Building coating method based on point cloud visual imaging technology

Abstract

The invention relates to the technical field of architectural coating, and discloses an architectural coating method based on a point cloud visual imaging technology, which comprises the following steps: S1, acquiring original point cloud data; s2, constructing a final point cloud model; s3, preparing an unmanned aerial vehicle system; s4, positioning an unqualified area; s5, targeted coating operation is carried out; and S6, operation verification and ending are carried out. According to the method, full-process accurate control of collection, modeling, comparison and adjustment is achieved through the point cloud vision imaging technology, the coating thickness deviation is controlled within + / -0.1 mm through real-time point cloud point-by-point comparison and dynamic adjustment in operation, the surface flatness error is smaller than or equal to 0.3 mm, the quality problems that traditional manual coating precision is affected by experience, coatings are uneven, coating missing is caused and the like are solved, and the coating quality is improved. The stability and the consistency of the coating quality are greatly improved; and secondly, the unmanned aerial vehicle can autonomously plan an operation path based on the point cloud model, targeted coating is realized in combination with real-time dynamic point cloud comparison, and the operation efficiency is remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of architectural coating technology, and in particular to an architectural coating method based on point cloud visual imaging technology. Background Technology

[0002] As a key part of building construction and maintenance, architectural coating not only affects the aesthetic appearance of buildings, but also plays an important role in the durability, corrosion resistance and safety of building structures. With the acceleration of urbanization and the intelligent upgrading of the construction industry, the traditional coating mode that relies on manual operation is gradually transforming towards automation and digitalization.

[0003] In the current stage of exterior wall painting operations, workers generally use hanging baskets to carry out the exterior wall painting work, while steel structure painting operations are generally carried out by workers using scaffolding and other tools. With the development of automation technology, some fields have begun to use robotic arms or drones for painting operations.

[0004] However, the above construction methods have the following drawbacks: First, traditional manual painting relies on high-altitude work platforms, which poses safety risks such as falls from heights and electric shocks. Moreover, the accuracy of manual operation is affected by experience, which can easily lead to quality problems such as uneven coating thickness and missed coatings. Second, existing automated painting equipment is mostly based on preset paths and lacks the ability to perceive and dynamically adjust the actual surface morphology of buildings in real time. Especially in the scenario of repairing old buildings, it is difficult to accurately identify the damaged areas of the coating, resulting in material waste and low efficiency. Summary of the Invention

[0005] To overcome the above shortcomings, this invention provides a building coating method based on point cloud visual imaging technology, which solves the quality problems of uneven coating thickness and missed coating in manual coating, as well as the problem that existing automated coating equipment lacks the ability to perceive and dynamically adjust the actual surface morphology of buildings in real time.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a building coating method based on point cloud visual imaging technology, comprising the following steps:

[0007] Step S1: Use a point cloud acquisition system that includes an imaging device and a GNSS module to perform a three-dimensional scan of the outer surface of the building or component to obtain raw point cloud data containing spatial coordinates, geometric shape and surface features;

[0008] Step S2: Perform noise filtering, hole filling and coordinate calibration preprocessing on the original point cloud data to generate a basic model. Then, add preset coating parameters to the basic model to construct the final point cloud model.

[0009] Step S3: Import the final point cloud model into a drone equipped with a point cloud acquisition subsystem and a paint spraying device, and complete the coordinate system matching and calibration between the drone and the model;

[0010] Step S4: After the drone takes off, it acquires dynamic point cloud data of the work area in real time through the point cloud acquisition subsystem, compares it point by point with the final point cloud model, and locates areas that do not meet the standards.

[0011] Step S5: The drone controls the paint spraying device to adjust the spray flow rate, angle and movement path according to the spatial coordinates of the non-compliant area and the preset coating parameters, and performs targeted coating operations.

[0012] Step S6: During the painting process, continuously collect dynamic point cloud data and compare it with the final point cloud model until all areas meet the preset standards, then stop the operation and retrieve the drone.

[0013] As a further description of the above technical solution: the imaging device in step S1 includes an industrial camera and a laser scanning rangefinder, which work synchronously. The ranging accuracy of the laser scanning rangefinder is ≤ ±2mm, and the image resolution of the industrial camera is ≥ 12 million pixels to ensure the accuracy of the three-dimensional reconstruction of the original point cloud data.

[0014] As a further description of the above technical solution: In step S2, a statistical filtering algorithm is used to remove outlier noise points. The core steps are as follows:

[0015] (1) Calculate the target point around Average distance between neighboring points: ,(Pick (Adapted to laser scanning accuracy)

[0016] (2) Calculate the standard deviation of the average distance between all points: ,(in (where N is the global average distance and N is the total number of points).

[0017] (3) Elimination Noise points.

[0018] As a further description of the above technical solution: In step S2, the void filling adopts a region-growing-based algorithm, the core steps of which are:

[0019] a. Select the curvature of the cavity edge Using the point as the seed point, the curvature is calculated as follows: ,( (Eigenvalues ​​of the neighborhood covariance matrix).

[0020] b. Expand the growth region according to the condition of "Euclidean distance < 1 mm and the angle between the normal vectors < 15°";

[0021] c. Use triangulation to fill the holes. The coordinates of the missing points are: , ( (The vertex of the triangular facet at the edge of the hole).

[0022] As a further description of the above technical solution: In step S2, coordinate calibration is achieved by aligning the local coordinate system with the global coordinate system using GNSS control points. The conversion formula is as follows: (R is the rotation matrix, T is the translation vector, solved using the least squares method);

[0023] The method for adding the preset coating parameters in step S2 is as follows: in the three-dimensional coordinate system of the basic model, the surface points of different regions are assigned corresponding coating thickness values ​​and coating type identifiers to form a final point cloud model with coating attributes.

[0024] As a further description of the above technical solution: the specific method of coordinate system matching and calibration in step S3 is as follows: the UAV obtains its own real-time position through the GNSS module and associates it with the GNSS coordinates of the final point cloud model, so that the spatial positioning error between the two is controlled within ±5cm.

[0025] As a further description of the above technical solution: the point-by-point comparison in step S4 includes geometric shape comparison and coating property comparison:

[0026] Geometric shape comparison (flatness identification): Calculate the Euclidean distance between dynamic points and model points: (A deviation > 0.3mm is considered substandard).

[0027] Coating property comparison (thickness identification): Coating thickness fitting based on laser echo intensity: ( This is the actual thickness. Echo intensity (Calibration factor), thickness deviation: (A deviation greater than 0.1mm is considered substandard); the difference threshold can be dynamically set according to the coating accuracy requirements.

[0028] As a further description of the above technical solution: the paint spraying device in step S5 includes: a corrosion-resistant storage unit that can hold at least 5L of paint; an atomizing spraying component with 360° rotation function, the diameter of the atomized particles can be adjusted in the range of 50 to 200um; and a flow control module linked to point cloud data, with a flow adjustment accuracy of ±0.1L / min.

[0029] Spraying flow rate calculation: Based on coating thickness and drone speed, dynamically adjusted, the formula is as follows: , ( For flow rate (L / min), The preset thickness (m) is used. The speed of the drone is (m / s). The spray width is in meters. The density of the coating (kg / m³) (for utilization rate)

[0030] Path planning: Generate an S-shaped path, the Y-coordinate of the k-th path is: , ( For path spacing, (Minimum Y coordinate of the region).

[0031] As a further description of the above technical solution: when the object to be painted is an old building, step S4 also includes a comparison step between the original point cloud data and the historical painting model, using the ICP iterative nearest point algorithm to achieve registration, with the objective function being: ,( For the current point cloud, (Based on historical models), iterates until the error is <0.1mm to quickly locate areas where the coating has peeled off, cracked, or aged, enabling repair coating only for damaged areas.

[0032] As a further description of the above technical solution: the preset standards in step S6 include flatness deviation ≤ 0.3mm and coating thickness deviation ≤ ±0.1mm, and are determined by the overall compliance rate. , ( The number of points that did not meet the standards. (The total number of points is used to determine whether a compliance rate of ≥99.9% is considered qualified). After the drone is recovered, the process also includes generating a coating quality report: the report includes a comparison diagram of the actual coating point cloud model and the preset model, coating thickness detection data for each area, and compliance rate statistics. The report can be stored in a cloud database for subsequent queries.

[0033] The present invention has the following beneficial effects:

[0034] 1. In this invention, point cloud visual imaging technology is used to achieve precise control of the entire process of "collection, modeling, comparison, and adjustment". During the operation, point cloud comparison and dynamic adjustment in real time ensure that the coating thickness deviation is controlled within ±0.1mm and the surface flatness error is ≤0.3mm. This solves the quality problems of traditional manual coating, such as the influence of experience on accuracy, uneven coating, and missed coating, and significantly improves the stability and consistency of coating quality. Secondly, the use of drones to replace the traditional manual high-altitude operation mode fundamentally avoids the safety risks of falling from heights and mechanical injuries. At the same time, it avoids the health problems such as skin diseases and respiratory damage caused by direct contact with paint and spraying dust, which significantly improves the safety of building coating operations.

[0035] 2. In this invention, the drone can autonomously plan the operation path based on the point cloud model and achieve targeted coating by combining real-time dynamic point cloud comparison. Compared with traditional manual and pre-set path automated equipment, the operation efficiency is significantly improved. Especially in the scenario of old building repair, by comparing the original point cloud data with the historical coating model, the damaged areas such as coating peeling and cracking can be quickly located, and repair work can be performed only on the target parts. Compared with overall recoating, material waste is reduced and construction costs are significantly reduced. Attached Figure Description

[0036] Figure 1 This is a schematic flowchart of a building coating method based on point cloud visual imaging technology according to the present invention.

[0037] Figure 2 This is a flowchart illustrating a building coating method based on point cloud visual imaging technology according to the present invention. Detailed Implementation

[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] Example 1, Reference Figure 1 and Figure 2 The present invention provides an embodiment of a building coating method based on point cloud visual imaging technology, comprising the following steps:

[0040] Step S1: Raw point cloud data acquisition uses a point cloud acquisition system that includes an imaging device and a GNSS module to perform 3D scanning of the outer surface of a building or component. The imaging device includes an industrial camera and a laser scanning rangefinder, which work synchronously. The ranging accuracy of the laser scanning rangefinder is ≤±2mm, and the image resolution of the industrial camera is ≥12 million pixels. The GNSS module is used to record the spatial coordinates of the scanned points, and finally obtains raw point cloud data containing spatial coordinates, geometric shape, and surface features (such as flatness and texture).

[0041] Step S2: Preprocessing of raw point cloud data for final point cloud model construction: using statistical filtering algorithms (taking...) 5% of noise points are removed, and holes caused by scaffolding obstruction are repaired using a region-growing-based hole-filling algorithm (curvature threshold 0.05). Then, global calibration is performed using GNSS coordinates to generate a basic 3D model of the building's exterior surface. Preset coating parameters are added to the basic model: in the 3D coordinate system, surface points in different regions are assigned corresponding coating thickness values ​​(range 0.1-5mm) and coating type identifiers (such as anti-corrosion coating, decorative coating) to form a final point cloud model with coating attributes.

[0042] Step S3: The UAV system prepares to import the final point cloud model into the UAV equipped with a point cloud acquisition subsystem and a paint spraying device; the parameters of the UAV's point cloud acquisition subsystem are consistent with those of the point cloud acquisition system in Step S1 to ensure data compatibility; the UAV's real-time position is associated with the GNSS coordinates of the final point cloud model through the GNSS module to complete the coordinate system matching calibration, so that the spatial positioning error between the two is controlled within ±5cm.

[0043] Step S4: After the drone takes off, the dynamic point cloud data of the work area is acquired in real time through the point cloud acquisition subsystem and compared point by point with the final point cloud model: geometric shape comparison (judgment threshold 0.3mm) identifies flatness deviation areas, and coating attribute comparison (judgment threshold 0.1mm) identifies areas with insufficient thickness; the difference threshold is dynamically set according to the coating accuracy requirements, and finally all substandard areas (including uncoated areas, areas with insufficient thickness, or areas with damaged coating) are located.

[0044] Step S5: The targeted coating operation drone controls the paint spraying device to perform the coating operation based on the spatial coordinates of the substandard areas and preset coating parameters. The paint spraying device includes: a corrosion-resistant storage unit that can hold 5L of paint; an atomizing spray component with 360° rotation function, and the atomized particle diameter can be adjusted within the range of 50-200um; a flow control module linked to point cloud data, with a flow adjustment accuracy of ±0.1L / min; and a spraying flow formula. Calculate the target flow rate, adjust the jet flow rate, angle, and S-shaped movement path of the drone to ensure that the coating thickness and material of the substandard areas meet the preset requirements;

[0045] Step S6: During the operation verification and final coating process, the UAV continuously collects dynamic point cloud data and compares it with the final point cloud model until all areas meet the preset standards (surface flatness deviation ≤ 0.3mm, coating thickness deviation ≤ ±0.1mm, compliance rate ≥ 99.9%). The operation is then stopped and the UAV is retrieved. After retrieval, a coating quality report is generated, which includes a comparison diagram of the actual coating point cloud model and the preset model, coating thickness detection data for each area, and compliance rate statistics. The report is stored in a cloud database for subsequent query and traceability.

[0046] For the repair of old buildings, step S4 also includes comparing the original point cloud data with the historical coating model (such as the coating point cloud model when the building was completed) using the ICP algorithm, iterating until the error is <0.1mm, quickly locating the areas where the coating has peeled off, cracked or aged, and realizing repair coating only for the damaged parts.

[0047] Example 2: Exterior wall coating of newly constructed concrete buildings

[0048] Step S1: Using a point cloud acquisition system (including a 12-megapixel industrial camera, a laser scanning rangefinder with a ranging accuracy of ±1mm and a GNSS module), a 3D scan of the exterior wall of the newly built 6-story concrete building is performed to obtain raw point cloud data with a resolution of 0.5mm / point, including the spatial coordinates and surface flatness information of each area of ​​the exterior wall.

[0049] Step S2: Preprocess the raw point cloud data: through statistical filtering ( Remove 5% of noise points and fill the voids caused by scaffolding obstruction using the region growing method; add coating parameters to the base model, setting a wear-resistant coating with a thickness of 2mm for the first to third layers of the exterior walls (areas prone to collision), and setting a decorative coating with a thickness of 1mm for the fourth to sixth layers, to generate the final point cloud model.

[0050] Step S3: Import the final point cloud model into a quadcopter drone (equipped with the same type of point cloud acquisition subsystem and paint spraying device), and control the coordinate system error between the drone and the model within ±3cm through GNSS calibration;

[0051] Step S4: After the drone takes off, it cruises at a speed of 0.5 m / s, collects dynamic point cloud data in real time and compares it with the final model. In the initial state, all areas are "unpainted areas" (areas that do not meet the standards).

[0052] Step S5: Drone-controlled spraying device: Based on the flow rate formula, for the first to third layer areas ( Adjust the atomized particle diameter to 100µm and the flow rate to 0.8L / min; for the four to six layers ( Adjust the atomized particle diameter to 80µm and the flow rate to 0.5L / min, and spray in an "S" shaped path (0.18m spacing);

[0053] Step S6: Continuously compare the dynamic point cloud data with the final model. After 2 hours, the coating thickness deviation of all areas is ≤ ±0.08mm, the flatness error is ≤ 0.2mm, the compliance rate is 100%, and the operation is stopped. After the drone is recovered, a quality report is generated and uploaded to the cloud for storage.

[0054] Example 3: Coating Repair of Old Steel Structure Factory Buildings

[0055] Step S1: Use a point cloud acquisition system to scan the outer surface of the steel structure factory building that has been in use for 10 years to obtain raw point cloud data containing information on areas with rust and coating peeling.

[0056] Step S2: After preprocessing, the original data is compared with the historical coating point cloud model (2mm thick anti-corrosion coating) at the time of factory completion using the ICP algorithm (50 iterations, error <0.1mm) to generate the final point cloud model with "damaged area" markings.

[0057] Step S3: Import the model into the drone and complete the coordinate calibration, with the error controlled within ±4cm;

[0058] Step S4: Real-time comparison revealed that 30% of the coating on the bottom of the factory columns had peeled off (thickness 0mm), and there were 5 rust bulges at the beam connection (flatness deviation 1.2mm), all of which were determined to be substandard areas;

[0059] Step S5: The drone targets the area at the base of the column ( Adjust the spray angle to 45° and the flow rate to 1.0L / min; for the bulging area of ​​the crossbeam, first fill the depression with a low flow rate (0.3L / min), then spray according to the standard thickness to finally complete the repair;

[0060] Step S6: Verification shows that the thickness deviation of the repaired area is ≤ ±0.1mm, the flatness error is ≤ 0.3mm, the pass rate is 99.95%, and a quality report containing before and after comparison images is generated. The work efficiency is 60% higher than that of overall recoating, and the paint consumption is reduced by 70%.

[0061] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for architectural painting based on point cloud vision imaging technology, characterized in that: The method comprises the following steps: Step S1: using a point cloud acquisition system comprising an imaging device and a GNSS module, performing three-dimensional scanning on the outer surface of a building or component to obtain original point cloud data containing spatial coordinates, geometric shapes and surface characteristics; Step S2: performing noise filtering, hole filling and coordinate calibration preprocessing on the original point cloud data to generate a basic model, and then adding preset coating parameters to the basic model to construct a final point cloud model; Step S3: importing the final point cloud model into a UAV equipped with a point cloud acquisition subsystem and a paint spraying device, and completing the coordinate system matching and calibration of the UAV and the model; Step S4: after the UAV takes off, real-time acquisition of dynamic point cloud data of the work area by the point cloud acquisition subsystem, point-by-point comparison with the final point cloud model, and positioning of the non-compliant area; Step S5: the UAV controls the paint spraying device to adjust the spraying flow, angle and movement path according to the spatial coordinates of the non-compliant area and the preset coating parameters, and performs targeted coating work; Step S6: continuously collecting dynamic point cloud data during the coating process and comparing it with the final point cloud model until all areas meet the preset standard, stopping the work and recovering the UAV.

2. The building painting method based on point cloud vision imaging technology according to claim 1, characterized in that: The imaging device in step S1 includes an industrial camera and a laser scanning range finder, both of which work synchronously, wherein the ranging accuracy of the laser scanning range finder is ≤±2mm, and the image resolution of the industrial camera is ≥1200 million pixels, to ensure the three-dimensional reconstruction accuracy of the original point cloud data.

3. The building painting method based on point cloud vision imaging technology according to claim 1, characterized in that: In step S2, statistical filtering algorithm is used to remove outlier noise points, and the core step is: (1) Calculate target point Surroundings Average distance of neighboring points: , (take , adapt laser scanning accuracy); (2) Calculate the standard deviation of the average distance of all points: , where D is the global average distance and N is the total number of points. ​ (3) reject noise points.

4. The building painting method based on point cloud vision imaging technology according to claim 1, characterized in that: In step S2, the hole filling algorithm is based on region growing, and the core step is: a. Select the curvature of the cavity edge Using the point as the seed point, the curvature is calculated as follows: ,( (Eigenvalues ​​of the neighborhood covariance matrix). b. expand the growth region according to the condition of "Euclidean distance <1mm and normal vector angle <15°"; c. The holes are filled with triangulation, and the missing point coordinates are: , (the hole edge triangle vertex).

5. The method of claim 1, wherein the method is a method of architectural painting based on point cloud vision imaging technology. The coordinate calibration in the step S2: the local coordinate system is aligned with the global coordinate system through the GNSS control point, and the conversion formula is: , (R is a rotation matrix, T is a translation vector, and is solved through the least square method); In step S2, the additional method of preset coating parameters is to assign corresponding coating thickness values and paint type identifiers to surface points in different regions in the three-dimensional coordinate system of the basic model, forming a final point cloud model with coating attributes.

6. The method of claim 1, wherein the method is a method of architectural painting based on point cloud vision imaging technology. In step S3, the specific method of coordinate system matching and calibration is that the UAV obtains its real-time position through the GNSS module, and associates it with the GNSS coordinates of the final point cloud model, so that the spatial positioning error of the two is controlled within ±5cm.

7. The building painting method based on point cloud vision imaging technology according to claim 1, characterized in that: The point-by-point comparison in step S4 includes geometric shape comparison and coating attribute comparison: Geometric shape comparison (flatness recognition): calculate the Euclidean distance between dynamic points and model points: , (deviation > 0.3 mm is determined as unqualified); Coating property comparison (thickness identification): fitting coating thickness based on laser echo intensity: ( is the actual thickness, is the echo intensity, is the calibration coefficient), thickness deviation: , (deviation> 0.1mm is determined as unqualified); the difference threshold can be dynamically set according to the coating accuracy requirement.

8. The building painting method based on point cloud vision imaging technology according to claim 1, characterized in that: The paint spraying device in step S5 includes: a corrosion-resistant storage unit that can accommodate at least 5L of paint; an atomizing spraying assembly with 360° rotation function, the atomizing particle diameter can be adjusted within the range of 50-200um; a flow control module linked with the point cloud data, the flow adjustment accuracy is ±0.1L / min; Spraying flow rate calculation: Based on coating thickness and drone speed, dynamically adjusted, the formula is as follows: , ( For flow rate (L / min), The preset thickness (m) is used. The speed of the drone is (m / s). The spray width is in meters. The density of the coating (kg / m³) (for utilization rate) Path planning: generate S-shaped path, the kth path Y coordinate is: , is the path spacing, is the minimum Y coordinate of the region.

9. The method of claim 1, wherein the method is a method of architectural painting based on point cloud vision imaging technology. When the painting object is an old building, the step S4 further includes a comparison step of the original point cloud data and the historical painting model, the ICP iterative closest point algorithm is adopted to realize registration, and the objective function is: ,( is the current point cloud, is the historical model), iteration is performed until the error is less than 0.1mm, to quickly locate the area of coating peeling, cracking or aging, and realize repair painting only for the damaged parts.

10. The method of claim 1, wherein the method is a method of architectural painting based on point cloud vision imaging technology. The preset standard in the step S6 includes flatness deviation ≤ 0.3 mm, coating thickness deviation ≤ ± 0.1 mm, and the whole standard reaching rate is determined: , is the number of non-standard points, is the total number of points, and the standard reaching rate ≥ 99.9% is determined as qualified); after the unmanned plane is recovered, a step of generating a coating quality report is further included: the report includes a comparison graph of the actual coating point cloud model and the preset model, coating thickness detection data of each region and standard reaching rate statistics, and the report can be stored in a cloud database for subsequent query.

Images