A test box surface weld quality detection method and system

Abstract

The present invention relates to the field of image processing, and more specifically, the present invention relates to a method and system for detecting the quality of welds on the surface of a test box. The method comprises: obtaining a weld image of the test box to obtain a weld edge image, and segmenting it according to the weld direction to obtain an edge connected domain of a weld in each segment; calculating the deformation probability of each pixel in the edge connected domain; obtaining feature points on the edge connected domain of each weld, constructing a feature matrix, and obtaining a feature value of each weld; forming a feature value sequence with the feature values ​​of the welds in each segment of the weld image and clustering them, calculating the abnormal index of each category, obtaining the maximum value of the abnormal index in all categories as the abnormal index of the segmented weld image, and in response to the abnormal index being greater than or equal to the abnormal threshold, the weld shows abnormality, otherwise, the weld is qualified. The present invention obtains the feature value of each weld through the texture information of the weld, thereby improving the accuracy of weld detection.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and more specifically, to a method and system for detecting the quality of weld seams on the surface of a test box. Background Art

[0002] Test chambers are often used in engineering or scientific research. Test chambers refer to equipment or systems used for experiments or tests, usually used to simulate specific environmental conditions or conduct controlled experiments.

[0003] The existing Chinese patent application document with publication number CN115294034A discloses a weld quality detection method and training method. More information is obtained through the first normal angle map and the information carried itself is more accurate, which can improve the detection accuracy of weld quality detection and solve the problem of high misjudgment rate and missed judgment rate when performing weld quality detection on objects with high surface reflectivity.

[0004] In addition to the problems in the above solutions, in actual production, due to local disturbances on the robot arm, the weld may be partially skewed and the weld points may be uneven while the weld remains generally straight. Therefore, when the welds on the surface of the test chamber are quality inspected, poor weld formation often occurs, the weld formed by the welding interface presents a snake shape, and the fish scale pattern formed by the superposition of welds is uneven and irregular. The serpentine structure may cause uneven metal structure in the weld area, thereby affecting the mechanical properties of the welded structure and reducing its load-bearing capacity and fatigue resistance. For welded structures such as test chambers that require sealing, poor weld formation may cause leakage, posing a safety hazard.

[0005] In addition to the above solutions, the most common method for detecting whether the direction of the weld is straight is the Hough line detection algorithm. However, on the one hand, the Hough line detection cannot capture the local bend of the weld and is not sensitive to the local change of the weld direction. On the other hand, it cannot analyze the texture information of the weld point, resulting in inaccurate weld quality detection results. Summary of the invention

[0006] In order to solve the problem that Hough line detection cannot capture such local weld bends and cannot analyze the texture information of the weld, the present invention provides solutions in the following aspects.

[0007] In a first aspect, a method for detecting the quality of welds on the surface of a test box includes: obtaining a weld image of the test box and performing preprocessing to obtain a weld edge image, segmenting the weld edge image according to the weld direction, and obtaining an edge connected domain of a weld in each segment; calculating the difference of the curvature values ​​of each pixel in the edge connected domain to obtain the deformation probability of each pixel; obtaining feature points on the edge connected domain of each weld, constructing a feature matrix, performing data dimension reduction on the feature matrix, and obtaining a feature value of each weld; forming a feature value sequence from the feature values ​​of the welds in each segment of the weld image and performing clustering to obtain multiple categories, and calculating an abnormality index for each category. , where For the The abnormal index of each category, For the The maximum number of consecutive solder joints of each category, is the maximum continuous number mean of all categories, For the The mean of the solder joint feature values ​​of the largest continuous segment in each category, is the mean of the maximum continuous segment feature values ​​in all classes, is a normalized function, and the maximum value of the abnormal index in all categories is obtained as the abnormal index of the segmented weld image. When the abnormal index is greater than or equal to the abnormal threshold, the weld is abnormal, otherwise, the weld is qualified.

[0008] The effects are as follows: preprocessing the weld image of the test box helps to extract the geometric features and edge shape of the weld; segmenting the weld edge image helps to locate the weld and the surrounding local structure; based on the deformation probability of the weld, it helps to evaluate the deformation degree of the weld, thereby judging the weld quality; clustering the characteristic values ​​of the weld, obtaining the weld anomaly index, and realizing automatic detection and evaluation of the weld quality on the surface of the test box with high efficiency and accuracy.

[0009] In one embodiment, the weld image is preprocessed to obtain a weld edge image, the weld edge image is segmented according to the weld direction, and the edge connected domain of the weld point in each segment is obtained, including:

[0010] De-noising and gray-scaling the weld image, and performing edge detection on the gray-scale image to obtain a weld edge image;

[0011] The weld edge image is segmented according to the weld direction to obtain the weld image of each segment, and the weld image is masked to obtain the weld area of ​​the weld image;

[0012] The center line of the welding area is extracted, and the vertical direction of the welding direction corresponding to each welding point on the center line is obtained for edge detection to obtain the edge connected domain of each welding point.

[0013] The effect is to eliminate the noise in the weld image and convert the color image into a grayscale image. Denoising can use various filters, such as Gaussian filter or median filter, obtain the weld edge through edge detection, use skeletonization or other morphological operations to extract the center line of the weld area, where the center line represents the main axis of the weld, and perform edge detection along the direction perpendicular to the center to obtain the edge connected domain of the weld in each segment. It is helpful to further analyze the shape, size and distribution of the weld, as well as the quality and characteristics of the weld.

[0014] In one embodiment, obtaining the deformation probability includes:

[0015] The distance between each pixel point in the edge connected domain and the central coordinate is calculated, the difference in curvature between the left and right pixels adjacent to any pixel point in the edge connected domain is calculated as the curvature change value, and the product of the distance and the curvature change value is used as the deformation probability of the welding point.

[0016] The effect is that the above calculation can reflect the deformation around the solder joint. Pixels with larger distances and smaller curvature change values ​​may have lower deformation probabilities, while pixels with smaller distances and larger curvature change values ​​may have higher deformation probabilities, which helps to evaluate the deformation degree of the solder joint and further analyze the welding quality and possible defects.

[0017] In one embodiment, the distance between each pixel point in the edge connected domain and the center coordinates includes:

[0018] The mean value of the pixels in the edge connected domain is obtained as the center coordinate, and the Euclidean distance from each pixel in the edge connected domain to the center coordinate is calculated.

[0019] In one embodiment, the step of obtaining feature points on the edge connected domain of each solder joint and constructing a feature matrix includes:

[0020] The two pixel points with the largest deformation probability in the edge connected domain are obtained as feature points, wherein the feature points divide the edge connected domain into two segments, and the sequential midpoints in the two connected domains are obtained as the other two feature points, and four feature points of each welding point are extracted;

[0021] The horizontal coordinates and the vertical coordinates of the four feature points are obtained as two attribute values ​​of the feature, and a 4×2 feature matrix is ​​constructed for each welding point, wherein the feature matrix is ​​the four feature points and their two attribute values.

[0022] The effect is that the two pixels with the highest probability of deformation usually represent the maximum deformation or inflection point of the edge connected domain. Combined with the scene in this embodiment, the edge of each solder joint is roughly crescent-shaped, and the two ends of the crescent are usually the two pixels with the highest probability of deformation of the solder joint. The average position of the pixels in each connected domain is obtained as two additional feature points. These feature points represent the center of the weld around each solder joint, which helps to analyze the shape and position characteristics of each solder joint, thereby obtaining the feature value of each solder joint.

[0023] In one embodiment, the step of forming a feature value sequence from the feature values ​​of the welding points in each weld image and clustering the feature values ​​comprises:

[0024] Normalizing the eigenvalue sequence, and classifying the normalized eigenvalue sequence;

[0025] A histogram is constructed with the characteristic value sequence as the horizontal coordinate and the number as the vertical coordinate, and the number of peak points in the histogram is obtained as the number of clusters of the characteristic value sequence of the welding point to obtain multiple categories.

[0026] The effect is: by normalizing the eigenvalue sequence to ensure the same calculation scale, by clustering the eigenvalue sequence, the solder joints with similar eigenvalues ​​are grouped together, which helps to analyze the distribution of the solder joints.

[0027] In one embodiment, the calculating of the abnormality index of each category further includes:

[0028] Extract the length of the continuous segments in each category, calculate the average continuous segment length of the continuous segments in each category, calculate the square of the difference between the length of the continuous segment and the average continuous segment length, calculate the ratio of the sum of the square differences of all categories to the number of continuous segments, and obtain the continuous length variance;

[0029] The run-length variance is used as a parameter for calculating the anomaly index for each category.

[0030] The effect is that the variance of the continuous length is used as the adjustment parameter of the abnormal index of each category, and the variance of the continuous length indicates the degree of variation of the length of the continuous segments in the category. A large variance of the continuous length may indicate that the distribution of the solder joints in the category is uneven or unstable.

[0031] In a second aspect, a test box surface weld quality detection system includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned test box surface weld quality detection method is implemented.

[0032] The present invention has the following effects:

[0033] 1. The present invention divides the weld by utilizing the texture information of the weld spot and obtains the characteristic value of each weld spot, which is conducive to accurately identifying abnormal areas in the weld and improving the accuracy of weld quality detection.

[0034] 2. Compared with the traditional Hough line detection algorithm, the present invention is more sensitive to small changes in the local weld, and can capture the local bending of the weld and the uneven distribution of weld points, so as to better evaluate the weld quality.

[0035] 3. The present invention can quickly classify the characteristic values ​​of weld points and calculate the abnormal index of welds through the minimum and maximum value normalization method and the k-means algorithm, so as to quickly locate abnormal welds. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0037] Figure 1 It is a method flow chart of steps S1 to S4 in a method for detecting the quality of welds on the surface of a test box according to an embodiment of the present invention.

[0038] Figure 2 It is a method flow chart of steps S10 to S12 in a method for detecting the quality of welds on the surface of a test box according to an embodiment of the present invention.

[0039] Figure 3 It is a method flow chart of steps S30 to S32 in a method for detecting the quality of welds on the surface of a test box according to an embodiment of the present invention.

[0040] Figure 4 It is a schematic diagram of characteristic points of welding spots extracted in a method for detecting weld quality on the surface of a test box according to an embodiment of the present invention.

[0041] Figure 5 It is a method flow chart of steps S40 to S42 in a method for detecting the quality of welds on the surface of a test box according to an embodiment of the present invention.

[0042] Figure 6 The present invention is a structural block diagram of a test box surface weld quality detection system. DETAILED DESCRIPTION

[0043] 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 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0044] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0045] Reference Figure 1 A method for detecting the quality of weld seams on the surface of a test chamber includes steps S1 to S4, which are specifically as follows:

[0046] It should be noted that the specific scenario targeted by the present invention is: the weld surface in the test box has uniform light, the camera is perpendicular to the weld, and the captured weld image has uniform fish scale patterns and weld point edges.

[0047] S1: The weld image of the test box is obtained and preprocessed to obtain a weld edge image, and the weld edge image is segmented according to the weld direction to obtain the edge connected domain of the weld point in each segment.

[0048] To further explain, the technicians in this field segment the weld image. The reason for segmentation is: since the welding directions of the welding positions in the test box may be different, the edge detection result is not good, which affects the subsequent analysis of the abnormal conditions of each weld point and the determination of the welding direction of each segment.

[0049] Reference Figure 2 , including steps S10 to S12:

[0050] S10: De-noising and gray-scale processing are performed on the weld image, and edge detection is performed on the gray-scale image to obtain a weld edge image;

[0051] To further explain, common denoising methods include median filtering, Gaussian filtering, etc., to eliminate noise in the image, convert the denoised weld image into a single-channel grayscale image, and use an edge detection algorithm to obtain the edge image. The edge detection algorithm can use the Sobel operator, Canny edge detection, etc.

[0052] S11: segmenting the weld edge image according to the weld direction to obtain a welding image of each segment, performing a mask operation on the welding image to obtain a welding area of ​​the welding image;

[0053] To further explain, a mask operation is applied to the welding image, and the mask is to set the pixel value of the non-welding area to 0 (black) and the pixel area of ​​the welding area to 255 (white), thereby obtaining a binary image and retaining the welding area.

[0054] S12: extracting the center line of the welding area, obtaining the perpendicular direction of the welding direction corresponding to each welding point on the center line, performing edge detection, and obtaining the edge connected domain of each welding point.

[0055] S2: Calculate the difference of the curvature values ​​of each pixel in the edge connected domain to obtain the deformation probability of each pixel.

[0056] The mean value of the pixels in the edge connected domain is obtained as the center coordinate, the Euclidean distance from each pixel in the edge connected domain to the center coordinate is calculated, the difference in curvature between the adjacent left and right pixels of any pixel in the edge connected domain is calculated as the curvature change value, and the product of the distance and the curvature change value is taken as the deformation probability of the solder joint.

[0057] Specifically, get the first The center coordinates of the solder joints , calculate the The first The coordinates of the pixel points are The pixel to center coordinates The Euclidean distance .

[0058] Get the The first Pixels on both sides and The curvature value of the pixel point is calculated, and the curvature difference is obtained to obtain the curvature change value. .

[0059] Specifically, the curvature change value satisfies the following relationship:

[0060] ,

[0061] In the formula, Indicates The first The curvature change value of each pixel, The current Pixel on the left The curvature value of the pixel on the right The curvature value of a pixel, that is, the curvature change value of the current pixel is the difference between the curvatures of the adjacent left and right pixels.

[0062] The curvature value of the pixel point is calculated using the Hessian Matrix, which is a square matrix containing second-order partial derivative information and is usually used to describe the second-order derivative information of a multivariable function. This technology is well known to those skilled in the art and will not be described in detail.

[0063] Then according to the Euclidean distance With the curvature change value The product of the deformation probability .

[0064] It should be noted that the edge lengths of each solder joint are different. In order to unify the dimensions of the solder joint edges and reduce subsequent calculations, it is considered to further extract feature points on the solder joint edges.

[0065] S3: Obtain feature points on the edge connected domain of each solder point, construct a feature matrix, perform data dimension reduction on the feature matrix, and obtain the feature value of each solder point.

[0066] Reference Figure 3 , including steps S30 to S32:

[0067] S30: obtaining two pixel points with the largest deformation probability in all edge connected domains as feature points, wherein the feature points divide the edge connected domain into two segments, obtaining sequential midpoints in the two connected domains as another two feature points, and extracting four feature points for each welding point;

[0068] Reference Figure 4 , obtain the edge connected domain of some welds in the segment. Usually, the first weld is roughly circular, and the edge contours of other welds except the first weld are roughly crescent-shaped. Take a solder joint as an example and get the maximum deformation probability in the edge connected domain of the solder joint. Two pixel points, among which the maximum deformation probability usually refers to the maximum deformation of the pixel point curvature in the edge contour or is expressed as an inflection point, and obtains The midpoint of both sides gives , and then get the four feature points of the edge connected domain of each solder joint. The first solder joint obtains the four points of its circle, which are the four feature points of the solder joint.

[0069] S31: Obtain the horizontal coordinates and vertical coordinates of the four feature points as two attribute values ​​of the feature, and construct a 4×2 feature matrix for each solder joint, wherein the feature matrix is ​​the four feature points and their two attribute values.

[0070] S32: traverse each solder joint and reduce the dimension of the feature matrix of the solder joint, where one solder joint corresponds to one eigenvalue.

[0071] Further explanation, in this embodiment, the PAC (Patch-based Adaptive Compression) algorithm is used, which is an adaptive block compression algorithm for image compression. The PAC algorithm maps the feature matrix of each solder joint to a lower dimensional space and retains important feature information. In addition, principal component analysis (PCA) and linear discriminant analysis (LDA) can also be used to achieve dimensionality reduction.

[0072] S4: The characteristic values ​​of the welding points in each segment of the weld image are used to form a characteristic value sequence and clustered to obtain multiple categories, and the abnormal index of each category is calculated. The maximum value of the abnormal index in all categories is obtained as the abnormal index of the segmented weld image. When the abnormal index is greater than or equal to the abnormal threshold, the weld is abnormal, otherwise, the weld is qualified.

[0073] To further illustrate, in this embodiment, the abnormal threshold is set to 20.

[0074] Reference Figure 5 , including steps S40 to S42:

[0075] S40: normalizing the eigenvalue sequence, and classifying the normalized eigenvalue sequence;

[0076] In this embodiment, the maximum and minimum value normalization method is used for processing. If the solder joints with similar eigenvalues ​​are adjacent in sequence number before normalization, the solder joints with similar eigenvalues ​​are not adjacent in sequence number after normalization, but should be evenly dispersed in the sequence. Therefore, if a local continuous eigenvalue approximate sequence appears after normalization, and the sequence segment has a large difference with the overall eigenvalue, the segment is considered to be an abnormal part.

[0077] S41: constructing a histogram with the characteristic value sequence as the horizontal coordinate and the number as the vertical coordinate, obtaining the number of peak points in the histogram as the number of clusters of the characteristic value sequence of the solder joint, and obtaining multiple categories;

[0078] It is further explained that the number of clusters is adaptively determined according to the distribution of the eigenvalue sequence, which improves the accuracy and stability of clustering.

[0079] S42: Calculate the abnormality index of each category;

[0080] Specifically, the anomaly index of each category satisfies the following relationship:

[0081] ,

[0082] In the formula, For the The abnormal index of each category, For the The maximum number of consecutive solder joints of each category, is the maximum continuous number mean of all categories, For the The mean of the solder joint feature values ​​of the largest continuous segment in each category, is the mean of the maximum continuous segment feature values ​​in all classes, is the normalization function.

[0083] Further explanation: The larger the value is, the more continuous the number of the solder joint sequence fragments belongs to the same category, and the greater the probability of abnormality in the area; The larger the value, the greater the difference between the number of consecutive segments in the segment and the average number of consecutive segments; The larger the value, the greater the difference between the morphological feature value of the fragment and the overall feature; therefore, and The larger the value of , the more likely the largest contiguous segment in that category is an anomaly.

[0084] In addition, in another embodiment, the length of the continuous segments in each category is extracted, the average continuous segment length of the continuous segments in each category is calculated, the square of the difference between the length of the continuous segment and the average continuous segment length is calculated, the ratio of the sum of the square differences of all categories to the number of continuous segments is calculated, and the continuous length variance is obtained, and the continuous length variance is used as a parameter for calculating the abnormality index of each category.

[0085] Specifically, the anomaly index of each category satisfies the following relationship:

[0086] ,

[0087] In the formula, For the The abnormal index of each category, For the The maximum number of consecutive solder joints of each category, is the maximum continuous number mean of all categories, For the The mean of the solder joint feature values ​​of the largest continuous segment in each category, is the mean of the maximum continuous segment feature values ​​in all classes, is the normalization function, For the The variance of the maximum number of consecutive solder joints in each category.

[0088] The continuous length variance can be used to indicate the degree of variation in the length of continuous segments in this category. A larger continuous length variance may indicate that the distribution of welds in this category is uneven or unstable. A larger continuous length variance will lead to an increase in the anomaly index of the category, thereby identifying irregular or abnormal areas in the weld.

[0089] So far, the present invention is completed and the normal or abnormal condition of the weld is obtained.

[0090] The present invention also provides a test box surface weld quality detection system. Figure 6 As shown, the system includes a processor and a memory, the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a test box surface weld quality detection method according to the first aspect of the present invention is implemented.

[0091] The system also includes other components familiar to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.

[0092] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0093] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0094] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A method for detecting the quality of welds on the surface of a test chamber, characterized in that: include: The weld image of the test box is obtained and preprocessed to obtain a weld edge image, the weld edge image is segmented according to the weld direction, and the edge connected domain of the weld point in each segment is obtained; Calculate the distance between each pixel point in the edge connected domain and the central coordinate, calculate the difference in curvature between the left and right pixels adjacent to any pixel point in the edge connected domain as the curvature change value, and use the product of the distance and the curvature change value as the deformation probability of the welding point; The two pixel points with the largest deformation probability in the edge connected domain are obtained as feature points, wherein the feature points divide the edge connected domain into two segments, and the sequential midpoints in the two connected domains are obtained as the other two feature points, and the four feature points of each solder joint are extracted; the horizontal coordinates and the vertical coordinates of the four feature points are obtained as the two attribute values ​​of the feature, and a 4×2 feature matrix is ​​constructed for each solder joint, wherein the feature matrix is ​​the four feature points and their two attribute values, and the feature matrix is ​​subjected to data dimensionality reduction to obtain the feature value of each solder joint; The characteristic values ​​of the welding points in each weld image are used to form a characteristic value sequence and clustered to obtain multiple categories. The abnormal index of each category is calculated. , where For the The abnormal index of each category, For the The maximum number of consecutive solder joints of each category, is the maximum continuous mean of all categories, For the The mean of the solder joint feature values ​​of the largest continuous segment in each category, is the mean of the maximum continuous segment feature values ​​in all classes, is a normalized function, and the maximum value of the abnormal index in all categories is obtained as the abnormal index of the segmented weld image. When the abnormal index is greater than or equal to the abnormal threshold, the weld is abnormal, otherwise, the weld is qualified.

2. A test box surface weld quality detection method according to claim 1, characterized in that: Preprocessing the weld image to obtain a weld edge image, segmenting the weld edge image according to the weld direction, and obtaining an edge connected domain of the weld point in each segment, including: De-noising and gray-scaling the weld image, and performing edge detection on the gray-scale image to obtain a weld edge image; The weld edge image is segmented according to the weld direction to obtain the weld image of each segment, and the weld image is masked to obtain the weld area of ​​the weld image; The center line of the welding area is extracted, and the vertical direction of the welding direction corresponding to each welding point on the center line is obtained for edge detection to obtain the edge connected domain of each welding point.

3. A test box surface weld quality detection method according to claim 1, characterized in that: in, Obtaining the distance between each pixel point in the edge connected domain and the center coordinates includes: The mean value of the pixels in the edge connected domain is obtained as the center coordinate, and the Euclidean distance from each pixel in the edge connected domain to the center coordinate is calculated.

4. A test box surface weld quality detection method according to claim 1, characterized in that: in, The step of forming a feature value sequence from the feature values ​​of the welding points in each weld image and performing clustering includes: Normalizing the eigenvalue sequence, and classifying the normalized eigenvalue sequence; A histogram is constructed with the characteristic value sequence as the horizontal coordinate and the number as the vertical coordinate, and the number of peak points in the histogram is obtained as the number of clusters of the characteristic value sequence of the welding point to obtain multiple categories.

5. A test chamber surface weld quality detection system, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the test box surface weld quality detection method according to any one of claims 1 to 4 is implemented.

Images