Land consolidation data management method and system integrating GIS and BIM

Abstract

The invention discloses a GIS and BIM integrated land consolidation data management method and system, and the method comprises the steps: obtaining GIS information and a BIM model of a land, classifying the GIS information, and obtaining a digital elevation model and a digital orthophoto map; performing geometric correction, color correction, fuzzy detection and fuzzy processing on the digital orthophoto map to obtain a digital orthophoto clear map, processing the digital orthophoto clear map by adopting an image processing model to obtain GIS land features, and matching the BIM model with the GIS land features to obtain land management data. And forming a first data item by the land management data, forming a second data item by the GIS information and the BIM model, constructing a data set to store land consolidation data, and inputting query conditions to obtain the land consolidation data. The method not only can improve the efficiency and accuracy of land consolidation data management, but also has good interpretability, and can be directly applied to a land consolidation data management system integrating GIS and BIM.

Description

Technical Field

[0001] This invention relates to the field of data management, and in particular to a land consolidation data management method and system that integrates GIS and BIM. Background Technology

[0002] With the acceleration of urbanization and the increasing scarcity of land resources, land consolidation and management have become crucial aspects of urban planning and resource management. Geographic Information System (GIS) technology, through the collection and analysis of macro-geospatial data, can provide multi-dimensional geographic information; Building Information Modeling (BIM), centered on building models, can support three-dimensional spatial design and full lifecycle management. These two types of digital technologies have been widely applied in the fields of land planning and management, and are of great significance for achieving multi-dimensional management and analysis of land consolidation data and improving the efficiency and accuracy of land consolidation.

[0003] However, traditional land data management methods suffer from the following problems: GIS systems lack sufficient support for detailed building information, and BIM models are not widely used in land consolidation and management. The isolated storage of these two types of data also leads to data redundancy, information fragmentation, and difficulties in dynamic updates. In addition, the insufficient image precision of GIS data seriously affects the accuracy of land status assessment. At the same time, existing management systems mostly adopt a flat storage structure, which makes it difficult to meet the needs of dynamic organization and efficient retrieval of large-scale, multi-level land consolidation data.

[0004] To address these issues, this invention employs image processing technology to improve the image quality of GIS data, and leverages data matching and fusion technology to achieve accurate matching and intelligent fusion of multi-source heterogeneous data. Simultaneously, it optimizes the data organization architecture through a structured hierarchical coding mechanism, designing a land consolidation data management method and system integrating GIS and BIM. This overcomes the shortcomings of existing land consolidation data management methods and systems, providing reliable technical support for land resource planning, status monitoring, and utilization optimization. Summary of the Invention

[0005] The purpose of this invention is to provide a land consolidation data management method and system that integrates GIS and BIM.

[0006] To achieve the above objectives, the present invention is implemented according to the following technical solution:

[0007] This invention includes the following steps:

[0008] Obtain GIS information and BIM model of the land, and classify the GIS information to obtain digital elevation model and digital orthophoto map;

[0009] The digital orthophoto map is processed to obtain GIS land features, and the BIM model is matched with the GIS land features to obtain land management data; the land management data includes land topography data, land status data, and land utilization rate;

[0010] The land topography data, the land status data, and the space utilization rate are combined into a first data item, and the GIS information and the BIM model are combined into a second data item.

[0011] A dataset is constructed to store land consolidation data. Management path information is determined through the dataset and query conditions, and the land consolidation data is obtained based on the management path information. The land consolidation data includes the first data item and the second data item.

[0012] Furthermore, the method for processing the digital orthophoto map to obtain GIS land features includes:

[0013] Geometric correction is performed on the digital orthophoto image, the histogram of the geometrically corrected digital orthophoto image is calculated, and color correction is performed on the digital orthophoto image based on the histogram of the digital orthophoto image; the color correction is achieved by performing histogram equalization on each color channel.

[0014] Blur detection is performed on the color-corrected digital orthophoto image, and deblurring is performed based on the blur detection results to obtain a clear digital orthophoto image.

[0015] Image processing models are used to process clear digital orthophoto maps to obtain GIS land features, and timestamps are set.

[0016] Furthermore, the method for obtaining the sharp image of the digital orthophoto includes:

[0017] Blur detection is performed: the color-corrected digital orthophoto image is blurred and segmented, the comprehensive blur metric of each segmented image is calculated, and the blurred segmented image of the digital orthophoto image is determined based on the comprehensive blur metric result; the comprehensive blur metric includes a first blur metric and a second blur metric.

[0018] Blurring processing is performed: based on the blurred segmented image, a non-blind deconvolution is used to obtain a sharp segmented approximate image. A domain transform recursive filter is used to perform edge processing on the sharp segmented approximate image to obtain a sharp segmented image. The sharp segmented images are then stitched together to obtain a sharp digital orthophoto image.

[0019] The steps for obtaining the blurred segmented image include:

[0020] The digital orthophoto image is blurred and segmented according to its content. The corresponding scale weight w is determined based on the area of ​​the i-th segmented image block. iThe image is subjected to Discrete Fourier Transform to calculate the image power spectral difference. The power spectral difference is then converted into polar coordinates to determine the slope of the image power spectral density. Based on the slope of the image power spectral density, the first fuzzy metric q is determined. 1i The expected value of a dual Gaussian mixture model is used to represent the heavy-tailed distribution of the image gradient, and the second fuzzy metric q is determined based on the variance of the heavy-tailed distribution. 2i ;

[0021] The combined fuzzy metric q of the i-th segmented image is obtained based on the first and second fuzzy metrics. i =c1q 1i +c2q 2i c1 and c2 are fuzzy metric weights. The segmented images with a comprehensive fuzzy metric greater than the comprehensive fuzzy metric threshold are selected as fuzzy segmented images.

[0022] The iterative expression for obtaining a sharp segmented approximate image using non-blind deconvolution is as follows:

[0023]

[0024] in For the i-th blurred segmented image B i The sharp segmented approximate image for the (k+1)th iteration, with the initial iteration conditions being: K is the blur kernel corresponding to the point spread function, * represents the convolution operation, K H Let be the conjugate transpose of K, and λ1 and λ2 be regularization parameters. express The gradient;

[0025] The row and column signals of each band of the clearly segmented approximate image are combined to obtain a one-dimensional signal. The one-dimensional signal is then processed iteratively using a domain transform recursive filter, which can be expressed as:

[0026] J[k]=[1-(β t ) d S[k]+(β) t ) d J[k-1]

[0027]

[0028] Where J[k] is the result of the k-th filtering operation, S[k] is a one-dimensional signal, and the corresponding domain transform signal is... S0 is the value of the original signal at the starting position, |S j -S j-1 | represents the intensity difference of the signal at adjacent positions, j∈[1,i] is the index variable for summation, and ξ s ξ is the filter size control parameter. rHere, γ is the filter ambiguity control parameter, and Var(S) is the local variance weight. j-k；j+k ) as S j Let k be the variance of the central region and β be the region size. t Here, d represents the dynamic feedback coefficient, and d represents the two adjacent signals U. i and U i-1 The distance between them, Var[S(tM:t)] is the local variance of signal S from time tM to t, and M is the size of the local window.

[0029] Furthermore, a method for obtaining GIS land features by processing the digital orthophoto image with an image processing model includes:

[0030] Digitized orthophotos were divided into training and testing sets.

[0031] An image processing model is constructed, which includes a CNN network, a cross-attention map diffusion layer, a self-contrast loss function, and an output layer.

[0032] A CNN network extracts features from the sharp image of a digital orthophoto to obtain a digital orthophoto feature map; a cross-attention map diffusion layer enhances the digital orthophoto feature map F to obtain an enhanced feature map F. ' The self-contrast loss function calculates the image loss and updates the model parameters; the output layer enhances the feature map F. ' Output GIS land features through a fully connected layer;

[0033] The clear image of the digital orthophoto to be processed is input into the image processing model to obtain GIS land features;

[0034] The image enhancement steps specifically include: segmenting the digital orthophoto feature map F using the SLIC superpixel segmentation algorithm to obtain superpixel features; defining each superpixel as a node; obtaining node feature vectors from the superpixel features; and calculating the similarity A between the feature vectors of two superpixel nodes using a Gaussian kernel function. ij According to similarity A ij Construct an adjacency matrix A, determine the k-closest neighbor graph from the adjacency matrix A, and normalize the adjacency matrix A to obtain the normalized adjacency matrix. Using a self-attention mechanism to process the normalized adjacency matrix We obtain a self-attention map S, and then perform cross-attention processing on the k-nearest neighbor map and the self-attention map S to obtain a cross-attention map C. Finally, we process the cross-attention map C to obtain the cross-attention feature Ci. F The graph diffusion module is used to apply cross-attention features C. F Image aggregation operation is performed to obtain the stationary state H of the image. diff For the cross-attention graph C and the stationary state H diffInformation fusion is performed to obtain the enhanced feature map F';

[0035] The expression for the self-comparison loss function is as follows:

[0036] Loss=μ1l pic +μ2l str

[0037]

[0038] Where Loss is the self-comparative loss function, μ1 and μ2 are the loss weights, and l pic For the graph self-contrast loss function, l str Let z be the structural self-comparison loss function. i To enhance the samples of feature map F', z j For z i Random samples within the same superpixel region, forming a positive sample pair, z k For z i Random samples from different superpixel regions form negative sample pairs, where N is the number of samples, sim(·) represents the similarity between samples, and τ is the temperature parameter.

[0039] Furthermore, the method for matching the BIM model with the GIS land features to obtain land management data includes:

[0040] The BIM model is extracted to obtain BIM building information and BIM location information. Clustering is used to classify the BIM building information to obtain project information, building use, building materials and construction progress. Clustering is also used to classify GIS land features to obtain terrain information, building feature set, land use type and GPS information. The construction progress corresponds to different project information.

[0041] The BIM location information is coarsely matched with GPS information; the BIM location information is the building coordinate point; the GPS information is the area coordinate range; the GPS information is matched with multiple BIM location information.

[0042] Based on the timestamps of GIS land features and construction progress, information is located in a time sequence, and engineering information is matched with a set of building features. The engineering information includes the building information of a single construction project. The set of building features includes the building features of multiple construction projects within the regional coordinate range. The method of fine matching is to calculate the correlation between engineering information and building features, and select the building feature with the highest correlation with the engineering information as a pair of matching data within the regional coordinate range.

[0043] Based on the detailed matching results, land status data within the regional coordinate range is determined according to the time series; the land status data includes building use, construction period, and land use type.

[0044] Feature extraction is performed on the digital elevation model to obtain topographic elevation information and building elevation information. The topographic data and topographic elevation information are fused to obtain the first topographic information. Based on the fine matching results, the engineering information, building characteristics and building elevation information are fused to obtain the first building information. The first topographic information and the first building information are cross-referenced to obtain landform data.

[0045] Land utilization rate is determined based on land topography data; the land utilization rate includes spatial utilization rate and horizontal utilization rate.

[0046] Furthermore, the method for obtaining the land consolidation data based on the management path information includes:

[0047] Three-level path information is generated based on the data item category. The three-level path information is then input into a hash function to generate a three-level hash code. A category code is generated based on the index category quantity, data size, and three-level hash code of the data item. The three-level path information and the category code are uniquely corresponding.

[0048] The first and second data items within the same time period are combined into a data bundle. Secondary path information is generated based on the time interval of the data bundle. The secondary path information is input into a hash function to generate a secondary hash code. A time code is generated based on the time interval of the data bundle, the total number of index categories, and the secondary hash code. The secondary path information and the time code are uniquely corresponding.

[0049] Data bundles from different time periods within the same coordinate region are grouped into data blocks. First-level path information is generated based on the coordinate interval of the data blocks. The first-level path information is input into a hash function to generate a first-level hash code. A region code is generated based on the region size of the data block, the region boundary points, and the first-level hash code. The first-level path information and the region code are uniquely corresponding.

[0050] All data blocks within the managed area are combined into a dataset. The management path information consists of three-level path information, two-level path information, and one-level path information. The management code consists of a category code, a time code, and a region code. Data within the dataset is stored according to the management path information. The management path information and the management code are uniquely associated.

[0051] A query code is generated based on the query conditions, a management code is matched based on the query code, management path information is determined based on the management code, and land consolidation data is read based on the management path information; the land consolidation data consists of a first data item and a second data item.

[0052] Secondly, an integrated land consolidation data management system combining GIS and BIM includes:

[0053] Data acquisition module: used to acquire GIS information and BIM model of land, and classify the GIS information to obtain digital elevation model and digital orthophoto map;

[0054] Image processing module: used to perform geometric correction, color correction, blur detection and blur processing on the digital orthophoto image to obtain a clear digital orthophoto image, used to construct the image processing model, and input the clear digital orthophoto image to be processed into the image processing model to obtain GIS land features;

[0055] Data matching module: used to extract BIM building information and BIM location information from the BIM model, classify the BIM building information and the GIS land features, and match the BIM model and the GIS land features according to the classification results to obtain land management data;

[0056] Dataset module: used to construct a dataset based on the data item, the data bundle, and the data block, and to store, view, and manage the first data item and the second data item based on the dataset.

[0057] The beneficial effects of this invention are:

[0058] This invention relates to a land consolidation data management method and system integrating GIS and BIM. Compared with existing technologies, this invention has the following technical advantages:

[0059] This invention, through image correction, image feature extraction, data matching, data encoding, and dataset construction, enhances data preprocessing capabilities and improves model adaptability in land consolidation data management. It enables precise matching and intelligent fusion of multi-source heterogeneous data, thereby improving the efficiency and accuracy of land consolidation data management. Optimizing land consolidation data management technology can significantly save resources, improve work efficiency, and provide reliable technical support for land resource planning, status monitoring, and utilization optimization. This invention is of great significance for improving land consolidation data management and can adapt to different integrated GIS and BIM land consolidation data management systems, as well as the terminal management needs of different users for integrated GIS and BIM land consolidation data, demonstrating a certain degree of universality. Attached Figure Description

[0060] Figure 1 This is a flowchart illustrating the steps of a land consolidation data management method integrating GIS and BIM according to the present invention. Detailed Implementation

[0061] The present invention will be further described below through specific embodiments. The illustrative embodiments and descriptions herein are used to explain the present invention, but are not intended to limit the present invention.

[0062] The present invention discloses a land consolidation data management method and system integrating GIS and BIM, comprising the following steps:

[0063] like Figure 1 As shown, this embodiment includes the following steps:

[0064] Obtain GIS information and BIM model of the land, and classify the GIS information to obtain digital elevation model and digital orthophoto map;

[0065] The digital orthophoto map is processed to obtain GIS land features, and the BIM model is matched with the GIS land features to obtain land management data; the land management data includes land topography data, land status data, and land utilization rate;

[0066] The land topography data, the land status data, and the space utilization rate are combined into a first data item, and the GIS information and the BIM model are combined into a second data item.

[0067] A dataset is constructed to store land consolidation data. Management path information is determined through the dataset and query conditions, and the land consolidation data is obtained based on the management path information. The land consolidation data includes the first data item and the second data item.

[0068] In this embodiment, the method for processing the digital orthophoto map to obtain GIS land features includes:

[0069] Geometric correction of digital orthophotos: Obtain the camera's intrinsic and extrinsic parameter matrices, convert the world coordinates of the image to image coordinates based on the intrinsic and extrinsic parameter matrices, convert the image coordinates of the image to camera coordinates based on the intrinsic parameter matrix, and convert the camera coordinates of the image to world coordinates based on the extrinsic parameter matrix.

[0070] Geometric correction is performed on the digital orthophoto image, the histogram of the geometrically corrected digital orthophoto image is calculated, and color correction is performed on the digital orthophoto image based on the histogram of the digital orthophoto image; the color correction is achieved by performing histogram equalization on each color channel.

[0071] Blur detection is performed on the color-corrected digital orthophoto image, and deblurring is performed based on the blur detection results to obtain a clear digital orthophoto image.

[0072] Image processing models are used to process clear digital orthophoto maps to obtain GIS land features, and timestamps are set.

[0073] In this embodiment, the method for obtaining the sharp image of the digital orthophoto includes:

[0074] Blur detection is performed: the color-corrected digital orthophoto image is blurred and segmented, the comprehensive blur metric of each segmented image is calculated, and the blurred segmented image of the digital orthophoto image is determined based on the comprehensive blur metric result; the comprehensive blur metric includes a first blur metric and a second blur metric.

[0075] Blurring processing is performed: based on the blurred segmented image, a non-blind deconvolution is used to obtain a sharp segmented approximate image. A domain transform recursive filter is used to perform edge processing on the sharp segmented approximate image to obtain a sharp segmented image. The sharp segmented images are then stitched together to obtain a sharp digital orthophoto image.

[0076] The steps for obtaining the blurred segmented image include:

[0077] The digital orthophoto image is blurred and segmented according to its content. The corresponding scale weight w is determined based on the area of ​​the i-th segmented image block. i The image is subjected to Discrete Fourier Transform to calculate the image power spectral difference. The power spectral difference is then converted into polar coordinates to determine the slope of the image power spectral density. Based on the slope of the image power spectral density, the first fuzzy metric q is determined. 1i The expected value of a dual Gaussian mixture model is used to represent the heavy-tailed distribution of the image gradient, and the second fuzzy metric q is determined based on the variance of the heavy-tailed distribution. 2i The expression is:

[0078]

[0079]

[0080] Where q 1i The first blur metric for the i-th segmented image. The scale weight w of the i-th segmented image i S0 standard image area, S i Let be the area of ​​the i-th segmented image, δ be the scale weight sensitivity factor, and α be the area of ​​the segmented image. i Let α be the power spectral slope of the i-th segmented image, α0 be the power spectral slope of the entire image, and q be the power spectral slope of the i-th segmented image. 2i σ is the second blur metric for the i-th segmented image. 1i The standard deviation of the gradient magnitude of the i-th segmented image, S(f) is the sum of the power spectra S(f,θ) in all θ directions, and S(f,θ) is the image power spectral difference. The corresponding polar coordinates are u = f cosθ, v = f sinθ, N*N is the image size, A is the scaling factor of the amplitude in each direction, α is the slope of the image power spectrum, HTD(x) is the heavy-tailed distribution of the image gradient, x is the gradient amplitude value, π0 and π1 are Gaussian distribution mixing coefficients, and π0 + π1 = 1, G(·) is the Gaussian distribution function with mean μ0 = μ1 = 1, σ0 and σ1 are the standard deviations of the gradient amplitude, and σ0 < σ1;

[0081] The combined fuzzy metric q of the i-th segmented image is obtained based on the first and second fuzzy metrics. i =c1q 1i +c2q 2i c1 and c2 are fuzzy metric weights. The segmented images with a comprehensive fuzzy metric greater than the comprehensive fuzzy metric threshold are selected as fuzzy segmented images.

[0082] The iterative expression for obtaining a sharp segmented approximate image using non-blind deconvolution is as follows:

[0083]

[0084] in For the i-th blurred segmented image B i The sharp segmented approximate image for the (k+1)th iteration, with the initial iteration conditions being: K is the blur kernel corresponding to the point spread function, * represents the convolution operation, K H Let be the conjugate transpose of K, and λ1 and λ2 be regularization parameters. express The gradient;

[0085] The row and column signals of each band of the clearly segmented approximate image are combined to obtain a one-dimensional signal. The one-dimensional signal is then processed iteratively using a domain transform recursive filter, which can be expressed as:

[0086] J[k]=[1-(β t ) d S[k]+(β) t ) d J[k-1]

[0087]

[0088] Where J[k] is the result of the k-th filtering operation, S[k] is a one-dimensional signal, and the corresponding domain transform signal is... S0 is the value of the original signal at the starting position, |S j -S j-1 | represents the intensity difference of the signal at adjacent positions, j∈[1,i] is the index variable for summation, and ξ s ξ is the filter size control parameter. r Here, γ is the filter ambiguity control parameter, and Var(S) is the local variance weight. j-k；j+k ) as S j Let k be the variance of the central region and β be the region size. t Here, d represents the dynamic feedback coefficient, and d represents the two adjacent signals U. i and U i-1The distance between them, Var[S(tM:t)] is the local variance of signal S from time tM to t, and M is the size of the local window;

[0089] In the actual assessment, the GIS digital orthophoto map of area A (50 square kilometers) in a certain region was blurred into 8 blocks with areas of (6, 7.5, 5, 9, 7, 5.5, 6.5, and 8.5 square kilometers). The weight sensitivity factor δ = 5 and the standard area S0 = 50 were taken, and the scale weight of each blurred segmented image was calculated as (0.415, 0.427, 0.407, 0.440, 0.423, 0.411, 0.419, and 0.436).

[0090] The power spectrum slopes (2.1, 2.3, 2.0, 2.4, 2.2, 2.1, 2.3, 2.2) and standard deviations of gradient magnitudes (0.5, 0.6, 0.55, 0.65, 0.58, 0.52, 0.62, 0.57) of the eight images were calculated respectively. The power spectrum slope α0 of the entire image was 2.2, and the standard deviation of gradient magnitude σ1 was 0.6. The first fuzzy metric q was then calculated. 1i (0.604, 0.554, 0.629, 0.520, 0.576, 0.608, 0.562, 0.563) and the second fuzzy measure q 2i The values ​​are (0.654, 0.573, 0.627, 0.523, 0.591, 0.644, 0.567, 0.586). Taking c1 = c2 = 0.5, the comprehensive metric is (0.629, 0.564, 0.628, 0.522, 0.584, 0.626, 0.565, 0.575). The comprehensive fuzzy metric threshold is set to 0.6. Fuzzy processing is then applied to the fuzzy segmented images of blocks 1, 3, and 6.

[0091] In non-blind deconvolution, the regularization parameters λ1 = 0.05 and λ2 = 0.0005 are used. In domain transform recursive filtering, the filter size control parameter ξ... s =10. Filter ambiguity control parameter ξ r =0.3, after non-blind deconvolution and domain transformation recursive filtering, the 1st, 3rd and 6th clear segmented images are obtained, which are then stitched together with the original segmented images to obtain a clear digital orthophoto image.

[0092] In this embodiment, the method for obtaining GIS land features by processing the digital orthophoto image with an image processing model includes:

[0093] Digitized orthophotos were divided into training and testing sets.

[0094] An image processing model is constructed, which includes a CNN network, a cross-attention map diffusion layer, a self-contrast loss function, and an output layer.

[0095] A CNN network extracts features from the sharp image of a digital orthophoto to obtain a digital orthophoto feature map; a cross-attention map diffusion layer enhances the digital orthophoto feature map F to obtain an enhanced feature map F. ' The self-contrast loss function calculates the image loss and updates the model parameters; the output layer enhances the feature map F. ' Output GIS land features through a fully connected layer;

[0096] The clear image of the digital orthophoto to be processed is input into the image processing model to obtain GIS land features;

[0097] The image enhancement steps specifically include: segmenting the digital orthophoto feature map F using the SLIC superpixel segmentation algorithm to obtain superpixel features; defining each superpixel as a node; obtaining node feature vectors from the superpixel features; and calculating the similarity A between the feature vectors of two superpixel nodes using a Gaussian kernel function. ij According to similarity A ij Construct an adjacency matrix A, determine the k-closest neighbor graph from the adjacency matrix A, and normalize the adjacency matrix A to obtain the normalized adjacency matrix. Using a self-attention mechanism to process the normalized adjacency matrix We obtain a self-attention map S, and then perform cross-attention processing on the k-nearest neighbor map and the self-attention map S to obtain a cross-attention map C. Finally, we process the cross-attention map C to obtain the cross-attention feature Ci. F The graph diffusion module is used to apply cross-attention features C. F Image aggregation operation is performed to obtain the stationary state H of the image. diff For the cross-attention graph C and the stationary state H diff Information fusion is performed to obtain the enhanced feature map F';

[0098] The expression for the self-comparison loss function is as follows:

[0099] Loss=μ1l pic +μ2l str

[0100]

[0101] Where Loss is the self-comparative loss function, μ1 and μ2 are the loss weights, and l pic For the graph self-contrast loss function, l str Let z be the structural self-comparison loss function. i To enhance the samples of feature map F', z j For z i Random samples within the same superpixel region, forming a positive sample pair, z k For z iRandom samples from different superpixel regions form negative sample pairs, where N is the number of samples, sim(·) represents the similarity between samples, and τ is the temperature parameter.

[0102] In the actual assessment, the loss weights μ1 = 0.9 and μ2 = 0.1, and the temperature parameter τ = 0.03, were used to process the digital orthophoto image to obtain GIS land features: 1. Topographic information (the pixel areas of flat land, hills, mountains, and water bodies were 120,000, 90,000, 60,000, and 30,000, respectively, with corresponding percentages of 40%, 30%, 20%, and 10%); 2. Land use types (the pixel areas of farmland, factories, residential areas, roads, green spaces, and other land uses were 90,000, 60,000, 75,000, 30,000, 30,000, and 15,000, respectively, with corresponding percentages of 30%, 20%, 25%, and 10%). 3. Building characteristics (small buildings with a floor area of ​​less than 500 square meters have a total floor area of ​​50,000 square meters, medium-sized buildings with a floor area between 500 and 2,000 square meters have a total floor area of ​​60,000 square meters, and large buildings with a floor area greater than 2,000 square meters have a total floor area of ​​15,000 square meters, with corresponding percentages of 40%, 48%, and 12% respectively; low-rise buildings with a building height of less than 10 meters have a total height of 8,000 meters, mid-rise buildings with a building height between 10 and 30 meters have a total height of 20,000 meters, and high-rise buildings with a building height greater than 30 meters have a total height of 12,000 meters, with corresponding percentages of 20%, 50%, and 30% respectively).

[0103] In this embodiment, the method for obtaining land management data by matching the BIM model with the GIS land features includes:

[0104] The BIM model is extracted to obtain BIM building information and BIM location information. Clustering is used to classify the BIM building information to obtain project information, building use, building materials and construction progress. Clustering is also used to classify GIS land features to obtain terrain information, building feature set, land use type and GPS information. The construction progress corresponds to different project information.

[0105] The BIM location information is coarsely matched with GPS information; the BIM location information is the building coordinate point; the GPS information is the area coordinate range; the GPS information is matched with multiple BIM location information.

[0106] Based on the timestamps of GIS land features and construction progress, information is located in a time sequence, and engineering information is matched with a set of building features. The engineering information includes the building information of a single construction project. The set of building features includes the building features of multiple construction projects within the regional coordinate range. The method of fine matching is to calculate the correlation between engineering information and building features, and select the building feature with the highest correlation with the engineering information as a pair of matching data within the regional coordinate range.

[0107] Based on the detailed matching results, land status data within the regional coordinate range is determined according to the time series; the land status data includes building use, construction period, and land use type.

[0108] Feature extraction is performed on the digital elevation model to obtain topographic elevation information and building elevation information. The topographic data and topographic elevation information are fused to obtain the first topographic information. Based on the fine matching results, the engineering information, building characteristics and building elevation information are fused to obtain the first building information. The first topographic information and the first building information are cross-referenced to obtain landform data.

[0109] Land utilization rate is determined based on land topography data; the land utilization rate includes spatial utilization rate and horizontal utilization rate.

[0110] In actual assessment, taking an industrial / residential area in Block A of a certain region as an example, the coordinate range of the industrial / residential area is determined, and BIM building models located in the area are selected based on BIM location information to complete coarse matching. The correlation between the engineering information (building footprint, floor height, building use type, etc.) of each building project BIM model and the building characteristics (building area, building height) of each building project in the coarse matching results is calculated to complete fine matching.

[0111] Based on the matching results, land status data, land topography data, and land use rate are determined as follows: 1. Land status data (building use and proportion: building area: industrial buildings 40%, residential buildings 24%, commercial buildings 16%, public facilities 12%, other buildings 8%; construction cycle and proportion: new buildings with a construction cycle of less than 5 years account for 48%, mid-term buildings with a construction cycle of 5-20 years account for 32%, old buildings with a construction cycle of more than 20 years account for 20%; land use type and proportion: industrial land accounts for 40%, residential land accounts for 24%, commercial land accounts for 16%, public land accounts for 12%, other land accounts for 8%), 2. Land topography data (buildings are mainly distributed on 64% of flat land and 24% of hills: low-rise buildings account for 32%, mid-rise buildings for 48%, high-rise buildings for 20%, and the remaining terrain areas are: mountains account for 8%, water areas account for 4%), 3. Land use rate (spatial utilization rate 70%, horizontal utilization rate 88%).

[0112] In this embodiment, the method for obtaining the land consolidation data based on the management path information includes:

[0113] Three-level path information is generated based on the data item category. The three-level path information is then input into a hash function to generate a three-level hash code. A category code is generated based on the index category quantity, data size, and three-level hash code of the data item. The three-level path information and the category code are uniquely corresponding.

[0114] The first and second data items within the same time period are combined into a data bundle. Secondary path information is generated based on the time interval of the data bundle. The secondary path information is input into a hash function to generate a secondary hash code. A time code is generated based on the time interval of the data bundle, the total number of index categories, and the secondary hash code. The secondary path information and the time code are uniquely corresponding.

[0115] Data bundles from different time periods within the same coordinate region are grouped into data blocks. First-level path information is generated based on the coordinate interval of the data blocks. The first-level path information is input into a hash function to generate a first-level hash code. A region code is generated based on the region size of the data block, the region boundary points, and the first-level hash code. The first-level path information and the region code are uniquely corresponding.

[0116] All data blocks within the managed area are combined into a dataset. The management path information consists of three-level path information, two-level path information, and one-level path information. The management code consists of a category code, a time code, and a region code. Data within the dataset is stored according to the management path information. The management path information and the management code are uniquely associated.

[0117] A query code is generated based on the query conditions, a management code is matched based on the query code, management path information is determined based on the management code, and land consolidation data is read based on the management path information; the land consolidation data consists of a first data item and a second data item.

[0118] In actual evaluation, taking an industrial / residential area within Block A of a certain region as an example, the hash function used is SHA-256. The third-level path information of the data in this area is "industrial-residential area", with a third-level hash code of H3. The corresponding category code for an indicator category of 5 and a data size of 1000MB is Hash(5,1000,H3). The second-level path information of the above data throughout 2024 is "20240101-20241231", with a second-level hash code of H2. The corresponding time code for a total indicator category of 10 is Hash(20240101-20241231,10,H2). The first-level path information of the above data in the interval enclosed by the coordinates (0,0), (100,0), (100,100), (0,100) is "(X 0,100 ,Y 0,100The hash code is H1, and the corresponding area code is Hash[10000,(0,0),(100,0),(100,100),(0,100),H1]; thus forming the corresponding management path "Information Industry-Living Area / 20240101-20241231 / (X) 0,100 ,Y 0,100 )” and manage the encoding Hash{(5,1000,H3),(20240101-20241231,10,H2),[10000,(0,0),(100,0),(100,100),(0,100),H1]};

[0119] When querying the five categories of indicators for industrial-residential areas located between coordinates (0,0), (100,0), (100,100), and (0,100) in Block A throughout 2024, a query code Hash{(5,1000),(20240101-20241231),[10000,(0,0),(100,0),(100,100),(0,100)]} is generated. The corresponding management code and management path are matched, and the land consolidation data corresponding to the query conditions can be viewed according to the management path.

[0120] Secondly, an integrated land consolidation data management system combining GIS and BIM includes:

[0121] Data acquisition module: used to acquire GIS information and BIM model of land, and classify the GIS information to obtain digital elevation model and digital orthophoto map;

[0122] Image processing module: used to perform geometric correction, color correction, blur detection and blur processing on the digital orthophoto image to obtain a clear digital orthophoto image, used to construct the image processing model, and input the clear digital orthophoto image to be processed into the image processing model to obtain GIS land features;

[0123] Data matching module: used to extract BIM building information and BIM location information from the BIM model, classify the BIM building information and the GIS land features, and match the BIM model and the GIS land features according to the classification results to obtain land management data;

[0124] Dataset module: used to construct a dataset based on the data item, the data bundle, and the data block, and to store, view, and manage the first data item and the second data item based on the dataset.

[0125] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A land consolidation data management method integrating GIS and BIM, characterized in that, Includes the following steps: S1. Obtain GIS information and BIM model of the land, and classify the GIS information to obtain digital elevation model and digital orthophoto map; S2. Process the digital orthophoto map to obtain GIS land features, and match the BIM model with the GIS land features to obtain land management data; the land management data includes land topography data, land status data, and land utilization rate; S3. The land topography data, the land status data and the space utilization rate are combined into a first data item, and the GIS information and the BIM model are combined into a second data item; S4. Construct a dataset to store land consolidation data, determine management path information through the dataset and query conditions, and obtain the land consolidation data based on the management path information; the land consolidation data includes the first data item and the second data item.

2. The land consolidation data management method integrating GIS and BIM according to claim 1, characterized in that, A method for processing the digital orthophoto map to obtain GIS land features includes: Geometric correction is performed on the digital orthophoto image, the histogram of the geometrically corrected digital orthophoto image is calculated, and color correction is performed on the digital orthophoto image based on the histogram of the digital orthophoto image; the color correction is achieved by performing histogram equalization on each color channel. Blur detection is performed on the color-corrected digital orthophoto image, and deblurring is performed based on the blur detection results to obtain a clear digital orthophoto image. Image processing models are used to process clear digital orthophoto maps to obtain GIS land features, and timestamps are set.

3. The land consolidation data management method integrating GIS and BIM according to claim 2, characterized in that, The method for obtaining the sharp image of the digital orthophoto includes: Blur detection is performed: the color-corrected digital orthophoto image is blurred and segmented, the comprehensive blur metric of each segmented image is calculated, and the blurred segmented image of the digital orthophoto image is determined based on the comprehensive blur metric result; the comprehensive blur metric includes a first blur metric and a second blur metric. Blurring processing is performed: based on the blurred segmented image, a non-blind deconvolution is used to obtain a sharp segmented approximate image. A domain transform recursive filter is used to perform edge processing on the sharp segmented approximate image to obtain a sharp segmented image. The sharp segmented images are then stitched together to obtain a sharp digital orthophoto image. The steps for obtaining the blurred segmented image include: The digital orthophoto image is blurred and segmented according to its content. The corresponding scale weight w is determined based on the area of ​​the i-th segmented image block. i The image is subjected to Discrete Fourier Transform to calculate the image power spectral difference. The power spectral difference is then converted into polar coordinates to determine the slope of the image power spectral density. Based on the slope of the image power spectral density, the first fuzzy metric q is determined. 1i The expected value of a dual Gaussian mixture model is used to represent the heavy-tailed distribution of the image gradient, and the second fuzzy metric q is determined based on the variance of the heavy-tailed distribution. 2i ; The combined fuzzy metric q of the i-th segmented image is obtained based on the first and second fuzzy metrics. i =c1q 1i +c2q 2i c1 and c2 are fuzzy metric weights. The segmented images with a comprehensive fuzzy metric greater than the comprehensive fuzzy metric threshold are selected as fuzzy segmented images. The iterative expression for obtaining a sharp segmented approximate image using non-blind deconvolution is as follows: in For the i-th blurred segmented image B i The sharp segmented approximate image for the (k+1)th iteration, with the initial iteration conditions being: K is the blur kernel corresponding to the point spread function, * represents the convolution operation, K H Let be the conjugate transpose of K, and λ1 and λ2 be regularization parameters. express The gradient; The row and column signals of each band of the clearly segmented approximate image are combined to obtain a one-dimensional signal. The one-dimensional signal is then processed iteratively using a domain transform recursive filter, which can be expressed as: J[k]\[1-(β t ) d ]S[k]+·(β t ) d J[k-1] Where J[k] is the result of the k-th filtering operation, S[k] is a one-dimensional signal, and the corresponding domain transform signal is... S0 is the value of the original signal at the starting position, |S j -S j-1 | represents the intensity difference of the signal at adjacent positions, j∈[1,i] is the index variable for summation, and ξ s ξ is the filter size control parameter. r Here, γ is the filter ambiguity control parameter, and Var(S) is the local variance weight. j-k；j+k ) as S j Let k be the variance of the central region and β be the region size. t Here, d represents the dynamic feedback coefficient, and d represents the two adjacent signals U. i and U i-1 The distance between them, Var[S(tM:t)] is the local variance of signal S from time tM to t, and M is the size of the local window.

4. The land consolidation data management method integrating GIS and BIM according to claim 2, characterized in that, A method for obtaining GIS land features by processing the digital orthophoto image with an image processing model includes: Digitized orthophotos were divided into training and testing sets. An image processing model is constructed, which includes a CNN network, a cross-attention map diffusion layer, a self-contrast loss function, and an output layer. A CNN network extracts features from the sharp image of a digital orthophoto to obtain a digital orthophoto feature map; a cross-attention map diffusion layer enhances the digital orthophoto feature map F to obtain an enhanced feature map F. ' The self-contrast loss function calculates the image loss and updates the model parameters; the output layer enhances the feature map F. ' Output GIS land features through a fully connected layer; The clear image of the digital orthophoto to be processed is input into the image processing model to obtain GIS land features; The image enhancement steps specifically include: segmenting the digital orthophoto feature map F using the SLIC superpixel segmentation algorithm to obtain superpixel features; defining each superpixel as a node; obtaining node feature vectors from the superpixel features; and calculating the similarity A between the feature vectors of two superpixel nodes using a Gaussian kernel function. ij According to similarity A ij Construct an adjacency matrix A, determine the k-closest neighbor graph from the adjacency matrix A, and normalize the adjacency matrix A to obtain the normalized adjacency matrix. Using a self-attention mechanism to process the normalized adjacency matrix We obtain a self-attention map S, and then perform cross-attention processing on the k-nearest neighbor map and the self-attention map S to obtain a cross-attention map C. Finally, we process the cross-attention map C to obtain the cross-attention feature Ci. F The graph diffusion module is used to apply cross-attention features C. F Image aggregation operation is performed to obtain the stationary state H of the image. diff For the cross-attention graph C and the stationary state H diff Information fusion is performed to obtain the enhanced feature map F'; The expression for the self-comparison loss function is as follows: Loss=μ1l pic +μ2l str Where Loss is the self-comparative loss function, μ1 and μ2 are the loss weights, and l pic For the graph self-contrast loss function, l str Let z be the structural self-comparison loss function. i To enhance the samples of feature map F', z j For z i Random samples within the same superpixel region, forming a positive sample pair, z k For z i Random samples from different superpixel regions form negative sample pairs, where N is the number of samples, sim(·) represents the similarity between samples, and τ is the temperature parameter.

5. The land consolidation data management method integrating GIS and BIM according to claim 1, characterized in that, A method for obtaining land management data by matching the BIM model with the GIS land features includes: The BIM model is extracted to obtain BIM building information and BIM location information. Clustering is used to classify the BIM building information to obtain project information, building use, building materials and construction progress. Clustering is also used to classify GIS land features to obtain terrain information, building feature set, land use type and GPS information. The construction progress corresponds to different project information. The BIM location information is coarsely matched with GPS information; the BIM location information is the building coordinate point; the GPS information is the area coordinate range; the GPS information is matched with multiple BIM location information. Based on the timestamps of GIS land features and construction progress, information is located in a time sequence, and engineering information is matched with a set of building features. The engineering information includes the building information of a single construction project. The set of building features includes the building features of multiple construction projects within the regional coordinate range. The method of fine matching is to calculate the correlation between engineering information and building features, and select the building feature with the highest correlation with the engineering information as a pair of matching data within the regional coordinate range. Based on the detailed matching results, land status data within the regional coordinate range is determined according to the time series; the land status data includes building use, construction period, and land use type. Feature extraction is performed on the digital elevation model to obtain topographic elevation information and building elevation information. The topographic data and topographic elevation information are fused to obtain the first topographic information. Based on the fine matching results, the engineering information, building characteristics and building elevation information are fused to obtain the first building information. The first topographic information and the first building information are cross-referenced to obtain landform data. Land utilization rate is determined based on land topography data; the land utilization rate includes spatial utilization rate and horizontal utilization rate.

6. The land consolidation data management method integrating GIS and BIM according to claim 1, characterized in that, The method for obtaining the land consolidation data based on the management path information includes: Three-level path information is generated based on the data item category. The three-level path information is then input into a hash function to generate a three-level hash code. A category code is generated based on the index category quantity, data size, and three-level hash code of the data item. The three-level path information and the category code are uniquely corresponding. The first and second data items within the same time period are combined into a data bundle. Secondary path information is generated based on the time interval of the data bundle. The secondary path information is input into a hash function to generate a secondary hash code. A time code is generated based on the time interval of the data bundle, the total number of index categories, and the secondary hash code. The secondary path information and the time code are uniquely corresponding. Data bundles from different time periods within the same coordinate region are grouped into data blocks. First-level path information is generated based on the coordinate interval of the data blocks. The first-level path information is input into a hash function to generate a first-level hash code. A region code is generated based on the region size of the data block, the region boundary points, and the first-level hash code. The first-level path information and the region code are uniquely corresponding. All data blocks within the managed area are combined into a dataset. The management path information consists of three-level path information, two-level path information, and one-level path information. The management code consists of a category code, a time code, and a region code. Data within the dataset is stored according to the management path information. The management path information and the management code are uniquely associated. A query code is generated based on the query conditions, a management code is matched based on the query code, management path information is determined based on the management code, and land consolidation data is read based on the management path information; the land consolidation data consists of a first data item and a second data item.

7. A land consolidation data management system integrating GIS and BIM, for performing the method according to any one of claims 1-6, characterized in that, include: Data acquisition module: used to acquire GIS information and BIM model of land, and classify the GIS information to obtain digital elevation model and digital orthophoto map; Image processing module: used to perform geometric correction, color correction, blur detection and blur processing on the digital orthophoto image to obtain a clear digital orthophoto image, used to construct the image processing model, and input the clear digital orthophoto image to be processed into the image processing model to obtain GIS land features; Data matching module: used to extract BIM building information and BIM location information from the BIM model, classify the BIM building information and the GIS land features, and match the BIM model and the GIS land features according to the classification results to obtain land management data; Dataset module: used to construct a dataset based on the data item, the data bundle, and the data block, and to store, view, and manage the first data item and the second data item based on the dataset.

Images