Building outer wall surface structure safety assessment method based on unmanned aerial vehicle data acquisition

Abstract

The invention discloses a building outer wall surface structure safety assessment method based on unmanned aerial vehicle data acquisition, and the method comprises the steps: determining a target outer wall surface, and collecting the infrared image and three-dimensional point cloud data of the target outer wall surface through an unmanned aerial vehicle; performing optimization processing on the three-dimensional point cloud data of the target outer wall surface to obtain an accurate three-dimensional point cloud image, and aligning the accurate three-dimensional point cloud image with the infrared image; performing graying processing on the infrared image, analyzing the hollowing area of the target outer wall surface by using the gray value of the pixel in the infrared gray image, and extracting the boundary line of the hollowing area and the area of the hollowing area; and according to the obtained point cloud data in the hollowing region, analyzing and evaluating the structural stability of the target outer wall surface by using the point cloud data and the area of the hollowing region. The method realizes efficient and accurate identification of hollowing and scientific evaluation of building outer wall surface structure safety, and provides powerful technical support for building safety maintenance.

Description

Technical Field

[0001] The present invention relates to the field of building exterior wall safety analysis, and in particular to a building exterior wall structural safety assessment method based on drone data collection. Background Art

[0002] With the development of urbanization, more and more high-rise buildings are being constructed in cities. To ensure the aesthetics of these buildings, construction companies often use decorative panels (ceramic tiles, stone, etc.) to beautify the exterior surfaces of buildings. However, as the building ages or the construction quality is poor, hollowing can occur between the decorative panels and the building walls. If the hollowing is not treated for a long time, the decorative panels may fall off, which not only damages the building's appearance but also endangers the safety of residents. However, the current hollowing detection of building exterior decorative panels is difficult and inefficient, often involving high-altitude operations and high operational risks. With the development of drone technology and intelligent technology, remote hollowing detection using drones carrying professional hollowing detection equipment has become possible. However, the existing technology lacks accurate and precise hollowing identification methods, making it impossible to effectively assess the safety of building exterior wall structures. Therefore, it is urgent to propose a building exterior wall structure safety assessment method based on drone data collection. Summary of the Invention

[0003] In response to the above-mentioned shortcomings of the existing technology, the present invention provides a method for assessing the structural safety of building exterior walls based on drone data collection, which uses infrared imaging technology and three-dimensional laser scanning technology to comprehensively identify and locate hollows on building exterior walls, and scientifically evaluate the structural safety of building exterior walls.

[0004] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is:

[0005] A method for assessing the safety of building exterior wall structures based on drone data collection is provided, which includes the following steps:

[0006] S1: Identify the target exterior wall surface and use a drone equipped with an infrared thermal imager and a 3D laser scanner to fly over the wall surface to collect infrared images and 3D point cloud data of the target exterior wall surface.

[0007] S2: Optimize the 3D point cloud data of the target exterior wall to obtain an accurate 3D point cloud map, and align the accurate 3D point cloud map with the infrared image;

[0008] S3: grayscale processing is performed on the infrared image, and the grayscale values of the pixels in the infrared grayscale image are used to analyze the hollow area of the target exterior wall, and the boundary line and area of the hollow area are extracted;

[0009] S4: Based on the acquired point cloud data within the hollow area, the structural stability of the target exterior wall is analyzed and evaluated using the point cloud data and the area of the hollow area.

[0010] Furthermore, step S2 specifically includes:

[0011] S21: Obtain all point cloud datasets A={Q1,Q2,…,Q n}, n is the number of point cloud sets, Q n is the nth point cloud set;

[0012] S22: Using the coordinate features of the point clouds in the point cloud set to select the two most similar point cloud set groups;

[0013]

[0014] Among them, Q i , Q j are any two point cloud sets in the point cloud dataset A, Q j As the reference point cloud, Q i is the target point cloud, T i is the coordinate feature of the point cloud in the target point cloud set, T j is the coordinate feature of the point cloud in the reference point cloud set, i is the number of the point cloud in the target point cloud set, j is the number of the point cloud in the target point cloud set, and I is the number of point clouds in the target point cloud set;

[0015] S23: Filter out the most similar point cloud group (Q e ,Q u ) then, according to the point cloud group (Q e ,Q u ) to perform point cloud matching based on the coordinate features of the point cloud; specifically:

[0016] S231: Based on point cloud Q e The coordinate feature T of any point cloud e in e , in Point Cloud Q u The point cloud u closest to the point cloud e is selected internally for closest point cloud matching;

[0017]

[0018] S232: Point Cloud Collection Q e Point cloud and point cloud set Q u After the point cloud in the nearest point cloud is matched, the intermediate coordinate feature T after the nearest point cloud matching is successful is calculated. e～u ;

[0019]

[0020] S233: The intermediate coordinate feature T e～u As the point cloud features after the two closest point clouds are matched, the point cloud set Q is generated e With Point Cloud Q u The point cloud set Q0 after point cloud matching is performed; the three-dimensional point cloud image formed by the point cloud set Q0 is used as the accurate three-dimensional point cloud image of the target exterior wall;

[0021] S24: Flattening the precise three-dimensional point cloud image to form a planar point cloud image, where each point cloud data in the planar point cloud image represents the relative height of a corresponding position on the target exterior wall surface;

[0022] S25: Select a typical area on the target exterior wall as an alignment reference area, obtain coordinate features corresponding to the point cloud data within the alignment reference area, and calculate the center coordinate T0 of the alignment reference area;

[0023]

[0024] Among them, m is the number of the point cloud in the alignment reference area, T m is the coordinate feature of point cloud m, M is the number of point clouds in the alignment reference area;

[0025] S26: Set the standard temperature t0 of the alignment reference area. According to the area of the alignment reference area, select an infrared pixel area with the same area as the alignment reference area on the infrared image, and make the average temperature within the pixel area equal to the standard temperature. The infrared pixel area satisfies the constraint conditions:

[0026]

[0027] Where B is the number of infrared pixels in the infrared pixel area, χ is the full level of the infrared pixel, b is the number of infrared pixels in the infrared pixel area, b is the infrared pixel set of the infrared pixel area, B is the infrared pixel set of the infrared image, S b is the area of the alignment reference region, t b is the temperature value corresponding to the infrared pixel;

[0028] S27: establishing a coordinate system identical to that of the plane point cloud image on the infrared image, obtaining coordinate features of the infrared pixels, and calculating the center coordinates T0′ of the infrared pixel area based on the coordinate features of the infrared pixels within the infrared pixel area;

[0029]

[0030] Where w is the number of pixels in the infrared pixel area, W is the number of pixels in the infrared pixel area, T w ′ is the coordinate feature of pixel w;

[0031] S28: According to the relative position between the center coordinate T0′ and the center coordinate T0, the plane point cloud image and the infrared image are moved so that the center of the infrared pixel area is aligned with the center of the alignment reference area, thereby aligning the plane point cloud image and the infrared image.

[0032] Furthermore, step S3 includes:

[0033] S31: Grayscale the infrared image to obtain an infrared grayscale image, set the standard pixel grayscale value H of the hollow center on the infrared grayscale image when the wall has hollows, and extract the grayscale value A of each pixel in the infrared grayscale image. x ′, x is the pixel number in the complete wall infrared grayscale image;

[0034] S32: Using the standard pixel grayscale value H and grayscale value A x ′, filter the hollow center pixel x0 on the target wall, the gray value of the hollow center pixel x0 satisfy H 阈值 The allowable error between the grayscale value of the hollow center pixel and the grayscale value of the standard pixel;

[0035] S33: Obtain all hollow center pixels x0 on the target wall, take the hollow center pixel x0 as the pixel reference, and convert the grayscale value of the pixel x0' adjacent to the pixel reference Compared with the gray value H, if Then the pixel x0′ is determined to be in the hollow area. Otherwise, the pixel x0′ is not in the hollow area and is regarded as the boundary pixel of the hollow area. 阈值 is the allowable error between the grayscale value of the pixel in the hollow area and the grayscale value of the standard pixel;

[0036] S34: After obtaining the pixel x0′ adjacent to the hollow pixel x0 and located in the hollow area, return to step S33, use the pixel x0′ located in the hollow area as the pixel reference, and continue to screen the pixels adjacent to the pixel w0′ and located in the hollow area;

[0037] S35: until all pixels in the hollow area around the hollow center pixel w0 are screened out, a complete hollow area on the infrared image is obtained, and the coordinate feature corresponding to the hollow center pixel w0 is used as the center of the hollow area;

[0038] S36: According to the coordinate characteristics of the boundary pixels in the infrared image, all boundary pixels are smoothly connected to obtain the boundary line of the hollow area, and the hollow area is displayed in the infrared image. The area S of the hollow area is calculated according to the number κ of pixels in all hollow areas:

[0039]

[0040] Where s is the area of the target wall corresponding to a single pixel, and γ is the number of pixels in the complete wall infrared grayscale image.

[0041] Furthermore, step S4 includes:

[0042] S41: acquiring point cloud data in an area corresponding to the hollow area on the plane point cloud map according to the hollow area in the infrared image, and calculating a protrusion height ν of the hollow area;

[0043]

[0044] Among them, c is the number of the point cloud data in the area corresponding to the hollow area, ν c is the relative height of the corresponding position of the point cloud data c, ν0 is the reference benchmark of the point cloud data, and C is the number of point cloud data in the area corresponding to the hollow area;

[0045] S42: Calculate the structural stability coefficient δ of the target outer wall where the hollow area is located based on the protruding height ν and area S of the hollow area d ;

[0046] δ d =α1exp(ν)+α2exp(S);

[0047] Where d is the number of the hollow area on the target outer wall, α1 and α2 are the weights of the impact of the protruding height and area of the hollow area on the structural stability, α1+α2=1;

[0048] S43: Setting the threshold δ for evaluating the wall structure stability 阈值 If δ d ≤δ 阈值 , then the structure of the hollow area d on the target outer wall is stable; if δ d >δ 阈值 , then the structure of the hollow area d on the target outer wall is unstable.

[0049] The beneficial effects of the present invention are as follows: the present invention combines infrared imaging technology with three-dimensional laser scanning technology to screen, locate and identify hollow areas on the exterior walls of buildings, and uses the area and protrusion height of the identified hollow areas to comprehensively evaluate the structural stability of the hollow areas, and analyze the structural safety of the hollow areas on the target exterior walls. Infrared imaging technology identifies hollow areas based on the different effects of the hollow areas on the exterior walls on absorbing thermal radiation, and three-dimensional laser scanning technology obtains texture data on the building wall to identify the protruding parts of the hollow areas. The combination of the two can not only identify the area of the hollow areas but also obtain the protrusion height of the hollow areas, thereby achieving efficient and accurate identification of hollows and scientific evaluation of the structural safety of the exterior walls of buildings, and providing strong technical support for building safety maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 Flowchart of the building exterior wall structural safety assessment method based on drone data collection. DETAILED DESCRIPTION

[0051] The specific embodiments of the present invention are described below to facilitate understanding of the present invention by those skilled in the art. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the appended claims, these changes are obvious, and all inventions and creations utilizing the concepts of the present invention are protected.

[0052] like Figure 1 As shown, a building exterior wall structural safety assessment method based on drone data collection includes the following steps:

[0053] S1: Identify the target exterior wall surface and fly a drone equipped with an infrared thermal imager and a 3D laser scanner over it to collect infrared images and 3D point cloud data. Infrared thermal imaging technology reveals hollow areas formed between the building's exterior wall and the decorative panels, utilizing the differences in thermal radiation absorption. 3D point cloud data captures texture features and determines their height. 3D laser scanning technology can then determine the protrusion height of the hollow area, which can be used to assess the structural stability of the hollow area.

[0054] S2: Optimize the 3D point cloud data of the target exterior wall to obtain an accurate 3D point cloud map, and align the accurate 3D point cloud map with the infrared image. Step S2 specifically includes:

[0055] S21: Obtain all point cloud datasets A={Q1,Q2,…,Q n}, n is the number of point cloud sets, Qn is the nth point cloud set;

[0056] S22: Using the coordinate features of the point clouds in the point cloud set to select the two most similar point cloud set groups;

[0057]

[0058] Among them, Q i , Q j are any two point cloud sets in the point cloud dataset A, Q j As the reference point cloud, Q i is the target point cloud, T i is the coordinate feature of the point cloud in the target point cloud set, T j is the coordinate feature of the point cloud in the reference point cloud set, i is the number of the point cloud in the target point cloud set, j is the number of the point cloud in the target point cloud set, and I is the number of point clouds in the target point cloud set;

[0059] S23: Filter out the most similar point cloud group (Q e ,Q u ) then, according to the point cloud group (Q e ,Q u ) to perform point cloud matching based on the coordinate features of the point cloud; specifically:

[0060] S231: Based on point cloud Q e The coordinate feature T of any point cloud e in e , in Point Cloud Q u The point cloud u closest to the point cloud e is selected internally for closest point cloud matching;

[0061]

[0062] S232: Point Cloud Collection Q e Point cloud and point cloud set Q u After the point cloud in the nearest point cloud is matched, the intermediate coordinate feature T after the nearest point cloud matching is successful is calculated. e～u ;

[0063]

[0064] S233: The intermediate coordinate feature T e～u As the point cloud features after the two closest point clouds are matched, the point cloud set Q is generated e With Point Cloud Q u The point cloud set Q0 after point cloud matching is performed; the three-dimensional point cloud image formed by the point cloud set Q0 is used as the accurate three-dimensional point cloud image of the target exterior wall;

[0065] S24: Flattening the precise three-dimensional point cloud image to form a planar point cloud image, where each point cloud data in the planar point cloud image represents the relative height of a corresponding position on the target exterior wall surface;

[0066] S25: Select a typical area on the target exterior wall as an alignment reference area, obtain coordinate features corresponding to the point cloud data within the alignment reference area, and calculate the center coordinate T0 of the alignment reference area;

[0067]

[0068] Among them, m is the number of the point cloud in the alignment reference area, T m is the coordinate feature of point cloud m, M is the number of point clouds in the alignment reference area;

[0069] S26: Set the standard temperature t0 of the alignment reference area. According to the area of the alignment reference area, select an infrared pixel area with the same area as the alignment reference area on the infrared image, and make the average temperature within the pixel area equal to the standard temperature. The infrared pixel area satisfies the constraint conditions:

[0070]

[0071] Where B is the number of infrared pixels in the infrared pixel area, χ is the full level of the infrared pixel, b is the number of infrared pixels in the infrared pixel area, b is the infrared pixel set of the infrared pixel area, B is the infrared pixel set of the infrared image, S b is the area of the alignment reference region, t b is the temperature value corresponding to the infrared pixel;

[0072] S27: Establishing the same coordinate system as the plane point cloud image on the infrared image, wherein each point cloud data in the plane point cloud image includes the relative height of the corresponding position on the target outer wall, obtaining the coordinate features of the infrared pixels, and calculating the center coordinate T′0 of the infrared pixel area based on the coordinate features of the infrared pixels in the infrared pixel area;

[0073]

[0074] Where w is the number of pixels in the infrared pixel area, W is the number of pixels in the infrared pixel area, T′ w is the coordinate feature of pixel w;

[0075] S28: According to the relative position between the center coordinate T′0 and the center coordinate T0, the plane point cloud image and the infrared image are moved so that the center of the infrared pixel area is aligned with the center of the alignment reference area, thereby aligning the plane point cloud image and the infrared image.

[0076] S3: grayscale the infrared image, analyze the hollow area of the target exterior wall using the grayscale value of the pixels in the infrared grayscale image, and extract the boundary line and area of the hollow area. Step S3 specifically includes:

[0077] S31: Grayscale the infrared image to obtain an infrared grayscale image, set the standard pixel grayscale value H of the hollow center on the infrared grayscale image when the wall has hollows, and extract the grayscale value A of each pixel in the infrared grayscale image. x ′, x is the pixel number in the complete wall infrared grayscale image;

[0078] S32: Using the standard pixel grayscale value H and grayscale value A x ′, filter the hollow center pixel x0 on the target wall, the gray value of the hollow center pixel x0 satisfy H 阈值 The allowable error between the grayscale value of the hollow center pixel and the grayscale value of the standard pixel;

[0079] S33: Obtain all hollow center pixels x0 on the target wall, take the hollow center pixel x0 as the pixel reference, and convert the grayscale value of the pixel x0' adjacent to the pixel reference Compared with the gray value H, if Then the pixel x0′ is determined to be in the hollow area. Otherwise, the pixel x0′ is not in the hollow area and is regarded as the boundary pixel of the hollow area. 阈值 is the allowable error between the grayscale value of the pixel in the hollow area and the grayscale value of the standard pixel;

[0080] S34: After obtaining the pixel x0′ adjacent to the hollow pixel x0 and located in the hollow area, return to step S33, use the pixel x0′ located in the hollow area as the pixel reference, and continue to screen the pixels adjacent to the pixel w0′ and located in the hollow area;

[0081] S35: until all pixels in the hollow area around the hollow center pixel w0 are screened out, a complete hollow area on the infrared image is obtained, and the coordinate feature corresponding to the hollow center pixel w0 is used as the center of the hollow area;

[0082] S36: According to the coordinate characteristics of the boundary pixels in the infrared image, all boundary pixels are smoothly connected to obtain the boundary line of the hollow area, and the hollow area is displayed in the infrared image. The area S of the hollow area is calculated according to the number κ of pixels in all hollow areas:

[0083]

[0084] Where s is the area of the target wall corresponding to a single pixel, and γ is the number of pixels in the complete wall infrared grayscale image.

[0085] S4: Based on the acquired point cloud data within the hollow area, the structural stability of the target exterior wall is analyzed and evaluated using the point cloud data and the area of the hollow area. Step S4 specifically includes:

[0086] S41: acquiring point cloud data in an area corresponding to the hollow area on the plane point cloud map according to the hollow area in the infrared image, and calculating a protrusion height ν of the hollow area;

[0087]

[0088] Among them, c is the number of the point cloud data in the area corresponding to the hollow area, ν c is the relative height of the corresponding position of the point cloud data c, ν0 is the reference benchmark of the point cloud data, and C is the number of point cloud data in the area corresponding to the hollow area;

[0089] S42: Calculate the structural stability coefficient δ of the target outer wall where the hollow area is located based on the protruding height ν and area S of the hollow area d ;

[0090] δ d =α1exp(ν)+α2exp(S);

[0091] Where d is the number of the hollow area on the target outer wall, α1 and α2 are the weights of the impact of the protruding height and area of the hollow area on the structural stability, α1+α2=1;

[0092] S43: Setting the threshold δ for evaluating the wall structure stability 阈值 If δ d ≤δ 阈值 , then the structure of the hollow area d on the target outer wall is stable; if δ d >δ 阈值 , then the structure of the hollow area d on the target outer wall is unstable. 阈值 It represents the maximum value allowed by the structural stability coefficient. The larger the structural stability coefficient is, the more unstable the target exterior wall structure where the hollow area is located is, and vice versa. The stability of the target exterior wall structure is directly related to the safety of the target exterior wall structure.

[0093] The present invention combines infrared imaging technology with three-dimensional laser scanning technology to screen, locate and identify hollow areas on building exterior walls, and uses the area and protrusion height of the identified hollow areas to comprehensively evaluate the structural stability of the hollow areas, and analyze the structural safety of the hollow areas on the target exterior walls. Infrared imaging technology identifies hollow areas based on the different effects of the hollow areas on the exterior walls on absorbing thermal radiation, and three-dimensional laser scanning technology obtains texture data on the building wall to identify the protruding parts of the hollow areas. The combination of the two can not only identify the area of the hollow areas but also obtain the protrusion height of the hollow areas, achieving efficient and accurate identification of hollows and scientific assessment of the structural safety of building exterior walls, providing strong technical support for building safety maintenance.

Claims

1. A building exterior wall structural safety assessment method based on drone data collection, characterized in that: The following steps are involved: S1: Identify the target exterior wall surface and use a drone equipped with an infrared thermal imager and a 3D laser scanner to fly over the wall surface to collect infrared images and 3D point cloud data of the target exterior wall surface. S2: Optimize the 3D point cloud data of the target exterior wall to obtain an accurate 3D point cloud map, and align the accurate 3D point cloud map with the infrared image; S3: grayscale processing is performed on the infrared image, and the grayscale values of the pixels in the infrared grayscale image are used to analyze the hollow area of the target exterior wall, and the boundary line and area of the hollow area are extracted; S4: Based on the acquired point cloud data within the hollow area, the structural stability of the target exterior wall is analyzed and evaluated using the point cloud data and the area of the hollow area.

2. The building exterior wall structure safety assessment method based on drone data collection according to claim 1 is characterized in that: The step S2 specifically includes: S21: Obtain all point cloud datasets A={Q1,Q2,…,Q n }, n is the number of point cloud sets, Q n is the nth point cloud set; S22: Using the coordinate features of the point clouds in the point cloud set to select the two most similar point cloud set groups; Among them, Q i , Q j are any two point cloud sets in the point cloud dataset A, Q j As the reference point cloud, Q i is the target point cloud, T i is the coordinate feature of the point cloud in the target point cloud set, T j is the coordinate feature of the point cloud in the reference point cloud set, i is the number of the point cloud in the target point cloud set, j is the number of the point cloud in the target point cloud set, and I is the number of point clouds in the target point cloud set; S23: Filter out the most similar point cloud group (Q e ,Q u ) then, according to the point cloud group (Q e ,Q u ) to perform point cloud matching based on the coordinate features of the point cloud; specifically: S231: Based on point cloud Q e The coordinate feature T of any point cloud e in e , in Point Cloud Q u The point cloud u closest to the point cloud e is selected internally for closest point cloud matching; S232: Point Cloud Collection Q e Point cloud and point cloud set Q u After the point cloud in the nearest point cloud is matched, the intermediate coordinate feature T after the nearest point cloud matching is successful is calculated. e～u ; S233: The intermediate coordinate feature T e～u As the point cloud features after the two closest point clouds are matched, the point cloud set Q is generated e With Point Cloud Q u The point cloud set Q0 after point cloud matching is performed; the three-dimensional point cloud image formed by the point cloud set Q0 is used as the accurate three-dimensional point cloud image of the target exterior wall; S24: Flattening the precise three-dimensional point cloud image to form a planar point cloud image, where each point cloud data in the planar point cloud image represents the relative height of a corresponding position on the target exterior wall surface; S25: Select a typical area on the target exterior wall as an alignment reference area, obtain coordinate features corresponding to the point cloud data within the alignment reference area, and calculate the center coordinate T0 of the alignment reference area; Among them, m is the number of the point cloud in the alignment reference area, T m is the coordinate feature of point cloud m, M is the number of point clouds in the alignment reference area; S26: Set the standard temperature t0 of the alignment reference area. According to the area of the alignment reference area, select an infrared pixel area with the same area as the alignment reference area on the infrared image, and make the average temperature within the pixel area equal to the standard temperature. The infrared pixel area satisfies the constraint conditions: Where B is the number of infrared pixels in the infrared pixel area, χ is the full level of the infrared pixel, b is the number of infrared pixels in the infrared pixel area, b is the infrared pixel set of the infrared pixel area, B is the infrared pixel set of the infrared image, S b is the area of the alignment reference region, t b is the temperature value corresponding to the infrared pixel; S27: establishing a coordinate system identical to that of the plane point cloud image on the infrared image, obtaining coordinate features of the infrared pixels, and calculating the center coordinates T0′ of the infrared pixel area based on the coordinate features of the infrared pixels within the infrared pixel area; Where w is the number of pixels in the infrared pixel area, W is the number of pixels in the infrared pixel area, T w ′ is the coordinate feature of pixel w; S28: According to the relative position between the center coordinate T0′ and the center coordinate T0, the plane point cloud image and the infrared image are moved so that the center of the infrared pixel area is aligned with the center of the alignment reference area, thereby aligning the plane point cloud image and the infrared image.

3. The building exterior wall structure safety assessment method based on drone data collection according to claim 2 is characterized in that: The step S3 comprises: S31: Grayscale the infrared image to obtain an infrared grayscale image, set the standard pixel grayscale value H of the hollow center on the infrared grayscale image when the wall has hollows, and extract the grayscale value A of each pixel in the infrared grayscale image. x ′, x is the pixel number in the complete wall infrared grayscale image; S32: Using the standard pixel grayscale value H and grayscale value A x ′, filter the hollow center pixel x0 on the target wall, the gray value of the hollow center pixel x0 H 阈值 The allowable error between the grayscale value of the hollow center pixel and the grayscale value of the standard pixel; S33: Obtain all hollow center pixels x0 on the target wall, take the hollow center pixel x0 as the pixel reference, and convert the grayscale value of the pixel x0' adjacent to the pixel reference Compared with the gray value H, if Then the pixel x0′ is determined to be in the hollow area. Otherwise, the pixel x0′ is not in the hollow area and is regarded as the boundary pixel of the hollow area. 阈值 is the allowable error between the grayscale value of the pixel in the hollow area and the grayscale value of the standard pixel; S34: After obtaining the pixel x0′ adjacent to the hollow pixel x0 and located in the hollow area, return to step S33, use the pixel x0′ located in the hollow area as the pixel reference, and continue to screen the pixels adjacent to the pixel w0′ and located in the hollow area; S35: until all pixels in the hollow area around the hollow center pixel w0 are screened out, a complete hollow area on the infrared image is obtained, and the coordinate feature corresponding to the hollow center pixel w0 is used as the center of the hollow area; S36: According to the coordinate characteristics of the boundary pixels in the infrared image, all boundary pixels are smoothly connected to obtain the boundary line of the hollow area, and the hollow area is displayed in the infrared image. The area S of the hollow area is calculated according to the number κ of pixels in all hollow areas: Where s is the area of the target wall corresponding to a single pixel, and γ is the number of pixels in the complete wall infrared grayscale image.

4. The building exterior wall structure safety assessment method based on drone data collection according to claim 3 is characterized in that: The step S4 comprises: S41: acquiring point cloud data in an area corresponding to the hollow area on the plane point cloud map according to the hollow area in the infrared image, and calculating a protrusion height ν of the hollow area; Among them, c is the number of the point cloud data in the area corresponding to the hollow area, ν c is the relative height of the corresponding position of the point cloud data c, ν0 is the reference benchmark of the point cloud data, and C is the number of point cloud data in the area corresponding to the hollow area; S42: Calculate the structural stability coefficient δ of the target outer wall where the hollow area is located based on the protruding height ν and area S of the hollow area d ; d d =α1exp(ν)+α2exp(S); Where d is the number of the hollow area on the target outer wall, α1 and α2 are the weights of the impact of the protruding height and area of the hollow area on the structural stability, α1+α2=1; S43: Setting the threshold δ for evaluating the wall structure stability 阈值 If δ d ≤δ 阈值 , then the structure of the hollow area d on the target outer wall is stable; if δ d >δ 阈值 , then the structure of the hollow area d on the target outer wall is unstable.

Images