Standardized batch processing method, system and medium for structural surface defect detection data

Abstract

The present invention relates to a standardized batch processing method, system and medium for structural surface defect detection data. Based on the structural defect data collected by manual detection, the standard image is divided into grids, and then batches are exported according to the bitmap image format standard to form a defect standard map. Based on the defect standard map, image graying, image binarization and connected domain search methods are used in sequence to obtain multiple connected domains contained in the image. Each connected domain is subjected to image refinement to extract its skeleton, thereby obtaining the complete characteristics of the connected domain. Through the connected domain determination method, the defects are informatized and classified, and standardized data that can be quickly analyzed and processed by a computer is obtained to perform defect data analysis. Compared with the prior art, the present invention realizes rapid and standardized processing and sorting of large data of manually detected defects, and provides an effective basic data processing method for related research and targeted guidance of bridge maintenance.

Description

Technical Field

[0001] The present invention relates to the field of structural defect detection, and in particular to a standardized batch processing method, system and storage medium for structural apparent defect detection data. Background Art

[0002] Manual inspection is one of the main tasks in the management and maintenance of in-service bridges. Various types of disease detection data generated by manual visual inspection or measurement are key basic data for guiding bridge operation and maintenance decisions. However, due to the lack of standardized rapid batch processing methods for manual disease detection data, such data are often not fully mined and utilized, presenting problems such as difficulty in data processing, high cost, and weak timeliness, making it difficult to support actual bridge management and maintenance and related research needs.

[0003] CN101894328A discloses a CAD-based bridge defect management system and a bridge defect management and maintenance method, which utilizes CAD-related technologies to record and process defects in bridge components, including bridge system inspection and error recording, bridge assessment and prediction, and defect maintenance. Although this technical solution makes it more convenient for bridge management units to inspect and maintain bridge defects, it still cannot provide inspiration on how to fully mine and utilize the massive amount of original defect data collected by manual detection. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a standardized batch processing method for structural surface disease detection data, so as to realize the rapid and standardized processing and collation of big data of manual disease detection, and provide an effective basic data processing method for related research and targeted guidance of bridge maintenance.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] The first purpose of this application is to protect a standardized batch processing method for structural surface disease detection data, comprising the following steps:

[0007] S1: The structural damage data collected based on manual inspection is divided into grids and then exported in batches according to the bitmap image format standard to form a standard damage map;

[0008] S2: Based on the disease standard image, the image is processed by graying, binarizing and connected domain searching methods in sequence to obtain multiple connected domains contained in the image. The skeleton of each connected domain is further extracted by image thinning to obtain the complete features of the connected domain. The disease is informatized and classified by the connected domain determination method to obtain standardized data that can be quickly analyzed and processed by a computer;

[0009] S3: Output standardized data based on grid labels that can be quickly analyzed and processed by computers to perform disease data analysis.

[0010] Furthermore, the standard image is divided into grids based on the structural damage data collected based on manual detection, with specific points in the specific structure as boundaries.

[0011] Furthermore, a CAD diagram of the damage distribution is generated from the structural damage data collected based on manual detection, and then a grid-like rasterized diagram is performed.

[0012] Furthermore, the disease standard map includes appearance features and location features identified in coded form, and also includes a disease description identified in a specific field.

[0013] Furthermore, the disease standard map includes primary grid division and secondary grid division, and the secondary grid division is based on the single grid obtained by the primary grid division to perform secondary division, and disease positioning and classification are achieved through multi-level grid division.

[0014] Furthermore, in the process of graying the image, the RGB three components of the original color image are weighted averaged according to formula (1) to obtain a grayscale image:

[0015] Gray(i,j)=0.299×R(i,j)+0.578×G(i,j)+0.114×B(i,j) (7)

[0016] Where (i, j) is a specific pixel.

[0017] Furthermore, the maximum inter-class difference method is used to binarize the grayscale image in the image binarization process, wherein according to the grayscale characteristics of the image, the inter-class variance function between the foreground and the background is obtained, and the foreground and background are segmented by taking the maximum value, thereby performing image binarization under a global threshold scenario.

[0018] Furthermore, in the process of searching for the connected domain, a binary image connected region labeling method is used to obtain connected domains with different labels:

[0019] Assume that the array X0 is the same size as the digital image A, and satisfies the requirement that the foreground pixels of A are marked as 1 and the background pixels are marked as 0. Iterate formula (2) to extract each connected domain:

[0020]

[0021] In the formula, Represents the structural element B to X k-1 The expansion operation is defined as in formula (15);

[0022]

[0023] In the formula, the structure element B slides along the digital image A in rows and columns, and each time the maximum value of the area where B is located is extracted, which is the calculation result of the point. The image processing here is pixel by pixel. The structure element B is generally a 3x3 or 5x5 area, and its center overlaps with the [pixel] to be processed. [The point] is the pixel being processed.

[0024] Furthermore, in the image thinning process for each connected domain:

[0025] The thinning operation of digital image A using structure element B is defined as formula (16);

[0026]

[0027] Where A#B is the hit-miss operation of structure element B on A, defined as in formula (17);

[0028]

[0029] In the formula, is the erosion operation of structure element B1 on A, is the erosion operation of the structure element B2 on Ac, defined as formula (18), Ac is the foreground and background of A interchanged;

[0030]

[0031] In the formula, the structural element B is slid along the digital image A in rows and columns, and the minimum value of the area where B is located is extracted each time, which is taken as the calculation result of the point.

[0032] Furthermore, the structural apparent disease is a bridge disease, and the connected domain determination method comprises the following steps:

[0033] 1) Connected domain attribute extraction: The SAUF algorithm is used to obtain the attribute information of the connected domain itself, and the connected domain skeleton is obtained by adding a thinning operation. The data obtained include length l, area a, average width w = a / l, centroid abscissa x w , centroid ordinate y w , the upper left corner point has a horizontal coordinate of x1, the upper left corner point has a vertical coordinate of y1, the lower right corner point has a horizontal coordinate of x2, and the lower right corner point has a vertical coordinate of y2;

[0034] 2) Determination of the boom area: Obtain the boom distribution parameters through the file name and obtain the boom starting number h start , end number h end ;

[0035] 3) Determination of the diaphragm area: Obtain the diaphragm spacing parameters within the corresponding boom interval through the drawings, and obtain the diaphragm starting number d start , end number d end ;

[0036] 4) Determination of stiffening rib area: Obtain the stiffening rib spacing parameters within the corresponding hanger range through the drawings, and obtain the stiffening rib starting number r start , end number r end ;

[0037] 5) Disease classification: The aspect ratio of the connected domain is used as the basis for classification. If the aspect ratio is not less than 3, it means that the area is linear and is classified as L category. Otherwise, it means that the area is planar and is classified as A category.

[0038] 6) Classification of diseases: For Class A diseases, the difference between the start and end numbers of the diaphragm is determined. If the difference is not less than 2, the disease passes through the diaphragm and the subcategory is A-1. Otherwise, the disease does not pass through the diaphragm and the subcategory is A-2. For Class L diseases, firstly, the pixel values ​​of the upper left corner, centroid, and lower right corner are obtained by performing pixel value operation I(x, y). If the value is greater than zero, it means that there is a disease point at the point and the disease is defined as a straight line type, otherwise it is defined as a curve type. Then, the difference between the start and end numbers of the diaphragm is determined. If the difference is not less than 2, the disease passes through the diaphragm, otherwise it does not pass through the diaphragm. Finally, the slope of the line connecting the upper and lower corners is determined. If it exceeds tan(65°), the disease is defined as longitudinal bridge direction, otherwise it is defined as transverse bridge direction.

[0039] The second purpose of this application is to protect a standardized batch processing system for manual detection data of structural surface diseases, including:

[0040] A storage module storing an executable program corresponding to the above-mentioned structural surface disease detection data standardization batch processing method;

[0041] The I / O module can input the structural damage data collected by manual inspection and output the damage data analysis results to the user end;

[0042] The microprocessor is electrically connected to the storage module and the I / O module through main lines, and can output standardized data for rapid computer analysis and processing based on the structural disease data collected by manual detection, and then obtain the disease data analysis results.

[0043] Furthermore, the microprocessor is an ARM processor or an x86 processor.

[0044] Furthermore, the storage module is an SSD hard disk and an HHD hard disk.

[0045] The third object of the present application is to protect a storage medium containing computer executable instructions, which, when executed by a computer processor, is used to execute the above-mentioned structural surface disease detection data standardization batch processing method.

[0046] Compared with the prior art, the present invention has the following technical advantages:

[0047] 1) This technical solution realizes the rapid and standardized processing and collation of big data of manually detected defects, and provides an effective basic data processing method for related research and targeted guidance of bridge maintenance.

[0048] 2) The standardized batch processing method proposed in this technical solution has strong portability and important reference significance for the processing and analysis of large-scale and wide-ranging other component disease data and other types of data in manual inspection. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a flow chart of a standardized batch processing method for structural surface disease detection data in this technical solution;

[0050] Figure 2 This is an example diagram of the grid-type rasterization method of the pavement disease distribution image in this technical solution;

[0051] Figure 3 It is a flowchart of the standard image processing method in this technical solution;

[0052] Figure 4 Schematic diagram of the flow of the connected domain determination algorithm in this technical solution;

[0053] Figure 5 This is a cross-sectional view of the main beam of an embodiment of the present technical solution;

[0054] Figure 6 The lane and U-rib distribution diagram of the embodiment of the present technical solution;

[0055] Figure 7 This is an example diagram of the pavement disease image processing process in the embodiment of this technical solution;

[0056] Figure 8 This is an example diagram of paving inspection records at the same location at different times in the embodiment;

[0057] Fig. 9 The longitudinal disease quantity distribution diagram of the embodiment from February to April 2011;

[0058] Fig.10 The longitudinal disease quantity distribution diagram from February to April 2013 in the embodiment;

[0059] Fig.11The horizontal disease quantity distribution diagram from February to April 2011 in the embodiment;

[0060] Fig.12 This is the horizontal disease quantity distribution diagram from February to April 2013 in the embodiment;

[0061] Fig.13 This is an example diagram of the distribution of disease types in the embodiment. DETAILED DESCRIPTION

[0062] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0063] In this embodiment, bridge disease data is used as an example for explanation, but this technical solution is also applicable to other types of structural diseases.

[0064] 1. Classify and code the defects detected by bridge workers according to their shape, location and other factors.

[0065] Taking pavement defects as an example, according to the morphology, they can be divided into surface defects and linear defects. For bridge structures, the morphology and location of pavement defects reflect the stress characteristics and heavy load distribution of bridge structures to a certain extent. Therefore, pavement defects are first classified and coded according to their appearance characteristics and location characteristics, as shown in Table 1.

[0066] Table 1 Classification and coding of pavement defects

[0067]

[0068] In addition, for long-span suspension bridges, the hangers, main beam diaphragms and main beam stiffeners are numbered to describe the location of the disease:

[0069] Driving direction: North, South

[0070] Hanger: 001, 002, 003, ...

[0071] Diaphram: 001, 002, 003, ...

[0072] Stiffening ribs (Rib): 001, 002, 003, ...

[0073] For a single disease detected at a time, use the fields in Table 2 to describe the disease and form a standardized plan for disease treatment results.

[0074] Table 2 Disease description fields

[0075]

[0076] 2. Grid-based rasterization method for disease distribution images

[0077] In order to extract and locate pavement defects in batches, this embodiment first obtains the CAD defect data recorded by the maintenance personnel's inspection results, and exports them in batches according to the bitmap image format standard to form a defect description standard image, such as Figure 2 As shown in (a), the anchor points of adjacent hangers or cables are used as the upper and lower boundaries, and the road centerline and guardrail line are used as the left and right boundaries. The bridge deck area is divided into grids to form a single standard image range, thereby realizing the regional allocation of the corresponding defects.

[0078] Therefore, the whole bridge paving range is divided into (n-1)×2 grid areas by the hangers and the road centerline, where n represents the number of hangers, thus obtaining a single standard diagram, namely Figure 2 (b) The standard image naming format is defined as: {year month}_{direction}_{boom start point}_{boom end point}.jpg, such as 201102_North_001_002.jpg.

[0079] For standard drawings, subgrid division is performed according to diaphragms and U-ribs for disease location and classification, such as Figure 2 (c) as shown.

[0080] 3. Key parameter extraction method of disease image

[0081] After obtaining the standard image, the key disease parameters are automatically extracted through the analysis algorithm based on digital image processing. The process is as follows: Figure 3 As shown. For a single standard image, image grayscale, image binarization and connected domain search algorithm are performed in sequence to obtain the connected domains contained in the image, which are matched one by one with the diseases. For each connected domain, its skeleton is further extracted through the image thinning algorithm to obtain the complete features of the connected domain. Finally, the disease is informatized and classified through the process-based connected domain determination algorithm.

[0082] Grayscale and Binarization

[0083] First, the original color image is grayed out, and the weighted average of the three components of RGB is performed according to formula (13) to obtain a grayscale image.

[0084] Gray(i,j)=0.299×R(i,j)+0.578×G(i,j)+0.114×B(i,j) (13)

[0085] Then, the maximum inter-class difference method (Otsu method) is used to binarize the grayscale image. This method segmentes the foreground and background by taking the maximum value according to the grayscale characteristics of the image by obtaining the inter-class variance function between the foreground and background, thereby binarizing the image under the global threshold scenario. This method is simple and fast to calculate, and is less affected by image brightness and contrast.

[0086] Connected domain search

[0087] The results of disease detection are reflected in the digital image as connected domains. Connected domains refer to image regions composed of pixels with the same pixel value and adjacent positions. They are often used for binary images with only one type of foreground information. Connected component labeling (CCL) of binary images is a common operation in image processing. Let the array X0 be the same size as the digital image A, and satisfy the requirement that the foreground pixels of A are marked as 1 and the background pixels are marked as 0. Iterate formula (14) to extract each connected domain.

[0088]

[0089] In the formula, Represents the structural element B to X k-1 The expansion operation is defined as in formula (15).

[0090]

[0091] In the formula, the structural element B slides along the digital image A in rows and columns, and the maximum value of the area where B is located is extracted each time, which is taken as the calculation result of the pixel point.

[0092] SAUF was proposed by Wu and is one of the efficiency-optimized algorithms of CCL. It is a two-pass scanning method that performs pixel labeling and region merging in sequence to quickly obtain connected domains with different labels.

[0093] Refinement operation

[0094] For digital images, the connected domain can only obtain the area, corner points, and centroid points. When the object shape is complex, its linear characteristics cannot be obtained, such as the length information of the connected domain of the curve. Therefore, image thinning operation is introduced, and the thinning operation of the digital image A with the structure element B is defined as formula (16).

[0095]

[0096] Where A#B is the hit-or-miss transformation of structure element B on A, which is defined as in formula (17).

[0097]

[0098] In the formula, is the erosion operation of structure element B1 on A, is the erosion operation of the structure element B2 on Ac, defined as formula (18), where Ac is the foreground and background of A interchanged.

[0099]

[0100] In the formula, the structural element B slides along the digital image A in rows and columns, and the minimum value of the area where B is located is extracted each time, which is taken as the calculation result of the pixel point.

[0101] The optimized parallel thinning method proposed by Zhang is used to efficiently extract the connected domain skeleton. Since the skeleton width is 1 pixel, the skeleton area is the actual length of the original connected domain.

[0102] Connectivity domain determination algorithm

[0103] After obtaining the connected domain, it is necessary to use algorithms to determine and classify parameters in order to accurately obtain the disease label corresponding to a single connected domain. Figure 4 As shown in the figure, it is divided into six sub-processes: connected domain attribute extraction, hanger area determination, diaphragm area determination, stiffening rib area determination, major disease classification and minor disease classification.

[0104] 1) Connected domain attribute extraction: The SAUF algorithm is used to obtain the attribute information of the connected domain itself, and the connected domain skeleton is obtained by adding a thinning operation. The data obtained include length l, area a, average width w = a / l, centroid abscissa x w , centroid ordinate y w , the upper left corner point has a horizontal coordinate of x1, the upper left corner point has a vertical coordinate of y1, the lower right corner point has a horizontal coordinate of x2, and the lower right corner point has a vertical coordinate of y2.

[0105] 2) Determination of the boom area: Obtain the boom distribution parameters through the file name and obtain the boom starting number h start , end number h end .

[0106] 3) Determination of the diaphragm area: Obtain the diaphragm spacing parameters within the corresponding boom interval through the drawings, and obtain the diaphragm starting number d start , end number d end .

[0107] 4) Determination of stiffening rib area: Obtain the stiffening rib spacing parameters within the corresponding hanger range through the drawings, and obtain the stiffening rib starting number r start , end number r end .

[0108] 5) Classification of diseases: The aspect ratio of the connected domain is used as the basis for classification. If the aspect ratio is not less than 3, it means that the area is linear and is classified as Class L. Otherwise, it means that the area is planar and is classified as Class A.

[0109] 6) Disease classification: For Class A diseases, determine the difference between the start and end numbers of the partition. If the difference is not less than 2, the disease passes through the partition and the sub-category is A-1. Otherwise, the disease does not pass through the partition and the sub-category is A-2. Figure 4 For L-type defects, firstly, the pixel values ​​of the upper left corner, centroid, and lower right corner are obtained through pixel value operation I(x,y). If the value is greater than zero, it means that there is a defect point at the point, and the defect is defined as a straight line type, otherwise it is defined as a curve type; then the difference between the start and end numbers of the diaphragm is determined. If the difference is not less than 2, the defect passes through the diaphragm, otherwise the defect does not pass through the diaphragm; finally, the slope of the line connecting the upper and lower corners is determined. If it exceeds tan(65°), the defect is defined as the longitudinal bridge direction, otherwise it is defined as the transverse bridge direction.

[0110] In order to verify and analyze the proposed standardized batch processing method for pavement defects, the data of pavement defects detected manually for many years of a large-span suspension bridge was taken as an example. The data was processed according to the above process. The feasibility of the proposed method and its advantages of rapidity and standardization were verified through result analysis and efficiency calculation.

[0111] The background project is a large-span suspension bridge in East China, with six lanes in both directions and emergency lanes on both sides. The main span adopts a 1395-meter flat steel box girder with a beam width of 36.9m and a beam height of 3.0m. The main beam section is as follows Figure 5 As shown in the figure, epoxy asphalt concrete bridge deck is used for paving. The horizontal distribution from outside to inside is temporary parking lane, heavy lane, change lane, fast lane and central dividing strip, and the corresponding U-rib numbers are 1#~3#, 4#~10#, 11#~16#, 17#~23# and 24#, as shown in the figure. Figure 6 shown.

[0112] This embodiment uses manual pavement inspection data from 2011 to 2014, with an inspection cycle of 2 months and an inspection range covering the entire bridge.

[0113] Pavement disease data processing

[0114] First, according to the aforementioned grid-type rasterization method of the defect image in the technical solution, the manual inspection results of the background suspension bridge deck pavement are divided into regions and exported in batches according to the cable anchor points and the road centerline to form a standard defect map with time series and location information.

[0115] Then, the image processing algorithm flow described above in this technical solution is used to perform rapid batch processing of pavement defects based on digital image processing. Taking the pavement inspection results in February 2014 as an example, the exported 201402_North_001_002.jpg is as follows: Figure 7 As shown in (a), the left boundary of the image is the center line of the road, the right boundary is the guardrail line, the upper boundary extension line passes through the 2# sling, and the lower boundary extension line passes through the 1# sling. The grayscale image is as follows Figure 7 As shown in (b), the original color image is converted into a grayscale image. The result after Otsu's global threshold binarization is shown in (c), where the foreground (disease) and background (disease-free area) are binary separated, the background area is processed as a white area with a brightness of 1, and the diseased area is processed as a black area with a brightness of 0. The bitmap coordinates in the figure are all pixel coordinates (unit: px), and the corresponding physical coordinate conversion result is 8.84mm / px.

[0116] Next, the SAUF algorithm is used to search for connected domains. In order to distinguish different diseases, different colors or different numbers are used to express them, as shown in Figures (d) and (e), thereby obtaining information such as the area and coordinates of the connected domain. In addition, the Zhang parallel refinement algorithm is used to refine the connected domain as shown in Figure (f). It can be seen that the algorithm extracts the skeleton of each connected domain, thereby obtaining information such as the length of the connected domain. Then, according to the above-mentioned disease classification coding rules in this technical solution, the type information of the disease corresponding to the connected domain is obtained. Finally, the results of each disease extraction are saved and stored to form digital disease information.

[0117] In addition, the paving area between the north 1# and 2# slings is counted in time series, such as

[0118] Figure 8 As shown in the figure, the process of disease development and increase in number can be seen. Therefore, in order to judge the evolution process of the disease, the diseases in the same location at different times are merged to form disease statistical information with time attributes.

[0119] Using a computer with Intel(R)Core(TM)i7-6800K CPU@3.40GHz, 32GB RAM, and NVIDIAGeForce GTX 1080Ti, batch image processing and disease data generation for 3 years of data took only 6min3s to complete, reflecting the obvious speed advantage and standardized results of this method compared with manual counting.

[0120] Disease Statistical Analysis

[0121] Quantity and location analysis

[0122] The longitudinal distribution of disease quantity from February to April 2011 and from February to April 2013 was analyzed and statistically analyzed. Fig. 9 and Fig.10 It can be seen that the pavement defects are evenly distributed along the longitudinal direction of the bridge, and the number of defects will increase significantly before artificial repair.

[0123] The longitudinal distribution of disease quantity from February to April 2011 and from February to April 2013 was analyzed and statistically analyzed. Fig.11 , Fig.12 It can be seen that the pavement defects are unevenly distributed along the transverse direction of the bridge. The number of defects in the emergency parking lane, fast lane and central dividing strip is very small, while most of them appear in the heavy lane and change lane. Among them, 5#, 10#, 11# and 16# are the most common rutting positions. Therefore, the defect detection results have a certain correlation with the bridge deck driving rules.

[0124] Type Analysis

[0125] The automatically identified pavement disease types are statistically analyzed, such as Fig.13 As shown in the figure, the number of linear diseases such as cracks is much greater than the number of surface diseases such as pits and bulges, and among the linear diseases, the L-3 type accounts for the majority, that is, the number of linear diseases that do not pass through the diaphragm is dominant.

[0126] The above description of the embodiments is to facilitate the understanding and use of the invention by those skilled in the art. It is obvious that those skilled in the art can easily make various modifications to these embodiments and apply the general principles described herein to other embodiments without creative work. Therefore, the present invention is not limited to the above embodiments, and improvements and modifications made by those skilled in the art based on the disclosure of the present invention without departing from the scope of the present invention should be within the scope of protection of the present invention.

Claims

1. A method for standardizing batch processing of structural apparent disease detection data, characterized in that: The following steps are involved: S1: Based on the structural damage image data collected by manual inspection, grid-type rasterization is performed, and then batch export is performed according to the bitmap image format standard to form a standard damage map; S2: Based on the disease standard image, the image is processed by graying, binarizing and connected domain searching methods in sequence to obtain multiple connected domains contained in the image. The skeleton of each connected domain is further extracted by image thinning to obtain the complete features of the connected domain. The disease is informatized and classified by the connected domain determination method to obtain standardized data that can be quickly analyzed and processed by a computer; S3: Output standardized data that can be quickly analyzed and processed by computers based on grid labels to analyze disease data; The apparent structural damage is a bridge damage, and the connected domain determination method comprises the following steps: 1) Connected domain attribute extraction: The SAUF algorithm is used to obtain the attribute information of the connected domain itself, and the connected domain skeleton is obtained by adding a thinning operation. The data obtained include length l, area a, average width w = a / l, centroid abscissa x w , centroid ordinate y w , the upper left corner point has a horizontal coordinate of x1, the upper left corner point has a vertical coordinate of y1, the lower right corner point has a horizontal coordinate of x2, and the lower right corner point has a vertical coordinate of y2; 2) Determination of the boom area: Obtain the boom distribution parameters through the file name and obtain the boom starting number h start , end number h end ; 3) Determination of the diaphragm area: Obtain the diaphragm spacing parameters within the corresponding boom interval through the drawings, and obtain the diaphragm starting number d start , end number d end ; 4) Determination of stiffening rib area: Obtain the stiffening rib spacing parameters within the corresponding hanger range through the drawings, and obtain the stiffening rib starting number r start , end number r end ; 5) Disease classification: The aspect ratio of the connected domain is used as the basis for classification. If the aspect ratio is not less than 3, it means that the area is linear and is classified as L category. Otherwise, it means that the area is planar and is classified as A category. 6) Classification of defects: For Class A defects, determine the difference between the start and end numbers of the diaphragm. If the difference is not less than 2, the defect passes through the diaphragm and the subcategory is A-1. Otherwise, the defect does not pass through the diaphragm and the subcategory is A-2. For Class L defects, firstly, the pixel values ​​of the upper left corner, centroid, and lower right corner are obtained by performing pixel value operation I(x, y). If the value is greater than zero, it indicates that there is a defect point at the point and the defect is defined as a straight line. Otherwise, it is defined as a curve. Then determine the difference between the start and end numbers of the diaphragm. If the difference is not less than 2, the defect passes through the diaphragm. Otherwise, the defect does not pass through the diaphragm. Finally, determine the slope of the line connecting the upper and lower corner points. If it exceeds tan65°, the defect is defined as longitudinal bridge direction. Otherwise, it is defined as transverse bridge direction.

2. A method for standardizing batch processing of structural surface defect detection data according to claim 1, characterized in that: The disease standard map includes appearance feature information and location feature information marked in coded form, and also includes disease description information marked in specific fields.

3. A method for standardizing batch processing of structural surface defect detection data according to claim 1, characterized in that: The disease standard map includes primary grid division and secondary grid division. The secondary grid division is based on a single grid obtained by the primary grid division to perform secondary division. Disease location and classification are achieved through multi-level grid division.

4. A method for standardizing batch processing of structural surface defect detection data according to claim 1, characterized in that: In the process of image graying, the RGB three components of the original color image are weighted averaged according to formula (1) to obtain a gray image: Gray(i,j)=0.299×R(i,j)+0.578×G(i,j)+0.114×B(i,j) (1) Where (i, j) is a specific pixel.

5. A method for standardizing batch processing of structural surface defect detection data according to claim 1, characterized in that: In the image binarization process, the maximum inter-class difference method is used to binarize the grayscale image, wherein according to the grayscale characteristics of the image, the inter-class variance function between the foreground and the background is obtained, and the foreground and background are segmented by taking the maximum value, thereby performing image binarization in a global threshold scenario.

6. A method for standardizing batch processing of structural surface defect detection data according to claim 1, characterized in that: In the process of searching for connected domains, a binary image connected region labeling method is used to obtain connected domains with different labels: Assume that the array X0 is the same size as the digital image A, and satisfies the requirement that the foreground pixels of A are marked as 1 and the background pixels are marked as 0. Iterate formula (2) to extract each connected domain: In the formula, Represents the structural element B to X k-1 The expansion operation is defined as in formula (3); In the formula, the structural element B is slid along the digital image A in rows and columns, and the maximum value of the area where B is located is extracted each time, which is taken as the calculation result of the selected pixel point.

7. A method for standardizing batch processing of structural surface defect detection data according to claim 6, characterized in that: The image thinning process for each connected domain is as follows: The thinning operation of digital image A using structure element B is defined as formula (4); Where A#B is the hit-miss operation of structure element B on A, defined as in formula (5); In the formula, is the erosion operation of structure element B1 on A, For the structural element B2 to A c The erosion operation is defined as in formula (6), A c To swap the foreground and background of A; In the formula, the structural element B is slid along the digital image A in rows and columns, and the minimum value of the area where B is located is extracted each time as the calculation result of the selected pixel point.

8. A standardized batch processing system for manual detection of structural surface defects, characterized in that: include: A storage module storing an executable program corresponding to the standardized batch processing method for structural surface defect detection data according to any one of claims 1 to 7; The I / O module can input the structural damage data collected by manual inspection and output the damage data analysis results to the user end; The microprocessor is electrically connected to the storage module and the I / O module through main lines, and can output standardized data for rapid computer analysis and processing based on the structural disease data collected by manual detection, and then obtain the disease data analysis results.

9. A storage medium containing computer executable instructions, characterized in that: The storage medium of the computer executable instructions is used to execute the standardized batch processing method for structural surface defect detection data according to any one of claims 1 to 7 when executed by a computer processor.

Images