Stainless steel butt weld identification method, system, medium and equipment

Abstract

The invention relates to the technical field of welding monitoring, and discloses a stainless steel butt weld identification method, system, medium and equipment, and the method comprises the steps: identifying a welding point frame and a weld frame for a to-be-identified weld image; for each welding seam frame, calculating a gray contribution degree score, an edge definition score, a position rationality score and a shape integrity score, and performing weighted summation to obtain a fitness value; and fitting a butt weld line in a region of interest by taking the weld frame with the maximum fitness value as the region of interest. Compared with the prior art, the interference of the welding points is eliminated, the interference of the identified bad welding seam frame is eliminated by selecting the optimal welding seam frame through the gray contribution degree score, the marginal definition score, the position rationality score and the shape integrity score, the precision of fitting the butt welding straight line is improved, and the welding seam tracking precision is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of welding monitoring. Specifically, it relates to a method, system, medium and device for identifying stainless steel butt welds. Background Technique

[0002] The statements in this part only provide background technical information related to the present invention and do not necessarily constitute prior art.

[0003] For the gapless butt weld of stainless steel thin plates without grooves, spot welding is usually used to fix the two stainless steel thin plates. Utilizing the smooth surface characteristics of stainless steel, under the action of the lighting system, a stainless steel weld image with bright bands is obtained from the camera. The region of interest is determined by the brightness difference on both sides of the weld, the weld feature points are extracted, and the weld line is fitted, thereby realizing the weld tracking technology.

[0004] However, in the above weld tracking technology, due to the lack or small change in the height of the stainless steel surface, the bright bands formed by the lighting device on the stainless steel surface make it difficult to distinguish the welding fixation points from the bright areas on both sides, resulting in inaccurate determination of the region of interest, greatly reducing the accuracy of fitting the weld line, and thus making the weld tracking effect poor. Summary of the Invention

[0005] In order to solve the above problems, the present invention provides a method, system, medium and device for identifying stainless steel butt welds. After identifying the weld frame and the welding fixation point frame, the optimal weld frame is selected as the region of interest, and the butt weld is fitted within the region of interest. This not only excludes the interference of the welding fixation points, but also, through the optimal weld frame selected by the gray contribution score, edge clarity score, position rationality score and shape integrity score, excludes the interference of the identified bad weld frames, and improves the accuracy of fitting the butt welding line.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] The first aspect of the present invention provides a method for identifying stainless steel butt welds, which includes:

[0008] Obtain the weld image to be identified;

[0009] For the weld image to be identified, identify the welding fixation point frame and the weld frame;

[0010] For each weld frame, calculate the gray contribution score, edge sharpness score, position rationality score, and shape integrity score, and perform weighted summation to obtain the fitness value. Among them, the gray contribution score is used to measure the gray contribution of the pixel points with gray values greater than or equal to the dynamic threshold within the weld frame to the entire weld frame. The edge sharpness score is obtained by comprehensively calculating the horizontal gradient and vertical gradient of each pixel point passed by the weld frame. The position rationality score is calculated based on the distance relationship between the center point of the current weld frame and the center point of the historical optimal weld frame. The shape integrity score is used to measure the closeness of the shape of the current weld frame to the standard weld shape in terms of aspect ratio and area.

[0011] Take the weld frame with the maximum fitness value as the region of interest, and fit the butt weld line within the region of interest.

[0012] Further, the fitting steps of the butt weld line include: after performing mean filtering on the region of interest, search for the minimum valley point within the region of interest, extract the weld feature points into the weld feature point array, and perform a straight line Hough transform on the weld feature point array to obtain the butt weld line.

[0013] Further, the gray contribution score is: Where T is the dynamic threshold, T = α * I + β, I is the ambient light intensity, and α and β are adjustment factors; Q ≥ T represents the number of pixel points with gray values greater than or equal to the dynamic threshold T within the weld frame; M * N is the pixel size of the weld frame; Gray_min is the lowest gray value benchmark, Gray_max is the highest gray value benchmark; Gray_p is the average gray value of the weld frame.

[0014] Further, the edge sharpness is: In the formula, W represents the total number of pixel points passed by the weld frame, (x, y) represents the coordinates of the pixel points passed by the weld frame, G x (x, y) represents the horizontal gradient of the pixel at (x, y), G y (x, y) represents the vertical gradient of the pixel at (x, y).

[0015] Further, the position rationality score is: Where (x c , y c ) is the coordinate of the center point of the current weld frame, (u x , u y ) is the coordinate of the center point of the historical optimal weld frame.

[0016] Further, the shape integrity score is: Wherein, AR is the aspect ratio of the minimum circumscribed rectangle of the current weld frame, AR0 is the aspect ratio of the standard weld frame shape, △A is the difference between the area of the current weld frame and the area of the standard weld frame (A - A0), and A0 is the area of the standard weld frame.

[0017] The second aspect of the present invention provides a stainless steel butt weld identification system, which includes:

[0018] An image acquisition module, which is configured to: acquire a weld image to be identified;

[0019] A weld frame identification module, which is configured to: for the weld image to be identified, identify the welding fixation point frame and the weld frame;

[0020] A fitness calculation module, which is configured to: for each weld frame, calculate the gray contribution score, the edge sharpness score, the position rationality score, and the shape integrity score, and perform weighted summation to obtain a fitness value; wherein, the gray contribution score is used to measure the gray contribution of the pixel points with gray values greater than or equal to the dynamic threshold within the weld frame to the entire weld frame; the edge sharpness score is obtained by comprehensively calculating the horizontal gradient and the vertical gradient of each pixel point passed by the weld frame; the position rationality score is calculated through the distance relationship between the center point of the current weld frame and the center point of the historical optimal weld frame; the shape integrity score is used to measure the closeness of the shape of the current weld frame to the standard weld shape in terms of aspect ratio and area;

[0021] A weld fitting module, which is configured to: use the weld frame with the maximum fitness value as the region of interest, and fit the butt weld line within the region of interest.

[0022] Further, the fitting step of the butt weld line includes: after performing mean filtering on the region of interest, searching for the minimum valley point within the region of interest, extracting the weld feature points into a weld feature point array, and performing a straight line Hough transform on the weld feature point array to obtain the butt weld line.

[0023] The third aspect of the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps in a stainless steel butt weld identification method as described above are implemented.

[0024] The fourth aspect of the present invention provides a computer device, including a memory, a processor, and a computer program stored on the memory and running on the processor. When the processor executes the program, the steps in a stainless steel butt weld identification method as described above are implemented.

[0025] Compared with the prior art, the beneficial effects of the present invention are:

[0026] The present invention provides a method for identifying stainless steel butt welds. After identifying the weld frame and the welding fixation point frame, the optimal weld frame is selected as the region of interest, and the butt weld is fitted within the region of interest. This not only excludes the interference of the welding fixation points, but also, the optimal weld frame selected through the gray contribution score, edge clarity score, position rationality score, and shape integrity score excludes the interference of the identified poor weld frames, improves the accuracy of fitting the butt welding line, and ensures the accuracy of weld tracking. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The accompanying drawings forming a part of this specification are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute a limitation to the present invention.

[0028] Figure 1 It is a block diagram of a method for identifying stainless steel butt welds according to Embodiment 1 of the present invention;

[0029] Figure 2 It is a schematic diagram of a weld image to be identified according to Embodiment 1 of the present invention;

[0030] Figure 3 It is a schematic diagram of the welding fixation point frame and the weld frame according to Embodiment 1 of the present invention;

[0031] Figure 4 It is a schematic diagram of the fitted welding line according to Embodiment 1 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0032] The present invention will be further described below in conjunction with the drawings and embodiments.

[0033] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0034] Without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. The present invention will be further described below in conjunction with the drawings and embodiments.

[0035] Embodiment 1

[0036] As introduced in the background art, in the butt welding of stainless steel thin plates without gaps and without grooves, due to the existence of welding fixation points, it affects the accuracy of weld identification. To solve the above technical problems, this embodiment provides a method for identifying stainless steel butt welds.

[0037] A method for identifying butt welds of stainless steel provided in this embodiment forms a reflection on the surface of a bright-surface stainless steel workpiece through the infrared light emitted by the lighting device, uses a camera to capture a stainless steel weld image with a bright band, distinguishes the weld from the welded point in the weld image obtained by the camera, selects the optimal weld area as the region of interest, obtains the weld feature points in the region of interest, and fits the weld trajectory line.

[0038] A method for identifying butt welds of stainless steel provided in this embodiment includes the following steps:

[0039] Step 1: Label the weld image to obtain a training set and a validation set.

[0040] Label the recognition frames for several collected weld images to distinguish the welded points and the welds; organize the labeled weld images and the recognition frame data into a data set; use the yolo-coco format conversion tool to convert the data set format into the coco format; divide the data set, with 70% as the training set for training and 30% as the validation set for validation.

[0041] During the process of labeling the recognition frames, use rectangular frames to label the welded points and the welds respectively.

[0042] Step 2: Construct a weld recognition model and train it using the training set.

[0043] Among them, the weld recognition model uses the YOLOV3 object detection network, uses the training set to perform recognition training on the weld recognition model so that it can recognize the welds and welded points in the weld image; and uses the validation set to verify the weight file of the optimal weld recognition model generated by training to ensure that the welds and welded points can be accurately recognized and meet the production application standard; export the weight file of the optimal weld recognition model that meets the production application standard to generate a dll file, deploy it to the industrial control computer of the production equipment control system, and embed it into the main weld tracking program.

[0044] After training and validating the weld recognition model, it is used for the recognition of butt welds in actual applications. As Figure 1 shown, the following introduces the specific steps.

[0045] Step 3: Obtain the weld image to be recognized.

[0046] In this embodiment, the lighting device is fixed together with the camera and the welding torch. The positions and angles of the lighting device and the camera are adjusted synchronously, and the height positions and horizontal positions of the lighting device, the camera and the welding torch are adjusted synchronously. The weld image to be recognized can be obtained through the lighting device and the camera. Specifically, it includes:

[0047] Step 301: Install a filter in front of the camera lens, adjust the angle between the lighting device and the camera so that a horizontal bright band is formed on the weld image to be recognized obtained through the camera;

[0048] Step 302: Adjust the heights of the lighting device, camera and welding torch so that an appropriate welding distance is maintained between the tungsten electrode of the welding torch and the stainless steel workpiece, and at the same time, align the tungsten electrode of the welding torch with the weld of the stainless steel workpiece.

[0049] Step 303: Adjust the angles of the lighting device and the camera so that the bright band formed on the weld image to be recognized obtained through the camera is located in the middle position on the image, and align the crosshair for tracking the weld with the weld. At the same time, display a partial image of the welding torch at the upper edge of the weld image, that is, the arc-shaped bright area in the upper edge of the picture, to observe the relative position between the bright band area and the welding torch and prevent tracking errors caused by the excessive distance between the weld recognition area and the welding torch.

[0050] Step 304: Adjust the aperture and focal length of the camera lens so that the bright band formed on the weld image to be recognized is clearly visible, as Figure 2 shown.

[0051] Step 4: For the weld image to be recognized, obtain a welding fixation point recognition frame and one or more weld recognition frames (abbreviation: weld frames) through the weld recognition model, as Figure 3 shown, and exclude the welding fixation point recognition frame.

[0052] Step 5: Select the optimal weld recognition frame as the region of interest for weld tracking.

[0053] Step 501: Set the minimum gray value reference Gray_min and the maximum gray value reference Gray_max, and use the floating-point algorithm to calculate the gray value Gray of each pixel point, Gray = 0.2989 * R + 0.5870 * G + 0.1140 * B, where R, G, and B respectively represent the values of the pixel point on the red, green, and blue color channels.

[0054] Step 502: For each weld recognition frame (assuming the size is M rows and N columns), calculate the average gray value Gray-p.

[0055] Among them, the average gray value: Among them, G i (x, y) represents the gray value of the i-th pixel point in the p-th weld recognition frame, Q represents the number of pixel points in the p-th weld recognition frame; (x, y) represents the position of the pixel point in the p-th weld recognition frame.

[0056] Step 503: Set the fitness function, assign weights to the four sub - indicators of the gray - level contribution score, edge sharpness score, position rationality score, and shape integrity score; and calculate the fitness value of each weld recognition box. Take the weld recognition box with the maximum fitness value as the region of interest (ROI region) for weld tracking recognition. Among them, the gray - level contribution score is used to measure the gray - level contribution of the pixel points with gray - level values greater than or equal to the dynamic threshold within the weld box to the entire weld box; the edge sharpness score is obtained by comprehensively calculating the horizontal gradient and vertical gradient of each pixel point passed by the weld box; the position rationality score is calculated through the distance relationship between the center point of the current weld box and the center point of the historical optimal weld box; the shape integrity score is used to measure the degree of closeness between the shape of the current weld box and the standard weld shape in terms of aspect ratio and area.

[0057] Among them, the fitness function: F = w1*s1 + w2*s2 + w3*s3 + w4*s4; where, F is the fitness value of the weld recognition box (the value range is 0 - 1); w i is the weight of the evaluation dimension, the gray - level contribution weight w1 = 0.4, the edge sharpness weight w2 = 0.3, the position rationality weight w3 = 0.2, w4 = 0.1; s i is the normalized score of the evaluation dimension.

[0058] Among them, the gray - level contribution score s1: Among them, T is the dynamic threshold, T = α*I + β, I is the ambient light intensity, α and β are adjustment factors; Q≥T represents the number of pixel points with gray - level values greater than or equal to the threshold T within the weld recognition box; M, N are the horizontal and vertical pixel sizes of the weld recognition box, and M*N is the pixel size of the weld recognition box.

[0059] Among them, the edge sharpness score In the formula, W represents the total number of pixel points at the position where the weld recognition box is located (that is, the pixel points passed by the line of the weld recognition box), (x,y) represents the coordinates of the pixel points at the position where the weld recognition box is located, G x (x,y) represents the horizontal gradient of the pixel at (x,y), G y (x,y) represents the vertical gradient of the pixel at (x,y).

[0060] Among them, the position rationality score Among them, (x c ,y c ) is the center - point coordinate of the current weld recognition box, (u x ,u y ) is the center - point coordinate of the historical optimal weld box.

[0061] Among them, the shape integrity score Wherein, AR is the aspect ratio of the minimum circumscribed rectangle of the current weld frame, AR0 is the aspect ratio of the standard weld frame shape, △A is the difference between the area of the current weld frame and the area of the standard weld frame (A - A0), and A0 is the area of the standard weld frame.

[0062] Step 6, Weld seam tracking: Perform 3×3 filtering on the ROI region selected in the weld image to remove noise points, extract weld feature points, and calculate the weld against the welding trajectory line, as Figure 4 shown, specifically:

[0063] Step 601: Perform mean filtering on the ROI region in the weld image to be recognized collected by the camera to remove noise points, search for the minimum trough point within the ROI region, search for one minimum trough point for each row within the ROI region, and the minimum trough point is the weld feature point. Extract the weld feature points into the weld feature point array;

[0064] Step 602: Perform a straight line Hough transform on the obtained weld feature point array to obtain the fitted welding straight line, which is the recognized weld seam, as Figure 4 shown.

[0065] In this embodiment, the weld seam and the welded joint area are recognized through deep learning, and the optimal weld seam area is selected as the region of interest. The butt weld seam is fitted within the region of interest, which not only excludes the interference of the welded joint points, but also, the weld seam area selected through the gray contribution score, edge sharpness score, position rationality score, and shape integrity score excludes the interference of the obtained bad weld recognition frames, improves the accuracy of fitting the butt welding straight line, and ensures the accuracy of weld seam tracking.

[0066] Embodiment 2

[0067] The purpose of this second embodiment is to provide a stainless steel butt weld seam recognition system, including:

[0068] An image acquisition module, which is configured to: acquire a weld image to be recognized;

[0069] A weld frame recognition module, which is configured to: for the weld image to be recognized, recognize the welded joint frame and the weld frame;

[0070] A fitness calculation module, which is configured to: for each weld frame, calculate the gray contribution score, edge sharpness score, position rationality score, and shape integrity score, and perform weighted summation to obtain the fitness value; wherein, the gray contribution score is used to measure the gray contribution of the pixel points with gray values greater than or equal to the dynamic threshold within the weld frame to the entire weld frame; the edge sharpness score is obtained by comprehensively calculating the horizontal gradient and vertical gradient of each pixel point passed by the weld frame; the position rationality score is calculated by the distance relationship between the center point of the current weld frame and the center point of the historical optimal weld frame; the shape integrity score is used to measure the closeness of the shape of the current weld frame to the standard weld shape in terms of aspect ratio and area.

[0071] A weld fitting module, which is configured to: use the weld frame with the maximum fitness value as the region of interest, and fit the butt weld line within the region of interest.

[0072] Further, the fitting steps of the butt weld line include: after performing mean filtering on the region of interest, searching for the minimum valley point within the region of interest, extracting the weld feature points into the weld feature point array, and performing a straight line Hough transform on the weld feature point array to obtain the butt weld line.

[0073] Further, the gray contribution score is: Wherein, T is the dynamic threshold, T = α * I + β, I is the ambient light intensity, α and β are adjustment factors; Q ≥ T represents the number of pixel points with gray values greater than or equal to the dynamic threshold T within the weld frame; M * N is the pixel size of the weld frame; Gray_min is the lowest gray value reference, Gray_max is the highest gray value reference; Gray_p is the average gray value of the weld frame.

[0074] Further, the edge sharpness is: In the formula, W represents the total number of pixel points passed by the weld frame, (x, y) represents the coordinates of the pixel points passed by the weld frame, G x (x, y) represents the horizontal gradient of the pixel at (x, y), G y (x, y) represents the vertical gradient of the pixel at (x, y).

[0075] Further, the position rationality score is: Wherein, (x c , y c ) is the coordinate of the center point of the current weld frame, (u x , u y ) is the coordinate of the center point of the historical optimal weld frame.

[0076] Further, the shape integrity score is: Wherein, AR is the aspect ratio of the minimum circumscribed rectangle of the current weld seam frame, AR0 is the aspect ratio of the standard weld seam frame shape, △A is the difference between the area of the current weld seam frame and the area of the standard weld seam frame (A - A0), and A0 is the area of the standard weld seam frame.

[0077] It should be noted here that each module in this embodiment corresponds one by one to each step in Embodiment 1, and the specific implementation process is the same, so it will not be repeated here.

[0078] Embodiment 3

[0079] This embodiment provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the steps in a stainless steel butt weld seam recognition method as described in Embodiment 1 above are implemented.

[0080] Embodiment 4

[0081] This embodiment provides a computer device, including a memory, a processor, and a computer program stored on the memory and running on the processor. When the processor executes the program, the steps in a stainless steel butt weld seam recognition method as described in Embodiment 1 above are implemented.

[0082] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

[0083] Although the specific implementation manners of the present invention have been described above in conjunction with the drawings, it is not a limitation to the protection scope of the present invention. Those skilled in the art should understand that based on the technical solutions of the present invention, various modifications or deformations that can be made without creative efforts by those skilled in the art are still within the protection scope of the present invention.

Claims

1. A method for identifying stainless steel butt welds, characterized in that, Including: Obtain the weld image to be recognized; For the weld image to be recognized, identify the welding fixation point frame and the weld frame; For each weld frame, calculate the gray contribution score, edge sharpness score, position rationality score, and shape integrity score, and perform weighted summation to obtain the fitness value; among them, the gray contribution score is used to measure the gray contribution of the pixel points with gray values greater than or equal to the dynamic threshold within the weld frame to the entire weld frame; the edge sharpness score is obtained by comprehensively calculating the horizontal gradient and vertical gradient of each pixel point passed by the weld frame; the position rationality score is calculated through the distance relationship between the center point of the current weld frame and the center point of the historical optimal weld frame; the shape integrity score is used to measure the closeness of the shape of the current weld frame to the standard weld shape in terms of aspect ratio and area; Use the weld frame with the maximum fitness value as the region of interest, and fit the butt weld line within the region of interest.

2. The stainless steel butt weld identification method according to claim 1, wherein The fitting steps of the butt weld line include: after performing mean filtering on the region of interest, search for the minimum valley point within the region of interest, extract the weld feature points into the weld feature point array, and perform a straight line Hough transform on the weld feature point array to obtain the butt weld line.

3. The method for identifying a stainless steel butt weld according to claim 1, characterized in that, The gray contribution score is as follows: where T is the dynamic threshold, T = α * I + β, I is the ambient light intensity, α and β are adjustment factors; Q ≥ T represents the number of pixels with gray values greater than or equal to the dynamic threshold T within the weld frame; M * N is the pixel size of the weld frame; Gray_min is the minimum gray value reference, Gray_max is the maximum gray value reference; Gray_p is the average gray value of the weld frame.

4. The stainless steel butt weld identification method according to claim 1, characterized in that, The edge sharpness is as follows: In the formula, W represents the total number of pixel points passed by the weld seam frame, (x, y) represents the coordinates of the pixel points passed by the weld seam frame, G x (x, y) represents the horizontal gradient of the pixel at (x, y), G y (x, y) represents the vertical gradient of the pixel at (x, y).

5. The stainless steel butt weld identification method according to claim 1, characterized in that, The position rationality score is as follows: where (x c , y c ) are the coordinates of the center point of the current weld seam frame, and (u x , u y ) are the coordinates of the center point of the historical optimal weld seam frame.

6. The stainless steel butt weld identification method according to claim 1, wherein The shape integrity score is as follows: where AR is the aspect ratio of the minimum circumscribed rectangle of the current weld seam frame, AR0 is the aspect ratio of the shape of the standard weld seam frame, △A is the difference between the area of the current weld seam frame and the area of the standard weld seam frame (A - A0), and A0 is the area of the standard weld seam frame.

7. A stainless steel butt weld identification system, characterized in that, Including: An image acquisition module, which is configured to: obtain the weld image to be recognized; A weld frame recognition module, which is configured to: for the weld image to be recognized, identify the welding fixation point frame and the weld frame; A fitness calculation module, which is configured to: for each weld frame, calculate the gray contribution score, edge sharpness score, position rationality score, and shape integrity score, and perform weighted summation to obtain the fitness value; among them, the gray contribution score is used to measure the gray contribution of the pixel points with gray values greater than or equal to the dynamic threshold within the weld frame to the entire weld frame; the edge sharpness score is obtained by comprehensively calculating the horizontal gradient and vertical gradient of each pixel point passed by the weld frame; the position rationality score is calculated through the distance relationship between the center point of the current weld frame and the center point of the historical optimal weld frame; the shape integrity score is used to measure the closeness of the shape of the current weld frame to the standard weld shape in terms of aspect ratio and area; A weld fitting module, which is configured to: use the weld frame with the maximum fitness value as the region of interest, and fit the butt weld line within the region of interest.

8. A stainless steel butt weld identification system according to claim 7, characterized in that, The fitting steps of the butt weld line include: after performing mean filtering on the region of interest, search for the minimum valley point within the region of interest, extract the weld feature points into the weld feature point array, and perform a straight line Hough transform on the weld feature point array to obtain the butt weld line.

9. A computer-readable storage medium having a computer program stored thereon, the program being executed by a processor, characterized in that, When the program is executed by the processor, it implements the steps in a stainless steel butt weld recognition method according to any one of claims 1-6.

10. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in a stainless steel butt weld recognition method according to any one of claims 1-6.

Images