Construction site modeling system and method based on images
By using an image-based construction site modeling system, a realistic model of the construction site is constructed, which solves the problems of low efficiency and information lag under traditional management methods, and realizes efficient visual management and real-time monitoring of the construction site.
Patent Information
- Application Number
- CN202511639788.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-11
- Publication Date
- 2026-03-06
AI Technical Summary
Traditional construction site management relies on manual inspections and paper records, which is inefficient, subjective, and has delayed information feedback, making it difficult to detect safety hazards and quality defects in real time.
An image-based construction site modeling system is adopted, including a spatial coordinate origin module, a video acquisition module, a white model generation module, and a texture module. A real-world model of the construction site is constructed using wireless positioning technology and video acquisition equipment, and the progress is calculated in conjunction with the BIM model.
It enables visualized management of construction sites, improves management efficiency, reduces subjective bias and information omissions, provides real-time dynamic monitoring and progress tracking, and supports efficient safety supervision and quality inspection.
Smart Images
Figure CN121616780A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of information technology for engineering management, specifically to an image-based construction site modeling system and method. Background Technology
[0002] With the development of the construction industry, large-scale and highly complex projects are becoming increasingly common, drastically increasing the difficulty of construction site management. Traditional inspection methods mainly rely on manual on-site inspections and paper records, which suffer from problems such as low efficiency, strong subjectivity, and delayed information feedback, making it difficult to achieve real-time detection and handling of safety hazards and quality defects. Summary of the Invention
[0003] The purpose of this invention is to provide an image-based construction site modeling system and method to visualize the construction site, making it easier for managers to intuitively view the progress and specific conditions of the construction.
[0004] To address the aforementioned technical problems, this invention provides an image-based construction site modeling system, comprising: The spatial coordinate origin module is used to provide position reference information; The video capture module is used to capture video from the construction site. The white model generation module is used to construct a white model of the construction site based on video and location reference information of the construction site. The texture module is used to process the video into textures based on the geometric features of the white model, and then overlay the textures onto the surface of the white model to form a realistic construction site model.
[0005] According to the above scheme, the spatial coordinate origin module includes a spatial positioning submodule and a visual positioning submodule; the spatial positioning submodule provides positioning information to the video acquisition module through one or more wireless positioning / communication technologies; the visual positioning submodule provides reference images for the high-altitude shooting equipment in the video acquisition module.
[0006] According to the above scheme, the video acquisition module includes a dedicated recording device, which includes a safety helmet, a positioning sub-module fixed to the safety helmet, and high-definition cameras and fisheye cameras arranged at intervals around the safety helmet; the positioning sub-module transmits data with the spatial coordinate origin module so that the captured video includes the pose information of the video acquisition module.
[0007] According to the above scheme, the video acquisition module includes autonomous inspection equipment, which includes unmanned equipment and video acquisition equipment and positioning sub-module mounted on the unmanned equipment.
[0008] According to the above scheme, the video acquisition module includes commercial acquisition equipment; the video acquired by the commercial acquisition equipment shall capture at least two or more spatial coordinate origin modules.
[0009] According to the above scheme, the construction site real-scene model constructed by dedicated recording equipment and autonomous inspection equipment has a higher priority than the construction site real-scene model constructed by commercial acquisition equipment.
[0010] According to the above scheme, a progress calculation module is included; the progress calculation module obtains the project completion progress at the construction site based on the white model and the pre-drawn BIM model.
[0011] According to the above scheme, the white model is stored according to the set white model storage period. Each white model storage period only stores the white model with the fastest project completion progress within the current period. When the white model is updated, the white model of the previous white model storage period is deleted, and the difference data between the white model of the current white model storage period and the white model of the previous white model storage period is saved.
[0012] According to the above scheme, the priority of textures is based on the video from which they originate, and they are arranged from high to low as follows: those collected manually or by personnel, those collected manually or automatically to supplement data, those collected by automatic inspection, and those collected by fixed camera equipment. Textures covering the surface of the white model are updated according to the set texture update cycle, and the video with the highest priority and the most recent collection time is selected as the texture source when updating.
[0013] This invention also provides an image-based construction site modeling method, comprising: S1. Collect videos of the construction site and obtain location reference information corresponding to the videos; S2. Construct a white model of the construction site based on the video and location reference information of the construction site; S3. Based on the geometric features of the white model, process the video into a texture and overlay the texture onto the surface of the white model to form a real-life model of the construction site.
[0014] Beneficial effects This invention provides a precise spatial reference for subsequent modeling by utilizing the location reference information offered by the spatial coordinate origin module. This effectively avoids the problem of model discrepancies with the actual construction site location that often occurs in traditional visual modeling, ensuring that both the subsequently acquired video data and the generated model correspond to the real spatial location of the construction site. The video acquisition module can directly capture video from the construction site. Compared to traditional manual site inspections and paper records, this not only captures the dynamic situation and detailed information of the construction site more efficiently and comprehensively, reducing subjective biases and information omissions in manual recording, but also retains first-hand visual data in real time, providing rich and realistic original evidence for subsequent model construction. The white model generation module combines construction site video with... The location reference information is used to construct the white model. On the one hand, it relies on video data to obtain the geometric shape information of the construction site. On the other hand, it uses location reference information to accurately anchor these geometric shapes to the spatial coordinates of the actual construction site. This gives the generated white model both a complete geometric framework and clear spatial relationships, providing a precise geometric carrier for subsequent real-world rendering. The texture module processes the video into textures based on the geometric characteristics of the white model and overlays them on the surface of the white model. This allows the white model to retain its accurate geometric structure while giving it realistic appearance details consistent with the construction site. The final real-world construction site model has both spatial location accuracy and geometric structural integrity that match the actual site, as well as visual information that intuitively reflects the real appearance of the site. This system, through the synergistic effect of four major modules, deeply integrates video data with spatial location information to construct a real-scene model. This model allows managers to intuitively and clearly view the specific conditions and geometric details of the construction site. It solves the problem of traditional construction site management relying on text records, drawings, or scattered photos, which makes the site conditions unintuitive and difficult to fully grasp. It significantly improves the visualization of construction site management, making it easier for managers to quickly and accurately understand the construction progress and site details. This reduces management delays or decision-making biases caused by unintuitive or inaccurate information transmission, providing strong support for efficient construction site management. Attached Figure Description
[0015] Figure 1 This is a flowchart illustrating the workflow of an image-based construction site modeling system according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the spatial coordinate origin module structure according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the QR code image area and the image distortion correction image area according to an embodiment of the present invention; Figure 4 This is a schematic diagram of a dedicated recording device according to an embodiment of the present invention.
[0016] In the diagram: 1. Spatial positioning equipment box; 2. Visual positioning sub-module; 3. Safety helmet; 4. High-definition camera; 5. Fisheye camera; 21. QR code image area; 22. Image distortion correction image area. Detailed Implementation
[0017] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the described embodiments of this disclosure without creative effort are within the scope of protection of this disclosure.
[0018] This embodiment discloses an image-based construction site modeling system, including: The spatial coordinate origin module is used to provide position reference information; The video capture module is used to capture video from the construction site. The white model generation module is used to construct a white model of the construction site based on video and location reference information of the construction site. The texture module is used to process the video into textures based on the geometric features of the white model, and then overlay the textures onto the surface of the white model to form a realistic construction site model.
[0019] Specifically, this system is suitable for information management of engineering projects, aiming to reduce the workload of management personnel and improve project management efficiency. It primarily uses image-based real-scene modeling of the construction site. When constructing the white model of the construction site, its accuracy needs to be verified. For white models with insufficient accuracy, a video acquisition module can be used to supplement the data, or radar point cloud data can be obtained using LiDAR to improve the accuracy of the white model. When processing video, the system first processes the video into a white model, then correctly places the white model in a pre-defined coordinate system based on its location information, and finally processes the video into corresponding textures based on the geometric features of the white model. As the project progresses, more and more video will be collected. After the same processing, the increase in the volume of the white model represents the construction progress of the project entity, and the textures are used for inspection and viewing of the project entity.
[0020] Furthermore, the spatial coordinate origin module includes a spatial positioning submodule and a visual positioning submodule 2; the spatial positioning submodule provides positioning information to the video acquisition module through one or more wireless positioning / communication technologies; the visual positioning submodule 2 provides reference images for the high-altitude shooting equipment in the video acquisition module.
[0021] Specifically, the spatial positioning submodule is a spatial positioning equipment box 1, which contains a spatial positioning device and a power supply. The spatial positioning device employs, but is not limited to, one or more technologies such as GNSS, TDOA, RSSI, LPWAN, and NB-IoT. Its function is to provide positioning information for the video acquisition module to determine its position (i.e., position reference information) in a set spatial coordinate system. The visual positioning submodule 2 provides an origin reference for the high-altitude shooting equipment and corrects image distortion, including a QR code image area 21 and an image distortion correction image area 22. The spatial origin module can be set individually at the engineering coordinate origin, arranged at multiple points along a grid, or arranged irregularly at multiple points. The spatial coordinate system is established by using the position of a certain spatial positioning submodule as the coordinate origin, with other spatial positioning submodules located at specific points in the spatial coordinate system to determine the position of the video acquisition module within the spatial coordinate system. It is understood that more spatial positioning submodules can provide higher positioning accuracy, but there is also a diminishing marginal utility effect; the decision should be made comprehensively based on the specific conditions of the project site.
[0022] Furthermore, the video acquisition module includes a dedicated recording device, which includes a safety helmet 3, a positioning sub-module fixed to the safety helmet 3, and high-definition cameras 4 and fisheye cameras 5 arranged at intervals around the safety helmet 3; the positioning sub-module transmits data with the spatial coordinate origin module so that the captured video includes the pose information of the video acquisition module.
[0023] Specifically, in this embodiment, the front of the safety helmet 3 is taken as the 0° direction. Starting from 0°, a camera is set every 45°. The cameras set at 0°, 90°, 180°, and 270° are high-definition cameras 4, and the cameras set at 45°, 135°, 225°, and 315° are fisheye cameras 5. The high-definition cameras 4 and fisheye cameras 5 are fixed to the safety helmet 3 with straps. The positioning submodule is a low-data positioning chip, which is located on the inner top of the safety helmet 3. Project managers wearing this safety helmet 3 can simultaneously complete video acquisition during daily inspections. In this embodiment, cameras with a resolution greater than 2K are defined as high-definition cameras.
[0024] Furthermore, the video acquisition module includes autonomous inspection equipment, which includes unmanned equipment as well as video acquisition equipment and positioning sub-module mounted on the unmanned equipment.
[0025] Specifically, in this embodiment, the autonomous inspection equipment adopts a shelf-based product + accessory architecture. It is based on mature industrial products and consists of a low-data positioning chip and a dedicated data transmission module. For example, after mounting the relevant components on drones, quadruped robots, etc., they can automatically collect video data according to a set program. It should be noted that the method of safe inspection of drones, quadruped robots, and other equipment in complex indoor environments refers to the patent literature: Method and System for Indoor Engineering Site Inspection Based on Commercial Drones (Application No.: 2024111847388).
[0026] Furthermore, the video acquisition module includes commercial acquisition equipment; the video acquired by the commercial acquisition equipment captures at least two or more spatial coordinate origin modules.
[0027] Specifically, commercial acquisition devices include mobile phones, 360-degree panoramic cameras, etc. The purpose of ensuring that the video acquired by commercial acquisition devices captures at least two or more spatial coordinate origin modules is to enable the system to perform positioning during subsequent video processing.
[0028] This section explains the video data processing. It's important to first clarify that the technical principles behind video processing are all based on aerial triangulation, a mature and widely used technology, so they will not be elaborated upon here. Only the unique aspects of this invention will be explained. Dedicated recording equipment, drones, quadruped robots, and other devices can exchange data with the spatial origin module. When recording video, they include data such as the spatial coordinates and shooting angle, so the models and other results obtained from the video can be matched with a pre-designed coordinate system. In other words, the biggest difference between this invention and traditional visual model generation technology is that it records and calculates the spatial information of the model while generating it. This allows the computer to match and unify videos taken by different devices, at different times, and in different ways using spatial location information, creating a dynamic model flow in the construction process of engineering projects, analogous to "time-lapse photography." When using mobile phones, 360-degree panoramic cameras, fixed cameras, etc., as video data sources, it is necessary to ensure that several QR codes on the spatial origin module are captured within the video stream. The system constructs independent spatial coordinates for the video at its beginning (e.g., video VID1) using aerial triangulation. Upon recognizing the QR code on the spatial origin module, aerial triangulation is also used to determine the positional relationship between the "videographer" and the spatial origin module. Furthermore, the QR code on the spatial origin module also contains the module's positional information. With these two sets of information, all positional information from VID1 to the generated model can be calculated, resolving the issue of inconsistent application across different videos. Moreover, the more positional information recorded in the video, the higher the overall system accuracy. However, since dedicated recording and autonomous inspection equipment collects more positional information than commercial acquisition equipment, the models obtained by dedicated recording and autonomous inspection equipment have higher priority. When the model information obtained by commercial acquisition equipment deviates significantly from these models, it will be corrected through algorithms or manually.
[0029] Furthermore, construction site reality models constructed using dedicated recording and autonomous inspection equipment have a higher priority than those constructed using commercial data acquisition equipment.
[0030] Furthermore, it includes a schedule calculation module; the schedule calculation module obtains the project completion schedule at the construction site based on the white model and the pre-drawn BIM model.
[0031] Specifically, by comparing the white model with the BIM model (by performing Boolean operations after aligning the white model and BIM model to the same coordinate system), the project completion progress can be obtained. This embodiment can also obtain the project's progress by comparing the white models at different points in time. It is understood that when a pre-drawn BIM model exists, the BIM model should be pre-loaded.
[0032] Furthermore, the white model is stored according to the set white model storage period. Each white model storage period only stores the white model with the fastest project completion progress within the current period. When the white model is updated, the white model of the previous white model storage period is deleted, and the difference data between the white model of the current white model storage period and the white model of the previous white model storage period is saved.
[0033] Specifically, in this embodiment, the white model storage period is 1 day; when storing the white model of the information, the difference is retained by deleting the original white model, which reduces the storage resource requirements.
[0034] Furthermore, the priority of textures is based on the video from which they originate, arranged from highest to lowest as follows: those collected manually or identified by personnel, those collected manually or automatically to supplement data, those collected through automatic inspection, and those collected by fixed camera equipment; textures covering the surface of the white model are updated according to the set texture update cycle, and the video with the highest priority and the most recent collection time is selected as the texture source during the update.
[0035] Specifically, in this embodiment, the priority of the textures can be labeled from high to low as 010, 020, 030, and 040, and the texture update cycle is half a day (00:00-12:00, 12:00-24:00). During the update, the file with the highest priority and the last time sequence is selected to construct the main texture. After the texture is generated, it overwrites the corresponding white model. If it cannot completely cover the model, other video data is called to supplement it according to the priority order and the reverse time sequence until full coverage is achieved. The above processing is also to reduce the storage resource requirements.
[0036] In this system, the white model combined with texture files constitutes a digital twin model of the engineering project, enabling remote project inspection using real-scene simulation technology. The "growth" of the white model represents the construction progress, and the corresponding textures achieve the same effect as "time-lapse photography," forming a digital twin time-series model of the entire project construction process. This model enables traceability of the construction process, providing a fast, efficient, and easy-to-use tool for problem tracing, maintenance, and repair. Compared to saving video files, it offers a generational advantage in terms of storage resources and ease of use. In this embodiment, an AI-edge computing unit is used as the hardware support for the aforementioned functions. The application platform of this system is the Yiye Digital Construction Platform V3.0 (Certificate No.: Software Copyright Registration No. 16260401), and the aforementioned functions will be implemented in its V3.4 version. This system can collect data in multiple ways, with simple data collection and good reusability. At the same time, by using innovative methods, it achieves functions such as safety supervision, quality inspection and construction process traceability while using less computing power and occupying less storage resources. It can quickly and effectively reduce the workload of personnel, and is an easy-to-learn and easy-to-use tool to improve the work efficiency of managers, thereby improving the management capabilities of engineering construction companies.
[0037] The construction site real-scene model obtained in this embodiment can be used to realize AR augmented reality, and can also be used for functions such as safety supervision, quality inspection, and construction process traceability to improve the visualization and manageability of the project. The construction site real-scene model can be integrated with AI to achieve automatic identification of safety risks, quality problems, etc.
[0038] This invention also provides an image-based construction site modeling method, comprising: S1. Collect videos of the construction site and obtain location reference information corresponding to the videos; S2. Construct a white model of the construction site based on the video and location reference information of the construction site; S3. Based on the geometric features of the white model, process the video into a texture and overlay the texture onto the surface of the white model to form a real-life model of the construction site.
[0039] The beneficial effects of the present invention include at least the following: 1. By innovating the algorithm structure, video files are first decoded into white models that can be incorporated into a unified spatial coordinate system. A method of recording differences in the white models is used to save only one set of model data, significantly reducing system computing power requirements and greatly minimizing storage resource needs. Furthermore, by employing a texture mapping approach, not only is the video function preserved, but AR augmented reality technology is also integrated to achieve viewing from any perspective, allowing for flexible and autonomous viewing based on work needs. This solves the technical challenges of high computing power requirements, huge storage resource consumption, and susceptibility to model integration errors in complex spatial conditions.
[0040] 2. By constructing a white model and assigning textures, a digital twin of the entire construction process of an engineering project with temporal attributes was achieved. This solved problems such as the difficulty in preserving engineering details and the limited reusability of data results.
[0041] 3. By recording the changes in the white model over time, the project's progress can be intuitively reflected. Furthermore, by integrating BIM technology, the project's progress can be identified more intelligently, solving the problem that traditional video computers cannot automatically recognize progress.
[0042] 4. An innovative method for establishing a spatial coordinate system was developed. Video data collected from multiple sources was standardized and structured using a spatial origin module, enabling comprehensive use. This solved the problem of poor video data reusability.
[0043] 5. By integrating and implementing the functional modules of this invention through the China Metallurgical Construction Data Platform, the invention achieves the goals of being easy to learn and use. This solves the problems of high learning costs, difficulty in use, and challenges in promoting and applying intelligent digital tools.
[0044] This invention offers a reliable technical approach with low computational and storage requirements, providing a method that can quickly and effectively reduce the workload of personnel and improve the management capabilities of engineering construction companies.
[0045] It should be noted that, depending on the implementation needs, the various steps / components described in this application can be broken down into more steps / components, or two or more steps / components or parts of the operation of steps / components can be combined into new steps / components to achieve the purpose of this invention.
[0046] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. An image-based construction site modeling system, characterized by, The application relates to a construction site real scene model construction method, which comprises the following steps: A space coordinate origin module is used to provide position reference information; A video acquisition module is used to acquire the video of a construction site; A white model generation module is used to construct a white model of a construction site according to the video of the construction site and the position reference information; A mapping module is used to process the video into a map according to the geometric characteristics of the white model, and the map is overlaid on the surface of the white model to form a construction site real scene model.
2. The image-based construction site modeling system of claim 1, wherein, The space coordinate origin module comprises a space positioning sub-module and a visual positioning sub-module; the space positioning sub-module provides positioning information for the video acquisition module through one or more wireless positioning / communication technologies; and the visual positioning sub-module provides a reference image for a high-altitude shooting device in the video acquisition module.
3. The image-based construction site modeling system of claim 1, wherein, The video acquisition module comprises a special recording device, the special recording device comprises a safety helmet, a positioning sub-module fixed to the safety helmet, and a high-definition camera and a fisheye camera arranged at intervals on the side of the safety helmet; the positioning sub-module transmits data with the space coordinate origin module, so that the pose information of the video acquisition module is attached to the acquired video.
4. The image-based construction site modeling system of claim 3, wherein, The video acquisition module comprises an autonomous inspection device, the autonomous inspection device comprises an unmanned device and a video acquisition device and a positioning sub-module carried on the unmanned device.
5. The image-based construction site modeling system of claim 4, wherein, The video acquisition module comprises a commercial acquisition device; the video acquired by the commercial acquisition device at least shoots two or more space coordinate origin modules.
6. The image-based construction site modeling system of claim 5, wherein, Compared with a construction site real scene model constructed by a commercial acquisition device, a construction site real scene model constructed by a special recording device and an autonomous inspection device has higher priority.
7. The image-based construction site modeling system of claim 1, wherein, The application further comprises a progress calculation module; the progress calculation module acquires the project completion progress of a construction site according to a white model and a pre-drawn BIM model.
8. The image-based construction site modeling system of claim 1, wherein, The white model is stored in a set white model storage period, and only the white model with the fastest project completion progress in the current period is stored in each white model storage period; when the white model is stored and updated, the white model in the previous white model storage period is deleted, and the difference data between the white model in the current white model storage period and the white model in the previous white model storage period is saved.
9. The image-based construction site modeling system of claim 1, wherein, The priority of the map is arranged from high to low based on the video source, and the arrangement is as follows: manually acquired or marked by personnel, manually acquired or automatically acquired for data supplement, automatically inspected, and acquired by a fixed position device; the map overlaid on the surface of the white model is updated in a set map update period, and the video with the highest priority and the latest acquisition time is selected as the map source during the update.
10. An image-based construction site modeling method, characterized by, The application further comprises the following steps: S1, acquiring the video of a construction site and acquiring the position reference information corresponding to the video; S2, constructing a white model of a construction site according to the video of the construction site and the position reference information; S3, processing the video into a map according to the geometric characteristics of the white model, and overlaying the map on the surface of the white model to form a construction site real scene model.