Image enhancement method for GIS shell welding seam detection

Abstract

The invention relates to the technical field of image processing, in particular to an image enhancement method for GIS shell welding seam detection, and the method comprises the steps: screening all edges according to the welding seam contrast, and obtaining suspected welding seam edges; acquiring a matching characteristic value between any two suspected weld seam edges according to the weld seam contrast ratio; according to an area formed between the suspected welding seam edge and the to-be-matched suspected welding seam edge, the matching reliability between the suspected welding seam edge and the to-be-matched suspected welding seam edge is obtained; according to the matching reliability, the authenticity of each suspected welding seam edge is obtained; acquiring the weld edge according to the authenticity of the suspected weld edge; and image enhancement is carried out according to the weld seam contrast of the weld seam edge to obtain an enhanced GIS shell surface welding position image, and the accuracy of a GIS shell weld seam detection result is improved.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and more particularly to an image enhancement method for GIS shell weld detection. Background Art

[0002] Image enhancement methods play a crucial role in GIS shell weld inspection, particularly in improving image quality and highlighting key details. Shell structures typically have complex geometries and surface features. As a crucial component of the structure, welds can be affected by environmental factors or deformation during use, impacting image quality. Furthermore, against complex backgrounds, weld edges can become blurred due to noise, illumination variations, and texture interference. Furthermore, cracks are often found in the weld joint area, often along the direction of the joint. Consequently, cracks can be mistaken for weld edges, leading to errors in GIS shell weld inspection. Summary of the Invention

[0003] In order to accurately identify and enhance the weld area, the present invention proposes an image enhancement method for GIS shell weld detection, which includes the following steps: Acquire images of the welds on the surface of the GIS shell; Obtain the grayscale difference characteristic value of each edge in the image of the weld on the GIS shell surface; obtain the weld contrast of each edge based on the grayscale difference characteristic value and the distribution of the gradient direction of the edge pixel points; filter all edges based on the weld contrast to obtain suspected weld edges; obtain the matching characteristic value between any two suspected weld edges based on the weld contrast; obtain the suspected weld edge to be matched for each suspected weld edge based on the matching characteristic value; Obtaining a matching reliability between the suspected weld edge and the suspected weld edge to be matched based on an area formed between the suspected weld edge and the suspected weld edge to be matched; obtaining the authenticity of each suspected weld edge based on the matching reliability; and obtaining the weld edge based on the authenticity of the suspected weld edge; Image enhancement is performed based on the weld contrast at the weld edge to obtain an enhanced image of the weld on the GIS shell surface.

[0004] Preferably, the step of obtaining the grayscale difference characteristic value of each edge in the image of the welded portion of the GIS shell surface includes: A neighborhood parameter U is preset. For any edge pixel point on any edge of the image at the welded portion of the GIS shell surface, a window with a size of U×U is obtained with the edge pixel point as the window center, and the window is recorded as the neighborhood pixel set of the edge pixel point. Where, Represents the grayscale distribution eigenvalue of the rth edge; Represents the number of all edge pixels on the rth edge; Represents the number of all pixels in the neighborhood pixel set of the i-th edge pixel on the r-th edge; Represents the grayscale value of the i-th edge pixel on the r-th edge; Represents the grayscale value of the jth edge pixel in the neighborhood pixel set of the i-th edge pixel on the r-th edge; represents the linear normalization function.

[0005] Preferably, obtaining the weld contrast of each edge according to the grayscale difference characteristic value and the distribution of the gradient direction of the edge pixel points includes: The sine value of the angle between the gradient direction of the i-th edge pixel on the r-th edge and the horizontal direction is recorded as the edge direction factor of the i-th edge pixel; the calculation expression for obtaining the weld contrast of the r-th edge is: ; Where, represents the weld contrast of the rth edge; Represents the grayscale distribution eigenvalue of the rth edge; Represents the number of all edge pixels on the rth edge; Represents the edge direction factor of the i-th edge pixel on the r-th edge; Represents the edge direction factor of the i+1th edge pixel on the rth edge; Indicates taking the absolute value; Represents an exponential function with a natural constant as its base.

[0006] Preferably, screening all edges according to weld contrast to obtain suspected weld edges includes: A significant threshold T1 is preset. If the weld contrast of the r-th edge is greater than or equal to the significant threshold T1, the r-th edge is recorded as a suspected weld edge.

[0007] Preferably, the specific formula for obtaining the matching characteristic value between any two suspected weld edges based on the weld contrast is: ; Where, represents the matching feature value between the c1th suspected weld edge and the c2th suspected weld edge; represents the mean value of the weld contrast between the c1th suspected weld edge and the c2th suspected weld edge; Represents the number of all edge pixels on the c1th suspected weld edge; Represents the number of all edge pixels on the c2th suspected weld edge; represents the Euclidean distance between the jth edge pixel on the c1th suspected weld edge and the c2th suspected weld edge; Represents an exponential function with a natural constant as its base.

[0008] Preferably, obtaining the suspected weld edge to be matched for each suspected weld edge according to the matching characteristic value includes: A matching parameter Z is preset, and the matching eigenvalues ​​between the c1th suspected weld edge and all other suspected weld edges are obtained. The first Z suspected weld edges with the largest matching eigenvalues ​​are used as the suspected weld edges to be matched with the c1th suspected weld edge.

[0009] Preferably, obtaining the matching reliability between the suspected weld edge and the suspected weld edge to be matched based on the area formed between the suspected weld edge and the suspected weld edge to be matched includes: The area formed by the vth suspected weld edge and the kth suspected weld edge to be matched is recorded as the kth suspected weld area to be matched; the sequence composed of all pixels on the circumscribed rectangle of the kth suspected weld area to be matched is obtained, and recorded as the background pixel sequence of the kth suspected weld area to be matched; for the ath edge pixel point on the kth suspected weld area to be matched, the distance between the ath edge pixel and all pixels in the background pixel sequence of the kth suspected weld area to be matched is obtained, and the pixel with the closest distance is recorded as the target pixel of the ath edge pixel point pixel point; the straight line passing through the centroid position of the kth suspected weld area to be matched and parallel to the vth suspected weld edge is recorded as the target straight line; on all edge pixel points of the kth suspected weld area to be matched, the pixel point symmetrical to the ath edge pixel point about the target straight line is obtained, and recorded as the comparison pixel point of the ath edge pixel point; according to the background pixel point sequence of the kth suspected weld area to be matched, as well as the distances between the edge pixel points, the comparison pixel points and the target pixel points, the matching reliability between the suspected weld edge and the suspected weld edge to be matched is obtained.

[0010] Preferably, the specific formula for obtaining the matching reliability between the suspected weld edge and the suspected weld edge to be matched is: ; Where, represents the matching reliability between the vth suspected weld edge and the kth suspected weld edge to be matched; represents the matching feature value between the vth suspected weld edge and the kth suspected weld edge to be matched, Represents the number of all pixels in the background pixel sequence of the kth suspected weld area to be matched; Represents the grayscale value of the mth pixel in the background pixel sequence of the kth suspected weld area to be matched; Represents the mean of the grayscale values ​​of all pixels in the background pixel sequence of the kth suspected weld area to be matched; Represents the number of all edge pixels on the kth suspected weld area to be matched; represents the Euclidean distance between the ath edge pixel point and the target pixel point in the kth suspected weld area to be matched; It represents the Euclidean distance between the comparison pixel point of the ath edge pixel point on the kth suspected weld area to be matched and the target pixel point; Indicates taking the absolute value; Represents an exponential function with a natural constant as its base.

[0011] Preferably, the specific formula for obtaining the authenticity of each suspected weld edge based on the matching reliability is: ; Where, Indicates the authenticity of the vth suspected weld edge; represents all the suspected weld edges to be matched for the vth suspected weld edge; represents the maximum value of the matching reliability between the vth suspected weld edge and all suspected weld edges to be matched; represents the matching reliability between the vth suspected weld edge and the bth suspected weld edge to be matched; Indicates taking the absolute value; represents the linear normalization function.

[0012] Preferably, obtaining the weld edge according to the authenticity of the suspected weld edge includes: A weld threshold T2 is preset. If the authenticity of the vth suspected weld edge is greater than or equal to the weld threshold T2, the vth suspected weld edge is recorded as the weld edge.

[0013] The present invention has the following beneficial effects: the present invention screens all edges according to weld contrast to obtain suspected weld edges; obtains matching characteristic values ​​between any two suspected weld edges according to the weld contrast; obtains the suspected weld edge to be matched of each suspected weld edge according to the matching characteristic values; obtains the matching reliability between the suspected weld edge and the suspected weld edge to be matched according to the area formed between the suspected weld edge and the suspected weld edge to be matched; obtains the authenticity of each suspected weld edge according to the matching reliability; obtains the weld edge according to the authenticity of the suspected weld edge; performs image enhancement according to the weld contrast of the weld edge to obtain an enhanced image of the weld on the surface of the GIS shell; thereby improving the accuracy of the GIS shell weld detection result. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a flowchart of the steps of an image enhancement method for GIS shell weld detection according to an embodiment of the present invention. DETAILED DESCRIPTION

[0015] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0016] See also Figure 1 , which shows a flowchart of an image enhancement method for GIS shell weld detection provided by one embodiment of the present invention, the method comprising the following steps: S001. Acquire images of welds on the GIS shell surface.

[0017] In the embodiment of the present invention, the specific method for obtaining the image of the weld on the surface of the GIS shell is as follows: Use a high-resolution digital camera (such as an industrial camera, a digital SLR camera, etc.) to capture images of the welds on the surface of the GIS shell, and perform median filtering, denoising, and grayscale operations to obtain images of the welds on the surface of the GIS shell.

[0018] Among them, median filtering and grayscale operations are existing technologies and will not be described in detail in this embodiment.

[0019] S002. Obtain the grayscale difference eigenvalue of each edge in the image of the weld on the GIS shell surface; obtain the weld contrast of each edge based on the grayscale difference eigenvalue and the distribution of the gradient direction of the edge pixels; filter all edges based on the weld contrast to obtain suspected weld edges; obtain the matching eigenvalue between any two suspected weld edges based on the weld contrast; and obtain the matching suspected weld edge for each suspected weld edge based on the matching eigenvalue.

[0020] It should be noted that the weld is usually the boundary between two metal surfaces, with relatively clear edge features that have significant contrast with the surrounding area. That is, there is a significant difference in grayscale between each edge pixel and the pixels in its neighborhood, and the gradient direction between the edge pixels is more consistent; when there is a significant difference in grayscale between the edge pixel and the pixels in its neighborhood, it means that the edge has a clearer edge structure.

[0021] In the embodiment of the present invention, the specific method for obtaining the grayscale difference characteristic value of each edge in the image of the weld on the surface of the GIS shell is: A neighborhood parameter U is preset, wherein this embodiment is described by taking U=10 as an example, and this embodiment is not specifically limited, wherein U is determined according to specific implementation conditions; The CANNY edge detection algorithm is used to obtain several edges in the image of the weld on the surface of the GIS shell. The CANNY edge detection algorithm is a prior art and will not be described in detail in this embodiment. For any edge pixel point on any edge of the image at the welded portion of the GIS shell surface, take the edge pixel point as the window center, obtain a window of size U×U, and record the window as the neighborhood pixel set of the edge pixel point; The calculation formula for obtaining the grayscale distribution eigenvalue of the rth edge is:

[0022] Where, Represents the grayscale distribution eigenvalue of the rth edge; Represents the number of all edge pixels on the rth edge; Represents the number of all pixels in the neighborhood pixel set of the i-th edge pixel on the r-th edge; Represents the grayscale value of the i-th edge pixel on the r-th edge; Represents the grayscale value of the jth edge pixel in the neighborhood pixel set of the i-th edge pixel on the r-th edge; represents the linear normalization function.

[0023] It should be noted that when the gradient directions of all pixels on the edge tend to be consistent and stable, that is, the edge has a significant change, and this change is continuous and uniform, then this type of edge has a more obvious weld structure.

[0024] In the embodiment of the present invention, the specific method for obtaining the weld contrast of each edge is as follows based on the grayscale difference characteristic value and the distribution of the gradient direction of the edge pixel points: The sine value of the angle between the gradient direction of the i-th edge pixel on the r-th edge and the horizontal direction is recorded as the edge direction factor of the i-th edge pixel; the calculation expression for obtaining the weld contrast of the r-th edge is:

[0025] Where, represents the weld contrast of the rth edge; Represents the grayscale distribution eigenvalue of the rth edge; Represents the number of all edge pixels on the rth edge; Represents the edge direction factor of the i-th edge pixel on the r-th edge; Represents the edge direction factor of the i+1th edge pixel on the rth edge; Indicates taking the absolute value; Represents an exponential function with a natural constant as its base.

[0026] In the embodiment of the present invention, all edges are screened according to the weld contrast, and the specific method for obtaining suspected weld edges is as follows: A significant threshold T1 is preset, wherein this embodiment is described by taking T1=0.75 as an example, and this embodiment does not make any specific limitation, wherein T1 is determined according to specific implementation conditions; If the weld contrast of the rth edge is greater than or equal to the significant threshold T1, the rth edge is recorded as a suspected weld edge.

[0027] It should be noted that since welds usually appear as areas with obvious edge contrast, that is, a clear boundary is formed by two edges; each suspected weld edge corresponds to a matching weld edge. If the number and distance of edge pixels between the two suspected weld edges are closer, and the weld contrast is greater, it is easier for them to match each other to form a suspected weld area.

[0028] In the embodiment of the present invention, the specific formula for obtaining the matching characteristic value between any two suspected weld edges according to the weld contrast is:

[0029] Where, represents the matching feature value between the c1th suspected weld edge and the c2th suspected weld edge; represents the mean value of the weld contrast between the c1th suspected weld edge and the c2th suspected weld edge; Represents the number of all edge pixels on the c1th suspected weld edge; Represents the number of all edge pixels on the c2th suspected weld edge; represents the Euclidean distance between the jth edge pixel on the c1th suspected weld edge and the c2th suspected weld edge; Represents an exponential function with a natural constant as its base.

[0030] In an embodiment of the present invention, a specific method for obtaining a suspected weld edge to be matched for each suspected weld edge according to the matching feature value is as follows: A matching parameter Z is preset, wherein this embodiment is described by taking Z=10 as an example, and this embodiment is not specifically limited thereto, wherein Z is determined according to specific implementation conditions; The matching eigenvalues ​​between the c1th suspected weld edge and all other suspected weld edges are obtained, and the first Z suspected weld edges with the largest matching eigenvalues ​​are used as the suspected weld edges to be matched with the c1th suspected weld edge.

[0031] At this point, the suspected weld edge to be matched for each suspected weld edge is obtained.

[0032] S003. Based on the area formed between the suspected weld edge and the suspected weld edge to be matched, obtain the matching reliability between the suspected weld edge and the suspected weld edge to be matched; based on the matching reliability, obtain the authenticity of each suspected weld edge; and based on the authenticity of the suspected weld edge, obtain the weld edge.

[0033] It should be noted that cracks in the weld area are usually prone to appear in the middle of the weld area, that is, the joint area of ​​the weld, and are usually cracks along the direction of the weld; the cracks may match the edge of the weld area, resulting in errors in the detection of the suspected weld edge, that is, the suspected weld edge to be matched obtained at this time is inaccurate; therefore, it is necessary to further analyze the area formed by the suspected weld edge and each suspected weld edge to be matched to obtain the true weld edge.

[0034] In an embodiment of the present invention, when the grayscale value difference between the edge pixels of the suspected weld area and its peripheral area is small, it indicates that the edge of the weld is relatively smooth and the transition with the surrounding area is relatively natural. That is, the closer the distance difference between the symmetrical pixel points at the edge position and the surrounding background pixel points is, the closer the distance difference is. This is consistent with the typical characteristics of the weld area. On the contrary, when the weld edge is matched with the crack edge, it is usually found that the grayscale difference between the edge of the crack area and the peripheral area is large because the edge of the crack is more obvious and the transition with the metal surface is more abrupt. Based on the area formed between the suspected weld edge and the suspected weld edge to be matched, the specific method for obtaining the matching reliability between the suspected weld edge and the suspected weld edge to be matched is: The area formed by the edge of the vth suspected weld and the edge of the kth suspected weld to be matched is recorded as the kth suspected weld area to be matched; Obtain a sequence of all pixels on the circumscribed rectangle of the kth suspected weld area to be matched, and record it as the background pixel sequence of the kth suspected weld area to be matched; For the ath edge pixel point on the kth suspected weld area to be matched, obtain the distance between the ath edge pixel and all pixels in the background pixel sequence of the kth suspected weld area to be matched, and record the pixel point with the closest distance as the target pixel point of the ath edge pixel point; The straight line passing through the centroid of the kth suspected weld area to be matched and parallel to the edge of the vth suspected weld is recorded as the target straight line; among all edge pixels of the kth suspected weld area to be matched, the pixel point symmetrical to the ath edge pixel about the target straight line is obtained and recorded as the comparison pixel point of the ath edge pixel; The specific formula for obtaining the matching reliability between the suspected weld edge and the suspected weld edge to be matched is:

[0035] Where, represents the matching reliability between the vth suspected weld edge and the kth suspected weld edge to be matched; represents the matching feature value between the vth suspected weld edge and the kth suspected weld edge to be matched, Represents the number of all pixels in the background pixel sequence of the kth suspected weld area to be matched; Represents the grayscale value of the mth pixel in the background pixel sequence of the kth suspected weld area to be matched; Represents the mean of the grayscale values ​​of all pixels in the background pixel sequence of the kth suspected weld area to be matched; Represents the number of all edge pixels on the kth suspected weld area to be matched; represents the Euclidean distance between the ath edge pixel point and the target pixel point in the kth suspected weld area to be matched; It represents the Euclidean distance between the comparison pixel point of the ath edge pixel point on the kth suspected weld area to be matched and the target pixel point; Indicates taking the absolute value; Represents an exponential function with a natural constant as its base.

[0036] It should be noted that if the suspected weld edge is a weld edge, there will only be one suspected weld edge that truly matches the suspected weld edge, indicating that the maximum value of the matching reliability of the suspected weld edge will be significantly different from that of other suspected weld edges to be matched; conversely, if the suspected weld edge is a crack edge, there will be no suspected weld edge that matches the suspected weld edge, and the matching reliability of the suspected weld edges will be relatively small, and the differences between them will be close.

[0037] In the embodiment of the present invention, the specific formula for obtaining the authenticity of each suspected weld edge according to the matching reliability is:

[0038] Where, Indicates the authenticity of the vth suspected weld edge; represents all the suspected weld edges to be matched for the vth suspected weld edge; represents the maximum value of the matching reliability between the vth suspected weld edge and all suspected weld edges to be matched; represents the matching reliability between the vth suspected weld edge and the bth suspected weld edge to be matched; Indicates taking the absolute value; represents the linear normalization function; In an embodiment of the present invention, a specific method for obtaining the weld edge according to the authenticity of the suspected weld edge is: A weld threshold value T2 is preset, wherein this embodiment is described by taking T2=0.7 as an example, and this embodiment is not specifically limited, wherein T2 is determined according to specific implementation conditions; If the authenticity of the vth suspected weld edge is greater than or equal to the weld threshold T2, the vth suspected weld edge is recorded as the weld edge.

[0039] At this point, the weld edge in the image of the weld on the GIS shell surface is obtained.

[0040] S004. Perform image enhancement based on the weld contrast at the weld edge to obtain an enhanced image of the weld on the GIS shell surface.

[0041] In the embodiment of the present invention, the specific method for performing image enhancement based on the weld contrast of the weld edge to obtain the enhanced image of the weld on the surface of the GIS shell is as follows: For any weld edge, obtain the matching reliability between the weld edge and all other weld edges, and use the weld edge with the greatest matching reliability as the matching weld edge of the weld edge; and use the area formed by each weld edge and its matching weld edge as the weld area; The average weld contrast of the two weld edges in each weld area is used as the enhancement coefficient of each weld area. Based on the enhancement coefficient of the weld area and using the adaptive Laplace sharpening algorithm, the specific method for enhancing the image of the weld on the surface of the GIS shell is as follows: The sharpening intensity of each pixel point not in the weld area is set to 0; the sharpening intensity of each pixel point in each weld area is set to the enhancement coefficient of each weld area; thus, the sharpening intensity of each pixel point in the image of the weld area on the GIS shell surface is obtained; The grayscale values ​​of all pixels in the image of the welded portion of the GIS shell surface after adaptive Laplace sharpening are used to form a sharpened and enhanced image, which is recorded as the sharpened and enhanced image of the image of the welded portion of the GIS shell surface.

[0042] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An image enhancement method for GIS shell weld detection, characterized in that: include: Acquire images of the welds on the surface of the GIS shell; Obtain the grayscale difference feature value of each edge in the image of the weld on the GIS shell surface; The weld contrast of each edge is obtained based on the grayscale difference characteristic value and the distribution of the gradient direction of the edge pixel points; all edges are screened based on the weld contrast to obtain suspected weld edges; According to the weld contrast, the matching feature value between any two suspected weld edges is obtained; Obtaining a suspected weld edge to be matched for each suspected weld edge according to the matching feature value; Obtaining a matching reliability between the suspected weld edge and the suspected weld edge to be matched based on an area formed between the suspected weld edge and the suspected weld edge to be matched; obtaining the authenticity of each suspected weld edge based on the matching reliability; and obtaining the weld edge based on the authenticity of the suspected weld edge; Image enhancement is performed based on the weld contrast at the weld edge to obtain an enhanced image of the weld on the GIS shell surface.

2. The image enhancement method for GIS shell weld detection according to claim 1 is characterized in that: The step of obtaining the grayscale difference characteristic value of each edge in the image of the welded portion of the GIS shell surface includes: A neighborhood parameter U is preset. For any edge pixel point on any edge of the image at the welded portion of the GIS shell surface, a window with a size of U×U is obtained with the edge pixel point as the window center, and the window is recorded as the neighborhood pixel set of the edge pixel point. Where, Represents the grayscale distribution eigenvalue of the rth edge; Represents the number of all edge pixels on the rth edge; Represents the number of all pixels in the neighborhood pixel set of the i-th edge pixel on the r-th edge; Represents the grayscale value of the i-th edge pixel on the r-th edge; Represents the grayscale value of the jth edge pixel in the neighborhood pixel set of the i-th edge pixel on the r-th edge; represents the linear normalization function.

3. The image enhancement method for GIS shell weld detection according to claim 1 is characterized in that: The step of obtaining the weld contrast of each edge based on the grayscale difference characteristic value and the distribution of the gradient direction of the edge pixel points includes: The sine value of the angle between the gradient direction of the i-th edge pixel on the r-th edge and the horizontal direction is recorded as the edge direction factor of the i-th edge pixel; the calculation expression for obtaining the weld contrast of the r-th edge is: ； Where, represents the weld contrast of the rth edge; Represents the grayscale distribution eigenvalue of the rth edge; Represents the number of all edge pixels on the rth edge; Represents the edge direction factor of the i-th edge pixel on the r-th edge; Represents the edge direction factor of the i+1th edge pixel on the rth edge; Indicates taking the absolute value; Represents an exponential function with a natural constant as its base.

4. The image enhancement method for GIS shell weld detection according to claim 1 is characterized in that: The method of screening all edges according to the weld contrast to obtain suspected weld edges includes: A significant threshold T1 is preset. If the weld contrast of the r-th edge is greater than or equal to the significant threshold T1, the r-th edge is recorded as a suspected weld edge.

5. The image enhancement method for GIS shell weld detection according to claim 1, characterized in that: The specific formula for obtaining the matching characteristic value between any two suspected weld edges based on the weld contrast is: ； Where, represents the matching feature value between the c1th suspected weld edge and the c2th suspected weld edge; represents the mean value of the weld contrast between the c1th suspected weld edge and the c2th suspected weld edge; Represents the number of all edge pixels on the c1th suspected weld edge; Represents the number of all edge pixels on the c2th suspected weld edge; represents the Euclidean distance between the jth edge pixel on the c1th suspected weld edge and the c2th suspected weld edge; Represents an exponential function with a natural constant as its base.

6. The image enhancement method for GIS shell weld detection according to claim 1, characterized in that: The step of obtaining a suspected weld edge to be matched for each suspected weld edge according to the matching feature value includes: A matching parameter Z is preset, and the matching eigenvalues ​​between the c1th suspected weld edge and all other suspected weld edges are obtained. The first Z suspected weld edges with the largest matching eigenvalues ​​are used as the suspected weld edges to be matched with the c1th suspected weld edge.

7. The image enhancement method for GIS shell weld detection according to claim 1, characterized in that: The obtaining, based on the area formed between the suspected weld edge and the suspected weld edge to be matched, the matching reliability between the suspected weld edge and the suspected weld edge to be matched includes: The area formed by the vth suspected weld edge and the kth suspected weld edge to be matched is recorded as the kth suspected weld area to be matched; the sequence composed of all pixels on the circumscribed rectangle of the kth suspected weld area to be matched is obtained, and recorded as the background pixel sequence of the kth suspected weld area to be matched; for the ath edge pixel point on the kth suspected weld area to be matched, the distance between the ath edge pixel and all pixels in the background pixel sequence of the kth suspected weld area to be matched is obtained, and the pixel with the closest distance is recorded as the target pixel of the ath edge pixel point pixel point; the straight line passing through the centroid position of the kth suspected weld area to be matched and parallel to the vth suspected weld edge is recorded as the target straight line; on all edge pixel points of the kth suspected weld area to be matched, the pixel point symmetrical to the ath edge pixel point about the target straight line is obtained, and recorded as the comparison pixel point of the ath edge pixel point; according to the background pixel point sequence of the kth suspected weld area to be matched, as well as the distances between the edge pixel points, the comparison pixel points and the target pixel points, the matching reliability between the suspected weld edge and the suspected weld edge to be matched is obtained.

8. The image enhancement method for GIS shell weld detection according to claim 7, characterized in that: The specific formula for obtaining the matching reliability between the suspected weld edge and the suspected weld edge to be matched is: ； Where, represents the matching reliability between the vth suspected weld edge and the kth suspected weld edge to be matched; represents the matching feature value between the vth suspected weld edge and the kth suspected weld edge to be matched, Represents the number of all pixels in the background pixel sequence of the kth suspected weld area to be matched; Represents the grayscale value of the mth pixel in the background pixel sequence of the kth suspected weld area to be matched; Represents the mean of the grayscale values ​​of all pixels in the background pixel sequence of the kth suspected weld area to be matched; Represents the number of all edge pixels on the kth suspected weld area to be matched; represents the Euclidean distance between the ath edge pixel point and the target pixel point in the kth suspected weld area to be matched; It represents the Euclidean distance between the comparison pixel point of the ath edge pixel point on the kth suspected weld area to be matched and the target pixel point; Indicates taking the absolute value; Represents an exponential function with a natural constant as its base.

9. The image enhancement method for GIS shell weld detection according to claim 1, characterized in that: The specific formula for obtaining the authenticity of each suspected weld edge based on the matching reliability is: ； Where, Indicates the authenticity of the vth suspected weld edge; represents all the suspected weld edges to be matched for the vth suspected weld edge; represents the maximum value of the matching reliability between the vth suspected weld edge and all suspected weld edges to be matched; represents the matching reliability between the vth suspected weld edge and the bth suspected weld edge to be matched; Indicates taking the absolute value; represents the linear normalization function.

10. The image enhancement method for GIS shell weld detection according to claim 1, characterized in that: The step of obtaining the weld edge according to the authenticity of the suspected weld edge includes: A weld threshold T2 is preset. If the authenticity of the vth suspected weld edge is greater than or equal to the weld threshold T2, the vth suspected weld edge is recorded as the weld edge.

Images