A high-rise building outer wall quality detection and evaluation method

Abstract

The application relates to the technical field of image feature processing, in particular to a high-rise building outer wall quality detection and evaluation method. The method acquires an edge image of a high-rise building outer wall detection image, and determines a plurality of edge regions surrounded by the edges; the overall width of the edge regions is used for primary screening to obtain a first suspected crack region; the overall flatness is used for secondary screening to obtain a second suspected crack region. The group features of the second suspected crack region are analyzed in depth, shape regularity is obtained according to the shape similarity between regions, and distribution regularity is obtained according to the distance distribution between regions; the regions with repeated texture features are removed by comprehensively considering the shape and distribution regularity, the real crack region is accurately screened out, and the crack defect degree is quantified according to the information proportion in the image. The application can effectively distinguish the real crack from the outer wall texture interference with similar shapes, and significantly improves the anti-interference ability and evaluation accuracy of the outer wall crack detection.

Description

Technical Field

[0001] This invention relates to the field of image feature processing technology, specifically to a method for quality inspection and evaluation of the exterior walls of high-rise buildings. Background Technology

[0002] With the acceleration of urbanization, the number of high-rise buildings has increased dramatically, making the maintenance and safety inspection of their exterior walls increasingly important. During the curing and use process, concrete exterior walls are prone to surface cracks due to factors such as shrinkage, temperature differences, or foundation settlement. If these cracks are not detected and repaired in time, they may pose safety hazards. Traditional manual inspection methods suffer from high risk, low efficiency, and limited coverage. In recent years, non-contact inspection technology based on machine vision has become mainstream. This involves using drones or wall-climbing robots equipped with high-definition cameras to capture images of exterior walls, extracting edge features through image processing algorithms, and automatically identifying and quantitatively assessing defects based on the grayscale differences and geometric shapes between the cracked area and the surrounding wall surface.

[0003] In real-world inspection scenarios, the surface environment of high-rise building exterior walls is quite complex, and existing visual inspection methods still face significant challenges, particularly in effectively distinguishing between genuine micro-cracks and the inherent texture features of the wall surface. Specifically, during the pouring process, concrete walls often retain wood grain imprints or regular joints left by wooden formwork. These exterior wall textures typically appear in images as thin, shallow edges with a certain degree of straightness, closely resembling the micro-network or crazing cracks that appear early in construction. Existing techniques largely rely on thresholding based on width, contrast, or single linear features. While this can remove most non-linear block noise (such as color spots), it cannot logically distinguish between "textures with repetitive patterns" and "randomly distributed cracks." This confusion makes it easy for detection algorithms to misjudge normal exterior wall textures as crack defects, resulting in a high false alarm rate, unreliable detection results, and consequently affecting the accurate assessment of the exterior wall quality level. Summary of the Invention

[0004] To address the technical problem that existing technologies for identifying cracks in building exterior walls are easily affected by the texture of the exterior wall itself, making it difficult to accurately and effectively distinguish crack defects, the present invention aims to provide a method for quality inspection and evaluation of exterior walls of high-rise buildings. The specific technical solution adopted is as follows: This invention proposes a method for quality inspection and evaluation of the exterior walls of high-rise buildings, the method comprising: Obtain the edge image of the detected exterior wall of a high-rise building; the edge image contains multiple edge regions enclosed by edges; A first suspected crack region is selected based on the overall width of the edge region; a second suspected crack region is selected based on the overall straightness of the first suspected crack region. For any second suspected crack region, the shape regularity of the second suspected crack region is obtained based on the shape similarity between the second suspected crack region and other second suspected crack regions; the distribution regularity of the second suspected crack region is obtained based on the distance distribution between the second suspected crack region and other second suspected crack regions; and the real crack regions are screened out based on the shape regularity and the distribution regularity. The degree of crack defect is obtained based on the proportion of the actual crack area in the detected images of the exterior walls of the high-rise building.

[0005] Furthermore, the method for quantifying the overall width includes: For each edge point in the edge region, draw a perpendicular line from the edge point to the tangent of the edge point, and obtain the intersection point of the perpendicular line with the edge of the edge region. Select the distance between the edge intersection point closest to the edge point and the edge point as the local width; calculate the average of the local widths of all edge points in the edge region to obtain the overall width.

[0006] Furthermore, the method for quantifying the overall flatness includes: For each edge point in the first suspected crack region, a predetermined number of other edge points are selected as neighboring edge points, centered on the edge point. The slope of the line connecting each neighboring edge point and the central edge point is obtained. The standard deviation of the line slope is negatively correlated to obtain the flatness confidence of each edge point. The average flatness confidence of all edge points in the first suspected crack region is taken as the overall flatness.

[0007] Furthermore, the method for quantifying the shape similarity includes: For any second suspected crack region, align the second suspected crack region with other second suspected crack regions based on their edges, rotate and translate the second suspected crack region until the degree of overlap between the edges of the second suspected crack region and other second suspected crack regions is maximized, and take the proportion of the area of ​​the overlapping area to the area of ​​the second suspected crack region as the shape similarity between the second suspected crack region and other second suspected crack regions.

[0008] Furthermore, the method for obtaining the shape regularity includes: For any second suspected crack region, the average shape similarity between it and all other second suspected crack regions is obtained as the shape regularity.

[0009] Furthermore, the method for obtaining the distribution regularity includes: The distance between the second suspected crack area and the nearest other second suspected crack area is used as the reference distance; For any second suspected crack region, calculate the reference distance difference between the second suspected crack region and each other second suspected crack region, perform negative correlation mapping on the accumulated value of the reference distance difference and normalize it to obtain the distribution regularity of the second suspected crack region.

[0010] Furthermore, the step of selecting the actual crack regions based on the shape regularity and the distribution regularity includes: Based on the shape regularity and the distribution regularity, an overall regularity is obtained, and the second suspected crack region with an overall regularity less than a preset regularity threshold is taken as the real crack region.

[0011] Furthermore, the method for obtaining the degree of crack defect includes: The degree of crack defect is obtained by using the total area of ​​the actual crack area as the numerator and the area of ​​the detected image of the high-rise building's exterior wall as the denominator.

[0012] Furthermore, after obtaining the degree of the crack defect, the process also includes: The area ratios of hollow areas and leakage areas are obtained, and the overall quality parameters are obtained based on the degree of crack defects, the area ratios of hollow areas, and the area ratios of leakage areas.

[0013] Furthermore, the overall quality parameter is the sum of the degree of the crack defect, the area ratio of the hollow area, and the area ratio of the leakage area.

[0014] The present invention has the following beneficial effects: This invention utilizes the overall width and flatness of the edge region for initial screening. Based on the physical characteristics of cracks being long and narrow with a certain extension direction, it can quickly eliminate large, flat background areas and irregularly shaped color noise in the image, initially identifying suspected areas with linear characteristics. Furthermore, it delves into the essential differences in the group distribution of cracks and textures. For texture interference that may still be confused after the initial screening, it obtains shape regularity by calculating the shape similarity between regions, taking advantage of the high shape repetition of wall textures due to construction templates, while cracks have random directions and unique shapes. It obtains distribution regularity by calculating the distance distribution between regions, utilizing the fact that texture arrangement often has fixed intervals or periodicity, while crack location distribution is highly random, ultimately achieving accurate identification of real crack areas. This invention elevates single geometric feature detection to group distribution pattern analysis, accurately identifying and eliminating texture artifacts with similar shapes and regular arrangements, thus filtering out real crack areas from complex backgrounds. Finally, it calculates the degree of defect based on the proportion of real crack areas, greatly improving the detection's anti-interference ability and accuracy, achieving an objective and accurate assessment of the exterior wall quality level. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A flowchart of a method for quality inspection and evaluation of the exterior walls of high-rise buildings provided in one embodiment of the present invention; Figure 2 An image of the exterior wall of a high-rise building provided in one embodiment of the present invention; Figure 3 An edge image provided in one embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method for quality inspection and evaluation of the exterior walls of high-rise buildings proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] The specific scheme of the method for quality inspection and evaluation of the exterior walls of high-rise buildings provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0020] Please see Figure 1 The diagram illustrates a flowchart of a method for quality inspection and evaluation of the exterior walls of high-rise buildings according to an embodiment of the present invention. The method includes: Step S1: Obtain the edge image of the detected image of the exterior wall of the high-rise building; there are multiple edge regions enclosed by edges in the edge image.

[0021] In this embodiment of the invention, a drone equipped with a camera can be used to capture images of the exterior walls of high-rise buildings, thereby obtaining inspection images of the exterior walls of high-rise buildings. For example... Figure 2 As shown, Figure 2This illustration shows an image of a high-rise building's exterior wall, provided by an embodiment of the present invention. The image includes not only clearly visible cracks but also regularly distributed textures and minute small cracks. Further conversion of this image into an edge image yields... Figure 3 The image shown has clearly defined edges.

[0022] It should be noted that, in order to obtain an edge image with clear edge features, the image of the high-rise building's exterior wall needs to be converted to grayscale. The grayscale image will contain a large number of spots with uneven grayscale, so a Gaussian filtering algorithm is needed for filtering. After obtaining the grayscale image, the final edge image can be obtained using the Canny edge detection operator. The above preprocessing techniques are well known to those skilled in the art and will not be elaborated upon here.

[0023] The resulting edge image contains numerous edge lines. This invention, based on the edge image obtained through edge detection, utilizes connected component analysis or region growing algorithms to acquire edge regions. Specifically, based on the characteristic that the grayscale within a crack region is relatively uniform, edge pixels are used as seed points. In the original high-rise building exterior wall detection image, adjacent pixels with grayscale differences within a preset range (e.g., within 10) are divided into the same connected component, serving as edge regions. In particular, if a large crack prevents complete information from being captured in the field of view of an image (i.e., the edge extends to the image boundary), the region enclosed by the corresponding edge truncation and the image boundary, or the set of connected pixels extending along the image boundary, is also considered as edge regions for subsequent processing. The specific implementation method is a well-known technique and will not be elaborated upon here.

[0024] Step S2: Select a first suspected crack area based on the overall width of the edge region; select a second suspected crack area based on the overall straightness of the first suspected crack area.

[0025] Firstly, crack areas typically have a certain width. A crack area is usually a narrow and elongated region consisting of two edges. The width of this region is not very large. If there is a wide edge area, it indicates that the area may be a coarse texture or dirt on the exterior wall. Therefore, the edge area can be initially screened based on its width to identify the first suspected crack area.

[0026] Furthermore, on the exterior wall surface, there are some small-width areas, such as color spots, that are clearly not characteristic of narrow cracks. Therefore, based on the shape characteristics of the first suspected crack area, the second suspected crack area can be screened out by the overall straightness, and these small-width areas, such as color spots, can be eliminated.

[0027] Preferably, in this embodiment of the invention, the method for quantifying the overall width includes: For each edge point in the edge region, draw a perpendicular line from the edge point to the tangent of the edge point, and obtain the intersection point of the perpendicular line with the edge of the edge region. Select the distance between the edge intersection point closest to the edge point and the edge point as the local width; calculate the average of the local widths of all edge points in the edge region to obtain the overall width.

[0028] In this embodiment of the invention, a width threshold is set, and edge regions with an overall width less than the width threshold are designated as first suspected crack regions. The width threshold is not fixed and needs to be set according to factors such as the actual image field of view and the accuracy requirements of the implementer. In this embodiment, considering that cracks appear as fine lines in images, their width is usually significantly smaller than large-area light and shadow variations on the wall or broad background textures. Therefore, the average width of all edge regions in the current image can be used as the width threshold. Regardless of the distance between the camera and the wall, cracks always belong to the finer end of the distribution relative to the overall background area (such as large areas of stains or light and shadow). Using the average value as the threshold can adaptively filter out regions with widths significantly lower than the average background level as first suspected crack regions.

[0029] Preferably, in this embodiment of the invention, the method for quantifying overall flatness includes: This embodiment of the invention takes into account that cracks may have branching structures, and these branching structures may have different orientations from the main trunk. Therefore, it is not possible to perform a linear fitting analysis on the whole. This embodiment of the invention starts from the local and analyzes each edge point on the first suspected crack area: For each edge point in the first suspected crack area, a predetermined number of other edge points centered on that edge point are selected as neighboring edge points. The definition of neighboring edge points enables local analysis of the first suspected crack area. By constructing a neighborhood range centered on each edge point, analyzing straightness within this neighborhood yields better statistical results. In this embodiment, the predetermined number is set to 10, meaning that five nearest edge points from the center edge point outwards on both sides of the same first suspected crack area are selected as neighboring edge points. It should be noted that if the number of edge points in the first suspected crack area is less than 11, it indicates that the area is significantly small and does not belong to the crack area; it may be a stain or noise on the wall surface and can be directly filtered out.

[0030] The slope of the line connecting each neighboring edge point to the central edge point is obtained. If the edge at the location of an edge pixel is straight, then the slope of the line connecting the corresponding neighboring edge points should be consistent. Therefore, the standard deviation of the slopes of all neighboring edge points is calculated. The smaller the standard deviation, the more the corresponding central edge point is in a straight segment. Therefore, the standard deviations of the line slopes are negatively correlated to obtain the straightness confidence of each edge point. Therefore, the average straightness confidence of all edge points in the first suspected crack region is calculated as the overall straightness.

[0031] In this embodiment of the invention, the reciprocal method can be selected for the negative correlation mapping of the standard deviation. To prevent the denominator from being 0, the result of adding the standard deviation to a positive integer is used as the denominator, and the positive integer 1 is used as the numerator to achieve the negative correlation mapping. It should be noted that other basic mathematical methods for negative correlation mapping can also be selected in other implementations of this invention, which will not be elaborated here.

[0032] In this embodiment of the invention, after obtaining the overall straightness of each first suspected crack region, a straightness threshold of 0.7 is set, and the first suspected crack regions with an overall straightness greater than the straightness threshold are designated as second suspected crack regions. This straightness threshold is statistically derived from external wall defect sample data. Disruptive items such as discoloration and mold spots often have circular or irregularly serrated edges, with drastic changes in the slope of their edge pixels; the calculated straightness is typically between 0.1 and 0.4. In contrast, while real cracks may exhibit local meandering, their overall extension direction is continuous, and their straightness is typically between 0.7 and 0.95. Selecting 0.7 as the threshold is the optimal critical point for maximizing the elimination of clumpy noise (high precision) while ensuring no slight bending cracks are missed (high recall). Those skilled in the art can adjust this threshold within the range of 0.6 to 0.8 according to the actual camera resolution or detection accuracy requirements. It should be noted that, in one specific implementation of this invention, since the number of exterior wall textures is significantly large, if the number of second suspected crack areas is less than a preset threshold, it is considered that there is no exterior wall texture in the current image, and all second suspected crack areas can be directly taken as the final real crack areas. In this embodiment of the invention, the threshold can be set to 3, and can be specifically set according to the specific implementation scenario such as image field of view and shooting angle.

[0033] Step S3: For any second suspected crack region, obtain the shape regularity of the second suspected crack region based on the shape similarity between the second suspected crack region and other second suspected crack regions; obtain the distribution regularity of the second suspected crack region based on the distance distribution between the second suspected crack region and other second suspected crack regions; and select the real crack regions based on the shape regularity and distribution regularity.

[0034] The exterior wall surface has some textures that are completely similar to micro-crack defects, both exhibiting characteristics such as being narrow and straight. Therefore, the second suspected crack area will still be affected by the exterior wall texture and requires further feature extraction for screening.

[0035] Considering the obvious shape consistency of the texture on the exterior wall surface and its uniform distribution across the wall panels, while the crack formation is related to the shrinkage and internal stress during the concrete curing process, the direction of the cracks is random. Therefore, the true crack areas can be identified by analyzing the regularity among the second suspected crack areas.

[0036] For exterior wall textures, there is a clear shape similarity between textures in different locations. Therefore, in this embodiment of the invention, the shape regularity of a second suspected crack area is obtained based on the shape similarity between the second suspected crack area and other second suspected crack areas. If a second suspected crack area is an exterior wall texture, since crack areas are relatively low-probability samples compared to edge areas, the exterior wall texture should exhibit a strong shape similarity to other second suspected crack areas as a whole. That is, the greater the shape similarity, the more likely the second suspected crack area is an exterior wall texture, and the stronger its shape regularity on the wall surface.

[0037] Secondly, the exterior wall texture has a uniform distribution characteristic on the exterior wall. Therefore, after considering the shape regularity, it is necessary to further analyze the distance distribution relationship between the second suspected crack areas. If a second suspected crack area has a significantly consistent distribution characteristic compared to other suspected crack areas, it indicates that the second suspected crack area belongs to the exterior wall texture. Therefore, the distribution regularity of the second suspected crack areas is obtained by further analyzing the distance distribution between the second suspected crack areas and other second suspected crack areas. Finally, the actual crack areas are screened out based on the shape regularity and the distribution regularity.

[0038] Preferably, in this embodiment of the invention, considering that the surface texture of the exterior wall detected by the image may not be complete, the lengths of different texture regions may be inconsistent. Therefore, it is not possible to directly compare the second suspected crack regions. In this embodiment of the invention, for any second suspected crack region, the second suspected crack region is aligned with other second suspected crack regions based on its edges. The second suspected crack region is then rotated and translated until the degree of overlap on the edges between the second suspected crack region and other second suspected crack regions is maximized. The proportion of the overlapping area to the area of ​​the second suspected crack region is taken as the shape similarity between the second suspected crack region and other second suspected crack regions. That is, when comparing two second suspected crack regions, this invention ensures that alignment is based on the edges, ensuring that each rotation and translation result has edge alignment. Furthermore, the rotation and translation stop condition is set to the maximum degree of overlap on the edges, avoiding obviously illogical results (such as inverted matching) from rotation and translation without a reference.

[0039] Preferably, in this embodiment of the invention, considering that the greater the shape similarity between a second suspected crack area and other second suspected crack areas, and the more significant the shape similarity, the more likely the second suspected crack area is to belong to the exterior wall texture, therefore, for any second suspected crack area, the average shape similarity between it and all other second suspected crack areas is obtained as the shape regularity.

[0040] Preferably, in this embodiment of the invention, the method for obtaining the distribution regularity includes: The distance between the second suspected crack region and the nearest other second suspected crack region is used as a reference distance. That is, this reference distance represents the distribution characteristics of the second suspected crack regions as described above. It should be noted that the distance between regions can be the distance between their centroids; specific details are not elaborated upon or limited.

[0041] For any second suspected crack region, the reference distance difference between the second suspected crack region and each other second suspected crack region is calculated. The smaller the reference distance difference, the more similar the distribution characteristics between the two second suspected crack regions. Therefore, the cumulative value of the reference distance difference is negatively correlated and normalized to obtain the distribution regularity of the second suspected crack regions.

[0042] In this embodiment of the invention, the reference distance difference is the absolute value of the difference between two reference distances. Similarly, the reciprocal form can be used for negative correlation processing. Likewise, the reciprocal of the result after adding the cumulative value of the reference distance difference to the positive integer 1 is used as the final negative correlation mapping result. Then, the negative correlation mapping result is normalized by range standardization to obtain the final distribution regularity.

[0043] In other implementations of this invention, the accumulated value of the reference distance difference can be substituted into the exp(-x) function, which is an exponential function with the natural constant as its base. The function output is directly the result of negative correlation mapping and normalization. Other implementations of this invention can also be implemented using other basic mathematical operations, which will not be elaborated or limited here.

[0044] Preferably, in this embodiment of the invention, the regularity of the shape and the regularity of the distribution are used to screen out the real crack regions, including: Based on the shape regularity and the distribution regularity, an overall regularity is obtained, and the second suspected crack region with an overall regularity less than a preset regularity threshold is taken as the real crack region.

[0045] It should be noted that, in one implementation of this invention, since both shape regularity and distribution regularity are normalized results, the average of shape regularity and distribution regularity can be directly used as the overall regularity. The resulting overall regularity is also a normalized data point between 0 and 1, therefore, a regularity threshold of 0.6 can be set. This threshold can be specifically set according to the specific implementation scenario, and will not be limited or elaborated upon here.

[0046] Step S4: Obtain the degree of crack defect based on the proportion of the actual crack area in the inspection image of the high-rise building exterior wall.

[0047] By statistically analyzing the proportion of all real crack areas in the inspection images of the exterior walls of high-rise buildings, the degree of crack defect can be obtained.

[0048] Preferably, in this embodiment of the invention, the total area of ​​the actual crack region is used as the numerator, and the area of ​​the detected image of the high-rise building's exterior wall is used as the denominator to obtain the degree of crack defect. That is, the degree of crack defect characterizes the proportion of information in an area dimension.

[0049] Furthermore, considering that the embodiments of the present invention are aimed at the quality inspection and evaluation of the exterior walls of high-rise buildings, the quality of the exterior walls should not only include the assessment of cracks but also the assessment of controlled areas and leakage areas. Therefore, after obtaining the degree of crack defects, the embodiments of the present invention also include: The area ratios of hollow and leaking areas are obtained. Based on the severity of crack defects, the area ratios of hollow and leaking areas, an overall quality parameter is derived. Since all three features are percentage information, the sum of the area ratios of crack defects, hollow areas, and leaking areas can be directly used as the overall quality parameter. The overall quality of the exterior wall location corresponding to the image is then comprehensively evaluated based on this overall quality parameter.

[0050] It should be noted that the detection of the controlled area and the leakage area can be performed using infrared thermal imaging images. Temperature difference segmentation is performed on the infrared thermal imaging images, and the Canny edge detection algorithm is used to identify abnormal areas with obvious temperature differences. The temperature gradient distribution changes are analyzed, and the degree of temperature gradient difference at the edges is quantified to obtain the variance of the temperature gradient difference in abnormal areas. The variance of the temperature gradient difference and shape regularity are used to distinguish between these areas. Clump-like or strip-like areas with large temperature gradient differences at the edges and clear edges are identified as hollow areas; cloud-like or band-like areas with small temperature gradient differences at the edges and blurred edges are identified as leakage areas. Furthermore, manual assistance and architectural structural drawings are used to locate and determine leakage points and hollow points. This process specifically includes: (1) Spatially align the infrared thermal imaging image with the high-rise building exterior wall detection image to ensure that the coordinates of the exterior wall areas covered by the two are consistent.

[0051] (2) The infrared thermal imaging image of the exterior wall is analyzed, and the Canny edge detection operator is used to extract the temperature jump edges to obtain the infrared anomaly areas with obvious temperature differences. In order to distinguish the anomaly type, the temperature change gradient of the edge of the anomaly area is further calculated. For each pixel point of the edge of the anomaly area, its temperature gradient value along the normal direction is calculated, and the standard deviation of the temperature gradient of all edge pixels in the area is counted, which is recorded as the degree of temperature change gradient difference.

[0052] (3) Calculate the variance of the temperature change gradient difference in each infrared anomaly region, and classify the defects using the following logic, combined with the geometric morphological characteristics of the region: (a) Determination of hollow areas: Because of the internal air insulation, the surface temperature of hollow areas rises (or falls) rapidly under sunlight, creating a significant and sharp temperature boundary with the surrounding walls. Therefore, if an abnormal area meets the following criteria: ① a large temperature gradient difference at the edge (i.e., a high degree of temperature gradient difference with a clear and sharp edge); ② a closed, clumpy or strip-shaped geometric shape, then the area is determined to be a hollow area.

[0053] (b) Determination of leakage area: Leaking areas typically exhibit abnormally low temperatures due to heat absorption from moisture evaporation or increased heat capacity. Furthermore, the diffusion of moisture within the wall results in a smooth transition of temperature boundaries without sharp, anisotropic edges. An abnormal area is identified as a leaking area if it meets the following criteria: ① the temperature gradient at the edge is relatively small (i.e., the degree of temperature gradient difference is low, and the edge is blurred); ② its geometric shape exhibits a cloud-like or irregular band-like pattern with diverging edges.

[0054] It should be noted that the identification of the hollow area and the leakage area mentioned above can be achieved using existing technologies. Specifically, each detailed step can be implemented by setting thresholds, shape fitting analysis and other algorithms. Since these are existing technologies and not the focus of this invention, they are only briefly described as described above, without further elaboration or limitation.

[0055] In summary, this invention acquires edge images of high-rise building exterior wall inspection images and identifies multiple edge regions enclosed by these edges. Initial screening is performed using the overall width of these edge regions to obtain first suspected crack regions. Further screening is conducted using overall straightness to obtain second suspected crack regions. Subsequently, the group characteristics of the second suspected crack regions are analyzed in depth. Shape regularity is obtained based on the shape similarity between regions, and distribution regularity is obtained based on the distance distribution between regions. Combining shape and distribution regularity, regions with repetitive texture features are eliminated, accurately identifying true crack regions. The degree of crack defect is quantified based on the proportion of information in the image. This invention can effectively distinguish between true cracks and interference from exterior wall textures with similar shapes, significantly improving the anti-interference capability and evaluation accuracy of exterior wall crack detection.

[0056] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0057] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for quality inspection and evaluation of the exterior walls of high-rise buildings, characterized in that, The method includes: Obtain the edge image of the detected exterior wall of a high-rise building; the edge image contains multiple edge regions enclosed by edges; The first suspected crack area is selected based on the overall width of the edge region; the second suspected crack area is selected based on the overall straightness of the first suspected crack area. For any second suspected crack region, the shape regularity of the second suspected crack region is obtained based on the shape similarity between the second suspected crack region and other second suspected crack regions; the distribution regularity of the second suspected crack region is obtained based on the distance distribution between the second suspected crack region and other second suspected crack regions; and the real crack regions are screened out based on the shape regularity and distribution regularity. The degree of crack defect is obtained based on the proportion of the actual crack area in the inspection image of the exterior wall of the high-rise building; The method for quantifying shape similarity includes: for any second suspected crack region, aligning the second suspected crack region with other second suspected crack regions based on the edge, rotating and translating the second suspected crack region until the overlap between the second suspected crack region and other second suspected crack regions on the edge is maximized, and taking the proportion of the overlapping area to the area of ​​the second suspected crack region as the shape similarity between the second suspected crack region and other second suspected crack regions. The method for obtaining the shape regularity includes: for any second suspected crack region, obtaining the average shape similarity between it and all other second suspected crack regions, as the shape regularity; The method for obtaining the distribution regularity includes: taking the distance between the second suspected crack region and the nearest other second suspected crack region as a reference distance; for any second suspected crack region, calculating the difference in reference distance between the second suspected crack region and each other second suspected crack region, performing negative correlation mapping and normalization on the accumulated value of the reference distance difference, and obtaining the distribution regularity of the second suspected crack region.

2. The method for quality inspection and evaluation of the exterior walls of high-rise buildings according to claim 1, characterized in that, The method for quantifying the overall width includes: For each edge point in the edge region, draw a perpendicular line from the edge point to the tangent of the edge point, and obtain the intersection point of the perpendicular line with the edge of the edge region. Select the distance between the edge intersection point closest to the edge point and the edge point as the local width; calculate the average of the local widths of all edge points in the edge region to obtain the overall width.

3. The method for quality inspection and evaluation of the exterior walls of high-rise buildings according to claim 1, characterized in that, The method for quantifying the overall flatness includes: For each edge point in the first suspected crack region, a predetermined number of other edge points are selected as neighboring edge points, centered on the edge point. The slope of the line connecting each neighboring edge point and the central edge point is obtained. The standard deviation of the line slope is negatively correlated to obtain the flatness confidence of each edge point. The average flatness confidence of all edge points in the first suspected crack region is taken as the overall flatness.

4. The method for quality inspection and evaluation of the exterior walls of high-rise buildings according to claim 1, characterized in that, The step of selecting the actual crack regions based on the shape regularity and the distribution regularity includes: Based on the shape regularity and the distribution regularity, an overall regularity is obtained, and the second suspected crack region with an overall regularity less than a preset regularity threshold is taken as the real crack region.

5. The method for quality inspection and evaluation of the exterior walls of high-rise buildings according to claim 1, characterized in that, The method for obtaining the degree of crack defect includes: The degree of crack defect is obtained by using the total area of ​​the actual crack area as the numerator and the area of ​​the detected image of the high-rise building's exterior wall as the denominator.

6. The method for quality inspection and evaluation of the exterior walls of high-rise buildings according to claim 5, characterized in that, After obtaining the degree of the crack defect, the following is also included: The area ratios of hollow areas and leakage areas are obtained, and the overall quality parameters are obtained based on the degree of crack defects, the area ratios of hollow areas, and the area ratios of leakage areas.

7. The method for quality inspection and evaluation of the exterior walls of high-rise buildings according to claim 6, characterized in that, The overall quality parameters are the sum of the degree of crack defects, the area ratio of the hollow area, and the area ratio of the leakage area.

Images