A Geographic Information Mapping Method Based on Remote Sensing Technology

Abstract

The present invention discloses a geographic information mapping method based on remote sensing technology, which relates to the field of remote sensing mapping technology. Remote sensing images of the area to be mapped are collected by remote sensing equipment for feature processing to obtain information features of the remote sensing images in different temporal dimensions. The spatio-temporal registration algorithm is used to unify the information features of different temporal dimensions into the same geographic coordinate system, and several buildings in the geographic coordinate system are identified, and ground object feature extraction is carried out to obtain the point cloud data and image data of each building. The corresponding data, point cloud data and image data obtained by ground object feature extraction are fused to construct a three-dimensional projection model of each building. Based on the three-dimensional projection model of each building, building projection information is obtained, and a feature extraction network for the corresponding building is established according to the building projection information. The three-dimensional projection model is processed through the feature extraction network to analyze the geographic element features of the corresponding building in the area to be mapped.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing mapping, and specifically to a geographic information mapping method based on remote sensing technology. Background Technique

[0002] Remote sensing technology refers to the technology of detecting the electromagnetic wave information of target objects from a long distance (such as platforms like satellites or drones) through non-contact sensors, and identifying, analyzing, and monitoring the features of ground objects, environmental states, or dynamic changes based on this electromagnetic wave information. Its core lies in "long-distance sensing", and data can be obtained without directly contacting the target.

[0003] In the existing geographic information mapping technology, the remote sensing images obtained through remote sensing technology often only have feature analysis in one dimension. This leads to insufficient extraction of remote sensing features, and the acquisition of the corresponding ground object features around buildings is also carried out only through data on a single level, resulting in inaccurate acquisition of information related to ground object features and affecting the mapping accuracy of buildings. How to integrate data in each dimension to effectively improve the mapping accuracy of buildings and obtain the corresponding geographic information of buildings has become an urgent problem to be solved at present. Summary of the Invention

[0004] In order to solve the above problems, the purpose of the present invention is to provide a geographic information mapping method based on remote sensing technology.

[0005] The purpose of the present invention can be achieved through the following technical solutions: A geographic information mapping method based on remote sensing technology, including the following steps:

[0006] Step S1: Collect remote sensing images of the area to be mapped through remote sensing equipment, perform feature processing on the remote sensing images to obtain the information features corresponding to the remote sensing images in different temporal sequence dimensions, and use a spatio-temporal registration algorithm to unify the information features in different temporal sequence dimensions into the same geographic coordinate system;

[0007] Step S2: Identify several buildings in the geographic coordinate system, extract the ground object features of each building to obtain the corresponding point cloud data and image data related to each building, and perform data fusion on the corresponding data, point cloud data, and image data extracted from the ground object features to construct a three-dimensional projection model of each building;

[0008] Step S3: Obtain building projection information based on the three-dimensional projection model of each building, establish a feature extraction network for the corresponding building according to the building projection information, process the three-dimensional projection model through the feature extraction network, and analyze the geographic element features of the corresponding building in the area to be mapped.

[0009] Further, the process of collecting remote sensing images of the area to be surveyed and mapped by a remote sensing device and performing feature processing on the remote sensing images to obtain information features corresponding to the remote sensing images in different temporal dimensions includes:

[0010] Select a remote sensing device and perform preparatory work on the remote sensing device. The remote sensing device includes a remote sensing satellite and a remote sensing drone. After the preparatory work of the remote sensing device is successfully completed, the remote sensing satellite collects the overall remote sensing image of the current area to be surveyed and mapped. Set several hovering points for the remote sensing drone to hover. At each hovering point, the remote sensing drone divides the area to be surveyed and mapped into sub-survey areas, and the remote sensing drone collects the branch detail remote sensing images of each sub-survey area and maps all the branch detail remote sensing images onto the overall remote sensing image;

[0011] Set the processing frequency corresponding to the feature processing, perform radiometric correction, geometric correction, and noise suppression on the remote sensing images at each processing frequency, complete the image preprocessing of the remote sensing images in each temporal dimension, analyze the contrast information, color transformation information, spectral feature information, texture feature information, and shape feature information of the remote sensing images in each temporal dimension, and integrate them as the information features of the remote sensing images in each temporal dimension.

[0012] Further, the process of unifying the information features of different temporal dimensions to the same geographic coordinate system by using a spatio-temporal registration algorithm includes:

[0013] Use the spatio-temporal registration algorithm to process the information features of different temporal dimensions, including time synchronization, spatial coarse registration, and dynamic error compensation, and set the processing duration of time synchronization, spatial coarse registration, and dynamic error compensation respectively;

[0014] Set a coordinate origin in the area to be surveyed and mapped, construct an X-axis, a Y-axis, and a Z-axis that are perpendicular to each other pairwise according to the coordinate origin, construct the corresponding geographic coordinate system of the area to be surveyed and mapped, and perform time synchronization, spatial coarse registration, and dynamic error compensation of the information features corresponding to different temporal dimensions in the geographic coordinate system in the order of their respective processing durations, thereby completing the operation of unifying the information features of different temporal dimensions to the same geographic coordinate system.

[0015] Further, the process of identifying several buildings in the geographic coordinate system and extracting the ground object features of each building to obtain the point cloud data and image data related to each building includes:

[0016] Set several types of building instance objects, where each type of building instance object is used to represent a type of building. Set an instance traversal window on the geographic coordinate system. According to the window size of the instance traversal window, divide several traversal regions to be traversed on the geographic coordinate system. Move the instance traversal window through each traversal region to be traversed in turn. If there is a building whose similarity with the building instance object exceeds the preset similarity threshold in the traversal region to be traversed, then make a mark;

[0017] Obtain the building height of the building. According to the building height, set the spatial normal vector and the vector radius of the spatial normal vector of each building on the geographic coordinate system, construct the point cloud spherical space of each building in the spatial coordinate system, extract the ground object features of each building, obtain the terrain features, building shape features of each building, and the object features of the objects distributed around the building, and obtain several point clouds in the point cloud spherical space through the point cloud spherical space. Statistically analyze the point cloud density of each distribution region in the point cloud spherical space, and retain the point cloud data in the distribution region where the point cloud density is greater than the preset density. Invoke a remote sensing drone to take image shots of each building to obtain the image data of each building.

[0018] Furthermore, the process of fusing the corresponding data, point cloud data, and image data extracted from the ground object features to construct the three-dimensional projection model of each building includes:

[0019] Construct data layers of different dimensions based on convolutional neural network technology;

[0020] The data layers of different dimensions include a top layer for mapping the terrain features, building shape features, and object features of the objects distributed around the building corresponding to each building, an intermediate layer for mapping the point cloud data, and a bottom layer for mapping the image data;

[0021] Obtain the number of layer channels of each data layer of different dimensions;

[0022] Select the data layer in any dimension as the reference layer, and uniformly adjust the number of layer channels of the data layers in other dimensions to the number of layer channels of the data layer in the current dimension, and then construct the building projection model corresponding to the building in the corresponding dimension;

[0023] Among them, construct the building projection models of corresponding high dimensions, general dimensions, and low dimensions based on the top layer, intermediate layer, and bottom layer respectively, and perform model fusion on the building projection models of high dimensions, general dimensions, and low dimensions to construct the three-dimensional projection model corresponding to the building.

[0024] Furthermore, the process of obtaining the building projection information based on the three-dimensional projection model of each building includes:

[0025] The information of each building is parsed by the three-dimensional projection model corresponding to each building, and then the geometric feature information, topological feature information, physical feature information, and spatio-temporal feature information corresponding to each building are obtained, which are aggregated and integrated as the building projection information corresponding to the respective building.

[0026] Further, the process of establishing the feature extraction network for the corresponding building according to the building projection information includes:

[0027] Based on the recurrent neural network, the preliminary feature extraction network of the building is constructed, and the number of network recurrent layers corresponding to the preliminary feature extraction network, the number of nodes in each network recurrent layer, as well as the activation function and loss function corresponding to each network recurrent layer are established. The building projection information is input into the preliminary feature extraction network to obtain the information extraction accuracy rate corresponding to the preliminary feature extraction network. When the information extraction accuracy rate does not meet the preset threshold, the preliminary feature extraction network is optimized and trained through a preselected training data set obtained from historical data until the information extraction accuracy rate meets the preset threshold, and the final feature extraction network of the building is constructed. When the information extraction accuracy rate meets the preset threshold, no operation is performed.

[0028] Further, the process of analyzing the geographical element features of the corresponding building in the area to be surveyed and mapped by processing the three-dimensional projection model through the feature extraction network includes:

[0029] The three-dimensional projection models of each building are sequentially input into their respective feature extraction networks. The feature extraction network processes all the building projection information recorded in the three-dimensional projection model related to the building, and the feature extraction network changes the information format of the building projection information to the element information format required to be output as predefined.

[0030] After the feature extraction network processes the three-dimensional projection model, it analyzes the building represented by the three-dimensional projection model in the area to be surveyed and mapped. The marked geographical element features of the building include the geographical coordinates of the building, the building elevation, the shape features of the building, the structural features of the building, the positions and sizes of each door and window in the building, the orientation of the building, and the spatial subordination relationship between buildings.

[0031] Compared with the prior art, the beneficial effects of the present invention are:

[0032] 1. By using a remote sensing device to collect the remote sensing image of the area to be surveyed and mapped, performing feature processing on the remote sensing image to obtain the information features of the remote sensing image corresponding to different time series dimensions, and using a spatio-temporal registration algorithm to unify the information features of different time series dimensions to the same geographical coordinate system, the unified fusion of information features in different time series dimensions is realized, so that the obtained remote sensing features can better reflect the characteristics of the buildings in the current area to be surveyed and mapped, and provide more comprehensive dimensional information for subsequent analysis.

[0033] 2. By identifying a number of buildings in the geographic coordinate system and extracting the ground object features of each building, the point cloud data and image data corresponding to each building are obtained. The corresponding data, point cloud data, and image data obtained by the ground object feature extraction are fused to construct a three-dimensional projection model for each building. Based on the three-dimensional projection model of each building, building projection information is obtained. According to the building projection information, a feature extraction network for the corresponding building is established. The three-dimensional projection model is processed through the feature extraction network to analyze the geographic feature elements of the corresponding building in the area to be surveyed and mapped. By integrating multi-level data information, the surveying and mapping accuracy of the building is improved, and the geographic feature elements for each building to reflect itself and the surrounding objects are obtained, realizing efficient surveying and mapping. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a flowchart of the present invention.

[0035] Figure 2 It is a working schematic diagram of the remote sensing UAV for preparation work.

[0036] Figure 3 It is a schematic diagram of adjusting the layer channel number of the data layer. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0037] As Figure 1 shown, a geographic information surveying and mapping method based on remote sensing technology includes the following steps:

[0038] Step S1: Collect remote sensing images of the area to be surveyed and mapped through remote sensing equipment, perform feature processing on the remote sensing images to obtain the information features of the remote sensing images corresponding to different time series dimensions, and use a spatio-temporal registration algorithm to unify the information features of different time series dimensions to the same geographic coordinate system;

[0039] Step S2: Identify a number of buildings in the geographic coordinate system and extract the ground object features of each building. The point cloud data and image data corresponding to each building are obtained. The corresponding data, point cloud data, and image data obtained by the ground object feature extraction are fused to construct a three-dimensional projection model for each building;

[0040] Step S3: Based on the three-dimensional projection model of each building, obtain building projection information. According to the building projection information, establish a feature extraction network for the corresponding building, and process the three-dimensional projection model through the feature extraction network to analyze the geographic feature elements of the corresponding building in the area to be surveyed and mapped.

[0041] It should be further noted that in the specific implementation process, the process of collecting remote sensing images of the area to be surveyed and mapped by remote sensing equipment and performing feature processing on the remote sensing images to obtain information features corresponding to the remote sensing images in different temporal dimensions includes:

[0042] Select remote sensing equipment and perform preparatory work on the remote sensing equipment. The remote sensing equipment includes remote sensing satellites and remote sensing drones. The preparatory work corresponding to the remote sensing satellite is: setting the imaging field of view range, imaging resolution, and imaging frequency corresponding to the remote sensing satellite, performing a test imaging by the remote sensing satellite, and determining whether the obtained remote sensing test image meets the pre-specified imaging clarity. If so, the remote sensing satellite is debugged; otherwise, change the imaging field of view range, imaging resolution, and imaging frequency of the remote sensing satellite, and synchronously re-perform the test imaging until the remote sensing test image meets the imaging clarity, and complete the preparatory work of the remote sensing satellite;

[0043] As Figure 2 shown, the preparatory work of the remote sensing drone is: planning the take-off path of the remote sensing drone, setting several hovering points of the remote sensing drone in the three-dimensional space corresponding to the area to be surveyed and mapped, setting the hovering time and hovering angle of the remote sensing drone at each hovering point, and planning the return path of the remote sensing drone to the ground at each hovering point. At each hovering point, the remote sensing drone performs a test shot to obtain a remote sensing image of the part of the area to be surveyed and mapped covered at the current hovering angle. The imaging clarity of the obtained remote sensing image needs to meet the preset clarity conditions. When it is met, the preparatory work of the remote sensing drone is completed;

[0044] Collect the overall remote sensing image of the current area to be surveyed and mapped by the remote sensing satellite;

[0045] The remote sensing drone at each hovering point refines the area to be surveyed and mapped into corresponding sub-survey areas, collects branch detail remote sensing images of each sub-survey area by the remote sensing drone, and maps all the branch detail remote sensing images onto the overall remote sensing image;

[0046] Set the processing frequency corresponding to the feature processing, perform radiometric correction, geometric correction, and noise suppression on the remote sensing image at each processing frequency, complete the image preprocessing of the remote sensing image in each temporal dimension, analyze the contrast information, color transformation information, spectral feature information, texture feature information, and shape feature information of the remote sensing image in each temporal dimension, and integrate all the above information in the same temporal dimension as the information feature of the remote sensing image in each temporal dimension.

[0047] It should be further noted that in the specific implementation process, the process of unifying the information features of different temporal dimensions to the same geographic coordinate system by using a spatio-temporal registration algorithm includes:

[0048] The spatio-temporal registration algorithm is used to process the information features in different temporal dimensions, specifically including time synchronization, spatial rough registration, and dynamic error compensation, and the processing durations of time synchronization, spatial rough registration, and dynamic error compensation are set respectively;

[0049] A coordinate origin is set in the area to be surveyed and mapped. According to the coordinate origin, an X-axis, a Y-axis, and a Z-axis that are perpendicular to each other are constructed, and then the geographical coordinate system corresponding to the area to be surveyed and mapped is constructed. According to the order of their respective processing durations, time synchronization, spatial rough registration, and dynamic error compensation of the information features corresponding to different temporal dimensions are performed in the geographical coordinate system, and then the information features of different temporal dimensions are unified to the same geographical coordinate system.

[0050] The content of the time synchronization is: comparing the time source clock corresponding to the information feature of each temporal dimension with the atomic clock, and then calculating the clock difference between each time source clock and the atomic clock. The time source clock corresponding to the temporal dimension with the clock difference within the preset difference range is not adjusted, and the time source clock corresponding to the temporal dimension with the clock difference exceeding the difference range is adjusted;

[0051] The content of the spatial rough registration is: extracting the SIFT feature points corresponding to the information feature of each temporal dimension, mapping the SIFT feature points into the geographical coordinate system, and screening out the position coordinates of the SIFT feature points in the geographical coordinate system;

[0052] The screening of the position coordinates is performed by calculating the affine transformation matrix corresponding to the SIFT feature points;

[0053] The content of the dynamic error compensation is: setting the standard spatial region range of each SIFT feature point in the current geographical coordinate system. If the position coordinate of the SIFT feature point is within the standard spatial region range, no operation is performed. Otherwise, the fitting vertical axis and the fitting horizontal axis of the position coordinate corresponding to the SIFT feature point are set, and the position of the SIFT feature point is changed in the Y-axis direction on the fitting vertical axis and the position of the SIFT feature point is changed in the X-axis direction on the fitting horizontal axis, so as to adjust the position coordinate of the SIFT feature point to be within the standard spatial region range.

[0054] It should be further noted that in the specific implementation process, the process of identifying several buildings in the geographical coordinate system and extracting the ground object features of each building to obtain the point cloud data and image data related to each building includes:

[0055] Set several types of building instance objects, where each type of building instance object is used to represent a building type. Set a corresponding instance traversal window on the geographic coordinate system, and package several types of building instance objects as the window comparison elements of the instance traversal window. The window comparison elements are used to compare with buildings to determine the similarity between the window comparison elements and the buildings. If the similarity exceeds a preset similarity threshold, then use the building type of the building instance object corresponding to the window comparison element used in the current instance traversal window as the building type of the traversed building;

[0056] According to the window size of the instance traversal window, divide several traversal regions to be traversed on the geographic coordinate system corresponding to the area to be surveyed and mapped. Move the instance traversal window through each traversal region to be traversed in turn. If there is a building with a similarity exceeding the similarity threshold with the building instance object in the traversal region to be traversed, then make a mark until several buildings under the geographic coordinate system are marked;

[0057] Obtain the building height of the building. According to the building height, set the spatial normal vector of each building on the geographic coordinate system and the vector radius corresponding to the spatial normal vector, and then construct the point cloud spherical space of each building in the spatial coordinate system. Extract the ground object features of each building through the point cloud spherical space to obtain the terrain features, building shape features, and object features of the objects distributed around the building corresponding to each building, and obtain several point clouds in the point cloud spherical space through the point cloud spherical space. Statistically calculate the point cloud density of each distribution area in the point cloud spherical space, and retain the point cloud data in the distribution area where the point cloud density is greater than 200 points / m 3 and retrieve the point cloud data in the distribution area where the point cloud density is greater than 200 points / m. Deploy a remote sensing drone to take images of each building to obtain the image data corresponding to each building.

[0058] It should be further noted that in the specific implementation process, the process of fusing the corresponding data, point cloud data, and image data extracted from the ground object features to construct the three-dimensional projection model of each building includes:

[0059] Construct data layers of different dimensions based on convolutional neural network technology;

[0060] The data layers of different dimensions include a top layer for mapping the terrain features, building shape features, and object features of the objects distributed around the building corresponding to each building extracted through the ground object features. The top layer represents the data features in high dimensions, including an intermediate layer for mapping the point cloud data. The intermediate layer represents the data features in general dimensions, and also includes a bottom layer for mapping the image data. The bottom layer represents the data features in low dimensions;

[0061] Obtain the layer channel numbers corresponding to the data layers of different dimensions;

[0062] Among them, the number of layer channels of the top layer, the middle layer, and the bottom layer are respectively denoted as:

[0063] The number of layer channels of the top layer is denoted as Rh1;

[0064] The number of layer channels of the middle layer is denoted as Rh2;

[0065] The number of layer channels of the bottom layer is denoted as Rh3;

[0066] Initially, numerically, Rh1 > Rh2 > Rh3;

[0067] As Figure 3 shown, select the data layer in any dimension as the reference layer, and uniformly adjust the number of layer channels of the data layers in other dimensions to the number of layer channels of the data layer in the current dimension, and then construct the building projection model corresponding to the building in the corresponding dimension;

[0068] Among them, based on the top layer, the middle layer, and the bottom layer, construct the building projection models corresponding to the high dimension, the general dimension, and the low dimension respectively, and perform model fusion on the building projection models of the high dimension, the general dimension, and the low dimension to construct the three-dimensional projection model corresponding to the building.

[0069] It should be further noted that in the specific implementation process, the process of obtaining the building projection information based on the three-dimensional projection model of each building and establishing the feature extraction network of the corresponding building includes:

[0070] Parse the information of each building by the three-dimensional projection model corresponding to each building, and then obtain the geometric feature information, topological feature information, physical feature information, and spatio-temporal feature information corresponding to each building respectively, and summarize and integrate them as the building projection information of the corresponding building;

[0071] Among them, the geometric feature information, topological feature information, physical feature information, and spatio-temporal feature information of the building are specifically described as follows:

[0072] Geometric feature information: Specifically includes the volume, surface area, building shape, length, width, and height of the building, including the surface curvature distribution histogram formed by each building node on the building, and also includes the building contour, enclosed space volume, and building area point cloud information corresponding to any building area on the building;

[0073] Topological feature information: Includes the number of adjacent buildings adjacent to the current building, the number of roads around the current building and the road connection relationship, the relative spatial position and spatial connection relationship of each sub-building space inside the current building;

[0074] Physical feature information: including the material thermal properties of the materials used for the building facade, the fundamental frequency of structural vibration, and the optical reflection characteristics;

[0075] Spatio-temporal feature information: the annual settlement rate of the building, the periodic surface temperature change information of the building facade, and the internal structure aging rate and internal temperature change information of each sub-building space on the building.

[0076] Based on a recurrent neural network, a preliminary feature extraction network for the building is constructed, and the number of network recurrent layers corresponding to the preliminary feature extraction network, the number of nodes in each network recurrent layer, as well as the activation function and loss function corresponding to each network recurrent layer are established. The building projection information is input into the preliminary feature extraction network to obtain the information extraction accuracy rate corresponding to the preliminary feature extraction network. When the information extraction accuracy rate does not meet the preset threshold, the preliminary feature extraction network is optimized and trained through a preselected training data set obtained from historical data until the information extraction accuracy rate meets the preset threshold, and the final feature extraction network for the building is constructed. When the information extraction accuracy rate meets the preset threshold, no operation is performed;

[0077] Specifically:

[0078] Denote the information extraction accuracy rate as Sc and the preset threshold as τ;

[0079] The information extraction accuracy rate Sc = the number of successful and accurate information extraction times n / the total number of information extraction times N, where both n and N are integers greater than 0, and numerically n ≤ N;

[0080] When Sc ≥ τ, perform network optimization training on the preliminary feature extraction network;

[0081] When Sc < τ, do not perform any operation.

[0082] It should be further noted that in the specific implementation process, the process of processing the three-dimensional projection model through the feature extraction network and analyzing the geographical element characteristics of the corresponding building in the area to be surveyed includes:

[0083] Sequentially input the three-dimensional projection model corresponding to each building into its respective corresponding feature extraction network. The feature extraction network processes all the building projection information recorded in the three-dimensional projection model related to the building, and the feature extraction network changes the information format of the building projection information to the element information format required to be output as predefined;

[0084] After the feature extraction network processes the three-dimensional projection model, analyze the buildings represented by the three-dimensional projection model in the area to be surveyed and mapped, mark the geographical coordinates of the buildings, the relative height of the buildings with respect to the preset reference elevation plane, that is, the building elevation, the shape features of the buildings, specifically including the planar shape and the three-dimensional shape, the dimensional features corresponding to the length, width, height, area, and volume of the buildings, the structural features of the buildings, specifically including the layout and dimensions of structural components such as walls, columns, beams, and floor slabs in the buildings, the positions and dimensions of each door and window in the buildings, including the orientation of the buildings, and also including the spatial subordination relationship between buildings, and the spatial subordination relationship includes adjacent, inclusion, and intersection.

[0085] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A geographic information mapping method based on remote sensing technology, characterized in that, Including the following steps: Step S1: Collect remote sensing images of the area to be surveyed and mapped through remote sensing equipment, perform feature processing on the remote sensing images to obtain information features corresponding to the remote sensing images in different temporal dimensions, and use a spatio-temporal registration algorithm to unify the information features in different temporal dimensions into the same geographic coordinate system; Step S2: Identify several buildings in the geographic coordinate system, extract ground object features for each building, obtain the corresponding point cloud data and image data related to each building, fuse the corresponding data, point cloud data, and image data extracted from the ground object features, and construct a three-dimensional projection model for each building; Step S3: Obtain building projection information based on the three-dimensional projection model of each building, establish a feature extraction network for the corresponding building according to the building projection information, process the three-dimensional projection model through the feature extraction network, and analyze the geographic element features of the corresponding building in the area to be surveyed and mapped; The process of identifying several buildings in the geographic coordinate system and extracting ground object features for each building includes: Set several types of building instance objects, each type of building instance object is used to represent a building type, set an instance traversal window on the geographic coordinate system, divide several areas to be traversed on the geographic coordinate system according to the window size of the instance traversal window, move the instance traversal window through each area to be traversed in turn, and if there is a building in the area to be traversed whose similarity to the building instance object exceeds a preset similarity threshold, then mark it; Obtain the building height of the building, set the spatial normal vector and the vector radius of the spatial normal vector of each building on the geographic coordinate system according to the building height, construct a point cloud spherical space of each building in the spatial coordinate system, extract ground object features for each building to obtain the terrain features, building shape features, and object features of the objects distributed around the building of each building, and obtain several point clouds in the point cloud spherical space through the point cloud spherical space, count the point cloud density of each distribution area in the point cloud spherical space, and retain the point cloud data in the distribution area where the point cloud density is greater than the preset density, and deploy a remote sensing UAV to take images of each building to obtain the image data of each building.

2. The method for mapping geographic information based on remote sensing technology according to claim 1, wherein, The process of collecting remote sensing images of the area to be surveyed and mapped through remote sensing equipment and performing feature processing on the remote sensing images to obtain information features corresponding to the remote sensing images in different temporal dimensions includes: Select remote sensing equipment and perform preparatory work on the remote sensing equipment. The remote sensing equipment includes remote sensing satellites and remote sensing UAVs. After the preparatory work of the remote sensing equipment is successfully completed, the remote sensing satellite collects the overall remote sensing image of the current area to be surveyed and mapped, set several hovering points for the remote sensing UAV to hover, the remote sensing UAV at each hovering point divides the area to be surveyed and mapped into sub-survey areas, the remote sensing UAV collects the branch detail remote sensing images of each sub-survey area, and maps all the branch detail remote sensing images onto the overall remote sensing image; Set the processing frequency corresponding to the feature processing, perform radiometric correction, geometric correction, and noise suppression on the remote sensing image at each processing frequency, complete the image preprocessing of the remote sensing image in each temporal dimension, analyze the contrast information, color transformation information, spectral feature information, texture feature information, and shape feature information of the remote sensing image in each temporal dimension, and integrate them as the information features of the remote sensing image in each temporal dimension.

3. A geographic information mapping method based on remote sensing technology according to claim 2, characterized in that The process of unifying the information features of different temporal dimensions to the same geographic coordinate system using the spatio-temporal registration algorithm includes: Use the spatio-temporal registration algorithm to process the information features of different temporal dimensions, including time synchronization, spatial rough registration, and dynamic error compensation, and set the processing duration of time synchronization, spatial rough registration, and dynamic error compensation respectively; Set a coordinate origin in the area to be surveyed, construct the X-axis, Y-axis, and Z-axis that are perpendicular to each other pairwise according to the coordinate origin, construct the geographic coordinate system corresponding to the area to be surveyed, and execute the time synchronization, spatial rough registration, and dynamic error compensation of the information features corresponding to different temporal dimensions in the geographic coordinate system in the order of their respective processing durations, thereby completing the operation of unifying the information features of different temporal dimensions to the same geographic coordinate system.

4. A geographic information mapping method based on remote sensing technology according to claim 3, characterized in that, The process of fusing the corresponding data, point cloud data, and image data extracted from the ground object features to construct the three-dimensional projection model of each building includes: Construct data layers of different dimensions based on convolutional neural network technology; The data layers of different dimensions include a top layer for mapping the terrain features, building shape features, and object features of the objects distributed around each building corresponding to each building, an intermediate layer for mapping the point cloud data, and a bottom layer for mapping the image data; Obtain the number of layer channels of each data layer of different dimensions; Select the data layer in any dimension as the reference layer, and uniformly adjust the number of layer channels of the data layers in other dimensions to the number of layer channels of the data layer in the current dimension, thereby constructing the building projection model corresponding to the building in the corresponding dimension; Among them, construct the building projection models of corresponding high dimensions, general dimensions, and low dimensions based on the top layer, intermediate layer, and bottom layer respectively, and perform model fusion on the building projection models of high dimensions, general dimensions, and low dimensions to construct the three-dimensional projection model corresponding to the building.

5. A geographic information mapping method based on remote sensing technology according to claim 4, characterized in that, The process of obtaining the building projection information based on the three-dimensional projection model of each building includes: Parse the information of the corresponding building by the three-dimensional projection model corresponding to each building, and then obtain the geometric feature information, topological feature information, physical feature information, and spatio-temporal feature information corresponding to each building respectively, and summarize and integrate them as the building projection information of the corresponding building.

6. A geographic information mapping method based on remote sensing technology according to claim 5, characterized in that The process of establishing the feature extraction network of the corresponding building according to the building projection information includes: Based on the construction of a recurrent neural network, a preliminary feature extraction network for buildings is completed, and the number of network recurrent layers corresponding to the preliminary feature extraction network, the number of nodes in each network recurrent layer, as well as the activation function and loss function corresponding to each network recurrent layer are established. The building projection information is input into the preliminary feature extraction network to obtain the information extraction accuracy rate corresponding to the preliminary feature extraction network. When the information extraction accuracy rate does not meet the preset threshold, the preliminary feature extraction network is optimized and trained through a preselected training data set obtained from historical data until the information extraction accuracy rate meets the preset threshold, and the final feature extraction network for buildings is completed. When the information extraction accuracy rate meets the preset threshold, no operation is performed.

7. A geographic information mapping method based on remote sensing technology according to claim 6, characterized in that The process of analyzing the geographical element features of the corresponding building in the area to be surveyed by processing the three-dimensional projection model through the feature extraction network includes: Sequentially input the three-dimensional projection model of each building into its respective feature extraction network. The feature extraction network processes all the building projection information related to the building recorded in the three-dimensional projection model, and the feature extraction network changes the information format of the building projection information to the element information format required by the predefined output. After the feature extraction network processes the three-dimensional projection model, it analyzes the building represented by the three-dimensional projection model in the area to be surveyed. The marked geographical element features of the building include the building's geographical coordinates, building elevation, shape features of the building, structural features of the building, positions and dimensions of each door and window in the building, orientation of the building, and spatial subordination relationship between buildings.

Images