Building remote sensing information processing system and method based on feature fusion

Abstract

The application discloses a kind of building remote sensing information processing system and method based on feature fusion, it is related to information processing technical field, the processing method includes the following steps: based on satellite equipment, the image information of each building is collected, the point of interest data is identified by third party data, and a plurality of images and effective point of interest data are generated by preprocessing;The height of ground surface ground object is inverted to obtain digital surface model data, data processing and operation operation are carried out to obtain normalized digital surface model data, and are combined with a plurality of images to carry out superposition processing, and multidimensional feature image is extracted;Building contour data is obtained by constructing instance segmentation model;Based on the preset roof style, the roof style features of each building are obtained;A plurality of class feature data are preset, and each type of feature data of each building is analyzed and calculated;The adjacent relationship between each building is obtained, and a distribution map containing each building is constructed, to obtain building function classification result.

Description

Technical Field

[0001] This invention relates to the field of information processing technology, specifically a building remote sensing information processing system and method based on feature fusion. Background Technology

[0002] The traditional way to obtain urban building data is through manual surveys or planning approval data. However, due to the lack of historical data and the time-consuming, labor-intensive, and untimely updates of manual surveys, it is difficult to meet the needs of urban management departments for dynamic acquisition and updating of urban building data during the rapid urban renewal process.

[0003] With the continuous development of remote sensing technology, massive amounts of multi-platform, multi-modal, and dynamically updated multi-source remote sensing data have been generated, providing data support for building extraction based on remote sensing data. At the same time, artificial intelligence technology, represented by deep learning, has developed rapidly and has been widely used in the field of intelligent processing of remote sensing images.

[0004] However, existing image segmentation models only utilize the visual features of high-resolution remote sensing images, and other features of multi-source remote sensing data are not effectively utilized. Existing building extraction and building function classification methods do not fully utilize the advantages of multi-source heterogeneity of remote sensing data. How to fully explore the information of multi-source and multi-modal remote sensing data and improve the accuracy of building extraction and building function classification remains an urgent problem to be solved in current research. Summary of the Invention

[0005] The purpose of this invention is to provide a building remote sensing information processing system and method based on feature fusion to solve the problems raised in the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for processing building remote sensing information based on feature fusion, the method comprising the following steps:

[0007] Step S100: Collect image information of each building based on several satellite devices, and identify the point of interest data of each building through third-party data; preprocess the collected image information and point of interest data to generate several images and effective point of interest data respectively;

[0008] Step S200: Based on the images, the height of surface features is inverted to obtain digital surface model data; the digital surface model data is processed and operated to obtain normalized digital surface model data; the normalized digital surface model data and the images are overlaid to extract several multi-dimensional feature images.

[0009] Step S300: Construct an instance segmentation model to extract building information from arbitrary multi-dimensional feature images, process the extracted building information to obtain the building outline data of each building; and obtain the roof style features of each building based on the preset roof style.

[0010] Step S400: Preset several types of feature data, spatially correlate and statistically analyze the building outline data of each building with the various types of data, and analyze and calculate any type of feature data of each building;

[0011] Step S500: Obtain the adjacency relationship between buildings, obtain the comprehensive features of each building based on various feature data, and construct a distribution map containing each building; construct a graph neural network to classify each building according to its function, and obtain the building function classification result.

[0012] Furthermore, step S100 includes the following steps:

[0013] Step S101: Acquire high-resolution stereo images from high-resolution stereo satellites, and perform preprocessing to correct and crop the obtained stereo images to generate high-resolution stereo image pairs; the preprocessing of stereo image pairs includes, but is not limited to, radiometric correction, atmospheric correction, geometric correction, cropping, etc.

[0014] Step S102: Acquire high-resolution nighttime light satellite images from high-resolution nighttime light satellites, and perform preprocessing to correct and crop the nighttime light satellite images to generate nighttime light images; the preprocessing of nighttime light images includes, but is not limited to, radiometric correction, atmospheric correction, geometric correction, cropping, etc.

[0015] Step S103: Collect geographic information of various buildings in the city monitored by satellite equipment through a map application, set the obtained geographic information data as points of interest (POIs), and generate several valid POIs after data correction. POIs are geographic information data that abstracts urban functions into spatial point-like forms. In addition to containing spatial information such as latitude and longitude coordinates and addresses of geographic entities, they also connect to various attribute information, such as POI name, category, and administrative division. POIs can empower research in spatiotemporal behavior, urban planning, and geographic information.

[0016] Because the original point-of-interest data collected has various problems such as classification errors and missing coordinate information, preprocessing such as data correction is required. Specifically, data correction refers to correcting the classification of incorrectly classified points, deleting points with missing information, and deleting points of interest with unclear functional characteristics or that cannot clearly represent the functional characteristics of buildings.

[0017] Furthermore, step S200 includes the following steps:

[0018] Step S201: Select a pair of stereo images that constitute the same region from all high-resolution stereo image pairs. The stereo image pair contains two images. Extract digital surface model data from the stereo image pair using a binocular stereo matching algorithm.

[0019] Step S202: Filter the extracted digital surface model data to remove surface features and retain ground height, and obtain digital elevation model data; subtract the digital surface model data from the digital elevation model data to obtain normalized digital surface model data.

[0020] Step S203: Acquire each orthorectified multispectral image, and correct the normalized digital surface model data by resampling with reference to the orthorectified multispectral image; overlay the corrected normalized digital surface model data with the orthorectified multispectral image to obtain a multidimensional feature image containing multispectral channels and height channels.

[0021] Furthermore, step S300 includes the following steps:

[0022] Step S301: Preset a fixed image cutting size, perform image segmentation on any multi-dimensional feature image to obtain an image slice data set of the multi-dimensional feature image; input the image slice data set into the instance segmentation model to extract the buildings contained in each image slice data to obtain a set of building extraction results corresponding to the image slice data set.

[0023] In terms of instance segmentation models, by constructing instance segmentation models, the attention modules of convolutional neural networks and Transformer models are combined in terms of feature extraction. The local feature extraction capabilities of convolutional neural networks and the global spatial context modeling capabilities of Transformer models are comprehensively utilized to improve the image feature extraction capabilities and effectively improve the accuracy of building outline extraction.

[0024] Step S302: Merge the set of building extraction results to obtain the building extraction results of the area presented in the multi-dimensional feature image; vectorize the building extraction results to obtain building outline data; regularize the building graphics contained in the building outline data to obtain regularized building outline data.

[0025] Step S303: Extract various roof styles from the building outline data, according to the formula:

[0026] ;

[0027] Among them, Confidence p→i C represents the confidence level that the p-th building in the region belongs to the i-th roof style.i () is a permutation and summarization function that summarizes the confidence scores of the p-th building belonging to each roof style category; thus, the roof style feature F of the p-th building in the region is obtained. p .

[0028] Furthermore, step S400 includes the following steps:

[0029] Step S401: Extract the building outline image of any building from the building outline data, obtain the building outline image of the p-th building, and set the pixel value of the j-th pixel in the building outline image of the p-th building to DN. j According to the formula:

[0030] ;

[0031] In this function, DESC() is a descending sorting function that sorts each pixel in the building outline image of the p-th building in descending order of pixel value. n is a positive number and n∈(0,100), representing the percentage of pixels selected after sorting. Mean() is an averaging function that calculates the average of the first n% of the selected pixel values. The building height feature H of the p-th building is then calculated. p ;

[0032] Step S402: Pre-set several types of valid point of interest (POI) data and several types of building function categories, and construct a mapping relationship between valid POIs and building functions, according to the formula:

[0033] ;

[0034] Where f is the mapping function, and type poi For the category of valid point of interest data, type building The system categorizes building functions; for any type of valid point of interest (POI) data, it matches the corresponding building function and assigns appropriate weights to each type of valid POI data; it reclassifies valid POI categories into categories that can be used to label building functions; for any valid POI data category, there exists a unique building function category that corresponds to it.

[0035] Step S403: Obtain the regularized building outline data of the p-th building; extract the building outline of the p-th building from the regularized building outline data; buffer the building outline of the p-th building by a preset buffer distance to obtain buffered building outline data; spatially connect the buffered building outline data of the p-th building with various types of valid interest point data to obtain various types of valid interest point data associated with the building outline of the p-th building; set the number of types of valid interest point data associated with the building outline of the p-th building to m. p The number of valid interest point data of the kth class is (uk ) p According to the formula:

[0036] ;

[0037] Among them, (w k ) p C represents the weight value of the k-th class of valid interest point data. k () is a permutation and aggregation function that summarizes the weight proportions of various valid points of interest data for the p-th building; the social function characteristic U of the p-th building is calculated. p ;

[0038] Step S404: Obtain the nighttime light image of the p-th building, perform spatial correlation analysis and surface statistical analysis with the regularized building outline data to obtain the building outline image contained within the building outline of the p-th building, obtain the number of pixels in the building outline image as r, and set the pixel value of the j1-th pixel as DN. j1 According to the formula:

[0039] ;

[0040] Among them, S p Let L be the outline area of ​​the p-th building; calculate the nighttime light index L of the p-th building. p The nighttime light index is used to characterize the human activity characteristics of buildings.

[0041] Furthermore, step S500 includes the following steps:

[0042] Step S501: Randomly select the p-th building and extract the roof style feature F of the p-th building. p Building height characteristics H p Social functional characteristics U p and nighttime light index L p The extracted features are concatenated according to the formula:

[0043] ;

[0044] CONTACT() is a concatenation operation; it obtains the comprehensive feature Z of the p-th building. p Connecting buildings can help enhance the distinction between them, which is beneficial for subsequent building classification.

[0045] Step S502: Obtain the centroid position of any building, construct a constrained Deloni triangulation with the center position of each building as the node, and calculate the correlation between each building through the Deloni triangulation; if there are two buildings that touch the same triangle edge, then establish an adjacency relationship between the nodes corresponding to the two buildings and generate a graph edge E between the two nodes.

[0046] Step S503: Take the nodes corresponding to each building as graph nodes V, and the graph edges E between two nodes to construct an undirected graph G(V,E), where the feature of any graph node V is set as the comprehensive feature Z of the corresponding building.

[0047] Step S504: Construct a graph attention network model to classify the functions of each building in the undirected graph and obtain the building function classification results; use the attention mechanism to automatically learn and optimize the connection relationship between nodes, and improve the computational efficiency by calculating the attention value between node pairs in parallel; the graph attention mechanism assigns different weights to each neighbor node, so that the model can better handle nodes with different weights.

[0048] To better implement the above methods, a building remote sensing information processing system is also proposed. The processing system includes a data acquisition and processing module, a data image analysis module, a building information extraction module, a feature data processing module, and a building function classification module.

[0049] The data acquisition and processing module is used to acquire image information and building data of various buildings through high-resolution stereo satellites, preprocess the acquired image information and building data, and generate several images and effective data.

[0050] The data image analysis module is used to invert the height of surface features based on the images to obtain digital surface model data, perform data processing and calculation operations on the digital surface model data to obtain normalized digital surface model data, and overlay the normalized digital surface model data and the images to extract several multi-dimensional feature images.

[0051] The building information extraction module is used to construct an instance segmentation model to extract building information from arbitrary multi-dimensional feature images, process the extracted building information to obtain the building outline data of each building, and obtain the roof style features of each building based on the preset roof style.

[0052] The feature data processing module is used to preset several types of feature data, spatially correlate and statistically analyze the building outline data of each building with various types of data, and analyze and calculate any type of feature data of each building.

[0053] The building function classification module is used to obtain the adjacency relationship between buildings, obtain the comprehensive characteristics of each building based on various feature data, and construct a distribution map containing each building; a graph neural network is constructed to classify each building according to its function, and the building function classification result is obtained.

[0054] Furthermore, the data image analysis module includes a model data generation unit and a feature image extraction unit;

[0055] The model data generation unit is used to invert the height of surface features based on the several images to obtain digital surface model data, and to perform data processing and calculation operations on the digital surface model data to obtain normalized digital surface model data; the feature image extraction unit is used to overlay the normalized digital surface model data and the several images to extract several multi-dimensional feature images.

[0056] Furthermore, the building information extraction module includes a contour data processing unit and a roof style analysis unit;

[0057] The contour data processing unit is used to construct an instance segmentation model to extract building information from arbitrary multi-dimensional feature images, and to process the extracted building information to obtain the building contour data of each building; the roof style analysis unit is used to obtain the roof style features of each building based on the preset roof style.

[0058] Furthermore, the building function classification module includes a building distribution analysis unit and a classification result generation unit;

[0059] The building distribution analysis unit is used to obtain the adjacency relationship between buildings, obtain the comprehensive characteristics of each building based on various feature data, and construct a distribution map containing each building; the classification result generation unit is used to construct a graph neural network to classify each building according to its function and obtain the building function classification result.

[0060] Compared with the prior art, the beneficial effects of the present invention are:

[0061] 1. This invention comprehensively utilizes multi-source and multi-modal remote sensing data, fully leverages the advantages of multi-source heterogeneity of remote sensing data, maximizes the mining of multi-source remote sensing information, and extracts multi-dimensional features of buildings such as visual features, morphological features, height features, socio-economic features, and human activity features, which can effectively improve the accuracy of building outline extraction and the accuracy of building function classification.

[0062] 2. This invention can efficiently and automatically extract urban building information, providing data support and technical guarantee for urban planning and management. It has the advantages of low production cost, high efficiency, and the ability to achieve periodic updates, and has broad application prospects. Attached Figure Description

[0063] Figure 1 This is a schematic diagram illustrating the steps of a feature fusion-based method for processing building remote sensing information.

[0064] Figure 2 This is a schematic diagram of a building remote sensing information processing system based on feature fusion.

[0065] Figure 3 A schematic diagram illustrating the process of retrieving the height of surface features;

[0066] Figure 4 This is a flowchart illustrating the process of extracting building information.

[0067] Figure 5 A schematic diagram of the graph attention network model used for building function classification;

[0068] Figure 6 This is a technical roadmap for a building remote sensing information processing system and method based on feature fusion. Detailed Implementation

[0069] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0070] Example: Figures 1 to 6 As shown, this invention provides a method for processing building remote sensing information based on feature fusion. The processing method includes the following steps:

[0071] Step S100: Collect image information of each building based on several satellite devices, and identify the point of interest data of each building through third-party data; preprocess the collected image information and point of interest data to generate several images and effective point of interest data respectively;

[0072] Step S100 includes the following steps:

[0073] Step S101: Acquire high-resolution stereo images from high-resolution stereo satellites, and perform preprocessing to correct and crop the obtained stereo images to generate high-resolution stereo image pairs;

[0074] Step S102: Acquire high-resolution nighttime light satellite images from high-resolution nighttime light satellites, and perform preprocessing to correct and crop the nighttime light satellite images to generate nighttime light images;

[0075] Step S103: Collect geographic information of various buildings in the city monitored by the satellite equipment through a map application, set the obtained geographic information data as point of interest data, and generate several valid point of interest data after data correction of the point of interest data.

[0076] Step S200: Based on the images, the height of surface features is inverted to obtain digital surface model data; the digital surface model data is processed and operated to obtain normalized digital surface model data; the normalized digital surface model data and the images are overlaid to extract several multi-dimensional feature images.

[0077] Step S200 includes the following steps:

[0078] Step S201: Select a pair of stereo images that constitute the same region from all high-resolution stereo image pairs. The stereo image pair contains two images. Extract digital surface model data from the stereo image pair using a binocular stereo matching algorithm.

[0079] Step S202: Filter the extracted digital surface model data to remove surface features and retain ground height, and obtain digital elevation model data; subtract the digital surface model data from the digital elevation model data to obtain normalized digital surface model data.

[0080] Step S203: Acquire each orthorectified multispectral image, and correct the normalized digital surface model data by resampling with reference to the orthorectified multispectral image; overlay the corrected normalized digital surface model data with the orthorectified multispectral image to obtain a multidimensional feature image containing multispectral channels and height channels.

[0081] Step S300: Construct an instance segmentation model to extract building information from arbitrary multi-dimensional feature images, process the extracted building information to obtain the building outline data of each building; and obtain the roof style features of each building based on the preset roof style.

[0082] Step S300 includes the following steps:

[0083] Step S301: Preset a fixed image cutting size, perform image segmentation on any multi-dimensional feature image to obtain an image slice data set of the multi-dimensional feature image; input the image slice data set into the instance segmentation model to extract the buildings contained in each image slice data to obtain a set of building extraction results corresponding to the image slice data set.

[0084] Step S302: Merge the set of building extraction results to obtain the building extraction results of the area presented in the multi-dimensional feature image; vectorize the building extraction results to obtain building outline data; regularize the building graphics contained in the building outline data to obtain regularized building outline data.

[0085] Step S303: Extract various roof styles from the building outline data, according to the formula:

[0086] ;

[0087] Among them, Confidence p→i C represents the confidence level that the p-th building in the region belongs to the i-th roof style. i () is a permutation and summarization function that summarizes the confidence scores of the p-th building belonging to each roof style category; thus, the roof style feature F of the p-th building in the region is obtained. p

[0088] Step S400: Preset several types of feature data, spatially correlate and statistically analyze the building outline data of each building with the various types of data, and analyze and calculate any type of feature data of each building;

[0089] Step S400 includes the following steps:

[0090] Step S401: Extract the building outline image of any building from the building outline data, obtain the building outline image of the p-th building, and set the pixel value of the j-th pixel in the building outline image of the p-th building to DN. j According to the formula:

[0091] ;

[0092] In this function, DESC() is a descending sorting function that sorts each pixel in the building outline image of the p-th building in descending order of pixel value. n is a positive number and n∈(0,100), representing the percentage of pixels selected after sorting. Mean() is an averaging function that calculates the average of the first n% of the selected pixel values. The building height feature H of the p-th building is then calculated. p ;

[0093] Step S402: Pre-set several types of valid point of interest (POI) data and several types of building function categories, and construct a mapping relationship between valid POIs and building functions, according to the formula:

[0094] ;

[0095] Where f is the mapping function, and type poiFor the category of valid point of interest data, type building The system categorizes building functions; for any type of valid point of interest data, it matches the corresponding building function and sets appropriate weights for each type of valid point of interest data.

[0096] Step S403: Obtain the regularized building outline data of the p-th building; extract the building outline of the p-th building from the regularized building outline data; buffer the building outline of the p-th building by a preset buffer distance to obtain buffered building outline data; spatially connect the buffered building outline data of the p-th building with various types of valid interest point data to obtain various types of valid interest point data associated with the building outline of the p-th building; set the number of types of valid interest point data associated with the building outline of the p-th building to m. p The number of valid interest point data of the kth class is (u k ) p According to the formula:

[0097] ;

[0098] Among them, (w k ) p C represents the weight value of the k-th class of valid interest point data. k () is a permutation and aggregation function that summarizes the weight proportions of various valid points of interest data for the p-th building; the social function characteristic U of the p-th building is calculated. p ;

[0099] Step S404: Obtain the nighttime light image of the p-th building, perform spatial correlation analysis and surface statistical analysis with the regularized building outline data to obtain the building outline image contained within the building outline of the p-th building, obtain the number of pixels in the building outline image as r, and set the pixel value of the j1-th pixel as DN. j1 According to the formula:

[0100] ;

[0101] Among them, S p Let L be the outline area of ​​the p-th building; calculate the nighttime light index L of the p-th building. p .

[0102] Step S500: Obtain the adjacency relationships between buildings, obtain the comprehensive features of each building based on various feature data, and construct a distribution map containing each building; construct a graph neural network to classify each building according to its function, and obtain the building function classification results;

[0103] Step S500 includes the following steps:

[0104] Step S501: Randomly select the p-th building and extract the roof style feature F of the p-th building. p Building height characteristics H p Social functional characteristics U p and nighttime light index L p The extracted features are concatenated according to the formula:

[0105] ;

[0106] CONTACT() is a concatenation operation; it obtains the comprehensive feature Z of the p-th building. p ;

[0107] Step S502: Obtain the centroid position of any building, construct a constrained Deloni triangulation with the center position of each building as the node, and calculate the correlation between each building through the Deloni triangulation; if there are two buildings that touch the same triangle edge, then establish an adjacency relationship between the nodes corresponding to the two buildings and generate a graph edge E between the two nodes.

[0108] Step S503: Take the nodes corresponding to each building as graph nodes V, and the graph edges E between two nodes to construct an undirected graph G(V,E), where the feature of any graph node V is set as the comprehensive feature Z of the corresponding building.

[0109] Step S504: Construct a graph attention network model to classify the functions of each building in the undirected graph and obtain the building function classification results;

[0110] Example: Figure 5 As shown, where, For the input feature vector; Output high-dimensional feature vectors; The importance of the feature relationships between nodes; is the weight matrix; contcat is the join operation; avg is the mean operation; softmax is the normalization exponential function; i∈1,2,3…N, N is the number of graph nodes, which in this invention refers to the number of building nodes; F is the number of features, which in this invention is 4.

[0111] A building remote sensing information processing system includes a data acquisition and processing module, a data image analysis module, a building information extraction module, a feature data processing module, and a building function classification module.

[0112] The data acquisition and processing module is used to acquire image information and building data of various buildings through high-resolution stereo satellites, preprocess the acquired image information and building data, and generate several images and effective data.

[0113] The data image analysis module is used to invert the height of surface features based on the images to obtain digital surface model data, perform data processing and calculation operations on the digital surface model data to obtain normalized digital surface model data, and overlay the normalized digital surface model data and the images to extract several multi-dimensional feature images.

[0114] The building information extraction module is used to construct an instance segmentation model to extract building information from arbitrary multi-dimensional feature images, process the extracted building information to obtain the building outline data of each building, and obtain the roof style features of each building based on the preset roof style.

[0115] The feature data processing module is used to preset several types of feature data, spatially correlate and statistically analyze the building outline data of each building with various types of data, and analyze and calculate any type of feature data of each building.

[0116] The building function classification module is used to obtain the adjacency relationship between buildings, obtain the comprehensive characteristics of each building based on various feature data, and construct a distribution map containing each building; a graph neural network is constructed to classify each building according to its function, and the building function classification result is obtained.

[0117] The data image analysis module includes a model data generation unit and a feature image extraction unit.

[0118] The model data generation unit is used to invert the height of surface features based on the several images to obtain digital surface model data, and to perform data processing and calculation operations on the digital surface model data to obtain normalized digital surface model data; the feature image extraction unit is used to overlay the normalized digital surface model data and the several images to extract several multi-dimensional feature images.

[0119] The building information extraction module includes a contour data processing unit and a roof style analysis unit.

[0120] The contour data processing unit is used to construct an instance segmentation model to extract building information from arbitrary multi-dimensional feature images, and to process the extracted building information to obtain the building contour data of each building; the roof style analysis unit is used to obtain the roof style features of each building based on the preset roof style.

[0121] The building function classification module includes a building distribution analysis unit and a classification result generation unit;

[0122] The building distribution analysis unit is used to obtain the adjacency relationship between buildings, obtain the comprehensive characteristics of each building based on various feature data, and construct a distribution map containing each building; the classification result generation unit is used to construct a graph neural network to classify each building according to its function and obtain the building function classification result.

[0123] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for processing building remote sensing information based on feature fusion, characterized in that: The processing method includes the following steps: Step S100: Collect image information of each building based on several satellite devices, and identify the point of interest data of each building through third-party data; preprocess the collected image information and point of interest data to generate several images and effective point of interest data respectively; Step S200: Based on the images, the height of surface features is inverted to obtain digital surface model data; the digital surface model data is processed and operated to obtain normalized digital surface model data; the normalized digital surface model data and the images are overlaid to extract several multi-dimensional feature images. Step S300: Construct an instance segmentation model to extract building information from arbitrary multi-dimensional feature images, process the extracted building information to obtain the building outline data of each building; and obtain the roof style features of each building based on the preset roof style. Step S400: Preset several types of feature data, spatially correlate and statistically analyze the building outline data of each building with the various types of data, and analyze and calculate any type of feature data of each building; Step S500: Obtain the adjacency relationships between buildings, obtain the comprehensive features of each building based on various feature data, and construct a distribution map containing each building; construct a graph neural network to classify each building according to its function, and obtain the building function classification results; Step S200 includes the following steps: The corrected normalized digital surface model data is overlaid with orthorectified multispectral imagery to obtain a multidimensional feature image containing multispectral and height channels. Step S400 includes the following steps: Step S403: Extract the building outline image of any building from the building outline data, and obtain the building outline image of the p-th building; Obtain the regularized building outline data of the p-th building; extract the building outline of the p-th building from the regularized building outline data; buffer the building outline of the p-th building by a preset buffer distance to obtain buffered building outline data; spatially connect the buffered building outline data of the p-th building with various types of valid interest point data to obtain various types of valid interest point data associated with the building outline of the p-th building; set the number of types of valid interest point data associated with the building outline of the p-th building to m. p The number of valid interest point data of the kth class is (u k ) p According to the formula: ； Among them, (w k ) p C represents the weight value of the k-th class of valid interest point data. k () is a permutation and aggregation function that summarizes the weight proportions of various valid points of interest data for the p-th building; the social function characteristic U of the p-th building is calculated. p ; Step S500 includes the following steps: Step S501: Randomly select the p-th building and extract the roof style feature F of the p-th building. p Building height characteristics H p Social functional characteristics U p and nighttime light index L p The extracted features are concatenated according to the formula: ； CONTACT() is a concatenation operation; it obtains the comprehensive feature Z of the p-th building. p .

2. The method for processing building remote sensing information based on feature fusion according to claim 1, characterized in that: Step S100 includes the following steps: Step S101: Acquire high-resolution stereo images from high-resolution stereo satellites, and perform preprocessing to correct and crop the obtained stereo images to generate high-resolution stereo image pairs; Step S102: Acquire high-resolution nighttime light satellite images from high-resolution nighttime light satellites, and perform preprocessing to correct and crop the nighttime light satellite images to generate nighttime light images; Step S103: Collect geographic information of various buildings in the city monitored by the satellite equipment through a map application, set the obtained geographic information data as point of interest data, and generate several valid point of interest data after data correction of the point of interest data.

3. The method for processing building remote sensing information based on feature fusion according to claim 2, characterized in that: Step S200 includes the following steps: Step S201: Select a pair of stereo images that constitute the same region from all high-resolution stereo image pairs. The stereo image pair contains two images. Extract digital surface model data from the stereo image pair using a binocular stereo matching algorithm. Step S202: Filter the extracted digital surface model data to remove surface features and retain ground height, and obtain digital elevation model data; subtract the digital surface model data from the digital elevation model data to obtain normalized digital surface model data. Step S203: Acquire each orthophoto multispectral image, and correct the normalized digital surface model data by resampling with reference to the orthophoto multispectral image.

4. The method for processing building remote sensing information based on feature fusion according to claim 3, characterized in that: Step S300 includes the following steps: Step S301: Preset a fixed image cutting size, perform image segmentation on any multi-dimensional feature image to obtain an image slice data set of the multi-dimensional feature image; input the image slice data set into the instance segmentation model to extract the buildings contained in each image slice data to obtain a set of building extraction results corresponding to the image slice data set. Step S302: Merge the set of building extraction results to obtain the building extraction results of the area presented in the multi-dimensional feature image; vectorize the building extraction results to obtain building outline data; regularize the building graphics contained in the building outline data to obtain regularized building outline data. Step S303: Extract various roof styles from the building outline data, according to the formula: ； Among them, Confidence p→i Let C be the confidence level that the p-th building in the region belongs to the i-th roof style. i () is a permutation and summarization function that summarizes the confidence scores of the p-th building belonging to each roof style category; thus, the roof style feature F of the p-th building in the region is obtained. p .

5. The method for processing building remote sensing information based on feature fusion according to claim 4, characterized in that: Step S400 further includes the following steps: Step S401: Extract the building outline image of any building from the building outline data, obtain the building outline image of the p-th building, and set the pixel value of the j-th pixel in the building outline image of the p-th building to DN. j DESC() is a descending sorting function that sorts each pixel in the outline image of the p-th building in descending order of pixel value, where n is a positive number and n∈(0,100), and represents the percentage of pixels selected after sorting. Mean() is an averaging function that calculates the average of the first n% of the selected pixel values. The building height feature H of the p-th building is then calculated. p ; Step S402: Pre-set several types of valid point of interest (POI) data and several types of building function categories, and construct a mapping relationship between valid POIs and building functions, according to the formula: ； Where f is the mapping function, and type poi For the category of valid point of interest data, type building The system categorizes building functions; for any type of valid point of interest data, it matches the corresponding building function and sets appropriate weights for each type of valid point of interest data. Step S404: Obtain the nighttime light image of the p-th building, perform spatial correlation analysis and surface statistical analysis with the regularized building outline data to obtain the building outline image contained within the building outline of the p-th building, obtain the number of pixels in the building outline image as r, and set the pixel value of the j1-th pixel as DN. j1 According to the formula: ； Among them, S p Let L be the outline area of ​​the p-th building; calculate the nighttime light index L of the p-th building. p .

6. The method for processing building remote sensing information based on feature fusion according to claim 5, characterized in that: The step S500 also Includes the following steps: Step S502: Obtain the centroid position of any building, construct a constrained Deloni triangulation with the center positions of each building as nodes, and calculate the correlation between each building through the Deloni triangulation. If two buildings touch the same triangle edge, then establish an adjacency relationship between the nodes corresponding to the two buildings and generate a graph edge E between the two nodes; Step S503: Take the nodes corresponding to each building as graph nodes V, and the graph edges E between two nodes to construct an undirected graph G(V,E), where the feature of any graph node V is set as the comprehensive feature Z of the corresponding building. Step S504: Construct a graph attention network model to classify the functions of each building in the undirected graph and obtain the building function classification results.

7. A building remote sensing information processing system, used to execute the building remote sensing information processing method based on feature fusion as described in any one of claims 1-6, characterized in that: The processing system includes a data acquisition and processing module, a data image analysis module, a building information extraction module, a feature data processing module, and a building function classification module; The data acquisition and processing module is used to acquire image information and building data of each building through high-resolution stereo satellite, preprocess the acquired image information and building data, and generate several images and effective data. The data image analysis module is used to invert the height of surface features based on the images to obtain digital surface model data, perform data processing and calculation operations on the digital surface model data to obtain normalized digital surface model data, and overlay the normalized digital surface model data and the images to extract several multi-dimensional feature images. The building information extraction module is used to construct an instance segmentation model to extract building information from arbitrary multi-dimensional feature images, and to process the extracted building information to obtain the building outline data of each building. Based on the preset roof style, the roof style characteristics of each building are obtained; The feature data processing module is used to preset several types of feature data, spatially correlate and statistically analyze the building outline data of each building with various types of data, and analyze and calculate any type of feature data of each building. The building function classification module is used to obtain the adjacency relationship between buildings, obtain the comprehensive features of each building based on various feature data, and construct a distribution map containing each building; a graph neural network is constructed to classify each building according to its function to obtain the building function classification result.

8. A building remote sensing information processing system according to claim 7, characterized in that: The data image analysis module includes a model data generation unit and a feature image extraction unit; The model data generation unit is used to invert the height of surface features based on the images to obtain digital surface model data, and to perform data processing and calculation operations on the digital surface model data to obtain normalized digital surface model data; the feature image extraction unit is used to overlay the normalized digital surface model data and the images to extract several multi-dimensional feature images.

9. A building remote sensing information processing system according to claim 7, characterized in that: The building information extraction module includes a contour data processing unit and a roof style analysis unit; The contour data processing unit is used to construct an instance segmentation model to extract building information from arbitrary multi-dimensional feature images, and to process the extracted building information to obtain the building contour data of each building; the roof style analysis unit is used to obtain the roof style features of each building based on a preset roof style.

10. A building remote sensing information processing system according to claim 7, characterized in that: The building function classification module includes a building distribution analysis unit and a classification result generation unit; The building distribution analysis unit is used to obtain the adjacency relationship between buildings, obtain the comprehensive characteristics of each building based on various feature data, and construct a distribution map containing each building; the classification result generation unit is used to construct a graph neural network to classify each building according to its function and obtain the building function classification result.

Images