Method and system for evaluating and detecting surface defects of large-span steel arch

Abstract

The invention belongs to the technical field of defect detection, and particularly relates to a large-span steel arch surface defect evaluation detection method and system.The large-span steel arch surface defect evaluation detection system comprises an image acquisition unit, an image processing unit, a defect screening unit, a defect analysis unit and a defect judgment unit. The comprehensiveness is ensured; through image preprocessing of graying and median filtering, the image quality is effectively improved, and a foundation is laid for subsequent analysis; detection is finer through image blocking processing, suspected defect areas are judged through gray average values and variances, and the probability of misjudgment and missed judgment is reduced; segmenting defect content by using a gray scale global threshold, and accurately extracting shape features; and the defect template library and a normalized cross-correlation algorithm are combined to reliably judge the defect type. The system improves the detection efficiency and precision, provides powerful support for the maintenance of the large-span steel arch, and guarantees the safe and stable operation of the large-span steel arch.

Description

Technical Field

[0001] The present invention belongs to the technical field of defect detection, and specifically relates to a method and system for evaluating and detecting surface defects of a long-span steel arch. Background Technique

[0002] As a key load-bearing component of a building, the structural integrity and surface quality of a long-span steel arch play a decisive role in the safety and stability of the building. However, during long-term use, the long-span steel arch will be affected by various complex factors.

[0003] From the perspective of natural environmental factors, the steel arch is exposed to the air for a long time, and will be eroded by rainwater, resulting in rust on the steel surface, thereby reducing the strength and durability of the steel; long-term exposure to ultraviolet rays will also gradually deteriorate the performance of the steel; drastic temperature changes will cause the steel arch to expand and contract thermally, generating stress inside the structure and accelerating the fatigue damage of the material. In terms of human factors, the surface of the steel arch may be damaged during the use of the building due to various construction operations, collisions, etc.; in addition, if the maintenance work of the steel arch is not in place, it will also lead to the gradual accumulation and deterioration of surface defects.

[0004] At present, there are many deficiencies in the detection methods for surface defects of long-span steel arches. Traditional manual detection methods mainly rely on naked-eye observation and simple tools, which not only have low efficiency, but are also greatly affected by the subjective factors of the detection personnel and the detection environment, and it is difficult to ensure the accuracy and comprehensiveness of the detection. For some defects with strong concealment, it is even more difficult to detect manually. And some existing automated detection technologies such as ultrasonic detection and ray detection, although improve the detection efficiency to a certain extent, these methods also have limitations. Ultrasonic detection has limited detection accuracy for surface defects and requires high professional skills of the detection personnel; ray detection has radiation hazards, has strict requirements for the detection environment and equipment, high costs, and it is difficult to achieve comprehensive and rapid detection of the surface of a long-span steel arch. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and system for evaluating and detecting surface defects of a long-span steel arch to solve the problems raised in the above background technique.

[0006] The purpose of the present invention can be achieved through the following technical solutions: A method for evaluating and detecting surface defects of a long-span steel arch includes: The first step, camera array layout: A plurality of industrial cameras are arranged around the long-span steel arch and distributed according to a preset angle and spacing. The second step, image acquisition: Industrial cameras are started simultaneously to capture the surface of the long-span steel arch in a synchronous manner and collect multiple original images from different perspectives; The third step is image preprocessing: Perform grayscale processing and denoising on multiple original images; The fourth step is image block processing: Divide the grayscale image after image preprocessing into several sub-image blocks of the same size; The fifth step is block feature extraction: Extract features from the sub-image blocks and obtain the grayscale mean and grayscale variance of the sub-image blocks; The sixth step is defect determination: Obtain the pre-set grayscale mean range and grayscale variance range under normal conditions; select a sub-image block. If its grayscale mean is not within the grayscale mean range or its grayscale variance is not within the grayscale variance range, it is determined that the area where the sub-image block is located may have defects and is marked as a suspected defect area; The seventh step is defect feature extraction: Obtain the pre-set global grayscale threshold under normal conditions; in the suspected defect area, obtain the grayscale values of each pixel point, and then combine it with the pre-determined global grayscale threshold to segment the defect content of the sub-image block corresponding to the suspected defect area. Through defect content segmentation, obtain the binary image corresponding to the suspected defect area, and then determine the shape features of the defect content from the binary image; The eighth step is defect feature determination: According to the pre-established defect template library, obtain the corresponding defect sample images of the steel arch surface of various known types and forms collected in advance; Then extract the shape features of the defect content in the binary image corresponding to the suspected defect area, and then match its shape features with various defect sample images in the defect template library one by one; among them, defect matching uses the normalized cross-correlation algorithm to calculate the normalized cross-correlation coefficient between the defect sample image and the shape features corresponding to the defect content; Next, extract the pre-set similarity threshold; when the normalized cross-correlation coefficient between the corresponding shape features of the defect content and the relevant defect sample image is greater than the similarity threshold, it is determined that there is a defect match between the shape features of the defect content in the suspected defect area and the defect sample image, and then determine the defect type of the suspected defect area according to the defect type of the defect sample image.

[0007] As a further solution of the present invention, the image preprocessing method is as follows: Step A1: Grayscale processing: First, convert the original image into a grayscale image; The conversion method is as follows: Obtain the RGB values of each pixel point in the original image on the red channel, green channel, and blue channel; Then, through Calculate the gray value H of the corresponding pixel point when it is converted into a grayscale image; Step A2: Denoising processing: Use the median filtering method to filter the grayscale image to remove the noise introduced during the acquisition of the original image. The method is as follows: For each pixel point in the grayscale image, select a 3×3 neighborhood centered on it. Then, arrange the gray values of these 9 pixel points in ascending order and take the middle value as the new gray value of the relevant pixel point.

[0008] As a further solution of the present invention, the size of each sub-image block is set to a×a pixels, where a is a preset value; the number of divided sub-image blocks is determined according to the resolution of the grayscale image and the size of the sub-image block. Specifically: Through Calculate the number k of horizontally divided sub-image blocks and the number f of vertically divided sub-image blocks, and k×f is the number of divided sub-image blocks; In the formula, M×N is the resolution of the grayscale image, and it is a preset value; is the floor symbol.

[0009] As a further solution of the present invention, the block feature extraction method is as follows: Mark the gray values of each pixel point in the sub-image block as g ij , where i = 1, 2,..., a, j = 1, 2,..., a, and g ij represents the gray value of the pixel point in the i-th row and j-th column of the sub-image block; Then, through Calculate the gray mean value H1 and gray variance H2 of the sub-image block respectively.

[0010] As a further solution of the present invention, the defect content segmentation method is as follows: Compare the gray values of each pixel point in the suspected defect area with the global gray threshold: If the gray value is greater than the global gray threshold, then determine the pixel point as a pixel in the defect area, and then assign the gray value of the pixel point to 255, that is, adjust the pixel point to white; If the gray value is less than or equal to the global gray threshold, then determine the pixel point as a pixel in the normal area, and then assign the gray value of the pixel point to 0, that is, adjust the pixel point to black; Among them, the white part in the binary image is the shape feature of the detected defective content.

[0011] As a further solution of the present invention, the normalized cross - correlation algorithm is as follows:

[0012] In the formula, R(p, q) is the normalized cross - correlation coefficient between the defective sample image and the corresponding shape feature of the defective content, that is, the similarity between them; (p, q) is the position of the defective sample image in the corresponding shape feature of the defective content; M and N are the width and height of the defective sample image respectively; T(x1, y1) and K(x1, y1) are the pixel values at the corresponding coordinates (x1, y1) in the defective sample image and the corresponding shape feature of the defective content respectively; T0 and K0 are the means of all corresponding pixel values in the defective sample image and the corresponding shape feature of the defective content respectively.

[0013] A large - span steel arch surface defect evaluation and detection system, which is used to implement a large - span steel arch surface defect evaluation and detection method, includes: An image acquisition unit, which is used to take pictures of the surface of the large - span steel arch through a plurality of industrial cameras set around the large - span steel arch and collect multiple original images from different perspectives; An image processing unit, which is used to perform grayscale processing and denoising processing on the multiple original images, and divide the grayscale images after image pre - processing into several sub - image blocks of the same size; A defect screening unit, which is used to perform feature extraction processing on the sub - image blocks, obtain the grayscale mean and grayscale variance of the sub - image blocks, and then combine them with the pre - set grayscale mean range and grayscale variance range under normal conditions to determine whether the area where the relevant sub - image blocks are located is a suspected defect area; A defect analysis unit, which is used to obtain the grayscale values of each pixel point in the suspected defect area, and then combine them with the pre - set global grayscale threshold under normal conditions to segment the defective content of the sub - image block corresponding to the suspected defect area, obtain a binary image corresponding to the suspected defect area, and then determine the shape feature of the defective content from the binary image; A defect determination unit, which is used to obtain the corresponding defective sample images of various known types and forms on the surface of the steel arch collected in advance from a pre - established defect template library, then extract the shape features of the defective content in the binary image corresponding to the suspected defect area, and then perform defect matching on its shape features with various defective sample images in the defect template library one by one.

[0014] Compared with the prior art, the beneficial effects of the present invention are: Multi-angle comprehensive detection: By setting up multiple industrial cameras around the large-span steel arch at pre-set angles and spacings, the surface of the steel arch can be photographed from different perspectives at the same time, and multiple original images from different perspectives can be collected, thus achieving comprehensive coverage detection of the large-span steel arch surface, avoiding detection blind spots caused by a single perspective, and improving detection accuracy and reliability.

[0015] Effective image preprocessing: grayscale processing and denoising are performed on the collected original image. Grayscale processing converts color images into grayscale images, simplifies image data, and facilitates subsequent processing; median filtering denoising effectively removes the noise introduced during the image acquisition process, improves image quality, and makes subsequent image analysis results more accurate.

[0016] Reasonable image segmentation: Divide the preprocessed grayscale image into several sub-image blocks of the same size. The number of divisions is determined according to the resolution of the grayscale image and the size of the sub-image blocks. This segmentation method can analyze the image more carefully, improve the accuracy of defect detection, and facilitate subsequent feature extraction and defect judgment of the sub-image blocks.

[0017] Accurate defect determination: By calculating the grayscale mean and grayscale variance of the sub-image block and comparing them with the pre-set grayscale mean range and grayscale variance range under normal conditions, the suspected defect area can be determined more accurately to identify areas where defects may exist, thereby reducing misjudgments and missed judgments.

[0018] Accurate defect feature extraction: In the suspected defect area, the defect content is segmented based on the predetermined grayscale global threshold to obtain a binary image, which clearly extracts the shape features of the defect content, providing an accurate basis for subsequent defect type determination.

[0019] Reliable defect type determination: Based on the pre-established defect template library, the normalized cross-correlation algorithm is used to calculate the similarity between the defect sample image and the shape features corresponding to the defect content. When the similarity is greater than the threshold, the defect is determined to match and the defect type is determined. This method can more reliably determine the type of defect, which helps to take corresponding maintenance and repair measures in a targeted manner, thereby improving the maintenance efficiency and safety of large-span steel arches.

[0020] Data support and adaptability: The grayscale mean range, grayscale variance range and grayscale global threshold are obtained through multiple experimental analyses of multiple normal large-span steel arch surface images. They have reliable data support and can adapt to the surface characteristics of different large-span steel arches, thereby improving the versatility and adaptability of the detection method. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The present invention will be further described below in conjunction with the accompanying drawings.

[0022] Figure 1 It is the system block diagram of a large-span steel arch surface defect evaluation and detection system of the present invention.

[0023] Figure 2 It is the process schematic diagram of a large-span steel arch surface defect evaluation and detection method of the present invention. Specific implementation manners

[0024] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0025] Embodiment 1 Please refer to Figure 1 and Figure 2 As shown, a large-span steel arch surface defect evaluation and detection method includes: Step 1, Camera array layout: A plurality of industrial cameras are arranged around the large-span steel arch and distributed at a preset angle and spacing; Among them, the angular interval between adjacent industrial cameras is set to u, and the spacing is d; In this embodiment, it is assumed that the large-span steel arch has a circular cross-section, and it is also assumed that there are a total of n industrial cameras; Then the central angle between adjacent industrial cameras is u = (2π) / n, and the spacing d of the industrial cameras along the circumferential direction of the steel arch is calculated according to the radius r of the steel arch, that is, d = 2r×sin(u / 2); In this embodiment, by arranging a plurality of industrial cameras around the large-span steel arch at a preset angle and spacing, the surface of the steel arch can be photographed from different perspectives, and multiple original images from different perspectives can be collected, ensuring the comprehensive acquisition of the surface information of the steel arch and providing a rich data basis for subsequent accurate defect detection.

[0026] Step 2, Image acquisition: The industrial cameras are started simultaneously to photograph the surface of the large-span steel arch in a synchronous manner and collect multiple original images from different perspectives; Step 3, Image block processing: The grayscale image after image preprocessing is divided into several sub-image blocks of the same size; Among them, the size of each sub-image block is set to a×a pixels, and a is a preset value; The number of divided sub-image blocks is determined according to the resolution of the grayscale image and the size of the sub-image block; specifically: By calculating the number k of sub - image blocks divided horizontally and the number f of sub - image blocks divided vertically, and k×f is the number of divided sub - image blocks; In the formula, M×N is the resolution of the grayscale image, and it is a preset value; is the floor symbol; In this embodiment, the grayscale image is divided into several sub - image blocks of the same size, and the number of divided sub - image blocks is scientifically determined according to the resolution of the grayscale image and the size of the sub - image blocks. This way of dividing blocks helps to perform refined processing on the image and improve the accuracy and pertinence of defect detection.

[0027] Step Four: Defect feature extraction: Obtain the preset global grayscale threshold under normal conditions; Among them, the global grayscale threshold is to determine a threshold in advance based on the grayscale distribution of multiple normal large - span steel arch surface images. In this embodiment, the grayscale histogram method is used to count the frequencies of different grayscale values in the image, draw the grayscale histogram, and then observe the concentrated interval of the grayscale distribution in the normal area on the histogram, and select a grayscale value that can distinguish the normal area and the defective area; In the suspected defect area, obtain the grayscale values of each pixel point, and then combine them with the preset global grayscale threshold to segment the defect content of the sub - image block corresponding to the suspected defect area. Through the defect content segmentation, a binary image corresponding to the suspected defect area is obtained, and the method is as follows: Compare the grayscale values of each pixel point in the suspected defect area with the global grayscale threshold: If the grayscale value is greater than the global grayscale threshold, then determine this pixel point as a defective area pixel, and then assign the grayscale value of this pixel point to 255, that is, adjust this pixel point to white; If the grayscale value is less than or equal to the global grayscale threshold, then determine this pixel point as a normal area pixel, and then assign the grayscale value of this pixel point to 0, that is, adjust this pixel point to black; In this embodiment, after this processing, the suspected defect area is segmented into a binary image of black and white parts; Among them, the white part in the binary image is the shape feature of the detected defective content; This embodiment uses the grayscale histogram method to determine the global grayscale threshold, which can accurately distinguish the normal area and the defective area according to the grayscale distribution of the normal large - span steel arch surface image, and then effectively segment the defect content of the suspected defect area to obtain a clear binary image, highlighting the shape feature of the defect and facilitating subsequent defect determination.

[0028] Step Five: Defect feature determination: According to a pre-established defect template library, obtain corresponding defect sample images of the steel arch surface of various known types and forms collected in advance therefrom; In this embodiment, the corresponding defects on the steel arch surface are such as cracks, holes, etc.; for the defect sample images corresponding to the crack defects, their shape contours are slender; for the defect sample images corresponding to the hole defects, their shape contours are circular or approximately circular; Then extract the shape features of the defective content existing in the binary image corresponding to the suspected defect area, and then match the shape features one by one with various defect sample images in the defect template library; Among them, defect matching uses the normalized cross-correlation algorithm to calculate the similarity between the defect sample image and the shape features corresponding to the defective content. The specific method is as follows:

[0029] In the formula, R(p, q) is the normalized cross-correlation coefficient between the defect sample image and the shape features corresponding to the defective content, that is, the similarity between them; (p, q) is the position of the defect sample image in the shape features corresponding to the defective content; M and N are the width and height of the defect sample image respectively; T(x1, y1) and K(x1, y1) are the pixel values at the corresponding coordinates (x1, y1) in the defect sample image and the shape features corresponding to the defective content respectively; T0 and K0 are the means of all corresponding pixel values in the defect sample image and the shape features corresponding to the defective content respectively; Extract a pre-set similarity threshold; When the normalized cross-correlation coefficient between the shape features corresponding to the defective content and the relevant defect sample image is greater than the similarity threshold, it is determined that the shape features of the defective content in the suspected defect area match the defects of the defect sample image, and then determine the defect type of the suspected defect area according to the defect type of the defect sample image; For example, if the normalized cross-correlation coefficient of a certain area with the crack template exceeds the similarity threshold, it is determined that there is a crack defect in this area and its position is recorded; if it matches the hole template, it is determined to be a hole defect; In this embodiment, according to the pre-established defect template library, the normalized cross-correlation algorithm is used to calculate the similarity between the defect sample image and the shape features corresponding to the defective content, which can accurately match the shape features of the suspected defect area with various defect sample images of known types and forms, so as to accurately determine the type and position of the defect, and improve the accuracy and reliability of defect detection.

[0030] Embodiment 2 The technical solution of the second embodiment is different from that of the first embodiment in that this embodiment further includes: an image preprocessing step, which is after the image acquisition step and before the image block processing step, and this step performs grayscale processing and denoising processing on multiple original images; Step A1: Grayscale processing: First, convert the original image into a grayscale image; The conversion method is as follows: Obtain the RGB values of each pixel point in the original image on the red channel, green channel, and blue channel; Then, through Calculate the grayscale value H of the corresponding pixel point when converting to a grayscale image; Step A2: Denoising processing: Use the median filtering method to filter the grayscale image to remove the noise introduced during the acquisition of the original image. The method is as follows: For each pixel point in the grayscale image, select a 3×3 neighborhood centered on it. Then, arrange the grayscale values of these 9 pixel points in ascending order and take the middle value as the new grayscale value of the relevant pixel point; Take a pixel point in the grayscale image as an example and mark its coordinates as (x, y); Among them, x is the horizontal coordinate value in the grayscale image, and y is the vertical coordinate value in the grayscale image; Centered on (x, y), the pixel point coordinates of its neighborhood are (x - 1, y - 1), (x - 1, y), (x - 1, y + 1), (x, y - 1), (x, y), (x, y + 1), (x + 1, y - 1), (x + 1, y), (x + 1, y + 1).

[0031] In this embodiment, an image preprocessing step including grayscale processing and denoising processing is added after the image acquisition step. Grayscale processing converts the original image into a grayscale image, simplifies the color information of the image, and facilitates subsequent analysis and processing; denoising processing uses the median filtering method to remove the noise introduced during the acquisition of the original image, improves the quality of the image, reduces the interference of noise on defect detection, and further improves the accuracy of defect detection.

[0032] Embodiment Three As the third embodiment of the present invention, in the specific implementation of this application, compared with the first and second embodiments, the technical solution of this embodiment is to combine and implement the solutions of the above first and second embodiments. The difference between the technical solution of this embodiment and the first and second embodiments is only that this embodiment further includes a block feature extraction and a defect feature screening step: The block feature extraction step is to perform feature extraction processing on the sub-image blocks and obtain the grayscale mean and grayscale variance of the sub-image blocks; The method is as follows: Mark the gray value of each pixel point in the sub-image block as g ij , where i = 1, 2,..., a, j = 1, 2,..., a, and g ij represents the gray value of the pixel point in the i-th row and j-th column of the sub-image block; Subsequently, through calculate the gray mean value H1 and gray variance H2 of the sub-image block respectively; In this embodiment, the gray mean value reflects the average gray level of the sub-image block, and the gray variance reflects the degree of dispersion of the pixel gray values within the sub-image block; In this embodiment, the block feature extraction step extracts features from the image in two dimensions of the gray mean value and gray variance by calculating the gray mean value and gray variance of the sub-image block, which can more comprehensively reflect the feature information of the sub-image block. The gray mean value reflects the average gray level, and the gray variance reflects the degree of dispersion of the pixel gray values, providing a richer feature basis for defect detection.

[0033] The defect feature screening step is to obtain the preset gray mean value range [H1min, H2max] and gray variance range [H2min, H2max] under normal conditions; Among them, the gray mean value range and gray variance range are determined by statistically analyzing the distribution of pixel value differences through multiple experiments on multiple normal large-span steel arch surface images, so as to determine a reasonable range; Select a sub-image block. If its gray mean value H1 ∉ [H1min, H2max] or its gray variance H2 ∉ [H2min, H2max], that is, its gray mean value is not within the gray mean value range or its gray variance is not within the gray variance range, it is determined that there may be a defect in the area where the sub-image block is located, and it is marked as a suspected defect area.

[0034] In this embodiment, the defect feature screening step can accurately screen out the area where the sub-image block where a defect may exist, that is, the suspected defect area, according to whether the gray mean value and gray variance of the sub-image block are within the corresponding ranges, reducing the subsequent defect detection range, improving the detection efficiency, and also improving the detection accuracy of the defect.

[0035] Embodiment 4 As Embodiment 4 of the present invention, when the present application is specifically implemented, compared with Embodiment 1, Embodiment 2, and Embodiment 3, the technical solution of this embodiment is to combine and implement the solutions of the above Embodiment 1, Embodiment 2, and Embodiment 3.

[0036] In this embodiment, the solutions of Embodiment 1, Embodiment 2, and Embodiment 3 are combined and implemented. It synthesizes the advantages of comprehensive image acquisition, scientific image segmentation, effective defect feature extraction and determination, image quality improvement, refined defect feature extraction, and accurate screening of suspected defect regions. It can more comprehensively, accurately, and efficiently evaluate and detect the defects on the surface of a long-span steel arch, maximizing the accuracy, reliability, and efficiency of defect detection, and having high practical value and application prospects.

[0037] The present invention also proposes a system for evaluating and detecting defects on the surface of a long-span steel arch. This system is used to implement a method for evaluating and detecting defects on the surface of a long-span steel arch, including: An image acquisition unit, which is used to take pictures of the surface of the long-span steel arch through a plurality of industrial cameras arranged around the long-span steel arch and collect multiple original images from different perspectives; An image processing unit, which is used to perform grayscale processing and denoising processing on the multiple original images, and divide the grayscale images after image preprocessing into several sub-image blocks of the same size; A defect screening unit, which is used to perform feature extraction processing on the sub-image blocks, obtain the grayscale mean and grayscale variance of the sub-image blocks, and then combine them with the pre-set grayscale mean range and grayscale variance range under normal conditions to determine whether the area where the relevant sub-image blocks are located is a suspected defect area; A defect analysis unit, which is used to obtain the grayscale values of each pixel point in the suspected defect area, and then combine them with the pre-set global grayscale threshold under normal conditions to segment the defect content of the sub-image block corresponding to the suspected defect area and obtain a binary image corresponding to the suspected defect area, and then determine the shape features of the defect content existing in the binary image; A defect determination unit, which is used to obtain the corresponding defect sample images of various known types and forms on the surface of the steel arch collected in advance from a pre-established defect template library, then extract the shape features of the defect content existing in the binary image corresponding to the suspected defect area, and then perform defect matching of its shape features with various defect sample images in the defect template library one by one.

[0038] The above formulas are all dimensionless and take their numerical calculations. The formulas are obtained by collecting a large amount of data for software simulation to obtain a formula that is closest to the actual situation. The preset parameters and threshold selection in the formulas are set by those skilled in the art according to the actual situation.

[0039] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed in the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for evaluating and detecting surface defects of a long-span steel arch, characterized in that, Including: First step: Take pictures of the surface of the long-span steel arch through multiple industrial cameras set around the long-span steel arch, and collect multiple original images from different perspectives; Second step: Perform grayscale processing and denoising processing on the multiple original images, and divide the grayscale images after image preprocessing into several sub-image blocks of the same size; Third step: Perform feature extraction processing on the sub-image blocks, and obtain the grayscale mean value and grayscale variance of the sub-image blocks. Then, combine them with the pre-set grayscale mean value range and grayscale variance range under normal circumstances to determine whether the area where the relevant sub-image blocks are located is a suspected defect area; Fourth step: In the suspected defect area, obtain the grayscale values of each pixel point therein. Then, combine them with the pre-set global grayscale threshold under normal circumstances to segment the defect content of the sub-image block corresponding to the suspected defect area, and obtain the binary image corresponding to the suspected defect area. Then, determine the shape features of the defect content existing in the binary image; Fifth step: According to the pre-established defect template library, obtain the corresponding defect sample images of various known types and forms collected in advance on the surface of the steel arch. Then, extract the shape features of the defect content existing in the binary image corresponding to the suspected defect area, and then match its shape features with various defect sample images in the defect template library one by one; 2. The method for evaluating and detecting surface defects of a long-span steel arch according to claim 1, wherein The grayscale processing method is as follows: First, convert the original image into a grayscale image; The conversion method is as follows: Obtain the RGB values of each pixel point in the original image on the red channel, green channel, and blue channel; Subsequently, through the gray value H of the corresponding pixel point is calculated when it is converted into a grayscale image.

3. A method for evaluating and detecting surface defects of a long-span steel arch according to claim 1, characterized in that, The denoising processing method is as follows: Adopt the median filtering method to filter the grayscale image to remove the noise introduced during the acquisition of the original image. The method is as follows: For each pixel point in the grayscale image, select a 3×3 neighborhood centered on it. Then, arrange the grayscale values of these 9 pixel points in ascending order, and take the middle value as the new grayscale value of the relevant pixel point.

4. A method for evaluating and detecting surface defects of a long-span steel arch according to claim 1, characterized in that, Among them, The size of each sub-image block is set to a×a pixels, where a is a preset value; the number of divided sub-image blocks is determined according to the resolution of the grayscale image and the size of the sub-image block. Specifically: By calculating the number k of sub-image blocks divided horizontally and the number f of sub-image blocks divided vertically, and k×f is the number of divided sub-image blocks; Wherein, M×N is the resolution of the grayscale image, and it is a preset value; is the floor symbol.

5. A method for evaluating and detecting surface defects of a long-span steel arch according to claim 1, characterized in that, The calculation methods of the grayscale mean value and grayscale variance are as follows: Mark the grayscale value of each pixel point in the sub-image block as g ij , where i = 1, 2,..., a, j = 1, 2,..., a, and g ij represents the grayscale value of the pixel point in the i-th row and j-th column of the sub-image block; Subsequently, through calculate the grayscale mean H1 and grayscale variance H2 of the sub-image blocks respectively; Among them, when the grayscale mean value of the sub-image block is not within the grayscale mean value range or its grayscale variance is not within the grayscale variance range, it is determined that the area where the relevant sub-image block is located is a suspected defect area.

6. A method for evaluating and detecting surface defects of a long-span steel arch according to claim 1, characterized in that, The defect content segmentation method is as follows: Compare the grayscale values of each pixel point in the suspected defect area with the global grayscale threshold: If the grayscale value is greater than the global grayscale threshold, the pixel point is determined as a pixel in the defect area, and then the grayscale value of the pixel point is assigned 255, that is, the pixel point is adjusted to white; If the grayscale value is less than or equal to the global grayscale threshold, the pixel point is determined as a pixel in the normal area, and then the grayscale value of the pixel point is assigned 0, that is, the pixel point is adjusted to black; Among them, the white part in the binary image is the shape feature of the detected defect content.

7. A method for evaluating and detecting surface defects of a long-span steel arch according to claim 1, characterized in that, Among them, For defect matching, the normalized cross-correlation algorithm is used to calculate the normalized cross-correlation coefficient between the defect sample image and the corresponding shape feature of the defect content.

8. A method for evaluating and detecting surface defects of a long-span steel arch according to claim 7, characterized in that, The normalized cross-correlation algorithm is as follows: Wherein, R(p, q) is the normalized cross-correlation coefficient between the defective sample image and the corresponding shape feature of the defective content, that is, the similarity therebetween; (p, q) is the position of the defective sample image in the corresponding shape feature of the defective content, M and N are the width and height of the defective sample image respectively, T(x1, y1) and K(x1, y1) are the pixel values at the corresponding coordinates (x1, y1) of the defective sample image and the corresponding shape feature of the defective content respectively, and T0 and K0 are the means of all corresponding pixel values in the corresponding shape features of the defective sample image and the defective content respectively.

9. A method for evaluating and detecting surface defects of a long-span steel arch according to claim 7, characterized in that, When the normalized cross-correlation coefficient between the corresponding shape feature of the defective content and the relevant defective sample image is greater than the preset similarity threshold, it is determined that the shape feature of the defective content in the suspected defective area matches the defect of the defective sample image, and then the defect type of the suspected defective area is determined according to the defect type of the defective sample image.

10. A surface defect evaluation and detection system for a long-span steel arch, which is used to implement the surface defect evaluation and detection method for a long-span steel arch described in any one of claims 1-9, characterized in that, Including: An image acquisition unit, configured to photograph the surface of the long-span steel arch through a plurality of industrial cameras arranged around the long-span steel arch, and acquire multiple original images from different perspectives; An image processing unit, configured to perform grayscale processing and denoising processing on the multiple original images, and divide the grayscale images after image preprocessing into a plurality of sub-image blocks of the same size; A defect screening unit, configured to perform feature extraction processing on the sub-image blocks, obtain the grayscale mean and grayscale variance of the sub-image blocks, and then combine them with the preset grayscale mean range and grayscale variance range under normal conditions to determine whether the area where the relevant sub-image blocks are located is a suspected defective area; A defect analysis unit, configured to obtain the grayscale values of each pixel point in the suspected defective area, and then combine them with the preset global grayscale threshold under normal conditions to segment the defective content of the sub-image block corresponding to the suspected defective area, and obtain a binary image corresponding to the suspected defective area, and then determine the shape feature of the defective content from the binary image; A defect determination unit, configured to obtain the corresponding defective sample images of the steel arch surface with various known types and forms collected in advance from a pre-established defect template library, then extract the shape features of the defective content in the binary image corresponding to the suspected defective area, and then perform defect matching on the shape features one by one with various defective sample images in the defect template library.

Images