An urban pipeline simulation data construction method based on intelligent remote sensing interpretation

Abstract

This invention discloses a method for constructing urban pipeline simulation data based on intelligent remote sensing interpretation, relating to the technical fields of intelligent remote sensing image interpretation and rapid generation of basic urban geographic information data. The method includes: acquiring original orthophotos of the city; preprocessing the original orthophotos; annotating the preprocessed orthophotos according to road contours using image annotation tools, and pre-creating sample training and validation sets; optimizing the encoding and decoding structures in the original U-Net model to establish an urban pipeline simulation data construction model and completing model training; extracting road lines using the trained model; segmenting the road lines; and assigning attributes to the segmented points and lines to complete the generation of pipeline data. This invention achieves rapid generation of urban pipeline network data by intelligently and automatically extracting roads from remote sensing images and constructing pipeline simulation data based on the extracted road data, providing pipeline simulation data under specific conditions.

Description

Technical Field

[0001] This invention relates to the field of intelligent interpretation of remote sensing images and rapid generation of basic urban geographic information data, and in particular to a method for constructing urban pipeline simulation data based on intelligent remote sensing interpretation. Background Technology

[0002] The construction of urban digital twins utilizes digital twin technology to create a digital twin city base that corresponds one-to-one with and maps to the physical city. This enables the digitization and virtualization of all urban elements, providing real-time and accurate decision support for city managers. The core of building this digital twin city base is acquiring basic surveying and mapping data. However, as urban digital twins are continuously developed, the cost of data acquisition is becoming increasingly high. For example, acquiring simulation data of the city's underground pipe network does not require precise data; only simulated data is needed.

[0003] Traditional methods for acquiring underground pipeline network data involve manual surveying, including field investigations, non-destructive testing techniques, underground pipeline surveys, and analysis of pipeline maps. While these methods provide relatively accurate network data, they are costly and inefficient. In certain scenarios using digital twins, where high data accuracy is not required for "non-interested areas," excessive precision may not be economically viable.

[0004] Therefore, proposing a method for constructing urban pipeline simulation data based on intelligent remote sensing interpretation to overcome the difficulties of existing technologies is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] In view of this, the present invention provides a method for constructing urban pipeline simulation data based on intelligent remote sensing interpretation. This method can intelligently and automatically extract roads from remote sensing images and construct pipeline simulation data based on the extracted road data, thereby realizing the rapid generation of urban pipeline network data and providing pipeline simulation data under specific circumstances.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for constructing urban pipeline simulation data based on intelligent remote sensing interpretation includes:

[0008] S1. Obtain the original orthophoto of the city;

[0009] S2. Preprocess the original orthophoto to obtain the preprocessed orthophoto.

[0010] S3. Use image annotation tools to annotate the preprocessed orthophotos according to the road contours, and pre-create sample training sets and validation sets;

[0011] S4. Based on the encoding and decoding structure in the original U-Net model, optimize it to establish a model for constructing urban pipeline simulation data and complete the training of the model;

[0012] S5. Use the trained model to extract road lines;

[0013] S6. Divide the road lines;

[0014] S7. Assign attributes to the segmented points and lines to complete the generation of pipeline data.

[0015] Optionally, in the above method, in S1, the original orthophoto of the city is acquired by using a drone to collect the original orthophoto data of the city.

[0016] Optionally, in the above method, in S2, preprocessing is performed on the original orthophoto image, specifically as follows:

[0017] The original orthophoto data of the city was collected, and the images were stitched and orthorectified. Aerial photos taken by the UAV during takeoff and landing were manually deleted, and photos with severe exposure and poor focus were removed. The orthophoto map was synthesized by following the process of importing the original images, aerial triangulation, model reconstruction and result input.

[0018] Optionally, in S3 of the above method, an image annotation tool is used to annotate the preprocessed orthophoto according to the road contour, and a sample training set and validation set are pre-made.

[0019] Target samples were extracted according to the geometric shape and spectral characteristics of the ground features. The open-source Labelme data annotation tool was used to perform semantic segmentation and dataset annotation on the road contour images. The annotated sample data was divided into training and validation sets at a ratio of 9:1.

[0020] Optionally, in S4, the above method is optimized based on the encoding and decoding structure of the original U-Net model to build a model from the urban pipeline simulation data, and the model training is completed. Specifically:

[0021] Optimization is performed based on the encoding and decoding structure of the original U-Net model: the original U-Net network is introduced into the wavelet transform module; a channel attention module is constructed and fused into the skip connections of the original network; a Res Net residual module is introduced to obtain the optimized model network structure and establish a model for urban pipeline simulation data construction.

[0022] During model training: The labeled orthophoto data is divided according to the Pascal VOC data format. The original size of the image is 512 pixels * 512 pixels. According to the optimized model network structure, the number of training epochs is set to 100, and the batch size of samples captured in a single batch is set to 4 to complete the model training.

[0023] Optionally, in S5 of the above method, a trained model is used to extract road lines: the image data of the area to be identified is input into the trained urban pipeline simulation data to build a model, the identification results of the road network are extracted, the identified raster results are converted into .shp format, and the center line of the road network is extracted based on the identification results.

[0024] Optionally, in S6, the road lines can be segmented using the above method. The specific method is as follows:

[0025] Based on road data to simulate pipeline data, long line segments are spatially decomposed into multiple smaller line segments with a length of 0-70 meters. Various business attributes are assigned to these smaller segments to form pipeline segments. The two endpoints of each segment are extracted separately to generate pipeline points, resulting in geometric lines and geometric points that conform to pipeline standards and specifications.

[0026] Optionally, in S7, the above method assigns attributes to the segmented points and lines to complete the generation of pipeline data, specifically including:

[0027] Attribute values ​​are assigned to the obtained geometric data of pipe segments and pipe points to complete the generation of pipeline data;

[0028] The pipe segment includes five basic attributes: unique pipeline number, starting pipe point number, ending pipe point number, starting point burial depth, and ending point burial depth.

[0029] The pipe point includes five attributes: unique pipe point number, pipe point elevation, accessories, pipe point characteristics, and manhole cover shape.

[0030] Among the above attributes, the unique point number attribute values ​​are sequentially specified as 0, 1, 2, 3, ... during the generation process. The starting point burial depth, ending point burial depth, and pipe point elevation are randomly selected from 0 to 1 meter. Other attribute values ​​are randomly specified or fixed through a value list. Finally, a point table and a line table containing the attributes are obtained.

[0031] As can be seen from the above technical solution, compared with the prior art, the present invention provides a method for constructing urban pipeline simulation data based on intelligent remote sensing interpretation, which has the following beneficial effects: 1) The present invention reduces data acquisition costs and improves work efficiency. It can automatically extract road line data from remote sensing images without a large amount of manual intervention, and can quickly and accurately generate pipeline data, significantly reducing manpower and material resources and accelerating data acquisition speed; 2) The generated data is consistent with the format, geometry and attributes of real data, without distortion; 3) By quickly constructing urban pipeline layout scenarios, it can help planners and designers simulate the operation of urban pipeline network systems, including water supply, drainage, gas, electricity and other pipeline systems; 4) Through simulation analysis, the operating efficiency, safety and reliability of pipeline systems can be evaluated, providing a scientific basis for urban planning and design. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0033] Figure 1 A flowchart of a method for constructing urban pipeline simulation data based on intelligent remote sensing interpretation provided by the present invention;

[0034] Figure 2 The preprocessed urban orthophoto provided by this invention;

[0035] Figure 3 This is a road surface image after road line extraction provided by the present invention;

[0036] Figure 4 The extracted road vector line image provided by this invention;

[0037] Figure 5 The 3D modeling effect of pipeline data in 3DMax provided by this invention

[0038] Figure 6 The application effect of the pipeline data provided by this invention in real-world scenarios.

[0039] Figure 7 The pipeline data provided by this invention is detailed Micro-features Effect;

[0040] Figure 8 This is a schematic diagram of the method for constructing urban pipeline simulation data based on intelligent remote sensing interpretation according to the present invention. Detailed Implementation

[0041] 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.

[0042] In this application, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. The terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0043] Reference Figure 1 As shown, this invention discloses a method for constructing urban pipeline simulation data based on intelligent remote sensing interpretation, including:

[0044] S1. Obtain the original orthophoto of the city;

[0045] S2. Preprocess the original orthophoto to obtain the preprocessed orthophoto.

[0046] S3. Use image annotation tools to annotate the preprocessed orthophotos according to the road contours, and pre-create sample training sets and validation sets;

[0047] S4. Based on the encoding and decoding structure in the original U-Net model, optimize it to establish a model for constructing urban pipeline simulation data and complete the training of the model;

[0048] S5. Use the trained model to extract road lines;

[0049] S6. Divide the road lines;

[0050] S7. Assign attributes to the segmented points and lines to complete the generation of pipeline data.

[0051] Furthermore, in S1, the original orthophotos of the city are acquired by using drones to collect the original orthophoto data of the city.

[0052] Furthermore, refer to Figure 2As shown in Figure S2, the original orthophoto image undergoes preprocessing, specifically as follows:

[0053] Because remote sensing images are affected by factors such as wind and lighting, aerial photos may have optical distortion and other problems. Therefore, it is necessary to perform image stitching and orthorectification on the original data, manually delete aerial photos taken by the drone during takeoff and landing, and also remove photos that are severely overexposed or out of focus. The orthorectified image is then synthesized by following the process of importing the original image, aerial triangulation, model reconstruction, and result input.

[0054] Furthermore, in S3, image annotation tools are used to annotate the preprocessed orthophotos according to road contours, and sample training and validation sets are pre-made:

[0055] Target samples were extracted according to the requirements of ground feature outline geometry, spectral features, etc., and the open-source Labelme data annotation tool was used to perform semantic segmentation dataset annotation on the road images respectively.

[0056] Furthermore, to ensure the accuracy and generalization ability of model training, the annotated sample data is divided into training and validation sets in a 9:1 ratio.

[0057] Furthermore, in S4, the encoding and decoding structures in the original U-Net model are optimized to establish a model for constructing urban pipeline simulation data, and the model training is completed. Specifically:

[0058] Optimizations were made based on the encoding and decoding structures of the original U-Net model: The original U-Net network was introduced into a wavelet transform module, which increases the information content (edges, textures) of the remote sensing images, resulting in higher quality images; a channel attention module was constructed and integrated into the skip connections of the original network, which helps the network better distinguish between targets and background, making the recognition results more accurate; a ResNet residual module was introduced, which improves the accuracy and stability of image segmentation, better handles complex image detection tasks, improves the model's generalization ability and performance, and significantly improves the efficiency of road network extraction, resulting in an optimized model network structure for building a model based on urban pipeline simulation data.

[0059] During model training: The labeled orthophoto data is divided according to the PascalVOC data format. The original size of the image is 512 pixels * 512 pixels. According to the optimized model network structure, the number of training epochs is set to 100, and the batch size of samples captured in a single batch is set to 4 to complete the model training.

[0060] Furthermore, refer to Figure 3As shown in S5, the trained model is used to extract road lines: the image data of the area to be identified is input into the trained urban pipeline simulation data to build the model, the identification results of the road network are extracted, the identified raster results are converted into .shp format, and the center line of the road network is extracted according to the identification results;

[0061] Furthermore, refer to Figure 4 As shown, the image data is analyzed and processed using the trained graph algorithm model weight file to extract the road network recognition results;

[0062] Recognition results are typically represented in raster form, which divides image data into a series of raster units, each representing a specific area. In road network recognition, raster units can be labeled as road network areas or non-road network areas. The results need to be expressed in .shp format according to different raster units, that is, raster data is converted into vector data. This process is achieved by converting raster units into polygon features.

[0063] To construct the centerline of the road network, methods based on computer graphics and road topology are used. By obtaining the edge lines on both sides of the surface, sampling each point, connecting them, and taking the midpoint, the centerline of the road is extracted from the road network surface elements. The extracted road centerline is then refined and smoothed to ensure that the centerline elements conform to the shape of the actual road.

[0064] Furthermore, in S6, road lines are divided using the following method:

[0065] Since urban pipelines are mostly built along roads, we can simulate pipeline data based on road data. The original line vector file generally contains several line segments with large spatial lengths (such as line segments of several kilometers). The basic idea of ​​simulating pipeline generation is to spatially decompose long line segments into multiple smaller line segments with a length range of 0-70 meters (refer to the national standard CJJ61-2017: Technical Specification for Urban Underground Pipeline Detection), and assign them various business attributes to form pipe segments. At the same time, the two ends of each line segment are extracted separately to form pipe points. Thus, we obtain geometric line (pipe segment) and geometric point (pipe point) data that conform to pipeline standards and specifications.

[0066] Furthermore, refer to Figure 5 , Figure 6 , Figure 7 As shown in S7, attributes are assigned to the segmented points and lines to complete the generation of pipeline data, specifically including:

[0067] Attribute values ​​are assigned to the obtained geometric data of pipe segments and pipe points to complete the generation of pipeline data;

[0068] The pipe segment includes five basic attributes: unique pipeline number, starting pipe point number, ending pipe point number, starting point burial depth, and ending point burial depth.

[0069] The pipe point includes five attributes: unique pipe point number, pipe point elevation, accessories, pipe point characteristics, and manhole cover shape.

[0070] Among the above attributes, the unique point number attribute values ​​are sequentially specified as 0, 1, 2, 3, ... during the generation process. The starting point burial depth, ending point burial depth, and pipe point elevation are randomly selected from 0 to 1 meter. Other attribute values ​​are randomly specified or fixed through a value list. Finally, a point table and a line table containing the attributes are obtained.

[0071] In one specific embodiment, refer to Figure 8 As shown: A method for constructing urban pipeline simulation data based on intelligent remote sensing interpretation, including:

[0072] Step 1: Data Collection

[0073] The original road image data was extracted using orthophotos taken by DJI drones. By using the drones to fly at low altitudes, the shooting accuracy was controlled to the centimeter level. During the process of using drones to collect images, in order to increase the image quality, it is necessary to reduce the impact of various negative factors on the acquisition, including environmental factors such as cloud cover and wind speed. Following the process of drone performance and safety checks, planning the aerial photography route, conducting on-site investigations and adjusting the aerial photography route, carrying out aerial photography operations, and saving the shooting data, the captured data was segmented to form a dataset (original images) to be labeled in deep learning.

[0074] Step 2: Image preprocessing:

[0075] Preprocessing of the dataset to be labeled (original images): Photos or videos captured by drones may be affected by weather conditions and lighting, resulting in inaccurate or unsatisfactory color representation. Color correction technology can adjust the color temperature, contrast, and saturation of the image to make it closer to the real scene. Distortion repair includes distortion repair and lens distortion correction. Perspective distortion correction is mainly performed based on the principle of perspective, with the aim of restoring the image to a standard perspective. Lens distortion correction is based on the specific properties of the lens, aiming to make straight lines in the image perpendicular or parallel again. Image qualification verification involves a series of checks and evaluations of the images to ensure that they are suitable for their intended use or application.

[0076] Step 3: Sample preparation:

[0077] The original image data was processed using the professional remote sensing image annotation tool LabelMe to manually annotate the road contours in the remote sensing image. The annotation rules were to draw the road contours one by one from the start and end of the road to ensure accuracy and completeness. The annotated road contours were quality checked to ensure that there were no omissions or errors. The annotated road contour data was organized into a sample set, including positive samples (containing road contours) and negative samples (not containing road contours). The entire sample set was divided into a training set, a validation set, and a test set, with 80% training set, 10% validation set, and 10% test set to ensure a balanced ratio of positive and negative samples and avoid the impact of imbalanced samples on the image training of the model.

[0078] Step 4: Deep Learning

[0079] 1. Optimization of network model structure:

[0080] During the training and testing of the model, it was found that the original U-Net semantic segmentation model suffers from problems such as vanishing gradients and blurred edge features. It also exhibits unremarkable performance in metrics such as mIOU, mPA, and F1 score, and the overall performance of the model is not very stable. Therefore, the following methods were used to address these issues:

[0081] (1) The ResNet residual module was introduced.

[0082] Since the upsampling operation of the decoder leads to an increase in the feature map size, it may lose some detailed information or cause a decrease in resolution. This may affect the model's ability to capture subtle structural and edge information, thus affecting the accuracy of the segmentation results. To address this issue, skip connections or more complex structures are introduced into the decoder to preserve the richness and detail of the features. ResNet34, as a representative network for solving the feature gradient problem, can be added to the decoder to solve the detail information problem. The ResNet34 residual module consists of two convolutional layers (constitutional layers) and uses a pair of 3x3 convolutional kernels. By introducing the ResNet residual module, the fineness and accuracy of feature representation are improved, and subtle features and structural information in the image can be captured better.

[0083] (2) Introducing wavelet transform layer and attention mechanism layer

[0084] Because drones are affected by lighting conditions, shooting angle, flight attitude, or speed during the acquisition of raw road images, problems such as missing local features in the acquired images can occur, leading to biases in the model's understanding of the images. Wavelet transform, as a powerful signal processing tool, utilizes wavelet functions to decompose and extract image information. A wavelet function is a finite-length waveform function with a unique ability to perform fine-grained analysis of information at different scales and locations, offering significant advantages, particularly for image processing of non-stationary remote sensing information. Discrete wavelet transform can be expressed as a wavelet series expansion using basis functions:

[0085]

[0086] Multi-scale analysis using wavelet transform can effectively extract multi-scale features of images, capture image details and overall features, thereby enhancing the model's ability to understand images.

[0087] Furthermore, to further enhance the model's focus on road areas, background information in the image is ignored. An attention mechanism layer is introduced on top of the existing improved model. Its working mechanism involves determining the input feature map size (including the number of channels (C), height (H), and width (W)), then processing the channel features of the input feature map through convolutional layers within the module. Attention weights are applied to each channel feature to obtain the channel feature results. By regulating the attention, the channel features that need to be focused on are determined, thereby increasing the model's generalization ability and task performance, while reducing redundant information and making the model's recognition and detection capabilities more accurate.

[0088] 2. Model Training and Prediction

[0089] The labeled data was strictly divided according to the Pascal VOC data format, with the original image size being 512 pixels * 512 pixels. The ratio of test set, validation set, and test set was 8:1:1. Following the improved model network structure, the Epoch (number of training iterations) was set to 100, and the Batch size (number of samples captured per batch) was set to 4. The model was trained under these parameters. Furthermore, to better qualitatively evaluate the optimized recognition and detection performance of the model, mPA (pixel accuracy ratio for each category), mIOU (mean Intersection over Union), and F1 score (harmonic mean of Recall and Precision) were used. In this experiment, mPA represents the overall accuracy of road network pixels, mIOU represents the average value of the intersection-union ratio of road labels and recognition and detection results, Recall represents the proportion of correctly predicted road contour pixels, and Precision represents the proportion of predicted pixels that match the actual road pixels.

[0090] The calculation methods for the above indicators are as follows:

[0091]

[0092]

[0093]

[0094]

[0095]

[0096]

[0097]

[0098] In the formula, TP and TN represent pixel predictions, T is True indicating correct classification, FP represents misclassified samples, FN represents missed samples, and F is Flax indicating misclassification.

[0099] The evaluation results are shown in Table 1 below:

[0100] Table 1. Performance Comparison of U-Net Models (%)

[0101] Model mPA mIOU Recall Precision F1 U-Net (Original) 82.36 83.41 84.65 84.21 84.43 U-Net (Optimized) 90.60 89.64 93.30 93.91 93.60

[0102] Step 5: Generate the pipeline network:

[0103] The system uses shapefile (SHP) to generate pipeline network data. The decoded raster data is used as input to a Python script. The script saves the predicted input image information (parameters), including coordinate and projection information. The raster data is converted to TIF standard format, and the input image information is fed into the converted raster data. Based on the attribute values ​​or other features of the raster pixels, the raster data is converted into corresponding vector features, generating the recognition result shapefile data. The shapefile data is then spatially decomposed into long line segments, breaking them down into multiple smaller line segments with a length range of 0-70 meters (referencing national standard CJJ61-2017: Technical Specification for Urban Underground Pipeline Detection). These smaller segments are assigned various business attributes to form pipe segments. Simultaneously, the endpoints of each line segment are extracted separately to generate pipe points. This yields geometric lines (pipe segments) and geometric points (pipe points) data that conform to pipeline standards and specifications, which are then applied in practice through the system.

[0104] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for system or system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and relevant parts can be referred to the descriptions in the method embodiments. The systems and system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0105] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing urban pipeline simulation data based on intelligent remote sensing interpretation, characterized in that, The method comprises the following steps: S1, acquiring original urban orthophoto; S2, preprocessing the original orthophoto, specifically: Image stitching and orthorectification are performed on the collected original urban orthophoto data, the aerial photographs taken by the unmanned aerial vehicle during takeoff and landing are manually deleted, the photographs with serious exposure and focusing failure are removed, the synthesis of the orthophoto map is completed according to the original image import, aerial triangulation, model reconstruction and result input process, and the preprocessed orthophoto is obtained; S3, using an image labeling tool to label the preprocessed orthophoto according to the road contour, and pre-preparing a sample training set and a verification set; S4, optimizing the encoding and decoding structure in the original U-Net model, establishing a city pipeline simulation data construction model, and completing the training of the model, specifically: Optimizing the encoding and decoding structure in the original U-Net model: introducing the original U-Net network into the wavelet transform module WaveletTransform; constructing a channel attention module and fusing the channel attention module into the skip connection of the original network; introducing a Res Net residual module to obtain the optimized model network structure and establish the city pipeline simulation data construction model; In the model training process: the labeled orthophoto data is divided according to the data format of PascalVOC, the original size of the image is 512 pixels*512 pixels, the training times Epoch are set to 100 according to the optimized model network structure, and the number of samples grabbed at a time Batch size is set to 4, and the training of the model is completed; S5, using the trained model to extract road lines: inputting the image data of the region to be recognized into the trained city pipeline simulation data construction model, extracting the recognition result of the road network, converting the recognized raster result into.shp format, and extracting the center line of the road network according to the recognition result; S6, segmenting the road lines, specifically: Based on the road data simulation pipeline data, the long line segment is decomposed and split into multiple small line segments with a length of 0-70 meters, and various business attributes are assigned to the line segments to form pipe segments, and the two end points of each line segment are extracted to generate pipe points, thereby obtaining geometric line and point data meeting the pipeline standard specification; S7, assigning attributes to the segmented points and lines to complete the generation of pipeline data, specifically including: Attribute assignment is performed on the obtained pipe segment and pipe point geometric data to complete the generation of pipeline data; The pipe segment includes five basic attributes: unique pipeline number, starting pipe point number, ending pipe point number, starting point burial depth, and ending point burial depth; The pipe point includes five attributes: unique pipe point number, pipe point elevation, accessory, pipe point feature, and manhole cover shape; In the above attributes, the unique point number attribute value is sequentially specified as 0, 1, 2, 3,..., the starting point burial depth, the ending point burial depth and the pipe point elevation are randomly taken in the range of 0-1 meters, and the other attribute values are randomly specified or fixed through a value list, and finally a point table and a line table containing attributes are obtained.

2. The urban pipeline simulation data construction method based on intelligent remote sensing interpretation according to claim 1, characterized in that, In S1, the original urban orthographic image is obtained, and the original urban orthographic image data is collected by using a UAV.

3. The urban pipeline simulation data construction method based on intelligent remote sensing interpretation according to claim 1, characterized in that, In S3, the preprocessed orthographic image is labeled according to the road contour by using an image labeling tool, and is precast into a sample training set and a verification set: According to the geometric shape and spectral feature requirements of the ground object contour, the target sample is extracted, the Labelme data labeling tool is used for semantic segmentation data set labeling of the road contour image, and the sample data with annotations is divided into a training set and a verification set at a ratio of 9:1.

Images