Quality detection method of steel components of steel structure workshop based on image data

Abstract

The present invention discloses a method for detecting the quality of steel components of a steel structure factory based on image data, which specifically includes the following steps: A1, using a two-dimensional camera to perform surface image acquisition operations on steel component products, and aggregating the images to form an image data set. The present invention relates to the field of data processing technology. The method for detecting the quality of steel components of a steel structure factory based on image data, through the grayscale processing operation on the component view, and through the defect feature part of the marked part after the grayscale processing, combined with the complete component view in different orientations, determines whether the defect feature part is a convex defect feature or a concave defect feature, and then determines whether the concave defect feature is a concave feature or a crack feature, thereby completing the detection operation of the surface quality of the steel component based on image data, not only can the defect feature be quickly found, but also the specific situation of the defect feature can be determined according to the processing of the image data, thereby improving the accuracy of product detection.

Description

Technical Field

[0001] The invention relates to the technical field of data processing, and in particular to a method for detecting quality of steel components of a steel structure plant based on image data. Background Art

[0002] The reference patent name is: Quality detection method for prefabricated steel structure components based on image data (patent publication number: CN116977333A, patent publication date: 2023-10-31), including: obtaining suspected concave areas and smooth areas according to the direction of illumination; obtaining the distribution trend of each pixel point in the suspected concave area and smooth area under each reachable radius according to the circular LBP operator of each pixel point in the suspected concave area and smooth area under each reachable radius; obtaining the distribution difference of each pixel point in the smooth area and the overall distribution trend level of the smooth area; obtaining the non-noise points in the smooth area and the overall distribution difference level of the smooth area; calculating the probability that each pixel point in the suspected concave area belongs to a concave pixel point; according to the probability of belonging to a concave pixel point, obtaining the concave area of ​​the surface image of the prefabricated steel structure component.

[0003] Based on the description in the above-mentioned documents, during the process of detecting surface defect features of steel components in existing steel structure factories, although the existence of defects can be known after image acquisition, the specific defect features such as protruding attachments, extruded concave grooves and generated cracks cannot be identified, so that the detection operation requires human judgment, which not only has the problem of time consumption, but also the accuracy of the detection results needs to be examined. For this reason, the present invention provides a quality detection method for steel components in steel structure factories based on image data. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a method for quality inspection of steel components of steel structure factories based on image data, which solves the problem that during the surface defect feature inspection process of steel components of existing steel structure factories, although the existence of defects can be known after image acquisition, the defect features cannot be identified as protruding attachments, extruded concave grooves and generated cracks.

[0005] To achieve the above objectives, the present invention is implemented by the following technical solutions: A method for quality inspection of steel components of a steel structure workshop based on image data, specifically comprising the following steps:

[0006] A1. Use a two-dimensional camera to collect images of the surface of the steel component product, and aggregate the images to form an image data set P;

[0007] A2, combining component images according to the image data set P to form complete component views in different orientations;

[0008] A3. By processing the component view, the product defects are identified according to the characteristic grayscale difference of the component view, specifically:

[0009] a31. Complete the grayscale processing operation on the component view, and mark the defect feature part after the grayscale processing;

[0010] a32. Determine whether the defect feature is a protruding defect feature or a concave defect feature by combining the complete component views at different orientations;

[0011] a33, then obtain multiple feature points by processing the concave defect feature, calculate the gray value change according to the feature points and compare them, determine whether the concave defect feature is a concave surface feature or a crack feature, and transmit the result;

[0012] A4. Determine the quality of steel structure products based on the structure obtained after image processing, and repair or scrap them based on the quality issues.

[0013] Preferably, the image combination operation of the image data set P in A2 to form a complete component view is:

[0014] a21, after classifying the image data subsets located at the same position in order, select two image data subsets having common feature parts in adjacent image data subsets;

[0015] a22, overlapping the parts with common features in adjacent image data subsets, and calculating the displacement difference between the adjacent image data subsets during image acquisition according to the vertical distance between the right boundary of the first image data subset and the left boundary of the second image data subset;

[0016] a23, and according to the displacement difference between the images, the latter image data subset is sequentially overlaid on the former image data subset, and the overlapping area on the former image data and located in the adjacent image data subset is filtered, and finally a complete component view about the same orientation is formed.

[0017] Preferably, the vertical distance calculation algorithm between the boundaries of the overlapping areas of adjacent image data subsets in a22 is:

[0018] b01. Establishing a coordinate system with the lower left corner node of the first image data as the origin according to the first image data subset, with the bottom horizontal boundary of the first image data subset as the X1 axis, and with the left vertical boundary of the first image data subset as the Y1 axis;

[0019] b02, using the coordinate system of b01, the upper right corner node position of the first image data subset is obtained as (x1, y1), and after the common feature parts of adjacent image data subsets overlap, the intersection coordinates of the upper boundary of the second image data subset and the first image data subset are determined to be (x2, y1);

[0020] b03. Calculate the vertical distance L based on the coordinate positions of the two points, and the calculation formula is: L = x1-x2.

[0021] Preferably, the grayscale processing operation of the component view in a31 is:

[0022] C1. Complete the image restoration operation of the component view by using the average value method;

[0023] C2. According to the gray value changes of each characteristic part in the image, the characteristic part with a large area is determined as the basic feature, and the characteristic part with a different gray value and a single characteristic part in the image is marked and determined as the defect characteristic part;

[0024] C3. Perform feature judgment processing on the image with the defect feature part marked.

[0025] Preferably, the operation of determining whether the defect feature part is a convex defect feature or a concave defect feature in a32 is:

[0026] D1. Establish a three-dimensional coordinate system based on a complete component view of one orientation, with the corner node of the complete component view as the origin, the long boundary direction of the complete component view as the X2 axis, the short boundary direction of the complete component view as the Y2 axis, and establish the Z axis based on the origin of the complete component view and at right angles to the X2 axis and the Y2 axis;

[0027] D2. Select a defect feature part at any location in an orientation, and determine the coordinates of the center point of the defect feature part as (x3, y2, 0);

[0028] D3. Select the adjacent complete component view according to the orientation in D2, and determine whether the upper boundary position at the x3 distance value or y2 distance value of the adjacent complete component view from the origin has a protruding feature part based on the x3 distance value or y2 distance value of the defect feature part in the orientation of D2.

[0029] Preferably, the characteristic part result of determining the adjacent azimuthally complete component views in D3 is:

[0030] Result 1: If the upper boundary position at the x3 distance value or y2 distance value from the origin of the adjacent azimuth complete component view has a protruding feature part, then the defect feature part in the D2 orientation is a protruding defect feature;

[0031] Result 2: On the contrary, if there is no protruding feature part at the upper boundary position of the adjacent orientation complete component view at the x3 distance value or the y2 distance value from the origin, then the defect feature part in the D2 orientation is a concave defect feature.

[0032] Preferably, the operation of determining whether the concave defect feature is a concave feature or a crack feature in a33 is:

[0033] E1. After determining that the defect feature is a concave defect feature in the D2-D3 operation, a straight line parallel to the X1 axis is made at the concave defect feature and marked as h1, and the straight line h1 intersects with the concave defect feature boundary to obtain an intersection node;

[0034] E2. The midpoint of the straight line h1 is marked as Q according to the intersection node. At the same time, a straight line parallel to the Y1 axis is made at point Q and marked as h2. The straight line h2 parallel to the Y1 axis intersects with the characteristic boundary of the concave defect to obtain two nodes;

[0035] E3. Determine the selected feature line segment region based on the comparison of the distances between nodes, and perform numerical comparison by setting three pixel points in the feature line segment region;

[0036] E4. Determine whether the concave defect feature is a concave surface feature or a crack feature based on the results of the numerical comparison.

[0037] Preferably, the operation of determining the selected characteristic line segment area according to the comparison of the distances between nodes in E3 is:

[0038] e31a, comparing the line segment distance between two nodes where the straight line h1 intersects the concave defect characteristic boundary with the line segment distance between two nodes where the straight line h2 intersects the concave defect characteristic boundary;

[0039] e32a. According to the comparison result, the line segment area with relatively shorter distance between the two is selected as the characteristic line segment area.

[0040] Preferably, the operation of performing numerical comparison on the three pixel points in the characteristic line segment area in E3 is:

[0041] e31b, by setting pixel points g1 and g2 at the center between two adjacent nodes and the midpoint in the feature line segment area, and setting the pixel midpoint g3 at the midpoint of the feature line segment area;

[0042] e32b, calculate the values ​​of pixel points g1 and g2 and pixel midpoint g3 and compare them, and the formula for calculating the pixel point is:

[0043] F = ∑(n1*R+n2*G+n3*B);

[0044] Where F is the grayscale value of any pixel, n1, n2 and n3 are the weight values ​​corresponding to R, G and B, and R, G and B represent the red, green and blue components of the pixel;

[0045] e33b. After calculating the results of pixel point g1, pixel point g2 and pixel point g3, compare the values ​​of each pixel point one by one.

[0046] Preferably, the result of the numerical comparison in E4 is:

[0047] e41. When F(g1)<F(g3) and F(g2)<F(g3), the concave defect feature is a concave feature;

[0048] e42. When F(g1)=F(g3)=F(g2), the concave defect characteristic is a crack characteristic.

[0049] The present invention provides a method for detecting the quality of steel components of a steel structure factory building based on image data. Compared with the prior art, it has the following beneficial effects:

[0050] (1) The method for detecting the quality of steel components of a steel structure workshop based on image data performs a grayscale processing operation on the component view, and determines whether the defect feature part is a convex defect feature or a concave defect feature by combining the complete component views in different orientations, and then determines whether the concave defect feature is a concave feature or a crack feature, thereby completing the detection operation of the surface quality of the steel component based on image data. Not only can the defect feature be found quickly, but also the specific situation of the defect feature can be determined according to the processing of the image data, thereby improving the accuracy of product detection.

[0051] (2) The steel component quality inspection method for steel structure workshops based on image data, after sequentially classifying the image data subsets of the same orientation, sequentially overlays the latter image data subsets on the former image data subsets according to the displacement difference between the images, and filters the overlapping areas on the former image data and located in the adjacent image data subsets, and finally forms a complete component view of the same orientation, thereby realizing the detection of the adjacent image data subsets. Figure 1 It can also avoid multiple data processing operations on the same feature and improve the detection efficiency.

[0052] (3) The method for detecting the quality of steel components of a steel structure factory building based on image data compares the line segment distance between the two nodes where the straight line h1 intersects the concave defect feature boundary with the line segment distance between the two nodes where the straight line h2 intersects the concave defect feature boundary. According to the comparison result, the relatively shorter line segment distance between the two is the selected characteristic line segment area. In this way, the problem of consistent pixel point values ​​in long line segments is avoided for both concave features and crack features, thereby improving the accuracy of the detection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a flow chart of the steel component quality detection method of the present invention;

[0054] Figure 2 A flowchart of the component view processing of the present invention;

[0055] Figure 3 It is a schematic diagram for judging the defect characteristic part of the present invention;

[0056] Figure 4 It is a schematic diagram for judging the characteristics of the concave defect of the present invention. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0058] See also Figure 1-Figure 4 The present invention provides a technical solution: a method for detecting the quality of steel components of a steel structure workshop based on image data, which specifically comprises the following steps:

[0059] A1. Use a two-dimensional camera to collect images of the surface of the steel component product, and aggregate the images to form an image data set P;

[0060] A2, combining component images according to the image data set P to form complete component views in different orientations;

[0061] A3. By processing the component view, the product defects are identified according to the characteristic grayscale difference of the component view, specifically:

[0062] a31. Complete the grayscale processing operation on the component view, and mark the defect feature part after the grayscale processing;

[0063] a32. Determine whether the defect feature is a protruding defect feature or a concave defect feature by combining the complete component views at different orientations;

[0064] a33, then obtain multiple feature points by processing the concave defect feature, calculate the gray value change according to the feature points and compare them, determine whether the concave defect feature is a concave surface feature or a crack feature, and transmit the result;

[0065] A4. Determine the quality of steel structure products based on the structure obtained after image processing, and repair or scrap them based on the quality issues.

[0066] Among them, by graying the component view, and through the defect feature part marked after graying, combined with the complete component views in different orientations, it is judged that the defect feature part is a convex defect feature or a concave defect feature, and then the concave defect feature is judged to be a concave feature or a crack feature, so as to complete the surface quality of the steel component and the detection operation based on the image data, which can not only quickly find the defect features, but also determine the specific situation of the defect features based on the processing of the image data, thereby improving the accuracy of product detection.

[0067] In the embodiment of the present invention, the image combination operation of the image data set P in A2 to form a complete component view is:

[0068] a21, after classifying the image data subsets located at the same position in order, select two image data subsets having common feature parts in adjacent image data subsets;

[0069] a22, overlapping the parts with common features in adjacent image data subsets, and calculating the displacement difference between the adjacent image data subsets during image acquisition according to the vertical distance between the right boundary of the first image data subset and the left boundary of the second image data subset;

[0070] a23, and according to the displacement difference between the images, the latter image data subset is sequentially overlaid on the former image data subset, and the overlapping area on the former image data and located in the adjacent image data subset is filtered, and finally a complete component view about the same orientation is formed.

[0071] Among them, after the image data subsets of the same orientation are classified in sequence, the latter image data subsets are sequentially covered on the former image data subsets according to the displacement difference between the images, and the overlapping areas on the former image data and located in the adjacent image data subsets are filtered, and finally a complete component view of the same orientation is formed, so that the orientation-dependent Figure 1 It can also avoid multiple data processing operations on the same feature and improve the detection efficiency.

[0072] In the embodiment of the present invention, the vertical distance calculation algorithm between the boundaries of the overlapping areas of adjacent image data subsets in a22 is:

[0073] b01. Establishing a coordinate system with the lower left corner node of the first image data as the origin according to the first image data subset, with the bottom horizontal boundary of the first image data subset as the X1 axis, and with the left vertical boundary of the first image data subset as the Y1 axis;

[0074] b02, using the coordinate system of b01, the upper right corner node position of the first image data subset is obtained as (x1, y1), and after the common feature parts of adjacent image data subsets overlap, the intersection coordinates of the upper boundary of the second image data subset and the first image data subset are determined to be (x2, y1);

[0075] b03. Calculate the vertical distance L based on the coordinate positions of the two points, and the calculation formula is: L = x1-x2.

[0076] In the embodiment of the present invention, the grayscale processing operation of the component view in a31 is:

[0077] C1. Complete the image restoration operation of the component view by using the average value method;

[0078] C2. According to the gray value changes of each characteristic part in the image, the characteristic part with a large area is determined as the basic feature, and the characteristic part with a different gray value and a single characteristic part in the image is marked and determined as the defect characteristic part;

[0079] C3. Perform feature judgment processing on the image with the defect feature part marked.

[0080] The average value method is to average the values ​​of the red, yellow and blue components in the image data to obtain the grayscale value of the image.

[0081] In the embodiment of the present invention, the operation of determining whether the defect feature portion is a convex defect feature or a concave defect feature in a32 is:

[0082] D1. Establish a three-dimensional coordinate system based on a complete component view of one orientation, with the corner node of the complete component view as the origin, the long boundary direction of the complete component view as the X2 axis, the short boundary direction of the complete component view as the Y2 axis, and establish the Z axis based on the origin of the complete component view and at right angles to the X2 axis and the Y2 axis;

[0083] D2. Select a defect feature part at any location in an orientation, and determine the coordinates of the center point of the defect feature part as (x3, y2, 0);

[0084] D3. Select the adjacent complete component view according to the orientation in D2, and determine whether the upper boundary position at the x3 distance value or y2 distance value of the adjacent complete component view from the origin has a protruding feature part based on the x3 distance value or y2 distance value of the defect feature part in the orientation of D2.

[0085] In the embodiment of the present invention, the characteristic part result of determining the adjacent orientation complete component views in D3 is:

[0086] Result 1: If the upper boundary position at the x3 distance value or y2 distance value from the origin of the adjacent azimuth complete component view has a protruding feature part, then the defect feature part in the D2 orientation is a protruding defect feature;

[0087] Result 2: On the contrary, if there is no protruding feature part at the upper boundary position of the adjacent orientation complete component view at the x3 distance value or the y2 distance value from the origin, then the defect feature part in the D2 orientation is a concave defect feature.

[0088] In the embodiment of the present invention, the operation of determining whether the concave defect feature is a concave feature or a crack feature in a33 is:

[0089] E1. After determining that the defect feature is a concave defect feature in the D2-D3 operation, a straight line parallel to the X1 axis is made at the concave defect feature and marked as h1, and the straight line h1 intersects with the concave defect feature boundary to obtain an intersection node;

[0090] E2. The midpoint of the straight line h1 is marked as Q according to the intersection node. At the same time, a straight line parallel to the Y1 axis is made at point Q and marked as h2. The straight line h2 parallel to the Y1 axis intersects with the characteristic boundary of the concave defect to obtain two nodes;

[0091] E3. Determine the selected feature line segment region based on the comparison of the distances between nodes, and perform numerical comparison by setting three pixel points in the feature line segment region;

[0092] E4. Determine whether the concave defect feature is a concave surface feature or a crack feature based on the results of the numerical comparison.

[0093] In the embodiment of the present invention, the operation of E3 determining the selected characteristic line segment area based on the comparison of the distances between nodes is:

[0094] e31a, comparing the line segment distance between two nodes where the straight line h1 intersects the concave defect characteristic boundary with the line segment distance between two nodes where the straight line h2 intersects the concave defect characteristic boundary;

[0095] e32a. According to the comparison result, the line segment area with relatively shorter distance between the two is selected as the characteristic line segment area.

[0096] In the embodiment of the present invention, the operation of performing numerical comparison on the three pixel points in the characteristic line segment area in E3 is:

[0097] e31b, by setting pixel points g1 and g2 at the center between two adjacent nodes and the midpoint in the feature line segment area, and setting the pixel midpoint g3 at the midpoint of the feature line segment area;

[0098] e32b, calculate the values ​​of pixel points g1 and g2 and pixel midpoint g3 and compare them, and the formula for calculating the pixel point is:

[0099] F = ∑(n1*R+n2*G+n3*B);

[0100] Where F is the grayscale value of any pixel, n1, n2 and n3 are the weight values ​​corresponding to R, G and B, and R, G and B represent the red, green and blue components of the pixel;

[0101] e33b. After calculating the results of pixel point g1, pixel point g2 and pixel point g3, compare the values ​​of each pixel point one by one.

[0102] Among them, by comparing the line segment distance between the two nodes where the straight line h1 intersects the concave defect feature boundary with the line segment distance between the two nodes where the straight line h2 intersects the concave defect feature boundary, and according to the comparison result, the relatively shorter line segment distance between the two is the selected feature line segment area, thereby avoiding the problem of consistent pixel point values ​​in long line segments for both concave surface features and crack features, thereby improving the accuracy of the detection results.

[0103] In the embodiment of the present invention, the result of the numerical comparison in E4 is:

[0104] e41. When F(g1)<F(g3) and F(g2)<F(g3), the concave defect feature is a concave feature;

[0105] e42. When F(g1)=F(g3)=F(g2), the concave defect characteristic is a crack characteristic.

[0106] Comparative experiment

[0107] Now, the detection equipment using the existing steel component quality detection method and the detection equipment using the steel component quality detection method of the present invention are used to perform detection operations on multiple groups of identical steel components, and the same image acquisition is performed on each group of steel components, and the time and accuracy of the detection of the concave feature part completed by the existing steel component quality detection method and the steel component quality detection method of the present invention are recorded, and the results are shown in Table 1:

[0108] Table 1 Experimental results comparison table

[0109] Time to complete the test Feature detection accuracy Existing steel component quality inspection methods 152s 76％ Steel component quality detection method of the present invention 89s 100％

[0110] In summary, the steel component quality inspection method of the present invention can complete the inspection operation on multiple groups of identical steel components in a shorter time, and the feature accuracy of the steel component quality inspection method of the present invention is higher.

[0111] Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0112] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0113] Although 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 the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for quality inspection of steel components of a steel structure workshop based on image data, characterized in that: The specific steps include: A1. Use a two-dimensional camera to collect images of the surface of the steel component product, and aggregate the images to form an image data set P; A2, combining component images according to the image data set P to form complete component views in different orientations; a21, after classifying the image data subsets located at the same position in order, select two image data subsets having common feature parts in adjacent image data subsets; a22, overlapping the parts with common features in adjacent image data subsets, and calculating the displacement difference between the adjacent image data subsets during image acquisition according to the vertical distance between the right boundary of the first image data subset and the left boundary of the second image data subset; a23, and sequentially overlaying the latter image data subset on the former image data subset according to the displacement difference generated between the images, and filtering the overlapping areas on the former image data and located in adjacent image data subsets, and finally forming a complete component view about the same orientation; A3. By processing the component view, the product defects can be distinguished according to the characteristic grayscale difference of the component view, specifically: a31. Complete the grayscale processing operation on the component view, and mark the defect feature part after the grayscale processing; a32. Determine whether the defect feature is a protruding defect feature or a concave defect feature by combining the complete component views at different orientations; a33, then obtain multiple feature points by processing the concave defect feature, calculate the gray value change according to the feature points and compare them, determine whether the concave defect feature is a concave surface feature or a crack feature, and transmit the result; A4. Determine the quality of steel structure products based on the structure obtained after image processing, and repair or scrap them based on the quality issues.

2. The method for quality inspection of steel components of steel structure workshops based on image data according to claim 1 is characterized by: The vertical distance calculation algorithm between the overlapping area boundaries of adjacent image data subsets in a22 is: b01. Establishing a coordinate system with the lower left corner node of the first image data as the origin according to the first image data subset, with the bottom horizontal boundary of the first image data subset as the X1 axis, and with the left vertical boundary of the first image data subset as the Y1 axis; b02, using the coordinate system of b01, the upper right corner node position of the first image data subset is obtained as (x1, y1), and after the common features of adjacent image data subsets overlap, the intersection coordinates of the upper boundary of the second image data subset and the first image data subset are determined to be (x2, y1); b03. Calculate the vertical distance L based on the coordinate positions of the two points, and the calculation formula is: L=x1-x2.

3. The method for quality inspection of steel components of steel structure workshops based on image data according to claim 1 is characterized in that: The grayscale processing operation of the component view in a31 is as follows: C1. Complete the image restoration operation of the component view by using the average value method; C2. According to the gray value changes of each characteristic part in the image, the characteristic part with a large area is determined as the basic feature, and the characteristic part with a different gray value and a single characteristic part in the image is marked and determined as the defect characteristic part; C3. Perform feature judgment processing on the image with the defect feature part marked.

4. The method for quality inspection of steel components of steel structure workshops based on image data according to claim 1 is characterized in that: The operation of determining whether the defect feature part is a convex defect feature or a concave defect feature in a32 is: D1. Establish a three-dimensional coordinate system based on a complete component view of one orientation, with the corner node of the complete component view as the origin, the long boundary direction of the complete component view as the X2 axis, the short boundary direction of the complete component view as the Y2 axis, and establish the Z axis based on the origin of the complete component view and at right angles to the X2 axis and the Y2 axis; D2. Select a defect feature part at any location in an orientation, and determine the coordinates of the center point of the defect feature part as (x3, y2, 0); D3. Select the adjacent complete component view according to the orientation in D2, and determine whether the upper boundary position at the x3 distance value or y2 distance value of the adjacent complete component view from the origin has a protruding feature part based on the x3 distance value or y2 distance value of the defect feature part in the orientation of D2.

5. The method for quality inspection of steel components of steel structure workshops based on image data according to claim 4 is characterized in that: The characteristic part result of determining the adjacent azimuth complete component view in D3 is: Result 1: If the upper boundary position at the x3 distance value or y2 distance value from the origin of the adjacent azimuth complete component view has a protruding feature part, then the defect feature part in the D2 orientation is a protruding defect feature; Result 2: On the contrary, if there is no protruding feature part at the upper boundary position of the adjacent orientation complete component view at the x3 distance value or the y2 distance value from the origin, then the defect feature part in the D2 orientation is a concave defect feature.

6. The method for quality inspection of steel components of steel structure workshops based on image data according to claim 4 is characterized in that: The operation of determining whether the concave defect feature is a concave feature or a crack feature in a33 is: E1. After determining that the defect feature is a concave defect feature in the D2-D3 operation, a straight line parallel to the X1 axis is made at the concave defect feature and marked as h1, and the straight line h1 intersects with the concave defect feature boundary to obtain an intersection node; E2. The midpoint of the straight line h1 is marked as Q according to the intersection node. At the same time, a straight line parallel to the Y1 axis is made at point Q and marked as h2. The straight line h2 parallel to the Y1 axis intersects with the characteristic boundary of the concave defect to obtain two nodes; E3. Determine the selected feature line segment region based on the comparison of the distances between nodes, and perform numerical comparison by setting three pixel points in the feature line segment region; E4. Determine whether the concave defect feature is a concave surface feature or a crack feature based on the results of the numerical comparison.

7. The method for quality inspection of steel components of steel structure workshops based on image data according to claim 6 is characterized by: The operation of determining the selected characteristic line segment area according to the comparison of the distances between nodes in E3 is: e31a, comparing the line segment distance between two nodes where the straight line h1 intersects the concave defect characteristic boundary with the line segment distance between two nodes where the straight line h2 intersects the concave defect characteristic boundary; e32a. According to the comparison result, the line segment area with relatively shorter distance between the two is selected as the characteristic line segment area.

Images