Steel structure defect identification system and method based on image analysis

Abstract

The invention relates to the technical field of steel structure image analysis, and discloses a steel structure defect recognition system and method based on image analysis, the system comprises a data acquisition module, an image data analysis module and a steel structure defect recognition and judgment module, and the data acquisition module comprises a region division unit and an image data acquisition unit. According to the system, abnormal point locations appearing after ultrasonic detection of a steel structure are obtained, region labeling is carried out, a clear image is analyzed, a corresponding region image influence coefficient and a local defect influence coefficient are output, and when the local defect influence coefficient exceeds a local defect threshold value, early warning is carried out; and when the local defect influence coefficient does not exceed the local defect threshold value, analyzing the overall image data, and judging whether to carry out early warning or not based on the overall image defect coefficient, so that the steel structure surface defect identification can be quickly and accurately completed, the situation that the equipment computing power is occupied by an area which does not need to be detected due to the overall image analysis is avoided, and the detection efficiency is improved. And the detection time is shortened.

Description

Technical Field

[0001] The present invention relates to the technical field of steel structure image analysis, and in particular to a steel structure defect recognition system and method based on image analysis. Background Art

[0002] Steel structure image recognition is of great significance in the fields of engineering and manufacturing. It can be used to detect surface defects, cracks, deformations, and other problems, assisting with quality control and timely repairs to ensure the safety and reliability of steel structures. Image recognition technology can automatically detect and identify steel structure components, improving production efficiency and product quality. It can also be used for positioning and identification in automated assembly processes, enabling real-time monitoring of steel structures and timely detection and identification of potential problems, allowing for early maintenance and overhaul, extending equipment life, and reducing maintenance costs.

[0003] However, existing image recognition technologies analyze and judge the overall image of the steel structure. When judging long steel structures, due to their limited width, the local clarity of the main steel structure in the acquired image is low, which may lead to misjudgment. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a steel structure defect identification system and method based on image analysis, which has the advantages of quickly and accurately identifying defects and solves the above-mentioned technical problems.

[0005] To achieve the above-mentioned object, the present invention provides the following technical solutions: a steel structure defect recognition system based on image analysis, comprising a data acquisition module, an image data analysis module and a steel structure defect recognition and judgment module;

[0006] The data acquisition module includes a region division unit and an image data acquisition unit. The region marking unit is used to obtain abnormal points appearing on the steel structure after ultrasonic testing, mark the regions, and transmit the divided regions to the image data acquisition unit. The image data acquisition unit is used to obtain image data of each region of the steel structure marked in the region marking unit and image data of the entire steel structure.

[0007] The image data analysis module includes an image blur degree analysis unit, a regional image analysis unit, and an overall image analysis unit. The image blur degree analysis unit is used to analyze the image data of each area of ​​the steel structure in the image data acquisition unit, and provide a clarity coefficient for each area. Based on the clarity coefficient of each area, it is determined whether to re-photograph the corresponding area. The regional image analysis unit is used to perform a secondary analysis on the area where the judgment has been completed, and output the corresponding regional image influence coefficient.

[0008] The steel structure defect recognition and judgment module performs a comprehensive calculation on the steel structure to be inspected based on all regional image influence coefficients to finally obtain a local defect influence coefficient. When the local defect influence coefficient exceeds the local defect threshold, an early warning is issued. When the local defect influence coefficient does not exceed the local defect threshold, the overall image analysis unit in the image data analysis module is called to analyze the overall image data and output the overall image defect coefficient. The steel structure defect recognition and judgment module determines whether to issue an early warning based on the overall image defect coefficient.

[0009] As a preferred technical solution of the present invention, the specific expression for the image data acquisition unit to acquire the image data of each area of ​​the steel structure marked in the area marking unit is as follows:

[0010] TXSJ=[TSXJ1,…,TSXJ i ,…,TSXJ I ]

[0011] Among them, TXSJ represents the regional image dataset, TSXJ1,…,TSXJ i ,…,TSXJ I They respectively represent the image data of the 1st region, ..., the image data of the i-th region, ..., the image data of the I-th region, i∈[1,I], and the image data of the i-th region is obtained by photographing with a mobile photographing device.

[0012] As a preferred technical solution of the present invention, the image blur degree analysis unit is used to analyze the image data of each area of ​​the steel structure in the image data acquisition unit, and provide a clarity coefficient for each area. The specific steps of determining whether to re-photograph the corresponding area based on the clarity coefficient of each area are as follows:

[0013] Step A1: Get the shooting factor PSYZ of the i-th area shooting i ;

[0014] Step A2: Get the image factor GRYZ captured in the i-th region i ;

[0015] Step A3: Comprehensively calculate the clarity coefficient QXXS of the i-th area i , the specific expression is as follows:

[0016] QXXS i =ω1*WJFSYZ i +ω2*GRYZ i

[0017] Among them, ω1 and ω2 represent two weight coefficients whose sum is 1;

[0018] Step A4: Determine the clarity coefficient QXXS of the i-th area i If it exceeds the clear threshold, then terminate step A4 and convert the image data TSXJ of the i-th region into i Send it to the regional image analysis unit. If it does not exceed the clear threshold, retake it and set the number of retakes CP for the i-th region. i = N, N represents the number of times step A4 is executed, when the number of retakes CP of the i-th region is i An error is reported when the maximum number of retakes is exceeded.

[0019] As a preferred technical solution of the present invention, the data acquisition module further stores a set of black area ratios after grayscale processing of the original image of the steel structure in each area;

[0020] In step A1, the shooting factor PSYZ of the i-th region is obtained. i The specific steps are as follows:

[0021] Step A1.1: Read the image data TSXj of the i-th region i , the image data TSXJ of the i-th region i Perform binarization processing;

[0022] Step A1.2: Get the proportion of all black areas in the i-th area MJZB i , and calculate the shooting factor PSYZ of the i-th area shooting i , the specific expression is as follows:

[0023]

[0024] Among them, |MJZB i -MJZB i,0 | indicates MJZB i -MJZB i,0 The absolute value of MJZB i,0 Indicates the proportion of the black area in the i-th area stored in the percentage set, MJZB i Indicates the proportion of black area in the i-th region.

[0025] As a preferred technical solution of the present invention, the image factor GRYZ of the i-th region is obtained in step A2. i The specific steps are as follows:

[0026] Step A2.1: Read the number of shakes of the mobile camera when shooting the i-th area DDCS i ;

[0027] Step A2.2: Calculate the image factor GRYZ captured in the i-th region i, the specific expression is as follows:

[0028]

[0029] Among them, e represents the natural constant, DDCS i represents the number of times the mobile shooting device shakes when shooting the i-th area, and σ represents the balance factor to avoid the denominator being 0.

[0030] As a preferred technical solution of the present invention, the specific steps of the regional image analysis unit for performing secondary analysis on the determined region and outputting the corresponding regional image influence coefficient are as follows:

[0031] Step B1: Read the image data TSXJ of the i-th area i , and crop the image to keep only the steel structure area;

[0032] Step B2: Identify the surface of the steel structure and perform image data TSXJ on the i-th region. i Perform grayscale processing;

[0033] Step B3: Identify all pixels in the non-black area in step B2, and cluster adjacent non-black pixels to obtain several cluster areas;

[0034] Step B4: Get the area JX of the minimum rectangular outline of the jth cluster area j and the area of ​​the minimum closed circle contour YX j ;

[0035] Step B5: Determine the area JX of the minimum rectangular outline of the jth cluster area j and the area of ​​the minimum closed circle contour YX j Whether the smaller one exceeds the area threshold, if not, the j+1th cluster area is judged, if it exceeds, the smaller area is stored until all cluster areas are judged, and the regional image influence coefficient QYYXXS of the i-th area is calculated i , the specific expression is as follows:

[0036]

[0037] Among them, min{JX j ,YX j} represents the area of ​​the minimum rectangular outline of the jth cluster area JX j and the area of ​​the minimum closed circle contour YX j The smaller one, QYYXXS i represents the regional image influence coefficient of the i-th region, ZMJ represents the total area of ​​the cropped image, It means summing the contour areas outputted by a total of J cluster regions, j∈[1,J].

[0038] As a preferred technical solution of the present invention, the steel structure defect recognition and judgment module performs a comprehensive calculation on the steel structure to be detected based on all regional image influence coefficients to finally obtain the specific expression of the local defect influence coefficient as follows:

[0039]

[0040] Among them, JBQX represents the local defect influence coefficient, It means summing the regional image influence coefficients of a total of I regions, i∈[1,I].

[0041] As a preferred technical solution of the present invention, the specific steps of the overall image analysis unit analyzing the overall image data and outputting the overall image defect coefficient are as follows:

[0042] Step C1: Obtain the overall image data of the steel structure and obtain the lengths of both sides of the steel structure respectively;

[0043] Step C2: Calculate the overall image defect coefficient CYZ of the steel structure. The specific expression is as follows:

[0044]

[0045] Among them, l a and l b Respectively represent the length of different sides of the steel structure, CYZ represents the overall image defect coefficient of the steel structure, |l a -l b | indicates l a -l b The absolute value of .

[0046] As a preferred technical solution of the present invention, the specific steps of the steel structure defect recognition and judgment module for judging whether to issue an early warning based on the overall image defect coefficient are: when the overall image defect coefficient CYZ of the steel structure exceeds the overall judgment threshold δ, an early warning is issued; when the overall image defect coefficient CYZ of the steel structure does not exceed the overall judgment threshold δ, no early warning is issued.

[0047] The present invention also provides a steel structure defect recognition method based on image analysis, which is based on the above-mentioned steel structure defect recognition system based on image analysis, comprising the following steps:

[0048] Step 1: Obtain abnormal points on the steel structure after ultrasonic testing and mark the areas;

[0049] Step 2: Obtain image data of each marked area of ​​the steel structure and image data of the entire steel structure;

[0050] Step 3: Analyze the image data of each area of ​​the steel structure, and give a clarity coefficient for each area. Based on the clarity coefficient of each area, determine whether to re-photograph the corresponding area;

[0051] Step 4: Perform secondary analysis on the determined area and output the corresponding regional image influence coefficient;

[0052] Step 5: Based on all the regional image influence coefficients, the steel structure to be inspected is comprehensively calculated to finally obtain the local defect influence coefficient;

[0053] Step 6: When the local defect influence coefficient exceeds the local defect threshold, an early warning is issued. When the local defect influence coefficient does not exceed the local defect threshold, the overall image data is analyzed and the overall image defect coefficient is output. At the same time, whether to issue an early warning is determined based on the overall image defect coefficient.

[0054] Compared with the existing technology, the present invention provides a steel structure defect identification system and method based on image analysis, which has the following beneficial effects:

[0055] The present invention obtains abnormal points that appear on the steel structure after ultrasonic testing, marks the area, takes images of the marked areas, gives a clarity coefficient for each area, and determines whether to re-shoot the corresponding area based on the clarity coefficient of each area. The clear image is analyzed to output the corresponding regional image influence coefficient and local defect influence coefficient. When the local defect influence coefficient exceeds the local defect threshold, an early warning is issued. When the local defect influence coefficient does not exceed the local defect threshold, the overall image data is analyzed, and it is determined whether to issue an early warning based on the overall image defect coefficient. This allows for rapid and accurate identification of surface defects in steel structures, avoids the situation where areas that do not need to be inspected occupy equipment computing power due to overall image analysis, and shortens the inspection time. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 Schematic diagram of the system framework of the present invention;

[0057] Figure 2 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0058] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0059] See also Figure 1 - Figure 2 , a steel structure defect recognition system based on image analysis, including a data acquisition module, an image data analysis module and a steel structure defect recognition and judgment module;

[0060] The data acquisition module includes a region division unit and an image data acquisition unit. The region labeling unit is used to obtain abnormal points that appear on the steel structure after ultrasonic testing, and to label the regions. The divided regions are then transmitted to the image data acquisition unit. The image data acquisition unit is used to obtain the image data of each region of the steel structure marked in the region labeling unit and the overall image data of the steel structure. The specific expression for the image data acquisition unit to obtain the image data of each region of the steel structure marked in the region labeling unit is as follows:

[0061] TXSJ=[TSXJ1,…,TSXJ i ,…,TSXJ I ]

[0062] Among them, TXSJ represents the regional image dataset, TSXJ1,…,TSXJ i ,…,TSXJ I They represent the image data of the 1st region, ..., the image data of the i-th region, ..., the image data of the I-th region, respectively, i∈[1,I], and the image data of the i-th region is captured by a mobile shooting device. The mobile shooting device here can be a crawler robot or a drone. Since the ultrasonic detection gives the distance of the specific abnormal point, the crawler robot or drone can be moved to the target point and capture the image after it stabilizes;

[0063] The image data analysis module includes an image blur degree analysis unit, a regional image analysis unit, and an overall image analysis unit. The image blur degree analysis unit is used to analyze the image data of each area of ​​the steel structure in the image data acquisition unit, and provide a clarity coefficient for each area. Based on the clarity coefficient of each area, it is determined whether to re-photograph the corresponding area. The regional image analysis unit is used to perform a secondary analysis on the area where the judgment has been completed, and output the corresponding regional image influence coefficient.

[0064] The image blur degree analysis unit is used to analyze the image data of each area of ​​the steel structure in the image data acquisition unit, and provide a clarity coefficient for each area. The specific steps of determining whether to re-photograph the corresponding area based on the clarity coefficient of each area are as follows:

[0065] Step A1: Get the shooting factor PSYZ of the i-th area shooting i ;

[0066] In step A1, the shooting factor PSYZ of the i-th area is obtained i The specific steps are as follows:

[0067] Step A1.1: Read the image data TSXJ of the i-th region i , the image data TSXJ of the i-th region i Perform binarization processing;

[0068] Step A1.2: Get the proportion of all black areas in the i-th area MJZB i , and calculate the shooting factor PSYZ of the i-th area shooting i , the specific expression is as follows:

[0069]

[0070] Among them, |MJZB i -MJZB i,0 | indicates MJZB i -MJZB i,0 The absolute value of MJZB i,0 Indicates the proportion of the black area in the i-th area stored in the percentage set, MJZB i represents the proportion of black area in the i-th region;

[0071] Step A2: Get the image factor GRYZ captured in the i-th region i ;

[0072] In step A2, the image factor GRYZ of the i-th region is obtained i The specific steps are as follows:

[0073] Step A2.1: Read the number of shakes of the mobile camera when shooting the i-th area DDCS i ;

[0074] Step A2.2: Calculate the image factor GRYZ captured in the i-th region i , the specific expression is as follows:

[0075]

[0076] Among them, e represents the natural constant, DDCS i represents the number of times the mobile camera shakes when shooting the i-th area, and σ represents the balance factor to avoid the denominator being zero;

[0077] Step A3: Comprehensively calculate the clarity coefficient QXXS of the i-th area i , the specific expression is as follows:

[0078] QXXS i=ω1*WJFSYZ i +ω2*GRYZ i

[0079] Among them, ω1 and ω2 represent two weight coefficients whose sum is 1;

[0080] Step A4: Determine the clarity coefficient QXXS of the i-th area i If it exceeds the clear threshold, then terminate step A4 and convert the image data TSXJ of the i-th region into i Send it to the regional image analysis unit. If it does not exceed the clear threshold, retake it and set the number of retakes CP for the i-th region. i = N, N represents the number of times step A4 is executed, when the number of retakes CP of the i-th region is i An error is reported when the maximum number of retakes is exceeded.

[0081] The data acquisition module also stores the set of black area ratios after grayscale processing of the original steel structure image of each area;

[0082] The specific steps of the regional image analysis unit for performing secondary analysis on the determined region and outputting the corresponding regional image influence coefficient are as follows:

[0083] Step B1: Read the image data TSXJ of the i-th area i , and crop the image to keep only the steel structure area;

[0084] Step B2: Identify the surface of the steel structure and perform image data TSXJ on the i-th region. i Perform grayscale processing;

[0085] Step B3: Identify all pixels in the non-black area in step B2, and cluster adjacent non-black pixels to obtain several cluster areas;

[0086] Step B4: Get the area JX of the minimum rectangular outline of the jth cluster area j and the area of ​​the minimum closed circle contour YX j ;

[0087] Step B5: Determine the area JX of the minimum rectangular outline of the jth cluster area j and the area of ​​the minimum closed circle contour YX j Whether the smaller one exceeds the area threshold, if not, the j+1th cluster area is judged, if it exceeds, the smaller area is stored until all cluster areas are judged, and the regional image influence coefficient QYYXXS of the i-th area is calculated i , the specific expression is as follows:

[0088]

[0089] Among them, min{JX j ,YX j} represents the area of ​​the minimum rectangular outline of the jth cluster area JX j and the area of ​​the minimum closed circle contour YX j The smaller one, QYYXXS i represents the regional image influence coefficient of the i-th region, ZMJ represents the total area of ​​the cropped image, It means summing the contour areas outputted by a total of J cluster regions, j∈[1,J].

[0090] The steel structure defect recognition and judgment module performs a comprehensive calculation on the steel structure to be inspected based on all regional image influence coefficients to finally obtain the local defect influence coefficient. When the local defect influence coefficient exceeds the local defect threshold, an early warning is issued. When the local defect influence coefficient does not exceed the local defect threshold, the overall image analysis unit in the image data analysis module is called to analyze the overall image data and output the overall image defect coefficient. The steel structure defect recognition and judgment module determines whether to issue an early warning based on the overall image defect coefficient.

[0091] The steel structure defect recognition and judgment module performs a comprehensive calculation on the steel structure to be inspected based on all regional image influence coefficients and finally obtains the specific expression of the local defect influence coefficient as follows:

[0092]

[0093] Among them, JBQX represents the local defect influence coefficient, It represents the summation of the regional image influence coefficients of a total of I regions, i∈p1,I]. The specific steps of the overall image analysis unit analyzing the overall image data and outputting the overall image defect coefficient are as follows:

[0094] Step C1: Obtain the overall image data of the steel structure and obtain the lengths of both sides of the steel structure respectively;

[0095] Step C2: Calculate the overall image defect coefficient CYZ of the steel structure. The specific expression is as follows:

[0096]

[0097] Among them, l a and l b Respectively represent the length of different sides of the steel structure, CYZ represents the overall image defect coefficient of the steel structure, |l a -l b | indicates l a -l bThe absolute value of .

[0098] The specific steps of the steel structure defect recognition and judgment module for judging whether to issue an early warning based on the overall image defect coefficient are as follows: when the overall image defect coefficient CYZ of the steel structure exceeds the overall judgment threshold δ, an early warning is issued; when the overall image defect coefficient CYZ of the steel structure does not exceed the overall judgment threshold δ, no early warning is issued;

[0099] When issuing an early warning, the photos taken will be sent to the staff for review.

[0100] Example 1:

[0101] The results of the ultrasonic testing conducted during this implementation revealed two abnormalities, which were captured using a drone;

[0102] When photographing the first abnormality, the clarity coefficient of the first area is determined as shown in Table 1 below:

[0103] Table 1

[0104] parameter Numerical <![CDATA[MJZB1]]> 80％ <![CDATA[MJZB 1,0 ]]> 83％ <![CDATA[DDCS1]]> 0 σ 0.45 <![CDATA[ω1]]> 0.56 <![CDATA[ω2]]> 0.44 Clear Threshold 0.90

[0105] PSYZ1 = 1-0.036 = 0.963, GRYZ1 = 0.891. At this time, the calculated QXXS1 = 0.93132 > 0.90 is obtained. At this time, the judgment is clear. Two clusters are obtained through the regional image analysis unit. The specific parameters are shown in Table 2 below;

[0106] Table 2

[0107] parameter Numerical ZMJ 122 <![CDATA[JX1]]> 16 <![CDATA[YX1]]> 12.56 <![CDATA[JX2]]> 12 <![CDATA[YX2]]> 28.27

[0108] At this time, the calculation results are QYYXXS1=0.201;

[0109] When photographing the second anomaly, the clarity coefficient of the second area is determined as shown in Table 3 below:

[0110] Table 3

[0111] parameter Numerical <![CDATA[MJZB2]]> 85％ <![CDATA[MJZB 2,0 ]]> 83％ <![CDATA[DDCS2]]> 0 σ 0.45 <![CDATA[ω1]]> 0.56 <![CDATA[ω2]]> 0.44 Clear Threshold 0.90

[0112] PSYZ2 = 0.976, GRYZ2 = 0.891. At this time, the calculated QXXS2 = 0.9386> 0.90 is obtained. At this time, the judgment is clear. One cluster is obtained through the regional image analysis unit. The specific parameters are shown in Table 4 below;

[0113] Table 4

[0114] parameter Numerical ZMJ 122 <![CDATA[JX1]]> 13.5 <![CDATA[YX1]]> 12.56

[0115] At this time, the calculation results are QYYXXS2=0.103;

[0116] At this time, it is determined that JBQX = 0.201 + 0.103 = 0.304, and an early warning is issued. At this time, the overall image data of the steel structure is not obtained, so that the surface defects of the steel structure can be identified quickly and accurately, avoiding the situation where the equipment computing power is occupied in the area that does not need to be inspected due to the overall image analysis, and shortening the inspection time;

[0117] Example 2:

[0118] In this embodiment, the local defect influence coefficient does not exceed the local defect threshold. At this time, the overall image data of the steel structure is obtained, and the lengths of both sides of the steel structure are obtained separately. See Table 5 below:

[0119] Table 5

[0120] parameter Numerical <![CDATA[l a ]]> 61 <![CDATA[l b ]]> 63.5 δ 0.05

[0121] The overall image defect coefficient of the steel structure CYZ = 0.04, which does not exceed the overall judgment threshold δ, and no warning is issued at this time;

[0122] The threshold values ​​in the above embodiments are set for ease of comparison. The threshold values ​​depend on the amount of sample data and the number of bases set by those skilled in the art for each set of sample data. As long as the proportional relationship between the parameter and the quantized value is not affected, the threshold values ​​can be determined by those skilled in the art based on each sample data set and multiple rounds of experiments.

[0123] The present invention also provides a steel structure defect recognition method based on image analysis, which is based on the above-mentioned steel structure defect recognition system based on image analysis, comprising the following steps:

[0124] Step 1: Obtain abnormal points on the steel structure after ultrasonic testing and mark the areas;

[0125] Step 2: Obtain image data of each marked area of ​​the steel structure and image data of the entire steel structure;

[0126] Step 3: Analyze the image data of each area of ​​the steel structure, and give a clarity coefficient for each area. Based on the clarity coefficient of each area, determine whether to re-photograph the corresponding area;

[0127] Step 4: Perform secondary analysis on the determined area and output the corresponding regional image influence coefficient;

[0128] Step 5: Based on all the regional image influence coefficients, the steel structure to be inspected is comprehensively calculated to finally obtain the local defect influence coefficient;

[0129] Step 6: When the local defect influence coefficient exceeds the local defect threshold, an early warning is issued. When the local defect influence coefficient does not exceed the local defect threshold, the overall image data is analyzed and the overall image defect coefficient is output. At the same time, whether to issue an early warning is determined based on the overall image defect coefficient.

[0130] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. Steel structure defect recognition system based on image analysis, characterized by: It includes data acquisition module, image data analysis module and steel structure defect recognition and judgment module; The data acquisition module includes a region division unit and an image data acquisition unit. The region marking unit is used to obtain abnormal points appearing on the steel structure after ultrasonic testing, mark the regions, and transmit the divided regions to the image data acquisition unit. The image data acquisition unit is used to obtain image data of each region of the steel structure marked in the region marking unit and image data of the entire steel structure. The image data analysis module includes an image blur degree analysis unit, a regional image analysis unit, and an overall image analysis unit. The image blur degree analysis unit is used to analyze the image data of each area of ​​the steel structure in the image data acquisition unit, and provide a clarity coefficient for each area. Based on the clarity coefficient of each area, it is determined whether to re-photograph the corresponding area. The regional image analysis unit is used to perform a secondary analysis on the area where the judgment has been completed, and output the corresponding regional image influence coefficient. The steel structure defect recognition and judgment module performs a comprehensive calculation on the steel structure to be inspected based on all regional image influence coefficients to finally obtain a local defect influence coefficient. When the local defect influence coefficient exceeds the local defect threshold, an early warning is issued. When the local defect influence coefficient does not exceed the local defect threshold, the overall image analysis unit in the image data analysis module is called to analyze the overall image data and output the overall image defect coefficient. The steel structure defect recognition and judgment module determines whether to issue an early warning based on the overall image defect coefficient.

2. The steel structure defect recognition system based on image analysis according to claim 1, characterized in that: The specific expression for the image data acquisition unit to acquire the image data of each area of ​​the steel structure marked in the area marking unit is as follows: TSXJ=[TSXJ1,…,TSXJ i ,…,TSXJ I ] Among them, TSXJI represents the regional image dataset, TSXJ1,…,TSXJ i ,…,TSXJ I They respectively represent the image data of the 1st region, ..., the image data of the i-th region, ..., the image data of the I-th region, i∈[1,I], and the image data of the i-th region is obtained by photographing with a mobile photographing device.

3. The steel structure defect recognition system based on image analysis according to claim 2, characterized in that: The image blur degree analysis unit is used to analyze the image data of each area of ​​the steel structure in the image data acquisition unit, and provide a clarity coefficient for each area. The specific steps of determining whether to re-photograph the corresponding area based on the clarity coefficient of each area are as follows: Step A1: Get the shooting factor PSYZ of the i-th area shooting i ; Step A2: Get the image factor GRYZ captured in the i-th region i ; Step A3: Comprehensively calculate the clarity coefficient QXXS of the i-th area i , the specific expression is as follows: QXXS i =ω1*WJFSYZ i +ω2*GRYZ i Among them, ω1 and ω2 represent two weight coefficients whose sum is 1; Step A4: Determine the clarity coefficient QXXS of the i-th area i Does it exceed the clarity threshold? If it does, then terminate step A4 and convert the image data TSXJ of the i-th region into i Send it to the regional image analysis unit. If it does not exceed the clear threshold, retake it and set the number of retakes CP for the i-th region. i = N, N represents the number of times step A4 is executed, when the number of retakes CP of the i-th region i An error is reported when the maximum number of retakes is exceeded.

4. The steel structure defect recognition system based on image analysis according to claim 3 is characterized in that: The data acquisition module also stores a set of black area ratios after grayscale processing of the original steel structure image of each area; In step A1, the shooting factor PSYZ of the i-th region is obtained. i The specific steps are as follows: Step A1.1: Read the image data TSXJ of the i-th region i , the image data TSXJ of the i-th region i Perform binarization processing; Step A1.2: Get the proportion of all black areas in the i-th area MJZB i , and calculate the shooting factor PSYZ of the i-th area shooting i , the specific expression is as follows: Among them, |MJZB i -MJZB i,0 | indicates MJZB i -MJZB i,0 The absolute value of MJZB i,0 Indicates the proportion of the black area in the i-th area stored in the percentage set, MJZB i Indicates the proportion of black area in the i-th region.

5. The steel structure defect recognition system based on image analysis according to claim 4, characterized in that: In step A2, the image factor GRYZ captured in the i-th region is obtained. i The specific steps are as follows: Step A2.1: Read the number of shakes of the mobile camera when shooting the i-th area DDCS i ; Step A2.2: Calculate the image factor GRYZ captured in the i-th region i , the specific expression is as follows: Among them, e represents the natural constant, DDCS i represents the number of times the mobile shooting device shakes when shooting the i-th area, and σ represents the balance factor to avoid the denominator being zero.

6. The steel structure defect recognition system based on image analysis according to claim 3, characterized in that: The specific steps of the regional image analysis unit for performing secondary analysis on the determined region and outputting the corresponding regional image influence coefficient are as follows: Step B1: Read the image data TSXJ of the i-th area i , and crop the image to keep only the steel structure area; Step B2: Identify the surface of the steel structure and perform image data TSXJ on the i-th region. i Perform grayscale processing; Step B3: Identify all pixels in the non-black area in step B2, and cluster adjacent non-black pixels to obtain several cluster areas; Step B4: Get the area JX of the minimum rectangular outline of the jth cluster area j and the area of ​​the minimum closed circle contour YX j ; Step B5: Determine the area JX of the minimum rectangular outline of the jth cluster area j and the area of ​​the minimum closed circle contour YX j Whether the smaller one exceeds the area threshold, if not, the j+1th cluster area is judged, if it exceeds, the smaller area is stored until all cluster areas are judged, and the regional image influence coefficient QYYXXS of the i-th area is calculated i , the specific expression is as follows: Among them, min{JX j ,YX j } represents the area of ​​the minimum rectangular outline of the jth cluster area JX j and the area of ​​the minimum closed circle contour YX j The smaller one, QYYXXS i represents the regional image influence coefficient of the i-th region, ZMJ represents the total area of ​​the cropped image, It means summing the contour areas outputted by a total of J cluster regions, j∈[1,J].

7. The steel structure defect recognition system based on image analysis according to claim 6, characterized in that: The steel structure defect recognition and judgment module performs a comprehensive calculation on the steel structure to be detected based on all regional image influence coefficients and finally obtains the specific expression of the local defect influence coefficient as follows: Among them, JBQX represents the local defect influence coefficient, It means summing the regional image influence coefficients of a total of I regions, i∈[1,I].

8. The steel structure defect recognition system based on image analysis according to claim 7, characterized in that: The specific steps of the overall image analysis unit analyzing the overall image data and outputting the overall image defect coefficient are as follows: Step C1: Obtain the overall image data of the steel structure and obtain the lengths of both sides of the steel structure respectively; Step C2: Calculate the overall image defect coefficient CYZ of the steel structure. The specific expression is as follows: Among them, l a and l b Respectively represent the length of different sides of the steel structure, CYZ represents the overall image defect coefficient of the steel structure, |l a -l b | indicates l a -l b The absolute value of .

9. The steel structure defect recognition system based on image analysis according to claim 8, characterized in that: The specific steps of the steel structure defect recognition and judgment module for judging whether to issue an early warning based on the overall image defect coefficient are as follows: when the overall image defect coefficient CYZ of the steel structure exceeds the overall judgment threshold δ, an early warning is issued; when the overall image defect coefficient CYZ of the steel structure does not exceed the overall judgment threshold δ, no early warning is issued.

10. A method for identifying defects in steel structures based on image analysis, based on the system for identifying defects in steel structures based on image analysis according to any one of claims 1 to 9, characterized in that: The following steps are involved: Step 1: Obtain abnormal points on the steel structure after ultrasonic testing and mark the areas; Step 2: Obtain image data of each marked area of ​​the steel structure and image data of the entire steel structure; Step 3: Analyze the image data of each area of ​​the steel structure, and give a clarity coefficient for each area. Based on the clarity coefficient of each area, determine whether to re-photograph the corresponding area; Step 4: Perform secondary analysis on the determined area and output the corresponding regional image influence coefficient; Step 5: Based on all the regional image influence coefficients, the steel structure to be inspected is comprehensively calculated to finally obtain the local defect influence coefficient; Step 6: When the local defect influence coefficient exceeds the local defect threshold, an early warning is issued. When the local defect influence coefficient does not exceed the local defect threshold, the overall image data is analyzed and the overall image defect coefficient is output. At the same time, whether to issue an early warning is determined based on the overall image defect coefficient.

Images