthermore, in practical applications, both static and dynamic correction strategies can be used for correction testing, and the morphological correction strategy with higher simulation accuracy can be selected.

[0049] Specific experimental tests were conducted using the method described in this embodiment. Six different CA models were constructed by combining three spatial partitioning methods (SNIC unit, road network unit, and fusion unit) with two morphological correction strategies (static and dynamic). The specific configurations of these models are shown in Table 2. Table 2 Specific configurations of the six CA models

[0050] These six models are compared and analyzed with traditional raster CA models from two dimensions: simulation accuracy and landscape morphology, to demonstrate the advantages of the proposed method. An urban growth simulation experiment was conducted using the above method as an example. The experiment was performed on a computer equipped with an Intel(R) Core™ i9-13900KF 3.0 GHz CPU, NVIDIA GeForce GTX4090 GPU, 32GB of memory, and Windows 11.

[0051] Regarding simulation accuracy, this example uses the commonly used Figure of Merit (FOM) metric, calculated using the following formula:

[0052] In the formula, It represents the actual number of urban growth that has been correctly simulated as the number of city cells. It represents the number of cells that actually experienced urban growth but were simulated as non-urban. FoM represents the number of cells that are incorrectly simulated as cities but have not actually experienced urban growth. A higher FoM value indicates higher model accuracy. This metric effectively avoids the problem of overestimating accuracy when unchanging areas dominate the landscape, making it particularly suitable for evaluating simulation scenarios characterized by relatively localized changes.

[0053] Regarding landscape morphology, to quantitatively assess the consistency between the simulated landscape morphology and the actual urban landscape, this example selected five landscape pattern indices. These indices were used to comprehensively measure the landscape morphology of the simulation results across multiple dimensions, including clustering, shape complexity, and fragmentation. Table 3 lists the specific meanings of these five indices: Table 3 Landscape pattern indices used to assess landscape morphology

[0054] Based on the above five landscape pattern indices, the landscape similarity index... It can be calculated as follows:

[0055] In the formula, The first part representing the simulation results A landscape pattern index, The first one represents the actual observation result. Landscape pattern index. The higher the value, the greater the similarity between the simulation results and the actual urban landscape.

[0056] The experimental data for this embodiment mainly includes land use raster data from 2000 and 2015, 2015 Landsat 8 remote sensing imagery data for image segmentation, traffic network data for street block division, point-of-interest data as a driver of urban growth, and topographic data. This embodiment simulates the urban growth process of a certain city from 2000 to 2015. Table 4 shows the simulation accuracy of the traditional CA model and the six models constructed by the method of this invention: Table 4 Simulation accuracy of each model

[0057] The results show that the accuracy of the CA model based on the dynamic morphological correction strategy (DM-CA) is generally lower than that of CA using the static correction strategy (SM-CA). This indicates that frequent application of morphological constraints during CA simulation iterations may interfere with the simulation process. Different spatial partitioning results have a significant impact on simulation accuracy. Under both static and dynamic correction modes, the simulation accuracy follows a consistent pattern: the fused unit model (FU-CA) outperforms the superpixel unit model (SU-CA), which in turn outperforms the road network unit model (RU-CA). This result indicates that the quality of spatial partitioning is a key factor determining the effectiveness of morphological correction. Due to insufficient road density in suburban areas, road network units result in coarse spatial partitioning, thus hindering effective morphological correction. Although superpixel units can finely depict natural boundaries, they cannot accurately identify road boundaries, limiting their correction effect. In contrast, fused units, by integrating natural boundaries and road networks, establish a more rational spatial unit system, ultimately achieving simulation accuracy superior to traditional raster CA models.

[0058] To verify the effectiveness of the proposed method in reproducing actual urban landscape morphology, this embodiment compares and analyzes the simulation results of each model with those of the traditional raster-based CA model. Five landscape pattern indices are calculated to comprehensively evaluate the similarity between the simulation results and the actual urban map from 2015. The results are shown in Table 5. Table 5 Landscape morphology results for each model

[0059] The results show that the landscape pattern similarity of the three models based on Dynamic Morphology Correction (DM-CA) exceeds 80%, which is superior to the traditional raster CA model. This result demonstrates that the dynamic correction method effectively improves the spatial morphological characteristics of the simulation results by continuously optimizing the morphology of newly developed urban patches in each iteration, confirming its role in enhancing morphological rationality.

[0060] In the static morphological correction models, SU-SM-CA and RU-SM-CA exhibited relatively poor performance, with a similarity of approximately 60%, while FU-SM-CA showed the highest similarity at 83.4%. This difference indicates that the effectiveness of static correction methods largely depends on the quality of spatial division—division based solely on SNIC is limited by its difficulty in accurately identifying road boundaries, while division relying solely on road networks is constrained by the sparse road networks in suburban areas, resulting in coarse divisions; neither of these methods can support effective morphological correction. In contrast, by integrating the advantages of natural boundary identification and road network structure information, the fusion unit establishes a more rational spatial unit system, thus achieving the best performance in reproducing landscape morphology.

[0061] To further analyze the performance differences of the models at the spatial detail level, this embodiment selected three representative local regions for comparison, and the comparison results are as follows: Figure 9 As shown.

[0062] like Figure 9 As shown, the local comparison results indicate that although traditional grid CA models often produce smooth but unrealistic urban boundaries due to their reliance on local self-organized growth mechanisms, morphological correction methods that combine natural boundaries and road network structures can effectively guide the boundaries to develop along geographical features, generating simulation results that are closer to the real urban outline.

[0063] Based on the above analysis, the FU-SM-CA model proposed in this embodiment of the invention outperforms the traditional CA model in terms of simulation accuracy and landscape morphology characterization.

[0064] Furthermore, this invention also provides a grid-vector collaborative urban growth simulation morphology correction system to implement the grid-vector collaborative urban growth simulation morphology correction method described in this invention. Figure 10 As shown, the system includes: The spatial unit fusion module is used to segment remote sensing image data into segmented patches, and combine them with street blocks obtained by processing vector road data to form fused spatial units. The raster simulation module is used to input land use and driving factor data into the raster CA model to obtain transformed urban cells and to obtain the urban growth simulation results map of the raster CA. The correction module is used to correct the urban growth simulation results of the raster CA using a morphological correction strategy to obtain the final corrected urban growth simulation results. Specifically, the morphological correction strategy divides and codes the urban growth simulation results of the raster CA according to the fused spatial units to obtain each patch with a unique code, calculates the urban land area ratio of each patch, and performs morphological correction on the cells in the corresponding patch based on the calculation results.

[0065] The various modules or mechanisms of the system are mainly used to implement the various steps of the above method embodiments, and will not be described in detail here.

[0066] The urban growth simulation morphology correction method and system based on grid-vector collaboration provided in this invention improves the precision of spatial unit division by combining segmented patches and street blocks to obtain a fused composite spatial unit with natural boundary recognition and planning semantic expression. This provides a more accurate spatial basis for morphology correction. Utilizing the collaborative mechanism of "grid simulation-vector correction," spatial division is performed based on the urban growth simulation result map and the fused spatial units. Morphology correction strategies are then used to perform hierarchical correction on the divided patches, improving the reliability and accuracy of the final urban growth simulation result map. Simultaneously, a two-layer structure separating the simulation calculation layer and the morphology control layer is formed, improving the simulation and correction efficiency of the model and achieving an effective balance between computational efficiency, process rationality, and morphological realism.

[0067] Furthermore, by designing static and dynamic correction strategies, this embodiment of the invention can correct the simulation results of the grid CA after iteration or the real-time simulation results during the iteration process according to the set fill threshold and trimming threshold, thus achieving an effective balance between the realism of the urban growth simulation morphology and the simulation accuracy.

[0068] It should be noted that, depending on the implementation needs, the various steps / components described in this application can be broken down into more steps / components, or two or more steps / components or parts of the operation of steps / components can be combined into new steps / components to achieve the purpose of this invention.

[0069] The order of the steps in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0070] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for morphological correction in urban growth simulation based on grid-vector coordination, characterized in that, The method includes: The remote sensing image data is segmented to obtain segmented patches, and combined with the street blocks obtained by processing vector road data to form a fused spatial unit; By inputting land use and driving factor data into the raster CA model, transformed urban cells are obtained, and the urban growth simulation results map of raster CA is obtained. The urban growth simulation results of the raster CA are corrected using a morphological correction strategy to obtain the final corrected urban growth simulation results. Specifically, the morphological correction strategy divides and codes the urban growth simulation results of the raster CA according to the fused spatial units to obtain each patch with a unique code. The proportion of urban land area of ​​each patch is calculated, and the cells in the corresponding patch are morphologically corrected according to the calculation results.

2. The method for urban growth simulation morphology correction based on grid-vector coordination according to claim 1, characterized in that, Land use, driving factors, and remote sensing image data are specifically preprocessed data, and the preprocessing specifically includes: Acquire initial and final land use data and classify them into urban and non-urban categories, respectively; Select urban growth drivers and standardize the driver data. The acquired remote sensing image data is exported in RGB three-band TIFF format, and the projection coordinate system, geographic range and spatial resolution of all data are unified.

3. The method for urban growth simulation morphology correction based on grid-vector coordination according to claim 1, characterized in that, The integrated spatial units are obtained in the following ways: Seed points are distributed on remote sensing image data, and each pixel in the image data is assigned to the nearest seed point by optimizing the distance function to obtain segmented patches; The vector road data is trimmed, extended, and the hanging lines are cleaned up. Based on the processed vector road data, road buffers and spatial division operations are performed to obtain the block units surrounded by the roads. By performing spatial overlay analysis on segmented patches and street units, and using the road network as a structural framework to optimize and reorganize the superpixel boundaries, a fused spatial unit is formed.

4. The method for urban growth simulation morphology correction based on grid-vector coordination according to claim 1, characterized in that, The raster CA model is specifically constructed based on preset transformation rules, as follows: Land use and driving factor data are input into the raster CA model for iteration. In each iteration, the urban conversion probability of each non-urban cell is calculated. Non-urban cells with a probability greater than a certain threshold are converted into urban cells until the newly added urban land area reaches a predetermined threshold, and the simulation result map of raster CA is obtained. Specifically, the city conversion probability is calculated by multiplying the development suitability probability of each cell, the neighborhood effect value, and the development restriction state.

5. The urban growth simulation morphology correction method based on grid-vector coordination according to claim 4, characterized in that, The development suitability probability is specifically calculated by the random forest model; the construction process of the random forest model is as follows: Acquire and preprocess initial and final land use data, overlay and analyze the preprocessed initial and final land use data, and identify and divide them into urban growth areas and unchanged areas. Positive samples are drawn from urban growth areas at a certain sampling rate, and an equal number of negative samples are drawn from unchanged areas. The driving factor values ​​of each sample point are obtained to construct a sample set. The sample set is input into the random forest model for training to obtain the relationship between urban growth and driving factors, resulting in the trained random forest model.

6. The urban growth simulation morphology correction method based on grid-vector coordination according to claim 4, characterized in that, The neighborhood effect value is specifically calculated based on the total number of cells in the neighborhood, the city status of neighboring cells at that time, and the weights of neighboring cells, where the weights of neighboring cells are calculated by the distance decay function.

7. The urban growth simulation morphology correction method based on grid-vector coordination according to claim 4, characterized in that, The morphological correction strategy includes static correction and dynamic correction. The static correction strategy corrects the morphology of cells in each patch based on the simulation results of the raster CA at this time after the newly added urban land area reaches a predetermined threshold and the iteration stops. The dynamic correction strategy corrects the morphology of cells in each patch in real time based on the simulation results of the raster CA in each iteration, and uses the corrected results as the input for the next iteration.

8. The method for urban growth simulation morphology correction based on grid-vector coordination according to claim 7, characterized in that, The static correction strategy is as follows: Spatial units are used to spatially divide the urban growth simulation results map of the raster CA after the iteration is completed, and each patch after division is coded. Traverse each patch and calculate the proportion of urban land area within each patch; Set a fill threshold and a trim threshold. If the proportion of urban land area is greater than or equal to the fill threshold, then all non-urban cells in the patch are converted to urban cells; if the proportion of urban land area is less than or equal to the trim threshold, then all urban cells in the patch are converted to non-urban cells; otherwise, maintain the original state.

9. The method for urban growth simulation morphology correction based on grid-vector coordination according to claim 7, characterized in that, The dynamic correction strategy is as follows: Spatial cells are used to spatially divide the simulation results of the raster CA in each iteration, and each patch after division is encoded. After each iteration, identify the newly added areas and link them with the patch codes. Traverse each patch containing the newly added areas and calculate the proportion of urban land area in each patch. Set a fill threshold and a trim threshold. If the proportion of urban land area is greater than or equal to the fill threshold, then all non-urban cells in the patch are converted to urban cells; if the proportion of urban land area is less than or equal to the trim threshold, then all urban cells in the patch are converted to non-urban cells; otherwise, maintain the original state. The iteration stops when the maximum number of iterations is reached, and the final city growth simulation result map is output.

10. A grid-vector collaborative urban growth simulation morphology correction system, characterized in that, The system includes: The spatial unit fusion module is used to segment remote sensing image data into segmented patches, and combine them with street blocks obtained by processing vector road data to form fused spatial units. The raster simulation module is used to input land use and driving factor data into the raster CA model to obtain transformed urban cells and to obtain the urban growth simulation results map of the raster CA. The correction module is used to correct the urban growth simulation results of the raster CA using a morphological correction strategy to obtain the final corrected urban growth simulation results. Specifically, the morphological correction strategy divides and codes the urban growth simulation results of the raster CA according to the fused spatial units to obtain each patch with a unique code, calculates the urban land area ratio of each patch, and performs morphological correction on the cells in the corresponding patch based on the calculation results.

Images