CN118052810A - Weld defect detection method, system and computer equipment - Google Patents
Weld defect detection method, system and computer equipment Download PDFInfo
- Publication number
- CN118052810A CN118052810A CN202410361909.3A CN202410361909A CN118052810A CN 118052810 A CN118052810 A CN 118052810A CN 202410361909 A CN202410361909 A CN 202410361909A CN 118052810 A CN118052810 A CN 118052810A
- Authority
- CN
- China
- Prior art keywords
- weld
- map
- gray
- defect
- depth map
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000007547 defect Effects 0.000 title claims abstract description 81
- 238000001514 detection method Methods 0.000 title claims description 28
- 238000000034 method Methods 0.000 claims abstract description 16
- 238000003466 welding Methods 0.000 claims description 18
- 238000013528 artificial neural network Methods 0.000 claims description 17
- 238000004590 computer program Methods 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 4
- -1 pits Substances 0.000 claims description 2
- 238000012216 screening Methods 0.000 claims description 2
- 230000011218 segmentation Effects 0.000 claims description 2
- 239000002893 slag Substances 0.000 claims description 2
- 230000002950 deficient Effects 0.000 claims 1
- 238000007689 inspection Methods 0.000 claims 1
- 238000007781 pre-processing Methods 0.000 claims 1
- 238000005516 engineering process Methods 0.000 abstract description 2
- 238000004458 analytical method Methods 0.000 description 5
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000013145 classification model Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005493 welding type Methods 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
The present invention relates to the field of defect identification technologies, and in particular, to a method, a system, and a computer device for detecting a weld defect; generating a weld depth map and a gray map through the segmented weld point cloud, simultaneously identifying and analyzing the weld depth map and the gray map by utilizing a multi-mode algorithm model, detecting suspected defect areas in the weld, and classifying the suspected defect areas to obtain real weld defects; therefore, by utilizing the weld depth map and the gray level map, the suspected weld area is identified through the multi-mode model, and finally the background and the actual defects are distinguished through the classification network, so that the omission of the weld defects can be effectively reduced under the condition of not depending on a large number of defect samples.
Description
Technical Field
The present invention relates to the field of defect identification technologies, and in particular, to a method, a system, and a computer device for detecting a weld defect.
Background
In general, the quality of a weld directly affects the performance of a welded product, at present, a 2D camera is used for acquiring a weld image for analysis aiming at a weld quality detection part, a weld defect is detected by judging the gray level change or the characteristic of the weld image, and a 3D camera is used for acquiring a weld point cloud image for analysis, and the weld defect is detected by identifying the point cloud image characteristic of the weld defect. In the actual identification process, the method for acquiring the weld image identification defects by the 2D camera does not fully utilize 3D depth information, the detection rate of the weld defects such as tiny air holes, pits and the like is not high, the method for acquiring the weld cloud image identification defects by the 3D camera adds 3D depth information into the 3D depth information for analysis, but a large amount of weld defect data is required to be sent into a neural network for training to acquire a better algorithm model, in the actual industrial production, so much defect data is difficult to provide, part of weld defects possibly cause model identification failure due to the fact that the model identification fails due to the fact that the weld defects are not detected before, and the existing weld defect detection method is low in detection accuracy.
Disclosure of Invention
The invention provides a weld defect detection method, a weld defect detection system and computer equipment, which are used for solving the problem that the existing weld defect detection method is low in detection precision.
In order to achieve the above object, the present invention is realized by the following technical scheme:
In a first aspect, the present invention provides a weld defect detection method, including:
S1: acquiring a 3D point cloud of a welding line, and generating a welding line depth map and a welding line gray level map based on the 3D point cloud;
s2: inputting the weld depth map and the weld gray map into a first neural network for training to obtain a multi-mode algorithm model;
S3: repeating the step S1, inputting the weld depth map and the weld gray map into a multi-modal algorithm model for recognition, and obtaining a suspected defect map output by the multi-modal algorithm model;
s4: and inputting the suspected defect map into a second neural network for classification to obtain n+1 images, wherein n is the defect type.
In a second aspect, the present application provides a weld defect detection system comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the method of the first aspect described above when the computer program is executed.
In a third aspect, the application also provides a computer device comprising a memory storing a computer program and a processor implementing the steps of the method of the first aspect when the processor executes the computer program.
Advantageous effects
According to the weld defect detection method provided by the invention, the weld depth map and the gray map are generated through the segmented weld point cloud, the multi-mode algorithm model is utilized to simultaneously identify and analyze the weld depth map and the gray map, the suspected defect area in the weld is detected, and the suspected defect area is classified to obtain the real weld defect; therefore, by utilizing the weld depth map and the gray level map, the suspected weld area is identified through the multi-mode model, and finally the background and the actual defects are distinguished through the classification network, so that the omission of the weld defects can be effectively reduced under the condition of not depending on a large number of defect samples.
Drawings
FIG. 1 is a flow chart of a weld defect detection method according to a preferred embodiment of the present invention.
Detailed Description
The following description of the present invention will be made clearly and fully, and it is apparent that the embodiments described are only some, but not all, of the embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Unless defined otherwise, technical or scientific terms used herein should be given the ordinary meaning as understood by one of ordinary skill in the art to which this invention belongs. The terms "first," "second," and the like, as used herein, do not denote any order, quantity, or importance, but rather are used to distinguish one element from another. Likewise, the terms "a" or "an" and the like do not denote a limitation of quantity, but rather denote the presence of at least one. The terms "connected" or "connected," and the like, are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "upper", "lower", "left", "right", etc. are used merely to indicate a relative positional relationship, which changes accordingly when the absolute position of the object to be described changes.
Referring to fig. 1, the method for detecting weld defects provided by the present application includes:
S1: acquiring a 3D point cloud of a welding line, and generating a welding line depth map and a welding line gray level map based on the 3D point cloud;
s2: inputting the weld depth map and the weld gray map into a first neural network for training to obtain a multi-mode algorithm model;
S3: repeating the step S1, inputting the weld depth map and the weld gray map into a multi-modal algorithm model for recognition, and obtaining a suspected defect map output by the multi-modal algorithm model;
s4: and inputting the suspected defect map into a second neural network for classification to obtain n+1 images, wherein n is the defect type.
In this embodiment, the first neural network may be PatchCore networks and the second neural network may be RepVGG networks, which are examples only and not limiting.
According to the weld defect detection method, the weld depth map and the gray map are generated through the segmented weld point cloud, the multi-mode algorithm model is utilized to simultaneously identify and analyze the weld depth map and the gray map, suspected defect areas in the weld are detected, and the suspected defect areas are classified to obtain real weld defects; therefore, by utilizing the weld depth map and the gray level map, the suspected weld area is identified through the multi-mode model, and finally the background and the actual defects are distinguished through the classification network, so that the omission of the weld defects can be effectively reduced under the condition of not depending on a large number of defect samples.
The steps of the above-described weld defect detection method are described in detail below with a complete example.
Step one: and 3D point clouds of the welding line are obtained, all the point clouds of the scanned object are obtained from the 3D line scanning camera, and as the welding line area only occupies a part of the scanning area, the welding line point cloud data are cut out from the scanning area by adopting a point cloud segmentation algorithm, the background interference is removed, and then the step two is carried out.
Step two: generating a weld depth image and a gray image, firstly, recording all Z-direction values of weld point clouds, stretching the Z-direction values to 0-255, carrying out rounding treatment, secondly, recording all intensity values of weld point clouds, stretching the intensity values to 0-255, carrying out rounding treatment, secondly, uniformly and orderly arranging the weld point clouds according to X and Y directions, counting the maximum points in the X and Y directions, creating blank depth images and blank gray images with the same specification according to the points, sequentially filling the Z-direction values of the stretched weld point clouds into the blank depth images, sequentially filling the intensity values of the stretched weld point clouds into the blank gray images to form the depth image and the gray image of the weld, and then entering a step III.
It is worth mentioning that before the weld depth map and the weld gray map are input into the first neural network for training, the method comprises the following steps:
and carrying out position one-to-one correspondence processing on the weld depth map and the weld gray map.
Specifically, the position one-to-one correspondence processing is performed on the weld depth map and the weld gray map, including:
And filling the pixel values of the defect cloud in the weld depth map and the weld gray map with the pixel values of 0 so as to realize one-to-one correspondence at the pixel positions.
In this embodiment, it should be noted that, the weld depth map is an image generated based on the 3D point cloud, the X and Y direction coordinates correspond to the pixel positions in the image according to the point cloud, the weld gray map is also an image generated based on the 3D point cloud, the X and Y direction coordinates correspond to the pixel positions in the image according to the point cloud, and the pixel values 0 are filled for both the empty point cloud pixel values, so that the pixel positions on the weld depth map and the weld gray map are in one-to-one correspondence.
Step three: and simultaneously, sending the weld seam depth map and the gray map into a neural network for training to obtain an algorithm model, firstly screening the weld seam depth map and the gray map data set once to ensure that the weld seam data set has no defects, and forming a positive sample library, wherein the pictures do not need to be marked, are carried out in an unsupervised mode, and then are preprocessed to meet the input requirements of the multi-mode algorithm model, and then are sent into the neural network for training to obtain an algorithm prediction model, and then enter step four.
Step four: and (3) repeating the work from the first step to the second step, inputting the obtained weld depth map and gray map into the algorithm model generated in the third step for identification analysis, and judging whether suspected defects exist, wherein small pictures are cut out from the same positions of the original weld depth map and the gray map respectively for the positions where the detection results are the suspected defects, and are respectively named and stored, and then step five is carried out.
Step five: and sending the cut pictures into a neural network for classification, and dividing the cut pictures into a background and defects, wherein the classified defects comprise six types of welding slag, pits, air holes, broken welding, missing welding and incomplete welding, and if the cut pictures are predicted as defect types by a classification model, judging that the welding line is unqualified, otherwise, the welding line is qualified.
The application also provides a weld defect detection system, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor realizes the steps of the weld defect detection method when executing the computer program. The weld defect detection system can realize each embodiment of the weld defect detection method and achieve the same beneficial effects, and the description is omitted here.
The application also provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor realizes the steps of the incremental layout method of the multi-camera when executing the computer program. The computer equipment can realize each embodiment of the weld defect detection method and achieve the same beneficial effects, and the description is omitted here.
The foregoing describes in detail preferred embodiments of the present invention. It should be understood that numerous modifications and variations can be made in accordance with the concepts of the invention by one of ordinary skill in the art without undue burden. Therefore, all technical solutions which can be obtained by logic analysis, reasoning or limited experiments based on the prior art by the person skilled in the art according to the inventive concept shall be within the scope of protection defined by the claims.
Claims (10)
1. A weld defect detection method, characterized by comprising:
s1: acquiring a 3D point cloud of a welding line, and generating a welding line depth map and a welding line gray scale map based on the 3D point cloud;
S2: inputting the weld depth map and the weld gray map into a first neural network for training to obtain a multi-mode algorithm model;
S3: repeating the step S1, inputting a weld depth map and a weld gray map into the multi-mode algorithm model for recognition, and obtaining a suspected defect map output by the multi-mode algorithm model;
S4: inputting the suspected defect map into a second neural network for classification to obtain n+1 images, wherein n is the defect type.
2. The weld defect detection method of claim 1, wherein the generating a weld depth map and a weld gray map based on the 3D point cloud comprises:
Stretching the Z-direction value of the 3D point cloud to 0-255, and stretching the intensity value of the 3D point cloud to 0-255;
And uniformly and orderly arranging the 3D point clouds in the X direction and the Y direction, counting the maximum points in the X direction and the Y direction, creating blank depth images and blank gray images with the same specification according to the maximum points, sequentially filling the Z direction values of the stretched 3D point clouds into the blank depth images, sequentially filling the intensity values of the stretched 3D point clouds into the blank gray images, and obtaining a weld depth image and a weld gray image.
3. The weld defect detection method of claim 1, wherein before the inputting the weld depth map and the weld gray map into a first neural network for training, the method comprises:
And carrying out position one-to-one correspondence processing on the weld depth map and the weld gray map.
4. The method for detecting a weld defect according to claim 3, wherein the performing the position-to-one correspondence processing on the weld depth map and the weld gray map includes:
And filling the pixel values of the defect cloud in the weld depth map and the weld gray map with the pixel value 0 so as to realize one-to-one correspondence at the pixel positions.
5. The weld defect detection method of claim 1, wherein before the inputting the weld depth map and the weld gray map into a first neural network for training, the method further comprises:
screening out defective images in the weld depth image and the gray level image, wherein the weld depth image and the gray level image without defects form a positive sample library;
and preprocessing the positive sample library, and then sending the preprocessed positive sample library into a neural network for training to obtain a multi-modal algorithm model.
6. The weld defect detection method of claim 1, wherein S4 comprises:
Determining the position of a suspected defect map, and cutting out a defect picture block from the same positions of an original weld depth map and a gray map respectively;
And sending the defect picture blocks into a second neural network for classification to obtain a background and n defects.
7. The method of claim 1, wherein the type of defect comprises flash slag, pits, air holes, broken welds, missing welds, or underfills.
8. The weld defect detection method of claim 1, wherein the acquiring a 3D point cloud of a weld comprises:
acquiring all point clouds of a scanned object from a 3D line scanning camera;
and cutting out 3D point clouds of the welding line by adopting a point cloud segmentation algorithm for the scanning area.
9. A weld defect inspection system comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor performs the steps of the method of any of claims 1 to 8 when the computer program is executed.
10. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any of claims 1 to 8 when the computer program is executed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410361909.3A CN118052810A (en) | 2024-03-28 | 2024-03-28 | Weld defect detection method, system and computer equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410361909.3A CN118052810A (en) | 2024-03-28 | 2024-03-28 | Weld defect detection method, system and computer equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118052810A true CN118052810A (en) | 2024-05-17 |
Family
ID=91044977
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410361909.3A Pending CN118052810A (en) | 2024-03-28 | 2024-03-28 | Weld defect detection method, system and computer equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118052810A (en) |
-
2024
- 2024-03-28 CN CN202410361909.3A patent/CN118052810A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP4280153A1 (en) | Defect detection method, apparatus and system | |
| CN111709948B (en) | Method and device for detecting defects of container | |
| CN109671058B (en) | Defect detection method and system for large-resolution image | |
| CN111507976B (en) | Defect detection method and system based on multi-angle imaging | |
| CN113362326A (en) | Method and device for detecting welding spot defects of battery | |
| CN111815555A (en) | Metal additive manufacturing image detection method and device combining anti-neural network with local binary | |
| TW201310359A (en) | System and method for identifying defects in a material | |
| KR20210150970A (en) | Detecting defects in semiconductor specimens using weak labeling | |
| CN112200790B (en) | Cloth defect detection method, device and medium | |
| CN113111875A (en) | Seamless steel rail weld defect identification device and method based on deep learning | |
| JP2021086379A (en) | Information processing apparatus, information processing method, program, and method of generating learning model | |
| CN115035092A (en) | Image-based bottle detection method, device, equipment and storage medium | |
| CN117252815A (en) | Industrial part defect detection method, system, equipment and storage medium based on 2D-3D multi-mode image | |
| CN112017154A (en) | Ray defect detection method based on Mask R-CNN model | |
| CN117274258A (en) | Method, system, equipment and storage medium for detecting defects of main board image | |
| CN113554645B (en) | Industrial anomaly detection method and device based on WGAN | |
| Jacobsen et al. | A comparison between neural networks and decision trees | |
| Kim et al. | Bi-directional convolutional recurrent reconstructive network for welding defect detection | |
| CN114331961A (en) | Method for defect detection of an object | |
| KR20210132566A (en) | Method and system for classifying defects in wafer using wafer-defect images, based on deep learning | |
| CN114596243A (en) | Defect detection method, device, equipment and computer readable storage medium | |
| Jie et al. | Improved bump detection and defect identification for hbms using refined machine learning approach | |
| CN118052810A (en) | Weld defect detection method, system and computer equipment | |
| CN116381053A (en) | Ultrasonic detection method and system for welding metal materials | |
| García et al. | A configuration approach for convolutional neural networks used for defect detection on surfaces |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |