Real-time external facade detection method and system based on point cloud

Abstract

The invention discloses a real-time external facade detection method and system based on point cloud, and belongs to the technical field of building measurement, and the method comprises the steps: S1, inputting external facade project data by a cloud end; s2, acquiring point cloud data of the building facade in real time through a scanner, and performing real-time display and dynamic cutting through a WebGL rendering engine; s3: through point cloud voxel segmentation, segmenting a wall surface and a ground with significance information; s4, simulating an artificial guiding rule, and fully measuring the perpendicularity and flatness of the outer vertical surface; and S5, automatically uploading a measurement result to a cloud system for storage and analysis. According to the invention, on the basis of the self-developed WebGL rendering engine, the point cloud semantic segmentation algorithm and the point cloud intelligent measurement algorithm, the point cloud of the building facade can be obtained in real time, point cloud cutting is carried out, and the measurement result of the facade is calculated and generated in real time.

Description

Technical Field

[0001] The present invention relates to the technical field of building measurement, and in particular to a point cloud-based real-time facade detection method and system. Background Art

[0002] With the development of cities, the renovation of old residential areas and urban beautification are constantly advancing. It is necessary to accurately obtain the three-dimensional spatial information of the building facades to assess their safety conditions and formulate renovation plans. For example, when renovating old buildings, it is necessary to accurately measure the size, shape and defects of the facades. During the construction process and after completion, the quality inspection of the facade is an important part of ensuring the quality of the building. Traditional inspection methods are often inefficient and not accurate enough. The point cloud-based inspection method can quickly and comprehensively obtain various data of the facade, such as the flatness and verticality of the wall, and detect quality problems in a timely manner.

[0003] Traditional facade measurement methods, such as the use of hanging baskets, rulers, total stations and other tools for manual measurement. Traditional manual methods face multiple challenges: First, the cost is high, because manual measurement is not only time-consuming and labor-intensive, but may also lead to repeated measurements due to human factors, further increasing time and labor costs. Secondly, there are limitations in accuracy. Simple tools are difficult to accurately capture the details of complex facades, and it is easy to introduce measurement errors. Furthermore, high-altitude operations in hanging baskets bring significant safety risks, especially in bad weather or complex environments. This risk is more prominent. In addition, manual methods are also incapable of data processing. Manually recording data is prone to errors or loss, which brings inconvenience to subsequent work. Finally, the overall work efficiency is low, which not only slows down the progress of inspection, but may also increase project costs.

[0004] Traditional manual methods also have many disadvantages. Not only are they costly and low in precision, but they also pose safety risks and low work efficiency. Moreover, it is difficult for traditional methods to obtain complete three-dimensional information of the facade. For some complex facades, the measurement results may have large errors. In addition, there are some hidden parts or high places that are difficult to reach on the facade of a building. Traditional detection methods are difficult to effectively detect these parts, while point cloud technology can obtain information about these parts through non-contact scanning. The existing point cloud-based facade detection software on the market has improved the detection efficiency to a certain extent, but it is usually still necessary to collect point cloud data on site and then export it for post-processing, which still limits the real-time and accuracy of the detection.

[0005] The information disclosed in this background technology section is only intended to enhance the understanding of the overall background of the invention and should not be regarded as an acknowledgment or any form of suggestion that the information constitutes the prior art already known to a person skilled in the art. Summary of the invention

[0006] The purpose of the present invention is to provide a point cloud-based real-time facade detection method and system, which can obtain the building facade point cloud in real time, perform point cloud clipping, and calculate and generate facade measurement results in real time based on a self-developed WebGL rendering engine, a point cloud semantic segmentation algorithm, and a point cloud intelligent measurement algorithm.

[0007] To achieve the above object, the present invention provides a real-time facade detection method based on point cloud, comprising the following steps:

[0008] S1: Input facade project data into the cloud;

[0009] S2: The point cloud data of the building facade is obtained in real time through the scanner, and the data is displayed and dynamically cropped in real time through the WebGL rendering engine;

[0010] S3: Segment the wall and ground with meaningful information through point cloud voxel segmentation;

[0011] S4: simulates manual ruler to measure verticality and flatness of the facade;

[0012] S5: The measurement results will be automatically uploaded to the cloud system for storage and analysis.

[0013] In one embodiment of the present invention, step S2 includes the following sub-steps:

[0014] S201: Control the scanner to scan, obtain the real-time point cloud image of the scanner, and transmit it to the 3D model display module based on WebGL through the WebSocket network protocol, and display it in real time in the 3D model display module;

[0015] S202: cropping the real-time point cloud image to crop the building to be measured or the wall to be measured, and feeding the cropping result back to the main program;

[0016] S203: Performing automated measurement of the point cloud.

[0017] In one embodiment of the present invention, step S3 includes the following sub-steps:

[0018] S301: setting the point cloud voxel resolution, using the octree structure to equally divide the input point cloud data, and calculating the normal vector, center point and standard deviation of each voxel;

[0019] S302: uniformly generating supervoxel seeds based on the equally divided voxels;

[0020] S303: Taking the generated supervoxel seed as the starting point, a breadth-first search is performed to cluster adjacent voxels according to feature distance and normal vector. The voxel surface distance of the same wall is less than 40 mm and the normal vector angle is less than 5°.

[0021] S304: Perform semantic analysis on the clustered point cloud groups, and use the spatial position and normal vector of the point cloud to distinguish between wall and non-wall surfaces; the judgment condition for a wall surface is that the angle with the vertical line is less than 5°, and the judgment condition for a non-wall surface is that the angle with the horizontal plane is less than 5°.

[0022] In one embodiment of the present invention, step S4 includes the following sub-steps:

[0023] S401: Layering the facade point cloud into different floor point clouds according to the floor information of the project;

[0024] S402: generating a burst point map for each floor according to the divided floors, and merging the burst point maps of each floor;

[0025] S403: Using the window opening or the set sub-wall width as the dividing line, divide the facade into areas to be measured, and measure the flatness and verticality of each floor.

[0026] In one embodiment of the present invention, in step S403, clustering is performed based on the consistency of window holes on each floor to remove holes that are detected incorrectly.

[0027] The present invention also provides a point cloud-based real-time facade detection system, comprising: an information input module, a scanning module, a semantic segmentation module, an automatic measurement module and a storage module;

[0028] The information input module is used to input facade project data in the cloud;

[0029] The scanning module is used to obtain point cloud data of the building facade in real time through a scanner, and to display and dynamically crop the point cloud data in real time through a WebGL rendering engine;

[0030] The semantic segmentation module is used to segment the wall and the ground with meaningful information through point cloud voxel segmentation;

[0031] The automatic measurement module is used to simulate a manual ruler to fully measure the verticality and flatness of the facade;

[0032] The storage module is used to automatically upload the measurement results to the cloud system for storage and analysis.

[0033] In one embodiment of the present invention, the scanning module includes: a point cloud acquisition submodule, a cropping submodule and an automatic measurement submodule;

[0034] The point cloud image acquisition submodule is used to control the scanner to scan, obtain the real-time point cloud image of the scanner, and transmit it to the WebGL-based three-dimensional model display module through the WebSocket network protocol, and display it in real time in the three-dimensional model display module;

[0035] The clipping submodule is used to clip the real-time point cloud image, clip out the building to be measured or the wall to be measured, and feed the clipping result back to the main program;

[0036] The automatic measurement submodule is used for automatic measurement of point clouds.

[0037] In one embodiment of the present invention, the semantic segmentation module includes: a voxel establishment submodule, a supervoxel seed generation submodule, a supervoxel segmentation submodule and a point cloud classification submodule;

[0038] The voxel establishment submodule is used to set the point cloud voxel resolution, use the octree structure to divide the input point cloud data into equal parts, and calculate the normal vector, center point and standard deviation of each voxel;

[0039] The supervoxel seed generation submodule is used to uniformly generate supervoxel seeds based on the equally divided voxels;

[0040] The supervoxel segmentation submodule is used to perform a breadth-first search starting from the generated supervoxel seed, clustering adjacent voxels according to feature distance and normal vector, and the voxel surface distance of the same wall is less than 40 mm and the normal vector angle is less than 5°;

[0041] The point cloud classification submodule is used to perform semantic analysis on the clustered point cloud group, and distinguish between wall and non-wall surfaces using the point cloud spatial position and normal vector; wherein, the judgment condition for a wall surface is that the angle with the vertical line is less than 5°, and the judgment condition for a non-wall surface is that the angle with the horizontal plane is less than 5°.

[0042] In one embodiment of the present invention, the automatic measurement module includes: a point cloud stratification submodule, a stratified burst point submodule and a vertical ruler submodule;

[0043] The point cloud stratification submodule is used to stratify the facade point cloud into different floor point clouds according to the floor information of the project;

[0044] The hierarchical burst point submodule is used to generate burst point maps for each floor according to the divided floors, and merge the burst point maps of each floor;

[0045] The verticality ruler module is used to divide the facade into areas to be measured using window openings or a set sub-wall width as a dividing line, and to measure the flatness and verticality of each floor.

[0046] In one embodiment of the present invention, in the vertical ruler module, the window holes are clustered according to their consistency on each floor, and the holes detected by mistake are removed.

[0047] Compared with the prior art, the point cloud-based real-time facade detection method and system according to the present invention has the following advantages:

[0048] 1. Reduce costs: By automatically collecting and processing point cloud data, the time and labor costs of manual measurement are reduced, and repeated measurements and errors caused by human factors are avoided, thereby reducing the overall detection cost.

[0049] 2. Improved accuracy: By using advanced point cloud processing technology, the present invention can accurately capture the details of complex facades and improve the accuracy and reliability of detection. This is of great significance for ensuring building safety and timely discovering potential problems.

[0050] 3. Reduce safety risks: By collecting and processing data in real time on site, the time and frequency of high-altitude operations are reduced, thereby reducing safety risks. Especially in bad weather or complex environments, this real-time detection method can better ensure the safety of workers.

[0051] 4. Improve work efficiency: Real-time collection, cutting and calculation make the detection process more efficient, and can quickly obtain detection results, providing timely data support for facade renovation. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 is a flow chart of a point cloud-based real-time facade detection method according to an embodiment of the present invention;

[0053] Figure 2 is a flow chart of on-site facade scanning of a real-time facade detection method based on point cloud according to an embodiment of the present invention;

[0054] Figure 3 is a WebGL communication flow chart according to an embodiment of the present invention;

[0055] Figure 4 is a schematic diagram of a flatness measurement method according to an embodiment of the present invention;

[0056] Figure 5 is a schematic diagram of a verticality measuring method according to an embodiment of the present invention;

[0057] Figure 6 4 is a schematic structural diagram of a point cloud-based real-time facade detection system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0058] The specific implementation modes of the present invention are described in detail below in conjunction with the accompanying drawings, but it should be understood that the protection scope of the present invention is not limited by the specific implementation modes.

[0059] Unless explicitly stated otherwise, throughout the specification and claims, the term “comprise” or variations such as “include” or “comprising”, etc., will be understood to include the stated elements or components but not to exclude other elements or components.

[0060] like Figures 1 to 5 As shown, according to a preferred embodiment of the present invention, a real-time facade detection method based on point cloud is performed, scanning operations are performed through scanning devices such as Leica brand RTC360, BLK360, MS60, etc., point cloud data is acquired and processed, and solid data support is provided for accurate evaluation of building facades and subsequent operations.

[0061] The method comprises the following steps:

[0062] S1: Input facade project data into the cloud.

[0063] Specifically, the facade project data includes project information, floor information, and measurement rules. Project information includes project name, number of buildings, address, etc.; floor information includes building height, floor height, floor elevation, etc.; measurement rules include facade flatness measurement rules, facade measurement avoidance, ruler length, etc., providing basic information for subsequent measurement work.

[0064] S2: The point cloud data of the building facade is acquired in real time through the scanner, and is displayed and dynamically cropped in real time through the WebGL rendering engine.

[0065] In this step, the scanner uses Leica brand RTC360, BLK360, MS60 and other devices. The user-side measurement program and the 3D model display module use the WebSocket network protocol for communication, and the 3D model display module is embedded with a WebGL rendering engine. The 3D model display module is set on the Pad side, which is convenient for users to view and set parameters on the move. The rendering and cropping of point cloud data are displayed in real time on the Pad side, ensuring that users can immediately observe the scanning progress and effect.

[0066] Specifically, this step S2 includes the following sub-steps:

[0067] S201: Control the scanner to scan, obtain the real-time point cloud image of the scanner, and transmit it to the 3D model display module based on WebGL through the WebSocket network protocol, and display it in real time in the 3D model display module.

[0068] S202: cropping the real-time point cloud image to crop the building to be measured or the wall to be measured, and transmitting the cropped area to the main program, which crops the point cloud according to the area to obtain the point cloud area to be measured.

[0069] Because the scanner scans 360 degrees, it will scan other buildings. Therefore, in the 3D model display module on the Pad, users can cut out the facade point cloud area to be measured, that is, the wall to be measured, through a simple operation interface according to actual needs. Specifically, select the area to be measured or select the area not to be measured in the 3D editor for cutting to obtain the point cloud area to be measured.

[0070] S3: Semantic segmentation of facade point cloud: Through point cloud voxel segmentation, the original meaningless discrete points are segmented into point clouds such as walls and ground with meaningful information.

[0071] In the original point cloud data, the original meaningless discrete points are just a set of coordinate data in three-dimensional space without any specific semantic information. For example, in the point cloud data of the building facade, the original discrete points are just a bunch of points in space, and it is impossible to directly distinguish which points belong to the wall, which belong to the ground, or other building structures. In this step S3, the self-developed point cloud semantic segmentation algorithm can automatically identify and extract key elements such as walls and windows. Subsequently, the intelligent measurement algorithm will automatically calculate key parameters such as the verticality and flatness of the facade.

[0072] Specifically, the point cloud intelligent measurement algorithm is called. The 3D model display module is an independent process based on three.js. The 3D model display module communicates with the user-side measurement program using the WebSocket communication protocol. The specific communication process is as follows: Figure 3 The measurement program on the user side is responsible for initiating operations and transmitting data, while the 3D model display module is responsible for receiving data, processing operations and feeding back results. The two interact through status information and data transmission to ensure the smooth progress of the entire process.

[0073] Specifically, this step S3 includes the following sub-steps:

[0074] S301: Establish voxel space: set the point cloud voxel resolution, use the octree structure to divide the input point cloud data into equal parts, and calculate the normal vector, center point, standard deviation and other information of each voxel.

[0075] A point cloud voxel is a cube of a set size, and the set point cloud voxel resolution is 200*200*200. The normal vector of a voxel can determine the direction of the geometric surface where the voxel is located. The center point coordinates of the voxel represent the position of the voxel in space. The standard deviation can reflect the degree of discreteness of the points within the voxel and is used to determine the consistency of the voxel and other characteristics.

[0076] S302: Generate supervoxel seeds: Generate supervoxel seeds uniformly based on the equally divided voxels.

[0077] The supervoxel seed is selected from the point cloud voxels constructed in step S301. In order to ensure that the seed is not selected on the edge of the wall or on the non-wall surface, the standard deviation of the selected supervoxel seed is less than 10 mm.

[0078] The location and number of these supervoxel seeds determine the granularity of the subsequent supervoxel segmentation. If the number of supervoxel seeds is large, the supervoxels segmented subsequently will be smaller and finer, and the granularity of the supervoxel segmentation will be finer; conversely, if the number of supervoxel seeds is small, the supervoxels will be larger and coarser.

[0079] S303: Supervoxel segmentation: Starting from the generated supervoxel seed, a breadth-first search (BFS) is performed to cluster adjacent voxels according to feature distance and normal vector. The voxel surface distance of the same wall is less than 40 mm and the normal vector angle is less than 5°.

[0080] S304: Point cloud classification: semantic analysis is performed on the clustered point cloud groups, and the spatial position and normal vector of the point cloud are used to distinguish between wall and non-wall surfaces. The judgment condition for a wall surface is that the angle with the vertical line is less than 5°, and the judgment condition for a non-wall surface (i.e., floor or ceiling) is that the angle with the horizontal plane is less than 5°.

[0081] S4: Automatic measurement of exterior wall surfaces: simulates a manual ruler to fully measure the verticality and flatness of the exterior wall surfaces.

[0082] By acquiring a large amount of point cloud information of the facade and analyzing the positional relationship of the point cloud in space, we can determine whether the wall is flat and vertical. Full measurement means measuring every part of the facade without missing any part. This is achieved by systematically analyzing the point cloud data of the entire facade.

[0083] Specifically, this step S4 includes the following sub-steps:

[0084] S401: Point cloud stratification: stratify the facade point cloud into different floor point clouds according to the floor information of the project.

[0085] S402: Layered burst points: according to the divided floors, a burst point map is generated for each floor, and the burst point maps of each floor are merged.

[0086] The burst point map can graphically display the distribution, density and other characteristics of the facade point cloud of the floor. The burst point maps generated on each floor are integrated together to form a complete map. The advantage of this is that the overall difference of the facade point clouds between different floors can be compared from a macro perspective, which is convenient for a comprehensive grasp of the facade situation of the entire building, and also helps to plan the measurement area based on the overall perspective in the future.

[0087] S403: Vertical measurement: Use the window opening or the set sub-wall width as the dividing line to divide the facade into the areas to be measured, and measure the flatness and verticality of each floor.

[0088] During the segmentation process, the window openings on the facade are first identified and clustered. Then, according to the longitudinal space where the door and window openings are located, the facade is segmented into sub-walls. Due to occlusion or obstacles, some point cloud holes may be mistakenly identified as window holes. Clustering is performed based on the consistency of the window holes on each floor to remove the holes that are detected incorrectly.

[0089] like Figure 4-5 As shown in the figure, the facade is divided into three large areas to be measured according to the position of the window opening (the position of the box in the picture), namely the left, middle and right walls. Since the wall in the middle is relatively wide, considering its large area and for more precise and accurate measurement, it is further divided into two sub-walls. Through such reasonable area division, just like manually determining each section of the wall to be measured with a ruler, the measurement work is more organized and detailed.

[0090] S5: Cloud result upload and storage: After the full test is completed, the measurement results will be automatically uploaded to the cloud system for storage and analysis.

[0091] Users can view historical measurement results in the cloud at any time for comparison and analysis, providing data support for subsequent maintenance, renovation or design.

[0092] In addition, the present invention further comprises the following steps:

[0093] S6: Cloud facade report: Automatically export reports such as the wall rectification project consumption of the facade building, the layered rectification project consumption, the wall vertical level data results, the layered vertical level data, etc.

[0094] The present invention can realize real-time, precise and efficient facade detection. Through on-site real-time collection, real-time cutting and real-time calculation, it can significantly improve detection efficiency, reduce manual intervention and reduce safety risks.

[0095] like Figure 6As shown, a point cloud-based real-time facade detection system according to a preferred embodiment of the present invention is connected to scanning devices such as Leica brand RTC360, BLK360, MS60, etc., and processes the point cloud data obtained by the scanning device during scanning operations, providing solid data support for accurate evaluation of building facades and subsequent operations.

[0096] The point cloud-based real-time facade detection system includes: an information input module 1, a scanning module 2, a semantic segmentation module 3, an automatic measurement module 4 and a storage module 5.

[0097] The information input module 1 is used to input facade project data in the cloud.

[0098] Specifically, the facade project data includes project information, floor information, and measurement rules. Project information includes project name, number of buildings, address, etc.; floor information includes building height, floor height, floor elevation, etc.; measurement rules include facade flatness measurement rules, facade measurement avoidance, ruler length, etc., providing basic information for subsequent measurement work.

[0099] The scanning module 2 is used to obtain point cloud data of the building facade in real time through a scanner, and to perform real-time display and dynamic cropping through a WebGL rendering engine.

[0100] Among them, the scanner uses Leica brand RTC360, BLK360, MS60 and other devices. The scanning module 2 includes a measurement program and a 3D model display module set on the user end. The measurement program and the 3D model display module use the WebSocket network protocol for communication, and the 3D model display module is embedded with a WebGL rendering engine. The scanning module 2 is set on the Pad end, which is convenient for users to view and set parameters on the move. The rendering and cropping of the point cloud data are displayed in real time on the Pad end, ensuring that users can immediately observe the scanning progress and effect.

[0101] Specifically, the scanning module 2 includes: a point cloud image acquisition submodule and a cropping submodule.

[0102] The point cloud acquisition submodule is used to control the scanner to scan, obtain the scanner's real-time point cloud map, and transmit it to the WebGL-based 3D model display module through the WebSocket network protocol for real-time display in the 3D model display module.

[0103] The cropping submodule is used to crop the real-time point cloud image, crop the building to be measured or the wall to be measured, and pass the cropped area to the main program. The main program crops the point cloud according to the area to obtain the point cloud area to be measured.

[0104] Because the scanner scans 360 degrees, it will scan other buildings. Therefore, in the 3D model display module on the Pad, users can cut out the facade point cloud area to be measured, that is, the wall to be measured, through a simple operation interface according to actual needs. Specifically, select the area to be measured or select the area not to be measured in the 3D editor for cutting to obtain the point cloud area to be measured.

[0105] The semantic segmentation module 3 is used to segment the point cloud voxel to segment the original meaningless discrete points into point clouds such as walls and ground with meaningful information.

[0106] In the original point cloud data, the original meaningless discrete points are just a set of coordinate data in three-dimensional space without any specific semantic information. For example, in the point cloud data of the building facade, the original discrete points are just a bunch of points in space, and it is impossible to directly distinguish which points belong to the wall, which belong to the ground, or other building structures. In the semantic segmentation module 3, the self-developed point cloud semantic segmentation algorithm can automatically identify and extract key elements such as walls and window openings. Subsequently, the intelligent measurement algorithm will automatically calculate key parameters such as the verticality and flatness of the facade.

[0107] Specifically, the semantic segmentation module 3 calls the point cloud intelligent measurement algorithm to perform automatic measurement of the point cloud. The 3D model display module is an independent process based on three.js. The 3D model display module on the user side communicates with the measurement program using the WebSocket communication protocol. The specific communication process is as follows: Figure 3 The measurement program on the user side is responsible for initiating operations and transmitting data, while the 3D model display module is responsible for receiving data, processing operations and feeding back results. The two interact through status information and data transmission to ensure the smooth progress of the entire process.

[0108] Specifically, the semantic segmentation module 3 includes: a voxel establishment submodule, a supervoxel seed generation submodule, a supervoxel segmentation submodule and a point cloud classification submodule.

[0109] The voxel establishment submodule is used to set the point cloud voxel resolution, use the octree structure to divide the input point cloud data into equal parts, and calculate the normal vector, center point, standard deviation and other information of each voxel.

[0110] A point cloud voxel is a cube of a set size, and the set point cloud voxel resolution is 200*200*200. The normal vector of a voxel can determine the direction of the geometric surface where the voxel is located. The center point coordinates of the voxel represent the position of the voxel in space. The standard deviation can reflect the degree of discreteness of the points within the voxel and is used to determine the consistency of the voxel and other characteristics.

[0111] The supervoxel seed generation submodule is used to uniformly generate supervoxel seeds based on the divided voxels.

[0112] The supervoxel seeds are selected from the constructed point cloud voxels. To ensure that the seeds are not selected on the edge of the wall or on non-wall surfaces, the standard deviation of the selected supervoxel seeds is less than 10 mm.

[0113] The location and number of these supervoxel seeds determine the granularity of the subsequent supervoxel segmentation. If the number of supervoxel seeds is large, the supervoxels segmented subsequently will be smaller and finer, and the granularity of the supervoxel segmentation will be finer; conversely, if the number of supervoxel seeds is small, the supervoxels will be larger and coarser.

[0114] The supervoxel segmentation submodule is used to perform a breadth-first search (BFS) starting from the generated supervoxel seed, clustering the adjacent voxels according to the feature distance and normal vector. The voxel surface distance of the same wall is less than 40 mm and the normal vector angle is less than 5°.

[0115] The point cloud classification submodule is used to perform semantic analysis on the clustered point cloud groups, and distinguish them into wall and non-wall surfaces using the spatial position and normal vector of the point cloud. The judgment condition for a wall surface is that the angle with the vertical line is less than 5°, and the judgment condition for a non-wall surface (i.e., floor or ceiling) is that the angle with the horizontal plane is less than 5°.

[0116] The automatic measurement module 4 is used to simulate a manual ruler to fully measure the verticality and flatness of the facade.

[0117] By acquiring a large amount of point cloud information of the facade and analyzing the positional relationship of the point cloud in space, we can determine whether the wall is flat and vertical. Full measurement means measuring every part of the facade without missing any part. This is achieved by systematically analyzing the point cloud data of the entire facade.

[0118] Specifically, the automatic measurement module 4 includes: a point cloud stratification submodule, a stratified burst point submodule and a vertical ruler submodule.

[0119] The point cloud stratification submodule is used to stratify the facade point cloud into different floor point clouds according to the floor information of the project.

[0120] The layered burst point submodule is used to generate burst point maps for each floor according to the divided floors, and merge the layered burst point maps into a complete map.

[0121] The burst point map can graphically display the distribution, density and other characteristics of the facade point cloud of the floor. The burst point maps generated on each floor are integrated together to form a complete map. The advantage of this is that the overall difference of the facade point clouds between different floors can be compared from a macro perspective, which is convenient for a comprehensive grasp of the facade situation of the entire building, and also helps to plan the measurement area based on the overall perspective in the future.

[0122] The vertical ruler module is used to divide the facade into areas to be measured using window openings or set sub-wall widths as dividing lines, and to measure the flatness and verticality of each floor.

[0123] During the segmentation process, the window openings on the facade are first identified and clustered. Then, according to the longitudinal space where the door and window openings are located, the facade is segmented into sub-walls. Due to occlusion or obstacles, some point cloud holes may be mistakenly identified as window holes. Clustering is performed based on the consistency of the window holes on each floor to remove the holes that are detected incorrectly.

[0124] like Figure 4-5 As shown in the figure, the facade is divided into three large areas to be measured according to the position of the window opening (the position of the box in the picture), namely the left, middle and right walls. Since the wall in the middle is relatively wide, considering its large area and for more precise and accurate measurement, it is further divided into two sub-walls. Through such reasonable area division, just like manually determining each section of the wall to be measured with a ruler, the measurement work is more organized and detailed.

[0125] The storage module 5 is used to automatically upload the measurement results to the cloud system for storage and analysis after the full measurement is completed.

[0126] Users can view historical measurement results in the cloud at any time for comparison and analysis, providing data support for subsequent maintenance, renovation or design.

[0127] In addition, the point cloud-based real-time facade detection system of the present invention further includes:

[0128] The facade report generation module 6 is used to automatically export reports such as the wall rectification project consumption of the facade building, the layered rectification project consumption, the wall vertical level data results, and the layered vertical level data.

[0129] The foregoing description of specific exemplary embodiments of the present invention is for the purpose of illustration and demonstration. These descriptions are not intended to limit the present invention to the precise form disclosed, and it is clear that many changes and variations can be made based on the above teachings. The purpose of selecting and describing the exemplary embodiments is to explain the specific principles of the present invention and its practical application, so that those skilled in the art can realize and utilize various different exemplary embodiments of the present invention and various different selections and changes. The scope of the present invention is intended to be limited by the claims and their equivalents.

Claims

1. A real-time facade detection method based on point cloud, characterized in that: The following steps are involved: S1: Input facade project data into the cloud; S2: The point cloud data of the building facade is obtained in real time through the scanner, and the data is displayed and dynamically cropped in real time through the WebGL rendering engine; S3: Segment the wall and ground with meaningful information through point cloud voxel segmentation; S4: simulates manual ruler to measure verticality and flatness of the facade; S5: The measurement results will be automatically uploaded to the cloud system for storage and analysis.

2. The real-time facade detection method based on point cloud according to claim 1, characterized in that: Step S2 includes the following sub-steps: S201: Control the scanner to scan, obtain the real-time point cloud image of the scanner, and transmit it to the 3D model display module based on WebGL through the WebSocket network protocol, and display it in real time in the 3D model display module; S202: cropping the real-time point cloud image to crop out the building to be measured or the wall to be measured, and feeding back the cropped area result to the main program, which crops according to the area to obtain the point cloud area to be measured.

3. The real-time facade detection method based on point cloud according to claim 2, characterized in that: Step S3 includes the following sub-steps: S301: setting the point cloud voxel resolution, using the octree structure to equally divide the input point cloud data, and calculating the normal vector, center point and standard deviation of each voxel; S302: uniformly generating supervoxel seeds based on the equally divided voxels; S303: Taking the generated supervoxel seed as the starting point, a breadth-first search is performed to cluster adjacent voxels according to feature distance and normal vector. The voxel surface distance of the same wall is less than 40 mm and the normal vector angle is less than 5°. S304: Perform semantic analysis on the clustered point cloud groups, and use the spatial position and normal vector of the point cloud to distinguish between wall and non-wall surfaces; the judgment condition for a wall surface is that the angle with the vertical line is less than 5°, and the judgment condition for a non-wall surface is that the angle with the horizontal plane is less than 5°.

4. The real-time facade detection method based on point cloud according to claim 3, characterized in that: Step S4 includes the following sub-steps: S401: Layering the facade point cloud into different floor point clouds according to the floor information of the project; S402: generating a burst point map for each floor according to the divided floors, and merging the burst point maps of each floor; S403: Using the window opening or the set sub-wall width as the dividing line, divide the facade into areas to be measured, and measure the flatness and verticality of each floor.

5. The real-time facade detection method based on point cloud according to claim 4, characterized in that: In step S403, clustering is performed based on the consistency of the window holes on each floor, and the wrongly detected window holes are removed.

6. A real-time facade detection system based on point cloud, characterized in that: include: Information entry module, scanning module, semantic segmentation module, automatic measurement module and storage module; The information input module is used to input facade project data in the cloud; The scanning module is used to obtain point cloud data of the building facade in real time through a scanner, and to display and dynamically crop the point cloud data in real time through a WebGL rendering engine; The semantic segmentation module is used to segment the wall and the ground with meaningful information through point cloud voxel segmentation; The automatic measurement module is used to simulate a manual ruler to fully measure the verticality and flatness of the facade; The storage module is used to automatically upload the measurement results to the cloud system for storage and analysis.

7. The point cloud-based real-time facade detection system according to claim 6, characterized in that: The scanning module includes: a point cloud acquisition submodule and a cropping submodule; The point cloud image acquisition submodule is used to control the scanner to scan, obtain the real-time point cloud image of the scanner, and transmit it to the WebGL-based three-dimensional model display module through the WebSocket network protocol, and display it in real time in the three-dimensional model display module; The cropping submodule is used to crop the real-time point cloud image, crop the building to be measured or the wall to be measured, and feed the cropping area result back to the main program. The main program crops according to the area to obtain the point cloud area to be measured.

8. The point cloud-based real-time facade detection system according to claim 7, characterized in that: The semantic segmentation module includes: a voxel establishment submodule, a supervoxel seed generation submodule, a supervoxel segmentation submodule and a point cloud classification submodule; The voxel establishment submodule is used to set the point cloud voxel resolution, use the octree structure to divide the input point cloud data into equal parts, and calculate the normal vector, center point and standard deviation of each voxel; The supervoxel seed generation submodule is used to uniformly generate supervoxel seeds based on the equally divided voxels; The supervoxel segmentation submodule is used to perform a breadth-first search starting from the generated supervoxel seed, clustering adjacent voxels according to feature distance and normal vector, and the voxel surface distance of the same wall is less than 40 mm and the normal vector angle is less than 5°; The point cloud classification submodule is used to perform semantic analysis on the clustered point cloud group, and distinguish between wall and non-wall surfaces using the point cloud spatial position and normal vector; wherein, the judgment condition for a wall surface is that the angle with the vertical line is less than 5°, and the judgment condition for a non-wall surface is that the angle with the horizontal plane is less than 5°.

9. The point cloud-based real-time facade detection system according to claim 8, characterized in that: The automatic measurement module includes: a point cloud stratification submodule, a stratified burst point submodule and a vertical ruler submodule; The point cloud stratification submodule is used to stratify the facade point cloud into different floor point clouds according to the floor information of the project; The hierarchical burst point submodule is used to generate burst point maps for each floor according to the divided floors, and merge the burst point maps of each floor; The verticality ruler module is used to divide the facade into areas to be measured using window openings or a set sub-wall width as a dividing line, and to measure the flatness and verticality of each floor.

10. The point cloud-based real-time facade detection system according to claim 9, characterized in that: In the vertical ruler module, the window holes are clustered according to their consistency on each floor, and the holes detected incorrectly are removed.

Images