Foundation pit scene automatic reconstruction method based on machine vision
Through the automated reconstruction method of foundation pit scenes based on machine vision, combined with high-level video processing and low-level laser scanning monitoring, high-precision automated monitoring and visualization of deep foundation pit projects are achieved, and the problem of insufficient construction process monitoring in the existing technology is solved. , improve the scientific nature of safety warning and emergency guidance.
Patent Information
- Application Number
- CN202411376177.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-30
- Publication Date
- 2024-11-29
AI Technical Summary
It is difficult for the existing technology to achieve high-precision automated monitoring and visualization of deep foundation pit construction processes, resulting in inaccurate and scientific safety warning and emergency guidance.
The automation reconstruction method of foundation pit scenes based on machine vision is adopted, and the three-dimensional surface model reconstruction and real-time update of the overall foundation pit scene is achieved through high-level behavior scene reconstruction and low-level detail displacement reconstruction, combined with multi-source point cloud data fusion. Specific steps include high-level reconstruction using a camera to shoot video, low-level monitoring by three-dimensional laser scanners, and data processing and model optimization using computer vision and deep learning technologies.
It realizes high-precision visualization of the entire process of foundation pit and deformation, improves the accuracy and scientificity of safety warning and emergency guidance, and reduces construction risks and costs.
Smart Images

Figure CN119048705A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of engineering information technology, and in particular to a method for automatic reconstruction of foundation pit scenes based on machine vision. Background Art
[0002] With the rapid development of urban construction, the increasing number of urban subways and super high-rise building projects, deep foundation pit projects have developed rapidly, and with them come the risk and safety issues of deep foundation pit projects. Deep foundation pit projects are relatively risky projects, so more stringent requirements are placed on the design, construction and monitoring process of deep foundation pit projects.
[0003] In order to improve the quality of engineering construction and the degree of automation of construction monitoring, machine vision technology is suitable for replacing human eyes to monitor and judge in dangerous working environments where manual work is required or where human eyes cannot meet the requirements. Machine vision has unique advantages in spatial measurement, which are mainly reflected in the fact that it does not need to contact the monitored object during measurement, can work stably and continuously for a long time, and has high measurement accuracy. In recent years, machine vision technology has begun to be applied to deep foundation pit monitoring. It can monitor continuously with high accuracy and has a monitoring function on the engineering site. It can effectively judge the safety status during the construction of deep foundation pits, and then guide the construction of deep foundation pits. Therefore, how to give full play to the advantages of machine vision measurement technology in deep foundation pit monitoring, and combine computer and artificial intelligence technology to process images and videos to realize the automatic reconstruction of foundation pit construction scenes is a new module in the monitoring and management of foundation pit projects. It uses intelligent monitoring equipment to continuously monitor the construction progress of the project, reconstruct the digital model in real time, and integrate the whole process of construction progress, deformation monitoring, dynamic feedback, etc. into the digital model, and realize dynamic update and reconstruction in the foundation pit scene to further improve the visualization level of foundation pit projects. It is a technical problem that needs to be solved urgently. Summary of the invention
[0004] In order to address the shortcomings of the prior art, the present invention provides an automatic reconstruction method for foundation pit scenes based on machine vision, which can visualize the progress and deformation of the entire foundation pit process with high precision, make safety warnings and emergency guidance more accurate and scientific, and have high application value.
[0005] The embodiment of the present invention provides the following solution: An embodiment of the present invention provides a method for automatic reconstruction of a foundation pit scene based on machine vision, the method comprising: Step 1: Reconstruct high-level behavior scenarios; Step 2: Low-level detail displacement reconstruction; Step 3: Fusion of multi-source point cloud data and reconstruction of the 3D surface model of the entire foundation pit scene; Step 4: Update in real time.
[0006] In an optional embodiment, step one includes the following process: S1.1. Use a camera to shoot the foundation pit construction video; S1.2. Process the foundation pit construction video based on the contents of different excavation stages; S1.3, using computer vision algorithms to convert video frame images into sparse scene point cloud data; S1.4. Convert the sparse scene point cloud data into dense point cloud data of high-rise foundation pit scene.
[0007] In an optional embodiment, the step S1.2 described in which the foundation pit construction video is processed based on the content of different excavation stages includes video segmentation, image denoising and image enhancement processing.
[0008] In an optional embodiment, step S1.3 uses motion recovery structure technology to infer the three-dimensional structure of the foundation pit scene and the camera position from the frame images of the video, establish a geometric relationship between the foundation pit scene and the camera, and obtain the camera position and sparse scene point cloud data.
[0009] In an optional embodiment, step S1.4 uses multi-view stereo vision technology to convert sparse scene point cloud data into dense point cloud data of a high-rise foundation pit scene.
[0010] In an optional embodiment, step 2 includes the following process: S2.1. Use 3D laser scanners and sensors to monitor the foundation pit; S2.2. Use point cloud comparison and registration algorithms to perform data analysis and obtain low-level foundation pit displacement point cloud data.
[0011] In an optional embodiment, step three includes the following process: S3.1. Data cleaning of dense point cloud data of high-rise foundation pit scenes and displacement point cloud data of low-rise foundation pits; S3.2. Use deep learning models to extract geometric and topological features from dense point cloud data of high-rise foundation pit scenes and displacement point cloud data of low-rise foundation pits, and classify and label the extracted features to identify key structural nodes; S3.3, using the iterative closest point algorithm to preliminarily register the dense point cloud data of the high-rise foundation pit scene and the displacement point cloud data of the low-rise foundation pit, and obtain a preliminarily registered reconstruction model; S3.4. Combine the preliminarily aligned reconstruction model with the extracted features and optimize it through deep learning to obtain an optimized reconstruction model.
[0012] In an optional embodiment, step S3.1 uses statistical filtering and normalization and standardization methods to clean the dense point cloud data of the high-rise foundation pit scene and the displacement point cloud data of the low-rise foundation pit.
[0013] In an optional embodiment, a time series analysis is performed on the optimized reconstruction model, and combined with real-time low-level foundation pit displacement point cloud data, a Kalman filter is used for dynamic update to obtain a dynamic reconstruction model to achieve real-time update.
[0014] In an optional embodiment, model verification is performed on the dynamically reconstructed model.
[0015] The beneficial effects of the present invention based on its technical solution are: The present invention provides a method for automatic reconstruction of foundation pit scenes based on machine vision, which divides scene reconstruction into two levels: high-level behavior scene reconstruction and low-level behavior details. The high-level scene reconstruction method relies on video technology, uses a camera to shoot a video of the entire foundation pit construction from a high altitude, extracts the overall shape, structure and information of different construction stages of the foundation pit through video frame processing, and then uses a computer vision algorithm to convert the video frame into three-dimensional point cloud data to form a point cloud model of the foundation pit scene. This step can obtain the scene information of the foundation pit and the staged tasks of the construction; the low-level detail reconstruction method uses a laser scanner to regularly monitor the foundation pit construction, obtain the displacement point cloud data, and then compare the point cloud data at different times through the point cloud comparison and registration algorithm to obtain the displacement change. This step is to reconstruct the corresponding foundation pit deformation details for each construction stage in the high-level scene. Finally, the high-level reconstruction model and the low-level displacement detail model are fused, and the multi-scale and multi-modal point cloud data are fused using artificial intelligence and other methods, and the deformation information is dynamically updated and feedback-adjusted in the foundation pit scene according to the displacement time series change information of the point cloud data monitored by the construction progress. The present invention can visualize the progress and deformation of the entire foundation pit process with high precision, making safety warning and emergency guidance more accurate and scientific, and has high application value. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0017] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION
[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention belong to the scope of protection of the embodiments of the present invention.
[0019] Reference Figure 1 The embodiment of the present invention provides a method for automatic reconstruction of a foundation pit scene based on machine vision, the method comprising: Step 1: Reconstruction of high-level behavior scenarios, including the following processes: S1.1. Use a camera to shoot the foundation pit construction video.
[0020] First, you need to select a high-definition, high-frame-rate industrial camera suitable for monitoring foundation pit construction. To obtain a panoramic view of the entire foundation pit, install the camera in a stable position that is not affected by construction, preferably a high place, such as a building or a hoisting device, to collect video sets of the foundation pit construction area from different angles, including scene information of the foundation pit construction and non-construction, as well as videos of each stage of the foundation pit construction. Based on the above considerations, the number of cameras and their layout settings can be determined according to the size and shape of the foundation pit. The position of the camera should be able to provide a comprehensive perspective to capture every detail of the foundation pit construction process.
[0021] The recorded video is then stored in a dedicated video storage device. Considering the long-term recording of the foundation pit construction video, the storage strategy should consider cloud storage or large-capacity storage devices, such as hard disk video recorders (DVRs), network video recorders (NVRs), cloud or local servers, etc.
[0022] S1.2. Process the foundation pit construction video based on the contents of different excavation stages.
[0023] Video preprocessing can improve the accuracy and stability of subsequent processing, including video segmentation, image denoising and image enhancement. First, the video is divided into multiple segments according to the construction of each stage. Since the excavation time of the foundation pit is relatively long and the excavation progress of most foundation pits is uneven, content-based segmentation is adopted: flexible segmentation according to the changes in the construction progress of each stage. This step can obtain the foundation pit scene information and information of different construction stages. Then, the noise type is preliminarily judged by observing the characteristics and histogram of the image for denoising. Denoising is to reduce the noise in the image and improve the quality of the image. Observe the video image and determine the possible noise type, such as Gaussian noise, salt and pepper noise, etc. According to the determined noise type, select a suitable denoising method. Common methods include mean filtering, median filtering, Gaussian filtering, etc. Evaluate the denoised video to check whether the denoising effect meets the requirements. Finally, use image processing software or image processing library to realize image enhancement of video data. Image enhancement is to improve the visual effect of the image, making the image clearer and easier to analyze. Common image enhancement methods include contrast enhancement, brightness enhancement, sharpening, etc. In a video, each frame of the image can be enhanced to improve the quality and clarity of the image.
[0024] Video preprocessing can be done using Python (a programming language) and the OpenCV library (open source computer vision library).
[0025] S1.3. Use Structure from Motion (SfM) technology to infer the three-dimensional structure of the foundation pit scene and the camera position from the frame images of the video, establish the geometric relationship between the foundation pit scene and the camera, and obtain the camera position and sparse scene point cloud data.
[0026] S1.4. Use Multi-View Stereo (MVS) technology to convert sparse scene point cloud data into dense point cloud data of high-rise foundation pit scene. This includes the following processes: S1.4.1, Image Segmentation: Image segmentation technology (Mask RCNN) is used to segment and classify foundation pit video images. This method uses FPN (Feature Pyramid Network) for feature fusion and RPN (Region Proposal Network) to mark different regions in the anchor image. Then, an image mask is generated and classification tasks are performed. Different objects in the image are segmented according to the classification and mask to identify different objects in the foundation pit scene (such as overall contours and supporting structures).
[0027] S1.4.2. Extract image feature points based on scale invariant feature transform (SIFT) algorithm: SIFT is an algorithm for describing local features of an image. It can detect and describe feature points in an image and has scale invariance. First, the difference of Gaussian (DOG) function is used to detect extreme values of the image, that is, the DOG operator is used to perform corresponding operations on each point in the image, and each pixel point is compared with all its neighboring points, so as to obtain local extreme points in the foundation pit video image, such as corner points, edge points, bright spots in dark areas or dark spots in bright areas.
[0028] S1.4.3, Feature Matching: In order to restore 3D information using multiple images, it is necessary to establish a connection between multiple images. The step is to match the common feature points contained in different images. Based on the feature point detection results, the Euclidean distance measurement method described by the high-dimensional vector of the feature points is used to match the common feature points in different images. The feature point descriptor in the reference image is ; The descriptor in the target graph is ; The Euclidean distance between two high-dimensional vector descriptors is:
[0029] It reflects the straight-line distance between two feature sub-vectors in n-dimensional space.
[0030] For a feature point in the reference image R i , search the target graph and calculate its R i The Euclidean distance of R i The nearest feature point is recorded as S i , the search distance R i The next closest feature point is denoted as S i–1 , in the feature point matching step of restoring the three-dimensional image of the foundation pit scene, The corresponding edge points are identifiable edge points of the actual foundation pit features; The corresponding point is the vertex or intersection of the supporting structure; Corresponding to the feature points at the bottom of the foundation pit, such as pits, protrusions, drainage ditches, etc., it can provide consistent geometric information in multi-view images; The corresponding points are the special concave-convex structural feature points on the side wall of the foundation pit; The corresponding ones are permanent or temporary marking points inside the foundation pit, such as survey marks fixed at certain positions in the foundation pit or fixed signs during construction; These points correspond to cracks or fracture points inside the foundation pit, which are significant and stable in the image through texture and morphological changes. In practical applications, more valuable feature points can be added for feature matching.
[0031] The conditions for feature point pairing are:
[0032] The matching threshold is set to determine whether two feature points are close enough to be considered a match. i With R i The similarity described by the high-dimensional vector of is obviously better than other feature points in the target graph, so it is believed that S i With R i The matching is successful. The common feature points that are successfully matched in the image and the invalid matching results after the matching algorithm attempts to search are marked with different colors. According to the trend of successful matching, it can be judged whether the matching results are consistent with the actual situation.
[0033] S1.4.4, Camera trajectory calculation: Based on the feature point matching results, the RANSAC (Random Sampling Consensus) algorithm is used to solve the camera trajectory, obtain the camera's position and posture information, and reconstruct the sparse point cloud of the foundation pit scene. If the camera trajectory solution result is good, the sparse point cloud can reflect the basic outline of the foundation pit scene and the depth of the longitudinal excavation.
[0034] S1.4.5. Generate dense point cloud data for high-rise foundation pit scenes: Dense reconstruction uses the camera pose obtained in the sparse reconstruction step and the sparse point cloud information of the foundation pit scene as input sources, and uses MVS (Multi-view Stereo) technology for dense reconstruction. ① Use the CMVS (Clustered Multi-view Stereo) algorithm to perform clustering operations and adaptively form a set of images with greater correlation to reduce the amount of peak operations; ② For each image set, the PMVS (Parallelized Multi-view Stereo) algorithm, polar constraints and correlation tests are used for stereo matching to reconstruct the dense point cloud of the foundation pit scene. In the initially established dense point cloud, there are usually some outliers (i.e. points that are obviously far away from the actual area of the foundation pit excavation), but they can be cleaned up by manual or programming methods to obtain the final high-rise foundation pit scene dense point cloud data, thereby improving the quality of subsequent model reconstruction.
[0035] Dense reconstruction significantly increases the number of point clouds, which can more clearly reflect the shape of the foundation pit scene. Details such as the supporting structure in the foundation pit and the changes in excavation depth of the video content in each construction stage have begun to appear.
[0036] Step 2: Low-level detail displacement reconstruction, including the following processes: S2.1. Use 3D laser scanners and sensors to monitor the foundation pit. You can choose a laser scanner with high resolution and high precision, and use its excellent performance and stability to collect point cloud data inside the foundation pit.
[0037] Fixed scanning positions are selected inside the foundation pit. Usually, fixed scanning points are set near the inner edge of the foundation pit, the supporting structure, and the wall. The position and angle of each scan are ensured to be consistent. The distance and angle between the scanning position and the monitoring point also need to be considered to ensure that sufficient information can be obtained to monitor and analyze displacement changes, which is convenient for subsequent point cloud data comparison and alignment.
[0038] S2.2. Use point cloud comparison and registration algorithms to perform data analysis and obtain low-level foundation pit displacement point cloud data.
[0039] Step 3: Fusion of multi-source point cloud data and reconstruction of the 3D surface model of the entire foundation pit scene, including the following processes: S3.1. Statistical filtering, normalization and standardization methods are used to clean the dense point cloud data of high-rise foundation pit scenes and the displacement point cloud data of low-rise foundation pits.
[0040] When implementing the data cleaning step, you can use point cloud processing libraries such as PCL (Point Cloud Library) or the Open3D library in Python (an open source library that supports rapid development of 3D data processing). The processed data will lay a solid foundation for subsequent feature extraction, registration, and fusion steps.
[0041] S3.2. Use the deep learning PointNet++ model to extract geometric and topological features from the dense point cloud data of high-rise foundation pit scenes and the displacement point cloud data of low-level foundation pits, and use the random forest algorithm to classify and label the extracted features to identify key structural nodes.
[0042] The extracted geometric features include: Normal vector: local surface orientation at each point, key to identifying surface inclination and directionality of structure; Curvature: The degree of curvature of the surface around each point, which helps to identify edges and corners, such as the corner of a wall or the edge of a foundation pit.
[0043] Topological features include: Density: The density of local points, which can be used to distinguish open spaces from densely structured areas.
[0044] Connectivity: Topological relationships between points, identifying supporting structures and other continuous structural elements.
[0045] The extracted features are used for the following classification and labeling using random forest: Structure type classification: Classify points in the point cloud into different structure types, such as supporting structures, walls, open areas, etc.
[0046] S3.3. Apply the Iterated Closest Points (ICP) algorithm to perform preliminary registration on the dense point cloud data of the high-rise foundation pit scene and the displacement point cloud data of the low-rise foundation pit to obtain a preliminary registered reconstruction model.
[0047] S3.4. Combine the reconstructed model of the preliminary registration with the extracted features, optimize through deep learning, align the main structural clues, and use genetic algorithms to optimize the parameter selection of ICP to improve the preliminary accuracy of the registration and obtain an optimized reconstructed model. Coarse registration can align the boundaries and corners of the foundation pit, the position of the supporting structure, and any significant vertical and horizontal surfaces in the foundation pit scene.
[0048] Step 4: Perform time series analysis on the optimized reconstruction model, combine it with the real-time low-rise foundation pit displacement point cloud data, use the Kalman filter to dynamically update it, obtain a dynamic reconstruction model, and realize real-time update.
[0049] Finally, the 3D modeling tool Open3D can be used to verify the dynamic reconstructed model.
[0050] The present invention provides an automatic reconstruction method of foundation pit scenes based on machine vision. By transforming the foundation pit engineering project from traditional manual investigation and simple local photography to the establishment of a complete digital information model, the machine vision measurement technology is combined with computer technology to realize high-precision three-dimensional scene data collection and reconstruction, and high-precision acquisition of foundation pit scene information, construction process, support structure, displacement and other information. The foundation pit scene model is then quickly constructed using high-precision three-dimensional point cloud data to realize real-time visualization of the foundation pit site, thereby effectively improving the construction monitoring efficiency and construction quality, thereby minimizing the risks and costs of the foundation pit project, timely discovering and solving problems, and ensuring the smooth progress of the foundation pit project.
[0051] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0052] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (modules, systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0053] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0054] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0055] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.
[0056] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.
Claims
1. A method for automatic reconstruction of foundation pit scenes based on machine vision, characterized in that: The method comprises: Step 1: Reconstruct high-level behavior scenarios; Step 2: Low-level detail displacement reconstruction; Step 3: Fusion of multi-source point cloud data and reconstruction of the 3D surface model of the entire foundation pit scene; Step 4: Update in real time.
2. The automatic reconstruction method of foundation pit scene based on machine vision according to claim 1 is characterized in that: Step 1 includes the following process: S1.
1. Use a camera to shoot the foundation pit construction video; S1.
2. Process the foundation pit construction video based on the contents of different excavation stages; S1.3, using computer vision algorithms to convert video frame images into sparse scene point cloud data; S1.
4. Convert the sparse scene point cloud data into dense point cloud data of high-rise foundation pit scene.
3. The automatic reconstruction method of foundation pit scene based on machine vision according to claim 2 is characterized in that: Step S1.2 processes the foundation pit construction video based on the content of different excavation stages, including video segmentation, image denoising and image enhancement processing.
4. The automatic reconstruction method of foundation pit scene based on machine vision according to claim 2 is characterized in that: Step S1.3 uses motion recovery structure technology to infer the three-dimensional structure of the foundation pit scene and the camera position from the frame images of the video, establishes the geometric relationship between the foundation pit scene and the camera, and obtains the camera position and sparse scene point cloud data.
5. The automatic reconstruction method of foundation pit scene based on machine vision according to claim 2 is characterized in that: Step S1.4 uses multi-view stereo vision technology to convert sparse scene point cloud data into dense point cloud data of high-rise foundation pit scenes.
6. The automatic reconstruction method of foundation pit scene based on machine vision according to claim 1 is characterized in that: Step 2 includes the following process: S2.
1. Use 3D laser scanners and sensors to monitor the foundation pit; S2.
2. Use point cloud comparison and registration algorithms to perform data analysis and obtain low-level foundation pit displacement point cloud data.
7. The method for automatic reconstruction of foundation pit scenes based on machine vision according to any one of claims 1 to 6, characterized in that: Step 3 The process includes: S3.
1. Data cleaning of dense point cloud data of high-rise foundation pit scenes and displacement point cloud data of low-rise foundation pits; S3.
2. Use deep learning models to extract geometric and topological features from dense point cloud data of high-rise foundation pit scenes and displacement point cloud data of low-rise foundation pits, and classify and label the extracted features to identify key structural nodes; S3.3, using the iterative closest point algorithm to preliminarily register the dense point cloud data of the high-rise foundation pit scene and the displacement point cloud data of the low-rise foundation pit, and obtain a preliminarily registered reconstruction model; S3.
4. Combine the preliminarily aligned reconstruction model with the extracted features and optimize it through deep learning to obtain an optimized reconstruction model.
8. The automatic reconstruction method of foundation pit scene based on machine vision according to claim 7 is characterized in that: Step S3.1 uses statistical filtering and normalization and standardization methods to clean the dense point cloud data of high-rise foundation pit scenes and the displacement point cloud data of low-rise foundation pits.
9. The automatic reconstruction method of foundation pit scene based on machine vision according to claim 7 is characterized in that: The optimized reconstruction model is subjected to time series analysis and combined with the real-time low-rise foundation pit displacement point cloud data. The Kalman filter is used for dynamic update to obtain a dynamic reconstruction model and achieve real-time update.
10. The method for automatic reconstruction of foundation pit scenes based on machine vision according to claim 7, characterized in that: Perform model validation on the dynamic reconstructed model.
Citation Information
Cited By
Deep foundation pit construction dynamic monitoring method and system based on BIM
CN120124166A
A Dynamic Monitoring Method and System for Deep Foundation Pit Construction Based on BIM
CN120124166B
Intelligent construction digital supervision method, system, equipment and medium
CN120725616A
