A method for collecting and analyzing surveying and mapping data for land use planning and investigation projects

Abstract

A method for collecting and analyzing surveying and mapping data for a land use planning and surveying project relates to the technical field of surveying and mapping data analysis. The method collects surveying and mapping data and additional attribute data within a preset range, performs weighted fusion on optical image data and microwave remote sensing data within the preset range, and obtains an optimized fused remote sensing image. A digital elevation model is constructed based on point cloud data, and a digital three-dimensional model is generated based on the digital elevation model combined with texture information of the optimized fused remote sensing image. A land use type classification model is constructed, and land type transfer hotspot grids in the digital three-dimensional model are obtained. Land use efficiency deviation analysis and planned ecological diversity analysis are performed on the land type transfer hotspot grids to obtain restricted land use types for inefficient use grids and recommended land use types or restricted land use types for efficient use grids, thereby significantly improving the accuracy and dynamics of land use analysis and the scientificity and practicality of land planning decision support.

Description

Technical Field

[0001] The present invention relates to the technical field of surveying and mapping data analysis, in particular to a method for collecting and analyzing surveying and mapping data for a land use planning and investigation project. Background Art

[0002] A Chinese patent with publication number CN118274805B discloses a surveying and mapping method based on remote sensing big data analysis, including: analyzing the current surveying and mapping operation environment suitability coefficient of the target task point from three dimensions: wind environment, atmospheric environment, and radiation environment, to determine whether the current environment of the target task point is suitable for surveying and mapping operations. If the judgment result is no, the target task point is marked as a delayed task point and its surveying and mapping task is suspended. If the judgment result is yes, a reference image is collected to further verify the accuracy of the current environmental suitability judgment of the target task point.

[0003] A Chinese patent with publication number CN118640866A discloses a land surveying and mapping data analysis method based on big data, including: obtaining high-precision remote sensing images within a preset range and preprocessing them to generate quantitative orthorectified images; automatically classifying and identifying land surveying and mapping data to extract land features; extracting and fusing the classified and identified land surveying and mapping data; using the generated high-level semantic information for mapping and visualization, and outputting standardized land surveying and mapping analysis map products; and evaluating the errors and uncertainties in the entire analysis process by comparing the data accuracy before and after analysis.

[0004] Existing surveying and mapping data collection and analysis technologies for land-use planning and survey projects suffer from problems such as single-source data collection, insufficient processing accuracy, limited analytical dimensions, and a lack of scientific planning and decision-making. Data collection relies on traditional ground-based technologies such as total stations and GPS, lacking the coordination of multi-source data such as meteorological and ecological data. Data processing is heavily manual, inefficient, and prone to errors. The fusion of optical and microwave remote sensing data often uses fixed weights, making it difficult to adapt to different scenarios. The analysis process is mostly based on two-dimensional planar data, lacking quantitative assessment of dynamic changes in land use. In planning and decision-making, land-use efficiency judgments often rely on experience, and ecological and economic factors are disconnected, making it difficult to achieve the precise allocation and sustainable use of land resources. Summary of the Invention

[0005] In order to solve the above technical problems, the purpose of the present invention is to provide a method for collecting and analyzing surveying and mapping data for land use planning and investigation projects, comprising the following steps:

[0006] Step s1: Collect surveying and mapping data and additional attribute data within a preset range and mark the collection period (usually half a day). The surveying and mapping data includes optical image data, microwave remote sensing data, and point cloud data. The additional attribute data includes meteorological data, biodiversity data, and socioeconomic data. The surveying and mapping data is preprocessed (including radiometric calibration, atmospheric correction, geometric correction, coordinate system unification, data format standardization, etc.).

[0007] Step s2: Obtain all land feature types within a preset range and the coverage area of ​​each land feature type, and perform weighted fusion of the optical image data and microwave remote sensing data within the preset range based on all land feature types within the preset range and the coverage area of ​​each land feature type to obtain an optimized fused remote sensing image;

[0008] Step s3: construct a digital elevation model based on the point cloud data, generate a digital three-dimensional model based on the digital elevation model and the texture information of the optimized fused remote sensing image, construct a land use type classification model, obtain the land use type and the area covered by each land use type for each triangular mesh in the digital three-dimensional model in the current acquisition cycle and several historical acquisition cycles, perform statistical analysis on the land use type and the area covered by each land use type, and obtain the land type transfer hotspot grid;

[0009] Step s4: Perform land use efficiency deviation analysis on the land type transfer hotspot grids, divide the transfer hotspot grids into high-efficiency utilization grids or low-efficiency utilization grids, obtain the restricted land use types of the low-efficiency utilization grids, perform planning ecological diversity analysis on the high-efficiency utilization grids, and obtain the recommended land use types or restricted land use types of the high-efficiency utilization grids.

[0010] Furthermore, the process of obtaining all land feature types within a preset range and the coverage area of ​​each land feature type includes:

[0011] Acquire surveying and mapping data and meteorological data within several collection periods, preset judgment index intervals for several meteorological indicators corresponding to clear weather (judgment indicators include cloud cover, visibility, temperature, and humidity, etc.), compare the meteorological indicators included in the meteorological data within the several collection periods with the corresponding judgment index intervals, and if the meteorological indicators included in the meteorological data within the collection period are all within the corresponding judgment threshold intervals, mark the collection period as a clear collection period;

[0012] Spectral feature extraction is performed on the optical image data within the clear acquisition period to obtain spectral reflectance data of different bands within the preset range. Based on the spectral reflectance data of different bands, all land feature types within the preset range and the coverage area of ​​each land feature type are obtained.

[0013] Furthermore, based on all land feature types within the preset range and the coverage area of ​​each land feature type, the optical image data and microwave remote sensing data within the preset range are weightedly fused to obtain an optimized fused remote sensing image, including:

[0014] A personalized weight feature model is constructed, and all land feature types within a preset range, the coverage area of ​​each land feature type, and several meteorological indicators included in the meteorological data in the current acquisition period are input into the personalized weight feature model. According to the personalized weight feature model, the optical image weight corresponding to the optical image data and the microwave remote sensing weight corresponding to the microwave remote sensing data are output. According to the optical image weight and the microwave remote sensing weight, each pixel in the optical image data and the microwave remote sensing data is weightedly fused to obtain the optimized fused remote sensing image of the current acquisition period.

[0015] Furthermore, the optical image data and the microwave remote sensing data are weightedly fused for each pixel according to the optical image weight and the microwave remote sensing weight to obtain the optimized fused remote sensing image. The process includes:

[0016] Assume that the pixel value of the optical image data is , the pixel value of microwave remote sensing data is , the optical image weight is , the microwave remote sensing weight is , pixel value after weighted fusion The calculation formula is: ;

[0017] For example, if the weight of the optical image data =0.7, the weight of microwave remote sensing data =0.3, a pixel value of optical image data =90, the corresponding pixel value of microwave remote sensing data =80, then the pixel value after weighted fusion =87.

[0018] Furthermore, the process of building a personalized weight feature model includes:

[0019] Based on the simulation learning concept, a simulator is used to generate a number of training data, wherein the training data include optical image data and microwave remote sensing data of different simulation areas under different meteorological data conditions. The simulation areas include several types of land objects and the coverage areas of several types of land objects. Different weight combinations of optical image data and microwave remote sensing data are preset (for example, the weights of optical image data are 0, 0.2, 0.4, 0.6, 0.8, and 1, and the corresponding weights of microwave remote sensing data are 1, 0.8, 0.6, 0.4, 0.2, and 0). According to different weight combinations, the optical image data and microwave remote sensing data of each simulation area under different meteorological data conditions are fused to obtain optimized fused remote sensing images of each simulation area under different weight combinations under different meteorological data conditions.

[0020] Preset standard image data for each simulation area, compare the standard image data for each simulation area with the optimized fused remote sensing images under different weight combinations of each simulation area under different meteorological data conditions, and obtain the error coefficients of the optimized fused remote sensing images under different weight combinations of each simulation area under different meteorological data conditions;

[0021] A personalized weight feature model is constructed, and the error coefficients of the optimized fused remote sensing images under different weight combinations of the several land feature types included in each simulation area, the coverage area of ​​the several land feature types, and the different meteorological data conditions are used as training sets and test sets. The training set is input into the personalized weight feature model for training until the loss function training is stable, and the model parameters are saved. The personalized weight feature model is tested by the test set until it meets the preset requirements, and the personalized weight feature model is output.

[0022] Furthermore, the process of constructing a digital elevation model based on point cloud data includes:

[0023] Preset an initial triangular mesh size, divide the preset range into several identical triangular meshes based on the initial triangular mesh size, obtain the elevation of each point in each triangular mesh based on the point cloud data, perform statistical analysis on the elevation of each point, and obtain the elevation standard deviation of each triangular mesh;

[0024] Construct a mesh size comparison table, wherein the mesh size comparison table includes triangular mesh sizes corresponding to different elevation standard deviations, wherein the larger the elevation standard deviation, the smaller the triangular mesh size; obtain the triangular mesh size corresponding to each initial triangular mesh according to the elevation standard deviation of each initial triangular mesh and the mesh size comparison table; divide each initial triangular mesh into a number of identical new triangular meshes according to the triangular mesh size corresponding to each initial triangular mesh; wherein, if the new triangular mesh size is consistent with the initial triangular mesh size, do not divide the initial triangular mesh, and mark the initial triangular mesh as a new triangular mesh; construct a digital elevation model according to each new triangular mesh and the elevation of each point in each new triangular mesh.

[0025] Furthermore, the process of generating a digital 3D model based on the digital elevation model combined with optimized texture information of the fused remote sensing image includes:

[0026] Obtain the coordinate matching relationship between the digital elevation model and the optimized fused remote sensing image, perform texture sampling on the optimized fused remote sensing image, and attach the sampled texture to the corresponding position on the surface of the digital elevation model according to the coordinate matching relationship to generate a digital three-dimensional model. For example, use the optimized fused remote sensing image to map the textures of buildings, roads, vegetation, etc. onto the digital elevation model, so that the three-dimensional terrain model can more intuitively and realistically reflect the distribution of forests, exposed rocks, valley rivers, roads, vegetation and other land features.

[0027] Furthermore, the process of constructing a land use type classification model and obtaining the land use type of each triangular mesh in the digital 3D model in the current acquisition cycle and several historical acquisition cycles, as well as the area covered by each land use type, includes:

[0028] A supervised classification method is used to construct a land use type classification model. A certain number of training samples are selected from the image in advance, and then a classification model is established based on the characteristics of the samples to classify the entire image. Common land use types include cultivated land, forest land, grassland, construction land, water area, etc. The classification accuracy of the land use type classification model should be guaranteed through field verification and accuracy evaluation. Generally, the classification accuracy is required to reach more than 85%. The terrain characteristics (terrain relief, slope, slope direction), land feature type and the occupied area of ​​the land feature type of each triangular grid in the digital three-dimensional model of the current acquisition period are extracted. The terrain characteristics, land feature type and the occupied area of ​​the land feature type of the triangular grid are input into the land use type classification model. According to the land classification model, the land use type in the triangular grid of the current acquisition period and the area covered by each land use type are output.

[0029] Furthermore, the process of performing statistical analysis on the land use types and the areas covered by each land use type to obtain land type transfer hotspot grids includes:

[0030] According to the land use types in each triangular mesh in the current collection period and several historical collection periods, as well as the area covered by each land use type, the land class transfer coefficient of each triangular mesh is obtained, and a coefficient threshold is preset. If the land class transfer coefficient of a triangular mesh is greater than the coefficient threshold, the triangular mesh is marked as a land class transfer hotspot mesh.

[0031] Furthermore, the calculation formula for obtaining the land class transfer coefficient of each triangular mesh is:

[0032] ;

[0033] in, represents the transfer coefficient from land use type i to land use type j, is the area converted from land use type i to land use type j, is the total area of ​​initial land use type i;

[0034] Furthermore, land use efficiency deviation analysis is performed on the land type transfer hotspot grids, and the transfer hotspot grids are divided into high-efficiency utilization grids or low-efficiency utilization grids. The process of obtaining the restricted land use type of the low-efficiency utilization grids includes:

[0035] Preset the degree classification index of different land use types, and based on the degree classification index of different land use types, the land use types of the class transfer hotspot grids in the current collection period and several historical collection periods, and the area covered by each land use type, obtain the land use index of the class transfer hotspot grids in the current collection period and several historical collection periods;

[0036] The calculation formula for obtaining the land use index is:

[0037] ;

[0038] in, is the land use index, is the grading index of land use degree of type i (e.g. unused land = 1, cultivated land = 2, construction land = 4), is the proportion of land use type i;

[0039] The land use index change rate of the land use hotspot grids within the current collection period and several historical collection periods is used to obtain the land use index change rate. At the same time, the economic output change rate of the land use hotspot grids is obtained based on the socioeconomic data.

[0040] The efficiency deviation index is obtained based on the land use index change rate and the economic output change rate. The efficiency deviation index threshold is preset (usually 1). If the efficiency deviation index of the land type transfer hotspot grid is less than or equal to the efficiency deviation index threshold, the transfer hotspot grid is marked as an efficient utilization grid. If the efficiency deviation index is greater than the efficiency deviation index threshold, the transfer hotspot grid is marked as an inefficient utilization grid.

[0041] The calculation formula for obtaining the efficiency deviation index based on the land use index change rate and economic output change rate is:

[0042] ;

[0043] in, is the efficiency deviation index, is the land use index change rate, is the rate of change of economic output;

[0044] According to the land type transfer coefficient greater than the coefficient threshold in the inefficient use grid, the restricted land use type of the inefficient use grid is obtained, such as the transfer coefficient of land use type i to land use type j. If is greater than the coefficient threshold, the land use type j is marked as a restricted land use type.

[0045] Furthermore, the process of conducting planning ecological diversity analysis on the high-efficiency utilization grid and obtaining the recommended land use type or restricted land use type of the high-efficiency utilization grid includes:

[0046] Obtain the planned ecological diversity threshold of each efficient utilization grid, compare the ecological diversity data of each efficient utilization grid with the planned ecological diversity threshold, and if the ecological diversity data is greater than or equal to the planned ecological diversity threshold, determine the recommended land use type of the efficient utilization grid based on the land class transfer coefficient greater than the coefficient threshold in the efficient utilization grid, for example If the land use type j is greater than the coefficient threshold, the land use type j is marked as the recommended land use type. If the ecological diversity data is less than the planned ecological diversity threshold, the restricted land use type of the efficient utilization grid is determined based on the land class transfer coefficient greater than the coefficient threshold in the efficient utilization grid.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] 1. Comprehensiveness and intelligence of data collection and processing

[0049] Collaborative multi-source data collection: This approach integrates multi-source mapping data, including optical imaging, microwave remote sensing, and point clouds, while simultaneously collecting additional attribute data such as meteorological, ecological, and socioeconomic data, breaking through the traditional single-source data collection model. This multi-source data fusion can comprehensively reflect the current land use situation from multiple dimensions, including spatial, temporal, environmental, and social dimensions, providing richer and more accurate foundational data for planning and avoiding planning deviations caused by data gaps.

[0050] Intelligent data preprocessing: Meteorological data is used to screen clear collection periods to ensure the quality of optical image data and reduce errors caused by weather factors; a personalized weight feature model is used to perform weighted fusion of optical and microwave remote sensing data. This model is based on the idea of ​​simulation learning and optimizes weight combinations through a large amount of training data. It can dynamically adjust the fusion weight according to different land feature types, coverage areas and meteorological conditions. Compared with traditional fixed weight fusion methods, it significantly improves the clarity of remote sensing images and the recognition of land features, providing high-quality data support for subsequent land feature identification and analysis.

[0051] 2. Accuracy and dynamism of land use analysis

[0052] High-precision 3D modeling: When constructing digital elevation models, a terrain-adaptive triangulated meshing method is used. The mesh size is dynamically adjusted based on the elevation standard deviation of the point cloud data. Small meshes are used in complex terrain to ensure accuracy, while large meshes are used in flat areas to improve computational efficiency. Compared to traditional fixed meshing methods, this method improves the vertical accuracy of digital elevation models to ±0.3m (plains) and ±0.5m (mountainous areas), while reducing data computation by 40%. The digital 3D model, generated by optimizing and integrating texture information from remote sensing imagery, can accurately restore the three-dimensional shape of the earth's surface and its features through precise coordinate matching and texture mapping, providing intuitive and accurate 3D visualization scenarios for land use planning.

[0053] Dynamic Change Monitoring and Hotspot Identification: By constructing a land use classification model, we conduct statistical analysis of the land use types and coverage areas of triangular meshes in the digital 3D model during the current and historical acquisition periods. We then use land class transition coefficients to accurately identify hotspot meshes where land class transitions occur. This time-series-based dynamic analysis can promptly capture land use change trends and hotspot areas. Compared to traditional static analysis, it can identify potential land use change issues in advance, providing a timely basis for planning adjustments and decision-making.

[0054] 3. Scientificity and practicality of planning and decision-making support

[0055] Accurate Land Use Efficiency Assessment: Based on the rate of change of the land use index and the rate of change of economic output, an efficiency deviation index is calculated. This is used to divide grids into high-efficiency and low-efficiency grids, and the restricted land use types of the low-efficiency grids are determined. This method quantifies the relationship between land use efficiency and economic output. Compared with traditional empirical judgment, it can more scientifically and accurately identify inefficient land use areas and issues, providing quantitative support for decision-making in planning, such as optimizing land resource allocation, guiding the intensive transformation of industrial land, and revitalizing existing land.

[0056] Coordinated Ecological and Economic Planning: Efficiently utilize grids for planned ecological diversity analysis. Combining ecological diversity data with planned ecological diversity thresholds, we determine recommended or restricted land use types, achieving a coordinated consideration of ecological protection and economic development in land use planning. This approach addresses the traditional separation between ecology and economy in planning, helping to rationally plan land use while safeguarding the ecological environment, improving the overall benefits of land use, and promoting regional sustainable development. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 This is a schematic diagram of a method for collecting and analyzing surveying data for a land use planning and investigation project according to an embodiment of the present application. DETAILED DESCRIPTION

[0058] The following is a clear and complete description of the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0059] like Figure 1 As shown, a method for collecting and analyzing surveying and mapping data for a land use planning and investigation project includes the following steps:

[0060] Step s1: Collect surveying and mapping data and additional attribute data within a preset range and mark the collection period (usually half a day). The surveying and mapping data includes optical image data, microwave remote sensing data, and point cloud data. The additional attribute data includes meteorological data, biodiversity data, and socioeconomic data. The surveying and mapping data is preprocessed (including radiometric calibration, atmospheric correction, geometric correction, coordinate system unification, data format standardization, etc.).

[0061] Step s2: Obtain all land feature types within a preset range and the coverage area of ​​each land feature type, and perform weighted fusion of the optical image data and microwave remote sensing data within the preset range based on all land feature types within the preset range and the coverage area of ​​each land feature type to obtain an optimized fused remote sensing image;

[0062] Step s3: construct a digital elevation model based on the point cloud data, generate a digital three-dimensional model based on the digital elevation model and the texture information of the optimized fused remote sensing image, construct a land use type classification model, obtain the land use type and the area covered by each land use type for each triangular mesh in the digital three-dimensional model in the current acquisition cycle and several historical acquisition cycles, perform statistical analysis on the land use type and the area covered by each land use type, and obtain the land type transfer hotspot grid;

[0063] Step s4: Perform land use efficiency deviation analysis on the land type transfer hotspot grids, divide the transfer hotspot grids into high-efficiency utilization grids or low-efficiency utilization grids, obtain the restricted land use types of the low-efficiency utilization grids, perform planning ecological diversity analysis on the high-efficiency utilization grids, and obtain the recommended land use types or restricted land use types of the high-efficiency utilization grids.

[0064] It should be further explained that, in a specific implementation process, the process of obtaining all land feature types within a preset range and the coverage area of ​​each land feature type includes:

[0065] Acquire surveying and mapping data and meteorological data within several collection periods, preset judgment index intervals for several meteorological indicators corresponding to clear weather (judgment indicators include cloud cover, visibility, temperature, and humidity, etc.), compare the meteorological indicators included in the meteorological data within the several collection periods with the corresponding judgment index intervals, and if the meteorological indicators included in the meteorological data within the collection period are all within the corresponding judgment threshold intervals, mark the collection period as a clear collection period;

[0066] Spectral feature extraction is performed on the optical image data within the clear acquisition period to obtain spectral reflectance data of different bands within the preset range. Based on the spectral reflectance data of different bands, all land feature types within the preset range and the coverage area of ​​each land feature type are obtained.

[0067] It should be further explained that, in a specific implementation process, the process of obtaining all ground object types within a preset range and the coverage area of ​​each ground object type based on spectral reflectance data of different bands includes:

[0068] Typical land features are selected from the image as training samples. For example, 100 vegetation sample points, 80 water sample points, 120 construction land sample points, and 60 bare land sample points are selected. A support vector machine (SVM) classification algorithm is used to train a model based on the spectral reflectance values ​​of the samples (such as red light band reflectance and near-infrared band reflectance). Each pixel in the image is classified as vegetation, water body, construction land, or bare land according to the model.

[0069] After classification, count the number of pixels of each land feature type. Assume that the vegetation class has 80,000 pixels, the water class has 20,000 pixels, the construction land class has 60,000 pixels, and the bare land class has 40,000 pixels.

[0070] If the image resolution is 10 meters, the area of ​​a single pixel is 10m×10m=100 square meters.

[0071] Vegetation coverage area = 80,000 × 100 square meters = 8,000,000 square meters = 800 hectares;

[0072] Water body coverage area = 20,000 × 100 square meters = 2,000,000 square meters = 200 hectares;

[0073] Construction land coverage area = 60,000 × 100 square meters = 6,000,000 square meters = 600 hectares;

[0074] Bare land coverage = 40,000 × 100 square meters = 4,000,000 square meters = 400 hectares.

[0075] It should be further explained that, in the specific implementation process, based on all the land feature types within the preset range and the coverage area of ​​each land feature type, the optical image data and microwave remote sensing data within the preset range are weightedly fused to obtain the optimized fused remote sensing image. The process includes:

[0076] A personalized weight feature model is constructed, and all land feature types within a preset range, the coverage area of ​​each land feature type, and several meteorological indicators included in the meteorological data in the current acquisition period are input into the personalized weight feature model. According to the personalized weight feature model, the optical image weight corresponding to the optical image data and the microwave remote sensing weight corresponding to the microwave remote sensing data are output. According to the optical image weight and the microwave remote sensing weight, each pixel in the optical image data and the microwave remote sensing data is weightedly fused to obtain the optimized fused remote sensing image of the current acquisition period.

[0077] In areas with sparse vegetation, bare ground, and clear, cloudless weather, optical imagery can clearly reveal terrain texture and details, such as rock distribution and soil type. In these cases, optical imagery can be given a relatively higher weight. For example, in desert areas, optical imagery can well reveal the shape and orientation of sand dunes, while radar imagery provides limited supplementary information on this relatively simple terrain without vegetation obstruction. Therefore, a weight of 70%-80% can be assigned to optical imagery, while 20%-30% can be assigned to radar imagery.

[0078] In densely vegetated areas, such as tropical rainforests, and in cloudy or foggy weather, radar imagery can penetrate the vegetation and capture the terrain beneath it. Because optical imagery is obscured by vegetation, it's difficult to directly observe the true terrain. Therefore, radar imagery should be given a relatively high weight. A weight of 60%-70% can be assigned to radar imagery, while optical imagery should be assigned a 30%-40% weight.

[0079] It should be further explained that, in the specific implementation process, the optical image data and microwave remote sensing data are weightedly fused for each pixel according to the optical image weight and the microwave remote sensing weight to obtain the optimized fused remote sensing image. The process includes:

[0080] Assume that the pixel value of the optical image data is , the pixel value of microwave remote sensing data is , the optical image weight is , the microwave remote sensing weight is , pixel value after weighted fusion The calculation formula is: ;

[0081] For example, if the weight of the optical image data =0.7, the weight of microwave remote sensing data =0.3, a pixel value of optical image data =90, the corresponding pixel value of microwave remote sensing data =80, then the pixel value after weighted fusion =87.

[0082] It should be further explained that, in the specific implementation process, the process of building a personalized weight feature model includes:

[0083] Based on the simulation learning concept, a simulator is used to generate a number of training data, wherein the training data include optical image data and microwave remote sensing data of different simulation areas under different meteorological data conditions. The simulation areas include several types of land objects and the coverage areas of several types of land objects. Different weight combinations of optical image data and microwave remote sensing data are preset (for example, the weights of optical image data are 0, 0.2, 0.4, 0.6, 0.8, and 1, and the corresponding weights of microwave remote sensing data are 1, 0.8, 0.6, 0.4, 0.2, and 0). According to different weight combinations, the optical image data and microwave remote sensing data of each simulation area under different meteorological data conditions are fused to obtain optimized fused remote sensing images of each simulation area under different weight combinations under different meteorological data conditions.

[0084] Preset standard image data for each simulation area, compare the standard image data for each simulation area with the optimized fused remote sensing images under different weight combinations of each simulation area under different meteorological data conditions, and obtain the error coefficients of the optimized fused remote sensing images under different weight combinations of each simulation area under different meteorological data conditions;

[0085] A personalized weight feature model is constructed, and the error coefficients of the optimized fused remote sensing images under different weight combinations of the several land feature types included in each simulation area, the coverage area of ​​the several land feature types, and the different meteorological data conditions are used as training sets and test sets. The training set is input into the personalized weight feature model for training until the loss function training is stable, and the model parameters are saved. The personalized weight feature model is tested by the test set until it meets the preset requirements, and the personalized weight feature model is output.

[0086] The present invention selects a multi-branch fully connected network as the deep learning architecture and the Mean Sequential Error (MSE) loss function. Training is then initiated by inputting the prepared training set into the selected deep learning model. During training, weights are continuously updated using a backpropagation algorithm, gradually reducing the loss function until a steady state is reached. During this process, techniques such as early stopping are utilized to prevent overfitting. In addition to the basic training process, various model parameters, including the learning rate, batch size, and regularization coefficient, are tuned using GridSearch.

[0087] After model training is complete and parameter adjustments are complete, a final evaluation is performed on the test set to obtain the model's evaluation results. These results include classification metrics such as accuracy, recall, and F1 score. Based on the evaluation results on the test set, it is determined whether the model meets the expected standards. If the requirements are met, the model parameters are saved and prepared for deployment. If not, it is necessary to return to a previous stage to re-examine issues such as data quality, model structure, or training strategy.

[0088] It should be further explained that, in the specific implementation process, the process of constructing a digital elevation model based on point cloud data includes:

[0089] Preset an initial triangular mesh size, divide the preset range into several identical triangular meshes based on the initial triangular mesh size, and when the preset range cannot be exactly divided into several identical triangular meshes, add some auxiliary triangular meshes of different sizes to the edge of the preset range to fill the gaps caused by the indivisibility. Obtain the elevation of each point in each triangular mesh based on the point cloud data, perform statistical analysis on the elevation of each point, and obtain the standard deviation of the elevation of each triangular mesh;

[0090] Construct a mesh size comparison table, wherein the mesh size comparison table includes triangular mesh sizes corresponding to different elevation standard deviations, wherein the larger the elevation standard deviation, the smaller the triangular mesh size; obtain the triangular mesh size corresponding to each initial triangular mesh according to the elevation standard deviation of each initial triangular mesh and the mesh size comparison table; divide each initial triangular mesh into a number of identical new triangular meshes according to the triangular mesh size corresponding to each initial triangular mesh; wherein, if the new triangular mesh size is consistent with the initial triangular mesh size, do not divide the initial triangular mesh, and mark the initial triangular mesh as a new triangular mesh; construct a digital elevation model according to each new triangular mesh and the elevation of each point in each new triangular mesh.

[0091] Breaking through the limitations of traditional fixed grid sizes (such as a uniform 5m×5m), the grid density is dynamically adjusted based on the standard deviation of point cloud elevation: small grids (1m×1m) are used in areas with dramatic terrain fluctuations (such as mountainous areas with a standard deviation >2m), while large grids (5m×5m) are used in flat areas (such as plains with a standard deviation <0.5m). This reduces data calculations by 40% while ensuring accuracy.

[0092] It should be further explained that, in the specific implementation process, the process of generating a digital 3D model based on the digital elevation model combined with optimized fusion of remote sensing image texture information includes:

[0093] Obtain the coordinate matching relationship between the digital elevation model and the optimized fused remote sensing image, perform texture sampling on the optimized fused remote sensing image, and attach the sampled texture to the corresponding position on the surface of the digital elevation model according to the coordinate matching relationship to generate a digital three-dimensional model. For example, use the optimized fused remote sensing image to map the textures of buildings, roads, vegetation, etc. onto the digital elevation model, so that the three-dimensional terrain model can more intuitively and realistically reflect the distribution of forests, exposed rocks, valley rivers, roads, vegetation and other land features.

[0094] It should be further explained that, in the specific implementation process, the process of constructing a land use type classification model and obtaining the land use type of each triangular mesh in the digital 3D model in the current acquisition cycle and several historical acquisition cycles, as well as the area covered by each land use type, includes:

[0095] A supervised classification method is used to construct a land use type classification model. A certain number of training samples are selected from the image in advance, and then a classification model is established based on the characteristics of the samples to classify the entire image. Common land use types include cultivated land, forest land, grassland, construction land, water area, etc. The classification accuracy of the land use type classification model should be guaranteed through field verification and accuracy evaluation. Generally, the classification accuracy is required to reach more than 85%. The terrain characteristics (terrain relief, slope, slope direction), land feature type and the occupied area of ​​the land feature type of each triangular grid in the digital three-dimensional model of the current acquisition period are extracted. The terrain characteristics, land feature type and the occupied area of ​​the land feature type of the triangular grid are input into the land use type classification model. According to the land classification model, the land use type in the triangular grid of the current acquisition period and the area covered by each land use type are output.

[0096] It should be further explained that, in the specific implementation process, the process of statistically analyzing the land use types and the areas covered by each land use type to obtain the land type transfer hotspot grid includes:

[0097] According to the land use types in each triangular mesh in the current collection period and several historical collection periods, as well as the area covered by each land use type, the land class transfer coefficient of each triangular mesh is obtained, and a coefficient threshold is preset. If the land class transfer coefficient of a triangular mesh is greater than the coefficient threshold, the triangular mesh is marked as a land class transfer hotspot mesh.

[0098] It should be further explained that, in the specific implementation process, the calculation formula for obtaining the land type transfer coefficient of each triangular grid is:

[0099] ;

[0100] in, represents the transfer coefficient from land use type i to land use type j, is the area converted from land use type i to land use type j, is the total area of ​​initial land use type i;

[0101] It should be further explained that, in the specific implementation process, land use efficiency deviation analysis is performed on the land type transfer hotspot grids, and the transfer hotspot grids are divided into high-efficiency utilization grids or low-efficiency utilization grids. The process of obtaining the restricted land use type of the low-efficiency utilization grids includes:

[0102] Preset the degree classification index of different land use types, and based on the degree classification index of different land use types, the land use types of the class transfer hotspot grids in the current collection period and several historical collection periods, and the area covered by each land use type, obtain the land use index of the class transfer hotspot grids in the current collection period and several historical collection periods;

[0103] The calculation formula for obtaining the land use index is:

[0104] ;

[0105] in, is the land use index, is the grading index of land use degree of type i (e.g. unused land = 1, cultivated land = 2, construction land = 4), is the proportion of land use type i;

[0106] The land use index change rate of the land use hotspot grids within the current collection period and several historical collection periods is used to obtain the land use index change rate. At the same time, the economic output change rate of the land use hotspot grids is obtained based on the socioeconomic data.

[0107] The efficiency deviation index is obtained based on the land use index change rate and the economic output change rate. The efficiency deviation index threshold is preset (usually 1). If the efficiency deviation index of the land type transfer hotspot grid is less than or equal to the efficiency deviation index threshold, the transfer hotspot grid is marked as an efficient utilization grid. If the efficiency deviation index is greater than the efficiency deviation index threshold, the transfer hotspot grid is marked as an inefficient utilization grid.

[0108] The calculation formula for obtaining the efficiency deviation index based on the land use index change rate and economic output change rate is:

[0109] ;

[0110] in, is the efficiency deviation index, is the land use index change rate, is the rate of change of economic output;

[0111] According to the land class transfer coefficient greater than the coefficient threshold in the inefficient use grid, the restricted land use type of the inefficient use grid is obtained, for example If is greater than the coefficient threshold, the land use type j is marked as a restricted land use type.

[0112] For example, the land utilization index of an industrial zone increased by 30% in ten years, but the GDP output per unit area decreased, indicating that there were idle factories or insufficient floor area ratio. The industrial land would be marked as restricted land use type, and the land use planning could guide the intensive transformation of industrial land.

[0113] It should be further explained that, in the specific implementation process, the process of conducting planning ecological diversity analysis on the high-efficiency utilization grid and obtaining the recommended land use type or restricted land use type of the high-efficiency utilization grid includes:

[0114] Obtain the planned ecological diversity threshold of each efficient utilization grid, compare the ecological diversity data of each efficient utilization grid with the planned ecological diversity threshold, and if the ecological diversity data is greater than or equal to the planned ecological diversity threshold, determine the recommended land use type of the efficient utilization grid based on the land type transfer coefficient greater than the coefficient threshold in the efficient utilization grid, such as the transfer coefficient of land use type i to land use type j. If the land use type j is greater than the coefficient threshold, the land use type j is marked as the recommended land use type. If the ecological diversity data is less than the planned ecological diversity threshold, the restricted land use type of the efficient utilization grid is determined based on the land class transfer coefficient greater than the coefficient threshold in the efficient utilization grid.

[0115] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A method for collecting and analyzing surveying and mapping data for a land use planning and investigation project, characterized in that: The following steps are involved: Step s1: collecting surveying and mapping data and additional attribute data within a preset range and marking the collection period. The surveying and mapping data includes optical image data, microwave remote sensing data, and point cloud data. The additional attribute data includes meteorological data, biodiversity data, and socioeconomic data. The surveying and mapping data is then preprocessed. Step s2: Obtain all land feature types within a preset range and the coverage area of ​​each land feature type, and perform weighted fusion of the optical image data and microwave remote sensing data within the preset range based on all land feature types within the preset range and the coverage area of ​​each land feature type to obtain an optimized fused remote sensing image; Step s3: Preset the initial triangular mesh size, divide the preset range into several identical triangular meshes according to the initial triangular mesh size, obtain the elevation of each point in each triangular mesh based on the point cloud data, perform statistical analysis on the elevation of each point, and obtain the elevation standard deviation of each triangular mesh; Constructing a mesh size comparison table, the mesh size comparison table including triangular mesh sizes corresponding to different elevation standard deviations, obtaining the triangular mesh size corresponding to each initial triangular mesh based on the elevation standard deviation of each initial triangular mesh and the mesh size comparison table, dividing each initial triangular mesh into a plurality of identical new triangular meshes based on the triangular mesh size corresponding to each initial triangular mesh, and constructing a digital elevation model based on the elevations of each new triangular mesh and each point within each new triangular mesh; Based on the digital elevation model combined with optimized fusion of remote sensing image texture information, a digital 3D model is generated, a land use type classification model is constructed, and the land use type of each triangular mesh in the digital 3D model in the current acquisition cycle and several historical acquisition cycles, as well as the area covered by each land use type, are obtained; Based on the land use types and the area covered by each land use type in each triangular mesh in the current collection cycle and several historical collection cycles, the land class transfer coefficient of each triangular mesh is obtained, and a coefficient threshold is preset. If the land class transfer coefficient of a triangular mesh is greater than the coefficient threshold, the triangular mesh is marked as a land class transfer hotspot mesh; Step s4: Preset the degree classification index of different land use types, and based on the degree classification index of different land use types, the land use types of the class transfer hotspot grids in the current collection period and several historical collection periods, and the area covered by each land use type, obtain the land use index of the class transfer hotspot grids in the current collection period and several historical collection periods; The land use index change rate of the land use hotspot grids within the current collection period and several historical collection periods is used to obtain the land use index change rate. At the same time, the economic output change rate of the land use hotspot grids is obtained based on the socioeconomic data. The efficiency deviation index is obtained based on the change rate of the land use index and the change rate of economic output. The efficiency deviation index threshold is preset. If the efficiency deviation index of the land type transfer hotspot grid is less than or equal to the efficiency deviation index threshold, the transfer hotspot grid is marked as an efficient utilization grid. If the efficiency deviation index is greater than the efficiency deviation index threshold, the transfer hotspot grid is marked as an inefficient utilization grid. According to the land type transfer coefficient greater than the coefficient threshold in the low-efficiency utilization grid, the restricted land use type of the low-efficiency utilization grid is obtained; At the same time, the planned ecological diversity threshold of each high-efficiency utilization grid is obtained, and the ecological diversity data of each high-efficiency utilization grid is compared with the planned ecological diversity threshold. If the ecological diversity data is greater than or equal to the planned ecological diversity threshold, the recommended land use type of the high-efficiency utilization grid is determined based on the land class transfer coefficient in the high-efficiency utilization grid that is greater than the coefficient threshold. If the ecological diversity data is less than the planned ecological diversity threshold, the restricted land use type of the high-efficiency utilization grid is determined based on the land class transfer coefficient in the high-efficiency utilization grid that is greater than the coefficient threshold.

2. The method for collecting and analyzing surveying and mapping data for a land use planning and investigation project according to claim 1, characterized in that: The process of obtaining all land feature types within a preset range and the coverage area of ​​each land feature type includes: Acquire surveying and mapping data and meteorological data within a number of collection periods, preset judgment index intervals for a number of meteorological indicators corresponding to clear weather, compare the meteorological indicators included in the meteorological data within the number of collection periods with the corresponding judgment index intervals, and mark the collection period as a clear collection period if the meteorological indicators included in the meteorological data within the collection period are all within the corresponding judgment threshold intervals; Spectral feature extraction is performed on the optical image data within the clear acquisition period to obtain spectral reflectance data of different bands within the preset range. Based on the spectral reflectance data of different bands, all land feature types within the preset range and the coverage area of ​​each land feature type are obtained.

3. The method for collecting and analyzing surveying and mapping data for a land use planning and investigation project according to claim 2, characterized in that: Based on all the land feature types within the preset range and the coverage area of ​​each land feature type, the optical image data and microwave remote sensing data within the preset range are weightedly fused to obtain the optimized fused remote sensing image. The process includes: A personalized weight feature model is constructed, and all land feature types within a preset range, the coverage area of ​​each land feature type, and several meteorological indicators included in the meteorological data in the current acquisition period are input into the personalized weight feature model. According to the personalized weight feature model, the optical image weight corresponding to the optical image data and the microwave remote sensing weight corresponding to the microwave remote sensing data are output. According to the optical image weight and the microwave remote sensing weight, each pixel in the optical image data and the microwave remote sensing data is weightedly fused to obtain the optimized fused remote sensing image of the current acquisition period.

4. The method for collecting and analyzing surveying and mapping data for a land use planning and investigation project according to claim 3, characterized in that: The process of building a personalized weighted feature model includes: Based on the concept of simulation learning, a simulator is used to generate a number of training data, wherein the training data includes optical image data and microwave remote sensing data of different simulation areas under different meteorological data conditions. The simulation areas include several types of land features and the coverage areas of the several types of land features. Different weight combinations of optical image data and microwave remote sensing data are preset. The optical image data and microwave remote sensing data of each simulation area under different meteorological data conditions are fused according to the different weight combinations to obtain optimized fused remote sensing images of each simulation area under different meteorological data conditions under different weight combinations. Preset standard image data for each simulation area, compare the standard image data for each simulation area with the optimized fused remote sensing images under different weight combinations of each simulation area under different meteorological data conditions, and obtain the error coefficients of the optimized fused remote sensing images under different weight combinations of each simulation area under different meteorological data conditions; A personalized weight feature model is constructed, and the error coefficients of the optimized fused remote sensing images under different weight combinations of the several land feature types included in each simulation area, the coverage area of ​​the several land feature types, and the different meteorological data conditions are used as training sets and test sets. The training set is input into the personalized weight feature model for training until the loss function training is stable, and the model parameters are saved. The personalized weight feature model is tested by the test set until it meets the preset requirements, and the personalized weight feature model is output.

5. The method for collecting and analyzing surveying and mapping data for land use planning and investigation projects according to claim 4, characterized in that: The process of generating a digital 3D model based on the digital elevation model combined with optimized texture information of remote sensing images includes: The coordinate matching relationship between the digital elevation model and the optimized fused remote sensing image is obtained, texture sampling is performed on the optimized fused remote sensing image, and the sampled texture is attached to the corresponding position on the surface of the digital elevation model according to the coordinate matching relationship to generate a digital three-dimensional model.

6. A method for collecting and analyzing surveying and mapping data for a land use planning and investigation project according to claim 5, characterized in that: The process of building a land use classification model and obtaining the land use type of each triangular grid and the area covered by each land use type includes: Construct a land use type classification model, extract the terrain characteristics, feature types and occupied areas of each triangular grid in the digital 3D model of the current acquisition period, input the terrain characteristics, feature types and occupied areas of the feature types of the triangular grid into the land use type classification model, and output the land use type in the triangular grid of the current acquisition period and the area covered by each land use type according to the land classification model.

Images