A visual real-time imaging method and system based on smart city CIM

Abstract

The application discloses a kind of visualization real-time imaging method and system based on smart city CIM, belong to image processing technical field, including obtaining N initial images that N camera real-time acquisition;According to whether the fire-fighting equipment is recorded in initial image, N initial images are divided into reference image set and background image set, and from reference image set, the equipment image of fire-fighting equipment under different angles is obtained to obtain equipment image set;According to equipment image set, complete equipment image of fire-fighting equipment under three-dimensional space is generated, and according to background image set, the three-dimensional space background where fire-fighting equipment is located is generated;Based on smart city CIM, the city three-dimensional model of target city under three-dimensional space is constructed, three-dimensional space background is mapped in city three-dimensional model, and complete equipment image is mapped in the three-dimensional space background of city three-dimensional model;The three-dimensional shape and structure of fire-fighting equipment are accurately reflected, and strong support is provided for the fine management of fire-fighting equipment.

Description

Technical Field

[0001] This invention belongs to the field of image processing technology, and specifically relates to a visualization real-time imaging method and system based on smart city CIM. Background Technology

[0002] With the acceleration of urbanization, the complexity and management difficulty of cities are constantly increasing. Traditional urban management models face numerous challenges in addressing urban public safety and resource management. For example, in the field of fire protection, the deployment and management of fire-fighting equipment is crucial, but traditional monitoring methods often only provide a two-dimensional, localized perspective, making it difficult to comprehensively and in real-time grasp the status of fire-fighting equipment and its specific location and surrounding environment in the urban space. Furthermore, there are shortcomings in the visual management of the overall urban space; there is a lack of an efficient visualization method that can closely integrate urban infrastructure with three-dimensional space.

[0003] Against this backdrop, visualization technology based on City Information Modeling (CIM) for smart cities has emerged. It can combine various urban information with a three-dimensional spatial model to provide more comprehensive and intuitive decision support for urban management and emergency response. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a real-time visualization imaging method and system based on Smart City Information Modeling (CIM). This application provides a real-time visualization imaging method based on Smart City Information Modeling (CIM), in which fire-fighting equipment and N cameras are deployed in the public space of the target city, where N is a positive integer greater than or equal to 1. The N cameras are deployed at different locations in the public space, including:

[0005] Acquire N initial images captured in real time by N cameras;

[0006] Based on whether the initial image records fire-fighting equipment, the N initial images are divided into a reference image set and a background image set. The equipment images of the fire-fighting equipment at different angles are extracted from the reference image set to obtain the equipment image set. The reference image set includes M reference images that record fire-fighting equipment, where M is a positive integer greater than or equal to 1. The background image set includes (NM) background images that do not record fire-fighting equipment.

[0007] Generate a complete image of the fire-fighting equipment in three-dimensional space based on the equipment image set, and generate a three-dimensional background of the fire-fighting equipment based on the background image set;

[0008] Based on the Smart City Information Modeling (CIM), a 3D model of the target city is constructed in three-dimensional space. The 3D spatial background is mapped onto the 3D urban model, and complete equipment images are mapped onto the 3D spatial background of the 3D urban model.

[0009] Furthermore, based on whether fire-fighting equipment was recorded in the initial images, the N initial images are divided into a reference image set and a background image set, including:

[0010] Scale the N initial images to fit the size of the device's recognition model, and then use the formula... Normalize the pixel values ​​of the N initial images. These are the pixel values ​​of the initial image. It is the mean of N initial images. It is the standard deviation of N initial images;

[0011] N initial images are sequentially input into the device recognition model to obtain the corresponding initial images, reference coordinates, and coordinate confidence scores output by the device recognition model. The reference coordinates are used to indicate the position of the fire-fighting equipment in the initial images, and the coordinate confidence scores are used to indicate the accuracy of the device recognition model in detecting the reference coordinates.

[0012] The coordinate confidence score of each initial image is compared with a confidence threshold.

[0013] Initial images with coordinate confidence scores greater than or equal to a confidence threshold are identified as reference images and added to the reference image set;

[0014] Initial images with coordinate confidence scores less than a confidence threshold are identified as background images and added to the background image set.

[0015] Furthermore, the device recognition model is trained through the following steps:

[0016] An initial recognition model, a first training set, and a second training set are constructed. The first training set includes multiple first training images with the image coordinates of fire-fighting equipment labeled, and the second training set includes multiple second training images without fire-fighting equipment recorded.

[0017] The first and second training sets are sequentially input into the initial recognition model to obtain the corresponding training images, training coordinates, and training confidence scores output by the initial recognition model.

[0018] use As the loss function of the initial recognition model, Used for balance and The weight, C represents the number of categories, which is 2. This is used to indicate that the i-th component containing fire-fighting equipment in the training image is 1, while the remaining components are 0. It is the probability of the i-th category predicted by the initial recognition model. , , It is the difference between the training coordinates corresponding to the i-th training image output by the initial recognition model and the real image coordinates corresponding to the i-th training image;

[0019] Through formula Backpropagation calculates the gradient of the loss function with respect to the model parameters. yes The gradient of the parameters of the initial recognition model, yes The gradient of the parameters of the initial recognition model;

[0020] Use optimizer The parameters of the initial recognition model are updated according to the gradient, where α is the learning rate used to control the step size of the parameter update.

[0021] Repeat the above steps until the initial recognition model completes the required number of training rounds on the first and second training sets to obtain the device recognition model.

[0022] Furthermore, equipment images of the fire-fighting equipment at different angles are extracted from the reference image set to obtain an equipment image set, including:

[0023] Obtain the reference coordinates corresponding to each background image in the background image set;

[0024] The device image set is obtained by extracting the device image from each background image according to the reference coordinates corresponding to each background image.

[0025] Furthermore, based on the equipment image set, a complete three-dimensional image of the fire-fighting equipment is generated, including:

[0026] Feature points are extracted from each reference image in the device image set;

[0027] pass By matching feature points from different reference images within the device image set and calculating the similarity metric between these feature points, corresponding feature point pairs between different reference images are determined, thereby establishing the spatial relationships and relative positions between them. and These are feature points in two reference images, where m is the dimension of the feature point. and Feature points and The coordinates in the k-th dimension;

[0028] Based on feature point pairs, through The point cloud data of fire-fighting equipment in three-dimensional space is calculated. The point cloud data includes the three-dimensional coordinate information of the fire-fighting equipment, which reflects the three-dimensional shape and structure of the equipment. and These are the rotation matrices for the two reference images, respectively;

[0029] pass Model fitting is performed on the point cloud data to fit a complete 3D model, resulting in a complete image of the fire-fighting equipment in 3D space. For point cloud data, The weighting coefficients are used to represent the contribution of each point.

[0030] Furthermore, based on the Smart City CIM, a three-dimensional urban model of the target city is constructed in three-dimensional space, including:

[0031] Collect geographic information data for the target city, including topographic elevation data, building outline data, and road network data;

[0032] The processed data is converted into a 3D model. A terrain mesh is generated using terrain elevation data. A building model is constructed using building outline data and height information. A road model is generated by combining road network data. Finally, the data are integrated to obtain a 3D urban model of the target city.

[0033] Mapping a 3D spatial background onto a 3D city model includes:

[0034] Boolean operations are used to merge the geometry of the 3D spatial background with that of the 3D city model, ensuring a seamless spatial connection between the two.

[0035] Mapping the complete equipment image onto the 3D spatial background of the city's 3D model includes:

[0036] Obtain the coordinates of the complete device image in 3D space , The two-dimensional coordinates of the fire-fighting equipment K represents the depth information of the fire-fighting equipment, and K represents the intrinsic parameters of the camera.

[0037] The calculated three-dimensional position Mapped onto the city's 3D model;

[0038] A texture fusion algorithm is used to smoothly blend the texture of the fire-fighting equipment with the three-dimensional spatial background.

[0039] This application provides a real-time visualization imaging system based on Smart City Information Modeling (CIM). The target city's public spaces are equipped with fire-fighting equipment and N cameras, where N is a positive integer greater than or equal to 1. The N cameras are deployed at different locations within the public spaces. The system includes:

[0040] The acquisition module is used to acquire N initial images captured in real time by N cameras;

[0041] The partitioning module is used to divide N initial images into a reference image set and a background image set based on whether the fire-fighting equipment is recorded in the initial image. The module also extracts equipment images of the fire-fighting equipment from different angles from the reference image set to obtain the equipment image set. The reference image set includes M reference images that record the fire-fighting equipment, where M is a positive integer greater than or equal to 1. The background image set includes (NM) background images that do not record the fire-fighting equipment.

[0042] The generation module is used to generate a complete image of the fire-fighting equipment in three-dimensional space based on the equipment image set, and to generate the three-dimensional spatial background of the fire-fighting equipment based on the background image set.

[0043] The building module is used to construct a 3D model of the target city in a 3D space based on Smart City CIM, map the 3D spatial background onto the 3D urban model, and map complete equipment images onto the 3D spatial background of the urban 3D model.

[0044] The beneficial effects of this invention compared with the prior art are: (1) By deploying multiple cameras in urban public spaces to collect images in real time and dividing them into reference image sets and background image sets, the status of fire-fighting equipment can be comprehensively monitored from different angles, realizing real-time visual management of urban fire-fighting equipment and timely detection of equipment abnormalities or potential risks; (2) By using equipment recognition models to accurately identify the position of fire-fighting equipment in images, and by using techniques such as feature point extraction, matching, and point cloud data calculation, a complete equipment image of fire-fighting equipment in three-dimensional space is generated, which accurately reflects the three-dimensional shape and structure of fire-fighting equipment and provides strong support for the refined management of fire-fighting equipment; (3) Collect geographic information data of the target city and convert it into a three-dimensional model. Seamlessly connect the three-dimensional spatial background with the city's three-dimensional model and map the complete equipment image into the three-dimensional spatial background. This achieves efficient integration of the overall urban space and fire equipment information, providing city managers with an intuitive and comprehensive three-dimensional visualization platform, which helps to optimize the urban spatial layout and resource allocation. (4) In the event of an emergency such as a fire, the system can quickly and accurately obtain the location and status information of fire equipment, as well as the three-dimensional spatial environment around it, providing emergency rescue personnel with more accurate navigation and decision-making basis, thereby improving the emergency response speed and rescue efficiency and reducing disaster losses. Attached Figure Description

[0045] Figure 1 This is a flowchart of a real-time visualization imaging method based on smart city CIM according to the present invention;

[0046] Figure 2 This is a schematic diagram of the structure of a real-time visualization imaging system based on smart city CIM according to the present invention. Detailed Implementation

[0047] Example 1: As Figure 1 As shown, a real-time visualization imaging method based on Smart City Information Modeling (CIM) is proposed. Firefighting equipment and N cameras are deployed in the public space of the target city, where N is a positive integer greater than or equal to 1. The N cameras are deployed at different locations in the public space, including:

[0048] Step S101: Acquire N initial images captured in real time by N cameras;

[0049] Step S102: Divide the N initial images into a reference image set and a background image set according to whether the fire-fighting equipment is recorded in the initial image. Extract the equipment images of the fire-fighting equipment from different angles from the reference image set to obtain the equipment image set. The reference image set includes M reference images that record the fire-fighting equipment, where M is a positive integer greater than or equal to 1. The background image set includes (NM) background images that do not record the fire-fighting equipment.

[0050] Step S103: Generate a complete image of the fire-fighting equipment in three-dimensional space based on the equipment image set, and generate a three-dimensional spatial background of the fire-fighting equipment based on the background image set;

[0051] Step S104: Construct a 3D model of the target city in 3D space based on Smart City CIM, map the 3D spatial background onto the 3D urban model, and map the complete equipment image onto the 3D spatial background of the urban model.

[0052] In this embodiment, N initial images are divided into a reference image set and a background image set based on whether fire-fighting equipment is recorded in the initial image. This includes: scaling the N initial images to fit the size of the equipment recognition model, and then using a formula... Normalize the pixel values ​​of the N initial images. These are the pixel values ​​of the initial image. It is the mean of N initial images. The standard deviation of N initial images is given. These N initial images are sequentially input into the device recognition model to obtain the corresponding initial images, reference coordinates, and coordinate confidence scores output by the model. The reference coordinates indicate the position of the fire-fighting equipment in the initial images, and the coordinate confidence scores indicate the accuracy of the device recognition model in detecting the reference coordinates. The coordinate confidence score of each initial image is compared with a confidence threshold. Initial images with a coordinate confidence score greater than or equal to the confidence threshold are identified as reference images and added to the reference image set. Initial images with a coordinate confidence score less than the confidence threshold are identified as background images and added to the background image set.

[0053] In this embodiment, the device recognition model is trained through the following steps: An initial recognition model, a first training set, and a second training set are constructed. The first training set includes multiple first training images with image coordinates of fire-fighting equipment labeled, and the second training set includes multiple second training images without fire-fighting equipment recorded. The first and second training sets are sequentially input into the initial recognition model to obtain the corresponding training images, training coordinates, and training confidence scores output by the initial recognition model. As the loss function of the initial recognition model, Used for balance and The weight, C represents the number of categories, which is 2. This is used to indicate that the i-th component containing fire-fighting equipment in the training image is 1, while the remaining components are 0. It is the probability of the i-th category predicted by the initial recognition model. , , It is the difference between the training coordinates of the i-th training image output by the initial recognition model and the real image coordinates of the i-th training image; expressed by the formula... Backpropagation calculates the gradient of the loss function with respect to the model parameters. yes The gradient of the parameters of the initial recognition model, yes Gradients of the parameters of the initial identification model; using the optimizer The parameters of the initial recognition model are updated according to the gradient, where α is the learning rate used to control the step size of the parameter update; the above steps are repeated until the initial recognition model completes the number of training rounds on the first and second training sets to obtain the device recognition model.

[0054] In this embodiment, extracting equipment images of fire-fighting equipment from a reference image set at different angles to obtain an equipment image set includes: obtaining the reference coordinates corresponding to each background image in the background image set; and extracting equipment images from each background image according to the reference coordinates corresponding to each background image to obtain the equipment image set.

[0055] In this embodiment, generating a complete three-dimensional image of the fire-fighting equipment based on the equipment image set includes: extracting feature points from each reference image in the equipment image set; and through... By matching feature points from different reference images within the device image set and calculating the similarity metric between these feature points, corresponding feature point pairs between different reference images are determined, thereby establishing the spatial relationships and relative positions between them. and These are feature points in two reference images, where m is the dimension of the feature point. and Feature points and The coordinates in the k-th dimension; based on the feature point pairs, through The point cloud data of fire-fighting equipment in three-dimensional space is calculated. The point cloud data includes the three-dimensional coordinate information of the fire-fighting equipment, which reflects the three-dimensional shape and structure of the equipment. and These are the rotation matrices for the two reference images; through Model fitting is performed on the point cloud data to fit a complete 3D model, resulting in a complete image of the fire-fighting equipment in 3D space. For point cloud data, The weighting coefficients are used to represent the contribution of each point.

[0056] Optionally, but not limited to, the three-dimensional spatial background of the fire-fighting equipment can be generated based on the background image set in the following ways:

[0057] 1. Feature extraction for each background image in the background image set: Feature extraction is performed on each background image to extract feature information including texture, edge, color, etc., for subsequent 3D reconstruction and background synthesis.

[0058] 2. Depth estimation of the background image is performed using the following formula: Assuming the background image is I(x,y), the depth value D(x,y) of each pixel is estimated using a monocular depth estimation network or stereo vision method:

[0059] Where f is the depth estimation function and θ are the model parameters. Depth estimation can be achieved using pre-trained deep learning models, such as monocular depth estimation networks or stereo matching networks.

[0060] 3. Generate a 3D point cloud based on the depth estimation results:

[0061] Using the depth value D(x,y) and camera intrinsic parameters K (including focal length and optical center coordinates), each pixel (x,y) is converted into a point P(x,y,z) in three-dimensional space:

[0062] in, and These are the focal lengths of the camera in the x and y directions, respectively. and Here are the coordinates of the optical center. Using the above formula, two-dimensional pixels are mapped into three-dimensional space, generating three-dimensional point cloud data of the background.

[0063] 4. Filter and denoise the generated 3D point cloud:

[0064] Voxel grid filtering or statistical filtering methods are used to filter and denoise point clouds, removing noisy points and outliers to improve the quality of point clouds.

[0065] Where P represents the original point cloud. The filtered point cloud is defined by leaf_size, mean_k, std_dev_mul_thresh, and std_dev_mul_thresh.

[0066] 5. Model the filtered point cloud using surface fitting or voxelization methods:

[0067] The filtered point cloud data is then subjected to surface fitting or voxelization to generate a 3D model of the background. For example, polynomial surface fitting or voxelization methods can be used.

[0068] in,

[0069] For the fitted surface model, For the voxelized model, degree is the polynomial order of the surface fitting, and voxel_size is the size of the voxel.

[0070] 6. Merge the generated 3D background model with the city's 3D model:

[0071] The generated 3D background model or Mapping this onto the city's 3D model allows for seamless integration with other parts of the model, creating a complete 3D background environment.

[0072] in, The final merged 3D city model This is the original 3D model of the city.

[0073] Through the above steps, a three-dimensional spatial background of the fire-fighting equipment is generated and integrated with the city's three-dimensional model to form a complete three-dimensional visualization environment.

[0074] In this embodiment, the construction of a three-dimensional urban model of the target city in three-dimensional space based on smart city CIM includes: collecting geographic information data of the target city, including terrain elevation data, building outline data, and road network data; converting the processed data into a three-dimensional model, generating a terrain grid through terrain elevation data, constructing a building model using building outline data and height information, generating a road model by combining road network data, and finally integrating to obtain the three-dimensional urban model of the target city.

[0075] In this embodiment, mapping the three-dimensional spatial background onto the city's three-dimensional model includes: merging the three-dimensional spatial background with the geometry of the city's three-dimensional model through Boolean operations to ensure seamless spatial integration between the two.

[0076] In this embodiment, mapping the complete device image onto the three-dimensional spatial background of the city's three-dimensional model includes: obtaining the coordinates of the complete device image in three-dimensional space. , The two-dimensional coordinates of the fire-fighting equipment The depth information of the fire-fighting equipment is given by K, where K is the intrinsic parameter of the camera; the calculated three-dimensional position is then used. Mapped into a 3D city model; texture fusion algorithms are used to smoothly blend the textures of fire-fighting equipment with the 3D spatial background.

[0077] Example 2: Figure 2 As shown, a real-time visualization imaging system based on Smart City Information Modeling (CIM) is described. Firefighting equipment and N cameras are deployed in the public spaces of the target city, where N is a positive integer greater than or equal to 1. The N cameras are deployed at different locations in the public spaces, including:

[0078] The acquisition module is used to acquire N initial images captured in real time by N cameras;

[0079] The partitioning module is used to divide N initial images into a reference image set and a background image set based on whether the fire-fighting equipment is recorded in the initial image. The module also extracts equipment images of the fire-fighting equipment from different angles from the reference image set to obtain the equipment image set. The reference image set includes M reference images that record the fire-fighting equipment, where M is a positive integer greater than or equal to 1. The background image set includes (NM) background images that do not record the fire-fighting equipment.

[0080] The generation module is used to generate a complete image of the fire-fighting equipment in three-dimensional space based on the equipment image set, and to generate the three-dimensional spatial background of the fire-fighting equipment based on the background image set.

[0081] The building module is used to construct a 3D model of the target city in a 3D space based on Smart City CIM, map the 3D spatial background onto the 3D urban model, and map complete equipment images onto the 3D spatial background of the urban 3D model.

[0082] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example.

Claims

1. A real-time visualization imaging method based on Smart City CIM, characterized in that, Firefighting equipment and N cameras, where N is a positive integer greater than or equal to 1, are deployed in public spaces of the target city. The N cameras are deployed at different locations within the public spaces. The method includes: Obtain N initial images captured in real time by the N cameras; Based on whether the fire-fighting equipment is recorded in the initial image, the N initial images are divided into a reference image set and a background image set. The equipment images of the fire-fighting equipment under different angles are extracted from the reference image set to obtain the equipment image set. The reference image set includes M reference images that record the fire-fighting equipment, where M is a positive integer greater than or equal to 1. The background image set includes (NM) background images that do not record the fire-fighting equipment. A complete image of the fire-fighting equipment in three-dimensional space is generated based on the equipment image set, and a three-dimensional spatial background of the fire-fighting equipment is generated based on the background image set. Based on the Smart City Information Modeling (CIM), a 3D model of the target city is constructed in 3D space. The 3D spatial background is mapped into the 3D urban model, and the complete equipment image is mapped into the 3D spatial background of the 3D urban model.

2. The real-time visualization imaging method based on smart city CIM according to claim 1, characterized in that: The step of dividing N initial images into a reference image set and a background image set based on whether the fire-fighting equipment is recorded in the initial image includes: The N initial images are scaled to fit the size of the device's recognition model, and then... (The formula is missing from the original text.) Normalize the pixel values ​​of the N initial images. These are the pixel values ​​of the initial image. It is the mean of the N initial images. It is the standard deviation of the N initial images; The N initial images are sequentially input into the device recognition model to obtain the corresponding initial images, reference coordinates, and coordinate confidence scores output by the device recognition model. The reference coordinates are used to indicate the position of the fire-fighting equipment in the initial images, and the coordinate confidence scores are used to indicate the accuracy of the device recognition model in detecting the reference coordinates. The confidence level of the coordinates for each initial image is compared with a confidence threshold. The initial images whose coordinate confidence is greater than or equal to the confidence threshold are identified as the reference images and added to the reference image set; The initial images whose coordinate confidence is less than the confidence threshold are identified as the background images and added to the background image set.

3. The real-time visualization imaging method based on smart city CIM according to claim 2, characterized in that, The device recognition model is trained using the following steps: An initial recognition model, a first training set, and a second training set are constructed. The first training set includes multiple first training images with the image coordinates of fire-fighting equipment labeled, and the second training set includes multiple second training images without fire-fighting equipment recorded. The first training set and the second training set are sequentially input into the initial recognition model to obtain the corresponding training images, training coordinates and training confidence scores output by the initial recognition model. use As the loss function of the initial recognition model, Used for balance and The weight, C represents the number of categories, which is 2. The r-th component used to indicate that fire-fighting equipment is included in the training image is 1, while the remaining components are 0. It is the probability of the i-th category predicted by the initial recognition model. , , It is the difference between the training coordinates corresponding to the kth training image output by the initial recognition model and the real image coordinates corresponding to the kth training image; Through formula Backpropagation calculates the gradient of the loss function with respect to the model parameters. yes The gradient of the parameters of the initial recognition model, yes The gradient of the parameters of the initial recognition model; Use optimizer The parameters of the initial recognition model are updated according to the gradient, where α is the learning rate used to control the parameter update step size; Repeat the above steps until the initial recognition model completes the required number of training rounds on the first training set and the second training set to obtain the device recognition model.

4. The real-time visualization imaging method based on smart city CIM according to claim 2, characterized in that, The step of extracting equipment images of the fire-fighting equipment from the reference image set at different angles to obtain an equipment image set includes: Obtain the reference coordinates corresponding to each reference image in the reference image set; The device image set is obtained by extracting the device image from each of the reference images according to the reference coordinates corresponding to each of the reference images.

5. The real-time visualization imaging method based on smart city CIM according to claim 1, characterized in that, The step of generating a complete three-dimensional image of the fire-fighting equipment based on the equipment image set includes: Feature points are extracted from each of the reference images in the device image set; pass Feature points from different reference images in the device image set are matched, and a similarity metric between feature points is calculated to determine corresponding feature point pairs between different reference images, thereby determining the spatial relationships and relative positions between the different reference images. and These are feature points in two reference images, where m is the dimension of the feature point. and Feature points and The coordinates in the k-th dimension; Based on the feature point pair, through The point cloud data of the fire-fighting equipment in three-dimensional space is calculated. The point cloud data includes the three-dimensional coordinate information of the fire-fighting equipment, which reflects the three-dimensional shape and structure of the fire-fighting equipment. and These are the rotation matrices for the two reference images, respectively; pass The point cloud data is fitted with a model to form a complete 3D model, resulting in a complete image of the fire-fighting equipment in 3D space. For point cloud data, The weighting coefficients are used to represent the contribution of each point.

6. The real-time visualization imaging method based on smart city CIM according to claim 1, characterized in that: The construction of a 3D city model of the target city in 3D space based on Smart City CIM includes: Collect geographic information data of the target city, including topographic elevation data, building outline data, and road network data; The processed data is converted into a 3D model. A terrain mesh is generated using terrain elevation data. A building model is constructed using building outline data and height information. A road model is generated by combining road network data. Finally, the three-dimensional model of the target city is obtained by integrating the data. The step of mapping the three-dimensional spatial background into the city's three-dimensional model includes: Boolean operations are used to merge the geometry of the three-dimensional spatial background with that of the three-dimensional city model, ensuring a seamless spatial connection between the two. The step of mapping the complete device image into the three-dimensional spatial background of the city's three-dimensional model includes: Obtain the coordinates of the complete device image in three-dimensional space. , Let the coordinates of the fire-fighting equipment be two-dimensional. Here, K represents the depth information of the fire-fighting equipment, and K represents the intrinsic parameters of the camera. The calculated three-dimensional position Mapped onto the city's 3D model; A texture fusion algorithm is used to smoothly blend the texture of the fire-fighting equipment with the three-dimensional spatial background.

7. A real-time visualization imaging system based on Smart City CIM, characterized in that, Firefighting equipment and N cameras, where N is a positive integer greater than or equal to 1, are deployed in public spaces of the target city. The N cameras are deployed at different locations within the public spaces. The system includes: The acquisition module is used to acquire N initial images captured in real time by the N cameras; The segmentation module is used to divide N initial images into a reference image set and a background image set according to whether the fire-fighting equipment is recorded in the initial image, and to extract equipment images of the fire-fighting equipment from the reference image set at different angles to obtain an equipment image set. The reference image set includes M reference images that record the fire-fighting equipment, where M is a positive integer greater than or equal to 1, and the background image set includes (NM) background images that do not record the fire-fighting equipment. The generation module is used to generate a complete image of the fire-fighting equipment in three-dimensional space based on the equipment image set, and to generate a three-dimensional spatial background of the fire-fighting equipment based on the background image set. The construction module is used to construct a three-dimensional urban model of the target city in three-dimensional space based on Smart City CIM, map the three-dimensional spatial background into the urban three-dimensional model, and map the complete equipment image into the three-dimensional spatial background of the urban three-dimensional model.

Images