Building surveying and mapping system based on remote sensing of unmanned aerial vehicle

Abstract

The invention discloses a building surveying and mapping system based on unmanned aerial vehicle remote sensing, and belongs to the technical field of building surveying and mapping. The system comprises a data acquisition module, a data preprocessing module, a feature extraction module, a spatial relationship analysis module and a dynamic change analysis module, and the data acquisition module is responsible for automatically acquiring multi-temporal remote sensing data covering a target area; the data preprocessing module is responsible for performing cleaning, geometric correction and multi-source data fusion on the acquired remote sensing data; the feature extraction module is responsible for extracting key features of the building from the multi-resolution remote sensing data; the spatial relationship analysis module analyzes the spatial distribution characteristics and the mutual relationship of the building group; and the dynamic change analysis module is responsible for monitoring and analyzing the dynamic change of the building group, including automatic identification of new building, dismantling and reconstruction conditions.

Description

Technical Field

[0001] The present application relates to the technical field of building surveying and mapping, and more specifically, to a building surveying and mapping system based on unmanned aerial vehicle remote sensing. Background Art

[0002] With the acceleration of urbanization, the scale of building complexes continues to expand, and traditional building surveying methods can no longer meet the needs of efficiency, accuracy, and dynamics. Traditional surveying methods often rely on manual operations, which are costly and inefficient. Especially in complex urban environments, the mutual occlusion and height differences between buildings further increase the difficulty of surveying. In addition, traditional surveying is usually based on data collection at a single point in time, and cannot effectively monitor the dynamic changes of building complexes, such as new construction, reconstruction, or demolition.

[0003] In recent years, the rapid development of drone technology has brought new opportunities for architectural surveying and mapping. Drones have the advantages of high flexibility, wide coverage, and no terrain restrictions. They can carry a variety of high-resolution sensors, such as optical cameras and lidar, to obtain high-quality remote sensing data, providing strong support for the stereoscopic mapping and change monitoring of buildings. At the same time, intelligent data processing and analysis technologies based on remote sensing data, such as multi-scale fusion, feature extraction, spatiotemporal analysis, etc., can greatly improve the accuracy of building identification and realize the monitoring of dynamic changes in building complexes. However, most of the existing drone remote sensing mapping systems focus on a single scene or short-term monitoring, and cannot meet the needs of long-term and cross-regional dynamic changes in buildings. In addition, how to effectively deal with noise, outliers in remote sensing data and the fusion of data with different resolutions is also a major challenge in achieving high-precision architectural surveying and mapping.

[0004] In summary, how to design a building surveying and mapping system based on UAV remote sensing to achieve efficient, accurate and dynamic building surveying and change monitoring has become a technical problem that needs to be solved urgently. Summary of the invention

[0005] In order to overcome a series of defects existing in the prior art, the purpose of this application is to provide a building surveying and mapping system based on UAV remote sensing in response to the above-mentioned problems, including a data acquisition module, a data preprocessing module, a feature extraction module, a spatial relationship analysis module and a dynamic change analysis module, wherein the data acquisition module is responsible for automatically acquiring multi-phase remote sensing data covering the target area; the data preprocessing module is responsible for cleaning, geometric correction and multi-source data fusion of the collected remote sensing data; the feature extraction module is responsible for extracting the key features of the building from the multi-resolution remote sensing data; the spatial relationship analysis module analyzes the spatial distribution characteristics of the building complex and their mutual relationships; the dynamic change analysis module is responsible for monitoring and analyzing the dynamic changes of the building complex, including automatic identification of new construction, demolition and reconstruction.

[0006] Furthermore, the data acquisition module includes the following components.

[0007] The UAV platform, as a carrier for data collection, is responsible for providing efficient flight capabilities and stability.

[0008] High-resolution sensors, mounted on UAV platforms, are used to obtain high-precision remote sensing images and spatial data.

[0009] The real-time positioning unit continuously outputs high-precision geographic coordinate information to ensure that data collection has accurate spatial reference.

[0010] Inertial measurement unit monitors the attitude parameters and motion status of the drone in real time.

[0011] The data storage unit is responsible for the real-time storage and backup of remote sensing images and related data.

[0012] Wireless communication unit, which enables two-way real-time data transmission between the UAV and the ground station.

[0013] The automatic navigation unit executes preset route planning and autonomous flight control to achieve full process automation of data collection.

[0014] Furthermore, the spatial resolution of the high-resolution sensor is better than 5 cm / pixel, the spectral range covers visible light to near-infrared band 400-900 nanometers, and the signal-to-noise ratio is greater than 45 decibels; the plane positioning accuracy of the real-time positioning unit is better than ±2 cm, and the elevation positioning accuracy is better than ±3 cm; the attitude measurement accuracy of the inertial measurement unit is better than 0.1 degree.

[0015] Furthermore, the data preprocessing module includes the following components.

[0016] Data cleaning unit, identifies and filters abnormal data points and noise.

[0017] The image registration unit performs spatial alignment processing on multi-temporal and multi-view remote sensing images to establish a unified coordinate reference system.

[0018] The radiation correction unit performs radiation calibration and atmospheric correction on remote sensing data to eliminate the influence of sensors and atmosphere.

[0019] The geometric correction unit optimizes the geometric accuracy of the image by compensating for terrain relief and correcting for perspective distortion.

[0020] The comprehensive data fusion unit performs standardization and multi-source fusion of raw data, including the integration of environmental factor data.

[0021] Furthermore, the image registration unit uses no less than 12 same-name points for registration, and the registration accuracy is better than 0.5 pixel; the correction accuracy of the geometric correction unit is better than 1 pixel; and the radiation resolution of the radiation correction unit is not less than 12 bits.

[0022] Furthermore, the feature extraction module includes the following components.

[0023] The deep convolutional network unit automatically extracts and expresses architectural features, and improves the abstraction level and generalization ability of feature representation through multi-layer convolution operations.

[0024] The multi-scale feature analysis unit realizes the multi-scale hierarchical expression of architectural features and ensures the integrity of feature representation.

[0025] Edge detection unit, accurately extracts the contour features of buildings.

[0026] The texture analysis unit, based on multi-dimensional texture descriptors, quantifies the texture features of building surfaces and extracts local structural information to assist building classification and recognition.

[0027] The monomer shape analysis unit generates the shape feature vector of the monomer building.

[0028] The feature selection and optimization unit constructs the optimal feature subset through information entropy evaluation and correlation analysis.

[0029] Furthermore, the deep convolutional network unit contains no less than 15 convolutional layers, and the convolution kernel size of each layer is 3×3 or 5×5 pixels; the detection accuracy of the edge detection unit is better than 1 pixel, and the edge positioning error is less than 0.8 pixel; the dimension of the shape feature vector extracted by the monomer shape analysis unit is no less than 128 dimensions.

[0030] Furthermore, the spatial relationship analysis module includes the following components.

[0031] The spatial distribution analysis unit of the building complex quantifies the distribution characteristics and spatial structure of the building complex and constructs the topological relationship network of the building complex.

[0032] The adjacent relationship detection unit establishes a topological connection matrix between buildings and optimizes the spatial association identification of building units.

[0033] The context information integration unit integrates multi-source environmental element data and constructs the spatial semantic network of the building.

[0034] The group spatial relationship modeling unit uses the geometric parameters and location characteristics of the building group to establish a quantitative spatial relationship model and realize the formal expression of the complex spatial relationship between the building groups.

[0035] The benchmark spatial feature analysis unit evaluates the statistical characteristics and clustering patterns of the current spatial distribution of the building complex, providing a reliable spatial reference benchmark for the dynamic change analysis module.

[0036] Furthermore, the dynamic change analysis module includes the following components.

[0037] The time series data management unit builds a spatiotemporal database of multi-temporal remote sensing data and building information to achieve full-cycle tracking and management of dynamic changes in building complexes.

[0038] The temporal feature extraction unit extracts the temporal change characteristics of buildings based on multi-temporal remote sensing data, providing effective feature support for change detection.

[0039] The change recognition and classification unit integrates a multi-dimensional feature matching algorithm to achieve automated detection and fine classification of new construction, demolition and renovation of buildings.

[0040] The spatiotemporal evolution analysis unit integrates the time dimension and spatial characteristics, constructs a spatiotemporal evolution model of building changes, and quantifies the change trends and their spatial correlations.

[0041] The building status update unit implements a dynamic update mechanism of the building database based on the change detection results.

[0042] Compared with the prior art, the present application has the following beneficial effects.

[0043] This application uses drone remote sensing technology and multi-scale fusion algorithms to achieve efficient collection and precise analysis of building information, and combines dynamic change detection and three-dimensional visualization to provide a comprehensive and accurate building surveying and mapping solution to support subsequent management and decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a structural diagram of a building surveying and mapping system based on drone remote sensing disclosed in an embodiment of the present application. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical scheme and advantages of the implementation of the present invention clearer, the technical scheme in the embodiment of the present invention will be described in more detail below in conjunction with the drawings in the embodiment of the present invention. In the drawings, the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions. The described embodiments are part of the embodiments of the present invention, not all of the embodiments.

[0046] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without making any creative work shall fall within the scope of protection of the present invention.

[0047] The embodiments and directional terms described below with reference to the accompanying drawings are exemplary and intended to be used to explain the present invention, but should not be construed as limiting the present invention.

[0048] like Figure 1 As shown, a building surveying and mapping system based on unmanned aerial vehicle remote sensing includes a data acquisition module, a data preprocessing module, a feature extraction module, a spatial relationship analysis module and a dynamic change analysis module, wherein the data acquisition module is responsible for automatically acquiring multi-phase remote sensing data covering the target area; the data preprocessing module is responsible for cleaning, geometric correction and multi-source data fusion of the collected remote sensing data; the feature extraction module is responsible for extracting key features of buildings from multi-resolution remote sensing data; the spatial relationship analysis module improves the accuracy and stability of building single body identification by analyzing the spatial distribution characteristics of building complexes and their mutual relationships; the dynamic change analysis module is responsible for monitoring and analyzing the dynamic changes of building complexes, including automatic identification of new construction, demolition and reconstruction.

[0049] As can be seen from the above, the above-mentioned building surveying and mapping system based on drone remote sensing can provide efficient and accurate building surveying and change monitoring services through the close cooperation of data collection, preprocessing, feature extraction, spatial relationship analysis and dynamic change analysis modules. Each module cooperates with each other, from the collection of original remote sensing data to the extraction of building features, and then to the real-time monitoring of dynamic changes, forming a closed-loop automation system. This system can not only improve the accuracy and efficiency of building surveying and mapping, but also provide strong technical support for building planning, management and urban development.

[0050] Furthermore, the data acquisition module includes the following components.

[0051] The UAV platform, as a carrier for data collection, provides efficient flight capabilities and stability to ensure full coverage of the target area.

[0052] High-resolution sensors, mounted on drone platforms, are used to obtain high-precision remote sensing images and spatial data, providing reliable data support for building information collection.

[0053] The real-time positioning unit continuously outputs high-precision geographic coordinate information to ensure that data collection has accurate spatial reference, providing a basic guarantee for subsequent analysis and processing.

[0054] The inertial measurement unit monitors the attitude parameters and motion status of the drone in real time, optimizes flight stability and improves the accuracy of collected data.

[0055] The data storage unit is responsible for the real-time storage and backup of remote sensing images and related data to ensure the integrity and reliability of the data.

[0056] The wireless communication unit realizes two-way real-time data transmission between the UAV and the ground station, and supports dynamic adjustment of mission monitoring and acquisition strategies.

[0057] The automatic navigation unit executes preset route planning and autonomous flight control, realizes full process automation of data collection, improves work efficiency and reduces human intervention.

[0058] The above modules work closely together to form an efficient and accurate data collection system. The drone platform provides flight capability and stability, high-resolution sensors provide an accurate basis for data, real-time positioning units and inertial measurement units ensure the accuracy and stability of spatial data, data storage units ensure the integrity and reliability of data, wireless communication units realize real-time transmission and dynamic regulation of data, and automatic navigation units improve work efficiency through intelligent control. The integration of these technologies enables building information collection to be carried out efficiently in complex environments and provides data support for subsequent building modeling, analysis and decision-making.

[0059] In this embodiment, the spatial resolution of the high-resolution sensor is better than 5 cm / pixel, the spectral range covers visible light to near-infrared band 400-900 nanometers, and the signal-to-noise ratio is greater than 45 decibels; the plane positioning accuracy of the real-time positioning unit is better than ±2 cm, and the elevation positioning accuracy is better than ±3 cm; the attitude measurement accuracy of the inertial measurement unit is better than 0.1 degrees. In addition, the maximum take-off weight of the UAV platform is 15 kg, the maximum load is 5 kg, the endurance time is not less than 45 minutes, and the wind resistance reaches level 6; the effective communication distance of the wireless communication unit is not less than 5 kilometers, and the data transmission rate is not less than 10 megabytes / second.

[0060] Furthermore, the data preprocessing module includes the following components.

[0061] The data cleaning unit, based on statistical analysis methods, identifies and filters abnormal data points and noise, improves data quality, and ensures the reliability of input data for subsequent analysis.

[0062] The image registration unit performs spatial alignment processing on multi-temporal and multi-view remote sensing images, establishes a unified coordinate reference system, and ensures the consistency of ground features in time and space dimensions.

[0063] The radiation correction unit performs radiation calibration and atmospheric correction on remote sensing data, eliminates the influence of sensors and atmosphere, and realizes standardized processing of radiation characteristics of multi-source data.

[0064] The geometric correction unit optimizes the geometric accuracy of the image by compensating for terrain undulations and correcting for perspective distortion, providing high-precision geospatial information.

[0065] The comprehensive data fusion unit performs standardized processing and multi-source fusion of raw data, including the integration of environmental factor data, to build a unified data foundation for subsequent analysis.

[0066] As can be seen from the above, the data preprocessing module plays a vital role in the entire remote sensing building surveying and mapping system. It ensures the quality, accuracy and consistency of the data through a series of technical means. The data cleaning unit improves the reliability of the data by identifying and removing abnormal data; the image registration unit ensures the spatial consistency of images at different times and different perspectives; the radiation correction unit eliminates the interference of the atmosphere and sensors on the data, so that the data has a unified radiation characteristic; the geometric correction unit optimizes the spatial accuracy of the image and ensures the high accuracy of the surveying and mapping; the comprehensive data fusion unit integrates multi-source data, providing a solid foundation for subsequent analysis and modeling. Through these processing steps, the data preprocessing module not only improves the quality of remote sensing data, but also provides reliable data support for subsequent tasks such as building feature extraction and change monitoring, ensuring the efficiency and accuracy of the building surveying and mapping system.

[0067] In this embodiment, the image registration unit uses no less than 12 same-name points for registration, and the registration accuracy is better than 0.5 pixel; the correction accuracy of the geometric correction unit is better than 1 pixel; and the radiation resolution of the radiation correction unit is not less than 12 bits.

[0068] Furthermore, the image coordinates after terrain undulation compensation and perspective distortion correction optimization are expressed as: (x′, y′) = (x0 + Δx(h0) + Δx distort (h camera ,θ0),y0+Δy(h0)+Δy distort (h camera ,θ0)), where (x′,y′) is the corrected image coordinate, which means the position of the image pixel in the geographic space after terrain compensation and perspective distortion correction; (x0,y0) is the original image coordinate, which refers to the coordinate of a pixel in the image without any compensation and correction; h camera is the camera height, which refers to the height of the camera on the drone relative to the ground; θ0 is the pitch angle of the camera, that is, the angle between the camera's shooting direction and the ground; Δx distort Indicates the horizontal position offset caused by perspective distortion; Δy distort represents the vertical position offset caused by perspective distortion; Δx(h0) represents the horizontal coordinate offset caused by terrain undulation; Δy(h0) represents the vertical coordinate offset caused by terrain undulation; Δx distort (h camera ,θ0) represents the camera height h cameraThe horizontal offset caused by the viewing angle θ0; Δy distort (h camera ,θ0) represents the height of the camera h camera and the vertical offset caused by viewing angle θ0.

[0069] Furthermore, the feature extraction module includes the following components.

[0070] The deep convolutional network unit builds an end-to-end deep learning framework to automatically extract and express architectural features, and improves the abstraction level and generalization ability of feature representation through multi-layer convolution operations.

[0071] The multi-scale feature analysis unit realizes the multi-scale hierarchical expression of architectural features, ensures the integrity of feature representation, and improves the accuracy and stability of recognition.

[0072] The edge detection unit uses advanced edge detection algorithms to accurately extract the contour features of buildings, providing basic support for subsequent geometric analysis and structural recognition.

[0073] The texture analysis unit, based on multi-dimensional texture descriptors, quantifies the texture features of building surfaces and extracts local structural information to assist building classification and recognition.

[0074] The single-body shape analysis unit extracts morphological features of independent buildings, calculates geometric parameters, and generates shape feature vectors of single buildings to provide support for building type classification.

[0075] The feature selection and optimization unit integrates feature screening and dimensionality reduction algorithms, constructs the optimal feature subset through information entropy evaluation and correlation analysis, and improves algorithm efficiency and performance.

[0076] As can be seen from the above, the feature extraction module effectively improves the accuracy and efficiency of building feature extraction through multiple units such as deep convolutional networks, multi-scale feature analysis, edge detection, texture analysis, monomer shape analysis, and feature selection and optimization. Each unit extracts the key features of buildings at different levels and dimensions to ensure that buildings can be fully and accurately identified and analyzed. These technologies not only improve the ability to classify, identify and monitor changes in buildings, but also enhance the adaptability and stability of the system in complex environments.

[0077] In this embodiment, the deep convolutional network unit includes no less than 15 convolutional layers, and the convolution kernel size of each layer is 3×3 or 5×5 pixels; the detection accuracy of the edge detection unit is better than 1 pixel, and the edge positioning error is less than 0.8 pixel; the dimension of the shape feature vector extracted by the monomer shape analysis unit is no less than 128 dimensions.

[0078] Furthermore, an end-to-end deep learning framework is constructed to automatically extract and express architectural features, and the abstraction level and generalization ability of feature representation are improved through multi-layer convolution operations, including the following steps.

[0079] In the convolution layer, the convolution kernel K (l) With input X (l-1) Perform convolution operation to extract local features of the building. The formula is: (l) =σ(X (l-1) *K (l) ), where Y (l) represents the output of the convolution operation of the lth layer, that is, the feature map of this layer; σ represents the activation function; * represents the convolution operation.

[0080] By stacking convolutional layers, architectural features from low-level to high-level are abstracted layer by layer to capture details and overall structure.

[0081] The multi-scale features of the building are extracted through convolution kernels of different scales to enhance the model's adaptability to buildings of different sizes and shapes. The formula is: Y multi-scale (l) =σ(X (l-1) *K scale (l) ), where Y multi-scale (l) Represents the output feature map after the l-th layer multi-scale convolution operation; K scale (l) Represents the multi-scale convolution kernel of the lth layer.

[0082] The convolutional features are converted into a fixed-length feature vector Z through a fully connected layer for building classification or recognition. The formula is: Z = W·A (L) +b, where W is the weight matrix in the fully connected layer, which is used to map the features of the convolutional layer output to a specific category or the final output space; A (L) is the activation output of the Lth layer, indicating the output of the last layer of the deep learning network, which is the feature representation after processing by all previous layers; b is the bias term, which is used to adjust the output result.

[0083] Use the back-propagation algorithm to optimize the network, update the weight parameters, and improve the accuracy and generalization ability of feature extraction through multiple iterations.

[0084] Furthermore, an advanced edge detection algorithm is used to accurately extract the contour features of the building, including the following steps.

[0085] Gaussian filtering is performed on the input image to remove noise and retain the main structural features, which can be expressed as: I′(i,j)=∑ m,nI(i+m,j+n)·G(m,n), where I′(i,j) represents the pixel value at position (i,j) of the image I′ after Gaussian filtering; m,n represent the index of the convolution kernel of the Gaussian filter; I(i+m,j+n) represents the neighborhood pixels around the pixel at coordinate (i,j) in the original image I, and (i+m,j+n) represents the offset in the m,n direction around the current position; G(m,n) is the weight value of the Gaussian filter, which is used for image convolution operations and defines the degree of influence of the filter in the m,n direction.

[0086] Using the gradient operator, the gradient of the image in the horizontal and vertical directions is calculated, which can be expressed as: G x =∑ i,j [I(i,j)·G x (i,j)],G y =∑ i,j [I(i,j)·G y (i,j)], where G x Represents the horizontal gradient of the image, describing the magnitude of the brightness change of the image in the horizontal direction; I(i,j) represents the pixel value of a specific position (i,j) in the original image I; G x (i, j) represents the horizontal gradient value of the image at position (i, j); G y Represents the vertical gradient of the image, describing the magnitude of the brightness change of the image in the vertical direction; G y (i, j) represents the vertical gradient value of the image at position (i, j).

[0087] Combine the calculated horizontal and vertical gradients to calculate the gradient magnitude and gradient direction of each pixel, which can be expressed as: G = √(G x 2 +G y 2 ), θ = tan -1 (G y / G x ), where G represents the gradient amplitude of the image at position (i, j), that is, the intensity of the brightness change; θ represents the gradient direction of the image at position (i, j), that is, the direction of the edge.

[0088] The gradient magnitude is binarized by a threshold value T to extract the edge, wherein when the gradient magnitude>T, the pixel is considered to be part of the edge.

[0089] The Canny algorithm is applied to accurately extract the building edges through Gaussian smoothing, gradient calculation, non-maximum suppression and double threshold processing.

[0090] The edge detection results are optimized to remove noise and breaks, and the continuity and accuracy of the contour are improved through edge connection and smoothing.

[0091] Furthermore, based on the multi-dimensional texture descriptor, the texture features of the building surface are quantified and the local structure information is extracted to assist the classification and recognition of the building, which includes the following steps.

[0092] The gray-level co-occurrence matrix of the building surface image is calculated to quantify the spatial relationship between pixels, and the local neighborhood of each pixel is encoded using the local binary pattern to extract local texture information, which is expressed as: P(e,f)=[∑ x,y δ(e(x,y)-e)δ(e(x+dx,y+dy)-f)] / N, where P(e,f) represents the element in the gray-level co-occurrence matrix, which represents the joint probability of the pixel pair with gray-level values ​​e and f in the image; δ(e(x,y)-e) is an indicative function used to calculate the matching between the pixel and the gray-level value e. When the gray-level value of the pixel e(x,y) is equal to e, the function value is 1, otherwise it is 0; e(x,y) represents the gray-level value of the pixel at position (x,y) in image e; δ(e(x+dx,y+dy)-f) represents whether the gray-level value of the pixel at position (x+dx,y+dy) in the image is equal to f. If so, it is 1, otherwise it is 0; e(x+dx,y+dy) represents the gray-level value of the pixel at position (x+dx,y+dy) in image e; N represents the total number of pixels in the image, which is used to normalize the frequency of the gray-level co-occurrence matrix to ensure that its sum is 1.

[0093] Calculate the energy value E of the co-occurrence matrix, quantify the texture intensity, and distinguish between smooth and rough areas. The formula is: E=∑ e,f P(e,f) 2 .

[0094] For each pixel, the texture pattern of its surrounding neighborhood is calculated to accurately capture the subtle local structural features of the building surface, which can be expressed as: LBP(x,y)=∑ p=0 P-1 s(e p -e c )·2 p , where LBP(x,y) is the result of the local binary pattern, which represents the binary encoding value based on its neighborhood at the pixel point (x,y); P is the number of neighborhood pixels, and P-1 represents the index value of the last neighborhood pixel; e p is the gray value of the neighborhood pixel at position p; e c is the gray value of the center pixel, that is, the gray value of the pixel at position (x, y); s(e p -e c ) is a symbolic function used to calculate e p With ec Is the difference between the two greater than zero?

[0095] The texture is analyzed at multiple scales through wavelet transform to capture texture features at different levels and process building surface details at different scales.

[0096] Apply edge detection algorithm to extract edge information of building surface and clarify the shape and structure of the building.

[0097] The extracted texture features and structural features are fused to generate a feature vector, and the support vector machine algorithm is used to classify the buildings, calculate the posterior probability, and predict the category to which the building belongs.

[0098] Furthermore, extracting morphological features of independent buildings, calculating geometric parameters, and generating shape feature vectors of individual buildings include the following steps.

[0099] According to the extracted building outline, the polygon area formula is used to calculate the building area A, which is expressed as: A=│∑ q=1 Q (x q y q+1 -x q+1 y q )│ / 2, where Q is the number of vertices of the building outline; (x q ,y q ) is the coordinate of the qth vertex on the contour, representing a point on the outer boundary of the building; (x q+1 ,y q+1 ) are the coordinates of the q+1th vertex on the contour.

[0100] The perimeter C is obtained by calculating the side length of the building outline, which can be expressed as: C = √[∑ q=1 Q (x q -x q+1 ) 2 +(y q -y q+1 ) 2 ].

[0101] Calculate the building's aspect ratio R, which is used to describe the length and width relationship of the building's shape. The formula is: R=L max / L min , where L max is the longest side length of the building's circumscribed rectangle; L min It is the length of the shortest side of the building's circumscribed rectangle.

[0102] The shape factor S is calculated by the area A and the perimeter C to measure the compactness of the building, which is expressed as: S=P 2 / (4πA).

[0103] Combine the area, perimeter, aspect ratio, and shape factor into a feature vector: f=[A,C,R,S].

[0104] Furthermore, the spatial relationship analysis module includes the following components.

[0105] The spatial distribution analysis unit of the building complex quantifies the distribution characteristics and spatial structure of the building complex and constructs the topological relationship network of the building complex.

[0106] The adjacent relationship detection unit establishes a topological connection matrix between buildings and optimizes the spatial association identification of building units.

[0107] The context information integration unit integrates multi-source environmental element data, builds a spatial semantic network of buildings, and improves the ability to interpret complex spatial relationships.

[0108] The group spatial relationship modeling unit uses the geometric parameters and location characteristics of the building group to establish a quantitative spatial relationship model and realize the formal expression of the complex spatial relationship between the building groups.

[0109] The benchmark spatial feature analysis unit evaluates the statistical characteristics and clustering patterns of the current spatial distribution of the building complex, providing a reliable spatial reference benchmark for the dynamic change analysis module.

[0110] As can be seen above, the spatial relationship analysis module effectively improves the ability to understand and interpret the spatial relationship of building complexes through multiple components such as building complex spatial distribution analysis, adjacent relationship detection, context information integration, group spatial relationship modeling, and baseline spatial feature analysis. Each unit analyzes the spatial relationship between buildings from different perspectives, whether from the perspective of geometry, environmental impact, or function and topology, providing an accurate quantitative description of the complex spatial relationship between buildings. This comprehensive spatial relationship analysis not only improves the accuracy of spatial understanding of building complexes, but also provides strong data support for tasks such as building identification, dynamic change monitoring, urban planning, and building management.

[0111] In this embodiment, the spatial clustering analysis of the building complex spatial distribution analysis unit adopts the DBSCAN algorithm, in which the neighborhood radius parameter is set to 20 meters and the minimum number of sample points is set to 4; the building spatial adjacency threshold set by the neighbor relationship detection unit is 10 meters, and the sparsity of the topological connection matrix does not exceed 0.15.

[0112] Furthermore, the distribution characteristics and spatial structure of the building complex are quantified, and a topological relationship network of the building complex is constructed, including the following steps.

[0113] The kernel density estimation method is used to analyze the spatial density distribution of building complexes in order to identify the spatial aggregation pattern and distribution law of building complexes.

[0114] The spatial distance matrix between buildings is constructed based on the actual distance of the road network to characterize the spatial proximity relationship between buildings.

[0115] The spatial association threshold is determined according to the characteristics of the target area and is used to define the effective spatial association between buildings.

[0116] Based on the determined spatial association threshold, buildings are taken as network nodes and spatial association relationships as network edges to construct a topological relationship network of the building complex.

[0117] The constructed topological relationship network is analyzed and the key indicators of the network are calculated, including degree distribution, clustering coefficient and centrality index, so as to evaluate the spatial organizational structure and functional association characteristics of the building complex.

[0118] As can be seen from the above, by adopting a series of steps such as kernel density estimation, actual distance calculation of road network, setting spatial association threshold, topological network construction and network analysis, the spatial distribution characteristics and structure of the building complex can be quantified and accurately modeled. These steps not only help to fully understand the spatial relationship between buildings in the building complex, but also provide strong data support for subsequent dynamic change analysis, urban planning decisions, building function analysis, etc. Through the analysis of the topological relationship network, key nodes, close areas and potential functional relationships in the building complex can be identified, providing a scientific basis for urban spatial layout and building management.

[0119] Furthermore, the dynamic change analysis module includes the following components.

[0120] The time series data management unit builds a spatiotemporal database of multi-temporal remote sensing data and building information to achieve full-cycle tracking and management of dynamic changes in building complexes.

[0121] The temporal feature extraction unit extracts the temporal change characteristics of buildings based on multi-temporal remote sensing data, providing effective feature support for change detection.

[0122] The change recognition and classification unit integrates a multi-dimensional feature matching algorithm to achieve automated detection and fine classification of new construction, demolition and renovation of buildings.

[0123] The spatiotemporal evolution analysis unit integrates the time dimension and spatial characteristics, constructs a spatiotemporal evolution model of building changes, and quantifies the change trends and their spatial correlations.

[0124] The building status update unit implements a dynamic update mechanism for the building database based on the change detection results to ensure the real-time and accuracy of the data status.

[0125] As can be seen above, the dynamic change analysis module provides full-cycle and all-round technical support for dynamic change monitoring of building complexes through the collaboration of sub-modules such as time series data management, time series feature extraction, change identification and classification, spatiotemporal evolution analysis, and building status update. Through precise spatiotemporal data analysis and the application of intelligent algorithms, it can monitor the changes of building complexes in real time, automatically identify dynamic changes such as new construction, demolition and reconstruction, and analyze their evolution trends and spatial correlations.

[0126] Furthermore, the building surveying and mapping system based on UAV remote sensing also includes a result output module responsible for generating a building surveying and mapping report and visualization results, and the result output module includes the following components.

[0127] The intelligent report generation unit integrates multi-dimensional analysis results, supports standardized and customized report templates, systematically presents the building's spatial characteristics, dynamic changes, and analysis and evaluation results, and provides flexible document support for building management decisions.

[0128] The three-dimensional model generation unit constructs a three-dimensional geometric model of the building complex based on multi-source remote sensing data and building feature parameters, achieving high-precision visual expression of the building space structure and layout.

[0129] The change map generation unit integrates the results of spatiotemporal change analysis to generate thematic maps of dynamic changes in building complexes, presenting the spatial distribution and evolution of change elements for easy intuitive understanding.

[0130] The data format conversion unit provides a multi-format data conversion interface to support the seamless migration of analysis results to mainstream GIS platforms and databases, ensuring data interoperability.

[0131] As can be seen above, the result output module realizes the comprehensive display and efficient sharing of architectural surveying and mapping results through four key units: intelligent report generation, 3D model construction, change map generation, and data format conversion. Through this module, not only can systematic and standardized reports be obtained, but also the spatial structure and change trend of the building can be deeply understood through 3D models and dynamic change maps. In addition, the data format conversion unit ensures the seamless integration and cross-platform application of the results.

[0132] Finally, it should be pointed out that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them. Although the present invention has been described in detail with reference to the above embodiments, a person skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A building surveying and mapping system based on drone remote sensing, characterized in that: It includes a data acquisition module, a data preprocessing module, a feature extraction module, a spatial relationship analysis module and a dynamic change analysis module, wherein the data acquisition module is responsible for automatically acquiring multi-phase remote sensing data covering the target area; the data preprocessing module is responsible for cleaning, geometric correction and multi-source data fusion of the collected remote sensing data; the feature extraction module is responsible for extracting key features of buildings from multi-resolution remote sensing data; the spatial relationship analysis module analyzes the spatial distribution characteristics of the building complex and their mutual relationships; the dynamic change analysis module is responsible for monitoring and analyzing the dynamic changes of the building complex, including automatic identification of new construction, demolition and reconstruction.

2. The building surveying and mapping system based on UAV remote sensing according to claim 1 is characterized in that: The data acquisition module includes the following components: The UAV platform, as a carrier for data collection, is responsible for providing efficient flight capabilities and stability; High-resolution sensors, mounted on UAV platforms, are used to obtain high-precision remote sensing images and spatial data; Real-time positioning unit, continuously outputs high-precision geographic coordinate information to ensure accurate spatial reference for data collection; Inertial measurement unit, real-time monitoring of the drone’s attitude parameters and motion status; Data storage unit, responsible for real-time storage and backup of remote sensing images and related data; Wireless communication unit, which enables two-way real-time data transmission between the UAV and the ground station; The automatic navigation unit executes preset route planning and autonomous flight control to achieve full process automation of data collection.

3. The building surveying and mapping system based on UAV remote sensing according to claim 2 is characterized in that: The spatial resolution of the high-resolution sensor is better than 5 cm / pixel, the spectral range covers visible light to near-infrared band 400-900 nanometers, and the signal-to-noise ratio is greater than 45 decibels; the plane positioning accuracy of the real-time positioning unit is better than ±2 cm, and the elevation positioning accuracy is better than ±3 cm; the attitude measurement accuracy of the inertial measurement unit is better than 0.1 degree.

4. The building surveying and mapping system based on UAV remote sensing according to claim 1 is characterized in that: The data preprocessing module includes the following components: Data cleaning unit, identifying and filtering abnormal data points and noise; Image registration unit, which performs spatial alignment processing on multi-temporal and multi-view remote sensing images to establish a unified coordinate reference system; Radiation correction unit, which performs radiation calibration and atmospheric correction on remote sensing data to eliminate the influence of sensors and atmosphere; The geometric correction unit optimizes the geometric accuracy of the image by compensating for terrain relief and correcting for perspective distortion; The comprehensive data fusion unit performs standardization and multi-source fusion of raw data, including the integration of environmental factor data.

5. The building surveying and mapping system based on UAV remote sensing according to claim 4 is characterized in that: The image registration unit uses no less than 12 same-name points for registration, and the registration accuracy is better than 0.5 pixel; the correction accuracy of the geometric correction unit is better than 1 pixel; the radiation resolution of the radiation correction unit is not less than 12 bits.

6. The building surveying and mapping system based on UAV remote sensing according to claim 1 is characterized in that: The feature extraction module includes the following components: Deep convolutional network units automatically extract and express architectural features, and improve the abstraction level and generalization ability of feature representation through multi-layer convolution operations; Multi-scale feature analysis unit, which realizes multi-scale hierarchical expression of building features and ensures the integrity of feature representation; Edge detection unit, accurately extracting the contour features of buildings; Texture analysis unit, based on multi-dimensional texture descriptors, quantifies the texture features of building surfaces and extracts local structural information to assist building classification and recognition; A monomer shape analysis unit generates a shape feature vector of a monomer building; The feature selection and optimization unit constructs the optimal feature subset through information entropy evaluation and correlation analysis.

7. The building surveying and mapping system based on UAV remote sensing according to claim 6 is characterized in that: The deep convolutional network unit contains no less than 15 convolutional layers, and the convolution kernel size of each layer is 3×3 or 5×5 pixels; the detection accuracy of the edge detection unit is better than 1 pixel, and the edge positioning error is less than 0.8 pixel; the dimension of the shape feature vector extracted by the monomer shape analysis unit is no less than 128 dimensions.

8. The building surveying and mapping system based on drone remote sensing according to claim 1 is characterized in that: The spatial relationship analysis module includes the following components: The building complex spatial distribution analysis unit quantifies the distribution characteristics and spatial structure of the building complex and constructs the topological relationship network of the building complex; Adjacent relationship detection unit, to establish the topological connection matrix between buildings and optimize the spatial association identification of building units; The context information integration unit integrates multi-source environmental element data to construct a spatial semantic network of the building; The group spatial relationship modeling unit uses the geometric parameters and location characteristics of the building group to establish a quantitative spatial relationship model to formalize the complex spatial relationship between the building groups; The benchmark spatial feature analysis unit evaluates the statistical characteristics and clustering patterns of the current spatial distribution of the building complex, providing a reliable spatial reference benchmark for the dynamic change analysis module.

9. The building surveying and mapping system based on UAV remote sensing according to claim 1 is characterized in that: The dynamic change analysis module includes the following components: The time series data management unit builds a spatiotemporal database of multi-temporal remote sensing data and building information to achieve full-cycle tracking and management of dynamic changes in building complexes; The temporal feature extraction unit extracts the temporal change characteristics of buildings based on multi-temporal remote sensing data, providing effective feature support for change detection; Change recognition and classification unit, integrating multi-dimensional feature matching algorithm to realize automatic detection and fine classification of new construction, demolition and reconstruction of buildings; The spatiotemporal evolution analysis unit integrates the time dimension and spatial characteristics to construct a spatiotemporal evolution model of building changes and quantify the change trend and its spatial correlation; The building status update unit implements a dynamic update mechanism of the building database based on the change detection results.

Images