Automated macrographic cross-sectional analysis of joints

Abstract

A method for analyzing a cross-section of a joint includes capturing an image of the cross-section of the joint with a camera, analyzing the image of the cross-section of the joint with one or more neural networks to classify two or more body segments that make up the joint, separating the body segments of the joint with the one or more neural networks, reassembling the body segments to form an assembled image of the joint, identifying key points in the assembled image of the joint, and using the key points to measure values ​​of properties of the joint. A system for automated cross-section analysis of a joint is also provided.

Description

[Technical Field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This PCT international patent application claims the benefit of and priority to U.S. Provisional Patent Application No. 63 / 427,115, filed November 22, 2022, entitled "Automated Macrographic Section Analysis Of Joints," the entire disclosure of which is incorporated herein by reference.

[0002] This disclosure relates generally to cross-sectional analysis of welds joining two or more pieces of metal. More particularly, this disclosure relates to automated systems and methods for analyzing such welds. [Background technology]

[0003] Periodic destructive inspection of joints between workpieces, such as self-piercing rivets (SPR), brazing, laser welding, spot welding, or gas metal arc welding (GMAW), is a common testing requirement to ensure the quality of such joints. Various parties, such as original equipment manufacturers (OEMs), may have performance requirements or standards for testing of GMAW joints.

[0004] The destructive testing process may involve extracting one or more axial cross sections of the weld, polishing them, and etching them to enhance the contrast between the weld and the base material, and then having a skilled operator inspect the polished and etched cross sections using a magnification imaging system. The inspection process often requires both measuring the weld dimensions and identifying defects within the weld, if any. The results of the test are included in a report and used as a quality control tool. This process is very time-consuming and labor-intensive, resulting in significant direct labor costs.

[0005] Traditional inspection processes require skilled operators, increasing operational costs. Due to the length of time required to process the large number of welds on a product, it can take hours or even days from the time the product is manufactured until a final quality report is generated. Such delays can result in hundreds of defective parts being manufactured before the problem is identified.

[0006] Additionally, some OEMs require that testing be completed and quality confirmed before the entire product batch can be shipped to the OEM for installation in an assembled product, such as an automobile. This process, known as "batch and hold," results in significant storage, management, and logistics costs.

[0007] Quality reports and associated measurement data are often presented and stored in a format that can be difficult to track over time and correlate with welding process data, which can limit the value of the data for process improvement.

[0008] Finally, as with any process that relies on humans, some variability in methods and procedures may exist, which may lead to inconsistent and / or erroneous test results. Summary of the Invention

[0009] The present disclosure provides a method for analyzing a cross-section of a joint between two workpieces, the method including capturing an image of the cross-section of the joint using a camera, analyzing the image of the cross-section of the joint using machine learning techniques to classify two or more body segments that make up the joint, separating the body segments of the joint using machine learning techniques, reassembling the body segments to form an assembled image of the joint, identifying keypoints in the assembled image of the joint, and using the keypoints to measure values ​​of properties of the joint.

[0010] The present disclosure also provides a method for analyzing a cross-section of a joint between two workpieces, the method including capturing an image of the cross-section of the joint using a camera, analyzing the image of the cross-section of the joint using machine learning techniques to classify two or more body segments that make up the joint, separating the body segments of the joint using machine learning techniques, reassembling the body segments to form an assembled image of the joint, and identifying characteristics indicative of defects in the joint based on the body segments.

[0011] The present disclosure also provides a system for automated cross-sectional analysis of a joint between two workpieces. The system includes a processor and a memory containing instructions that, when executed by the processor, cause the processor to analyze an image of the cross-section of the joint using machine learning techniques, classify two or more body segments that make up the joint, separate the body segments of the joint using machine learning techniques, reassemble the body segments to form an assembled image of the joint, identify keypoints in the assembled image of the joint, and use the keypoints to measure values ​​of properties of the joint.

[0012] Further details, features and advantages of the inventive design may be gleaned from the following description of exemplary embodiments with reference to the associated drawings. [Brief explanation of the drawings]

[0013] [Figure 1] FIG. 1 is a block diagram of a system for automated cross-sectional analysis of a weld in accordance with the present disclosure. [Figure 2] 1 is a flow chart illustrating steps in a method for cross-sectional analysis of a weld. [Figure 3] FIG. 1 shows a cross-sectional image of a weld connecting two metal pieces. [Figure 4A] FIG. 4 shows three separated body segments based on the cross-sectional image of FIG. 3. [Figure 4B] FIG. 4 shows three separated body segments based on the cross-sectional image of FIG. 3. [Figure 4C] FIG. 4 shows three separated body segments based on the cross-sectional image of FIG. 3. [Figure 5] FIG. 4D shows a reassembled image of the three separated body segments of FIGS. 4A-4C, with key points within the welds identified. [Figure 6] 6 shows the reassembled image of FIG. 5, with the lines representing the welding measurements. [Figure 7] 1 is a flowchart of steps in a method for automated inspection of welds according to the present disclosure. [Figure 8] 8 is a flowchart of substeps for automatic image processing in the method of FIG. 7. [Figure 9] FIG. 1 illustrates a series of training images used to train an artificial intelligence (AI) model, according to the present disclosure. [Figure 10] 1A-1C are cross-sectional images of a weld illustrating features of a semantic segmentation method according to the present disclosure. [Figure 11] FIG. 10 shows a cross-sectional image of a first weld illustrating weld properties measured by the method and system of the present disclosure. [Figure 12] FIG. 10 is a cross-sectional image of a second weld showing weld properties measured by the method and system of the present disclosure. [Figure 13A] FIG. 1 shows a cross-sectional image of a weld joining two metal pieces, where the lines show measurements identified by a manual inspection process. [Figure 13B] 13B shows a cross-sectional image of the weld of FIG. 13A, where the lines represent measurements identified by an automated system and method according to the present disclosure. [Figure 14] 1 is an architecture diagram of a system for automated cross-sectional analysis of joints, such as welds, in accordance with the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0014] With reference to the drawings, the present invention will be explained in detail in view of the following embodiments.

[0015] The objective of the disclosed systems and methods is to reduce the time and labor associated with inspecting and measuring weld cross-section images. The disclosed systems and methods can reduce the direct labor costs of image inspection and measurement and reduce the total time for report generation. The disclosed systems and methods can provide faster process control and reduce batch and hold costs compared to traditional manual processes. Additionally, the disclosed systems and methods improve data storage and availability to enable improved reporting, process monitoring, and correlation. By automating steps in the weld cross-section image inspection and measurement process, variability and human error can be reduced or eliminated, resulting in more consistent and reliable weld cross-section inspection and measurement.

[0016] The disclosed systems and methods can reduce cycle time for processing macrographic images of weld cross sections by automatically inspecting and measuring the specimen. Unless otherwise defined, a macrographic image is an image magnified or otherwise provided at a scale at which relevant features are visible to the naked eye. Macrographic images can have a magnification scale of 1.0x to 10x.

[0017] The disclosed method can include: 1. utilizing trained neural networks and machine vision algorithms to segment the weld pool and substrate from the image; 2. using the body with pixels / mm calibration to measure properties of the weld using algorithms; 3. using additional neural networks and machine vision algorithms to identify and measure weld defects (undercut, porosity, burn-through, etc.) as needed; 4. displaying predicted results to an operator for review and adjustment as needed; and 5. automatically generating a report and adding the weld information and measurements to a database for use in other dashboards and displays.

[0018] The disclosed systems and methods can use semantic segmentation neural network models in trained models used to predict bodies in weld cross-section images. There can be several different trained models for predicting bodies and detecting defects from various weld joint types, shapes, sizes, and materials.

[0019] The systems and methods of the present disclosure can use data annotations: training and test sets to train segmentation neural network models and evaluate their performance. A set of general logical rules can be used to measure weld properties based on the segmented bodies predicted by the neural network, and these may be referred to as a "semantic segmentation algorithm." There may be several different sets of rules that allow for measurement of different joint types, customer specifications, and the presence of defects. The semantic segmentation algorithm may, for example, attempt to predict weld properties.

[0020] The disclosed systems and methods can provide rapid sample processing. By automating the measurement of weld properties, a single operator can process samples much faster. The disclosed systems and methods can provide improved flexibility over conventional systems and methods. The same trained neural network can be used for a wide variety of weld joint types and products with little or no customization. As such, the disclosed systems and methods can provide greater cost savings with less upfront investment. The disclosed systems and methods can also provide improved robustness. The trained neural network is much more robust to changing image conditions than conventional machine vision and image processing algorithms. This reduces the opportunity and severity of errors in processing large volumes of images. Furthermore, the system can be deployed as new images and conditions are added to the training set.

[0021] As shown in FIG. 1 , an automated inspection system 20 is provided for automated analysis of a cross-section of a weld 10. The inspection system 20 includes a computing device 22 configured to perform some or all of the functions of the automated inspection system 20. The computing device 22 may include a tablet, such as an iPad, or an Android or Windows tablet device. The computing device 22 may be another type of device, such as a smartphone, smart glasses, a laptop, a netbook, or the like. In some embodiments, the computing device 22 may include a tablet due to its high processor performance, long battery life, ease of use, and relatively low cost.

[0022] As shown in the exemplary embodiment depicted in the block diagram of FIG. 1 , the computing device 22 includes a user interface 30 and a first processor 32 coupled to a first machine-readable storage memory 34. The user interface 30 comprises an output device 36 configured to present output data to a user and an input device 38 configured to receive input data from the user. The output device 36 may include a video display, such as a display screen, a projection display, or a virtual reality (VR) or augmented reality (AR) image. Alternatively or additionally, the output device 36 may include an audio output, such as one or more speakers that provide output in the form of an audible signal. The input device 38 may include a touchscreen, a keyboard, a mouse, a trackpad, a trackball, or gesture input. Alternatively or additionally, the input device 38 may include hardware and / or software that responds to verbal commands. The output device 36 may be combined with the input device 38 as, for example, a touchscreen.

[0023] The computing device 22 includes a camera 40 having a field of view 42 for viewing a cross-section of the weld 10. The camera 40 may be configured to capture images of the cross-section of the weld 10 in the visible light spectrum. Alternatively or additionally, the camera 40 may use other non-visible wavelengths, such as infrared (IR) and / or ultraviolet (UV). The camera 40 may be configured to capture video that can be presented on the output device 36 as a live image. Alternatively or additionally, the camera 40 may be configured to capture still images of the field of view 42, including images of the cross-section of the weld 10. The video and / or still images captured by the camera 40 may be stored in memory for future use. In some embodiments, an inverted microscope may be used in which the field of view 42 of the camera 40 faces upward toward the cross-section of the weld 10. Such an inverted microscope allows gravity to assist in holding the cross-section of the weld 10 on a flat surface.

[0024] 1, computing device 22 includes a first communication interface 48 configured to transmit data to and receive data from a server 60 over a network 50. First communication interface 48 may include, for example, a Universal Serial Bus (USB) or Ethernet interface, or a wired or wireless interface, such as Wi-Fi, ZigBee, or cellular data radio. Network 50 may include one or more wired and / or wireless segments, which may include, for example, Wi-Fi, ZigBee, Ethernet, infrared, etc.

[0025] The first machine-readable storage memory 34 may include one or more of RAM memory, ROM memory, flash, or DRAM, and may include magnetic, optical, semiconductor, or another type of machine-readable storage. The computing device 22 also includes first instructions 44 stored in the first machine-readable storage memory 34 that direct the first processor 32 to: cause the output device 36 to present particular output data to a user; cause the first processor 32 to receive feedback from the user via the input device 38; store data in a first data storage area 46 of the first storage memory 34; and transmit the data to the server 60. The first instructions 44 may include compiled or interpreted data instructions that cause the first processor 32 to perform operations that enable the functionality of the automated inspection system 20.

[0026] The server 60 includes a second communication interface 62 for communicating with and / or to the computing device 22. The second communication interface 62 may include one or more wired and / or wireless interfaces, which may be of the same type as or a different type from the first communication interface 48. The server 60 also includes a second processor 64 and a second machine-readable storage memory 66 including second instructions 68 and a second data storage area 52 for storing data. As shown in FIG. 1 , the second data storage area 52 may be organized as a database. Alternatively, or in addition, the data may be stored in an external database outside of the second machine-readable storage memory 66 of the server 60. For example, the data may be hosted on a dedicated database.

[0027] The second instructions 68 may be configured to cause the second processor 64 to store and analyze the data.

[0028] Either or both of first processor 32 and / or second processor 64 may be configured to process one or more images captured by camera 40 and perform one or more functions, such as machine learning (ML) and / or artificial intelligence (AI) processing for automated analysis of a cross-section of weld 10. For example, first processor 32 and / or second processor 64 may work in conjunction to analyze the images captured by the camera to generate a binary segmented image, locate key points, and calculate measurements of properties of weld 10 based on the binary segmented image. Either or both of first processor 32 and / or second processor 64 may be further configured to generate a report listing the measurements of several properties of weld 10.

[0029] 2 shows a flow diagram illustrating steps in a first method 70 for cross-sectional analysis of a weld. The first method 70 may be performed by one or more controllers, such as first processor 32 and / or second processor 64. As can be understood in light of the present disclosure, the order of operations in the method is not limited to sequential execution as shown in FIG. 2, but may be performed in one or more varying orders, as appropriate, in accordance with the present disclosure.

[0030] The first method 70 includes, at 72, performing a visual inspection of the weld 10. The visual inspection may be performed manually (i.e., by a human operator) and / or automatically (i.e., by a computer vision system). The visual inspection may be used to identify external defects in the weld 10, such as misalignments, cracks, etc.

[0031] The first method 70 also includes, at step 74, cutting and grinding the weld 10 to obtain one or more cross sections of the weld 10. Step 74 may include manual and / or automated operations.

[0032] The first method 70 also includes, at step 76, etching one or more cross sections of the weld 10 to produce an etched cross section. Step 76 may include manual and / or automated operations. The etched cross section may enhance the visibility of the boundaries between the constituent pieces of the weld 10, such as two or more pieces of metal and / or the weld material joining the two or more pieces of metal.

[0033] The first method 70 also includes, at step 78, imaging the etched cross-section of the weld 10. Step 78 may include using a camera, such as a digital camera, to generate image data representative of the etched cross-section of the weld 10. The image data may be used for further manual and / or automated inspection processing. In some embodiments, the image data may be stored for future use.

[0034] The first method 70 also includes cross-sectional measurements of the weld 10 at step 80. Step 80 may include automatically determining one or more measurements of the weld 10 based on the image data provided in step 78.

[0035] The first method 70 also includes generating a report about the weld 10 at step 82. Step 82 may include comparing one or more measurements of the weld 10 to predetermined values ​​to determine whether the weld 10 meets specifications or requirements. Generating the report may include generating one or more graphical representations of the measurements, such as lines overlaid on an image of a cross-section of the weld 10. The report may indicate one or more characteristics that are within or outside a predetermined range (e.g., pass / fail) for each of the one or more corresponding measurements.

[0036] In some embodiments, steps 72-78 of first method 70 may be performed by a human operator using a manual process. In some embodiments, steps 80-82 of first method 70 may be performed using an automated process. However, one or more of steps 72-82 of first method 70 may be performed in other ways in accordance with the present disclosure.

[0037] 3 shows a cross-sectional image of a weld 10 connecting two workpieces 12, 14, with a weld material 16 joining the two workpieces 12, 14. The weld material 16 can include materials added during the welding process and / or a combination of materials from one or both of the workpieces 12, 14 that are melted or otherwise transformed in forming the weld 10.

[0038] Figures 4A-4C show three separate body segments 12, 14, 16 based on the cross-sectional image of Figure 3. Figure 4A shows the first metal piece 12 of the workpieces 12, 14 to be joined at the weld 10. Figure 4B shows the weld material 16, and Figure 4C shows the second metal piece 14 of the workpieces 12, 14 to be joined at the weld 10.

[0039] The three body segments 12, 14, 16 can be separated using machine learning techniques, which can include, for example, supervised learning (SL) techniques, unsupervised learning (UNS) techniques, or reinforcement techniques. The machine learning techniques can be implemented using one or more machine learning frameworks, such as Tensorflow or PyTorch. In some embodiments, the machine learning techniques can include one or more neural networks, such as a convolutional neural network (CNN), a recurrent neural network (RNN), or a feedforward neural network. In some embodiments, the machine learning techniques can be implemented using a specific model of a neural network. Examples of such specific models include, for example, U-net, U-net++, FPN, PAN, DeepLapV3, and DeepLabV3+. In some embodiments, the machine learning techniques can be implemented using one or more supporting libraries. Examples of such supporting libraries include, for example, Numpy, Sympy, CV2, Scikit-Image, Pandas, Sklearn, Poutyne, Albumments, Torchmetrics, Imgaug, and Openpyxl.

[0040] In one exemplary embodiment, one or more neural networks, such as a convolutional neural network (CNN), can be used to classify the body segments 12, 14, 16 by recognizing features of the body segments 12, 14, 16. One or more neural networks can then segment or separate the body segments 12, 14, 16. A U-Net CNN can be used for this function in the automated inspection system 20. The U-Net CNN can locate and distinguish boundaries through classification at every pixel. An open-source Python library with a pre-trained backbone can be utilized for the model to train the U-Net CNN.

[0041] 5 shows a reassembled image including the three body segments 12, 14, 16 of FIGS. 4A-4C with key points 90 identified within the weld 10. The key points 90 represent intersections between the body segments 12, 14, 16 at predetermined locations. The key points 90 can be identified from the image of the weld 10 using a logic algorithm.

[0042] FIG. 6 shows the reassembled image of FIG. 5 with lines 92, 94 indicating measurements of properties of the weld 10. The measurements can be determined by converting the number of pixels in the image into a linear measurement, such as millimeters (mm). For example, the disclosed systems and methods can be configured to determine the weld penetration depth 92 by performing the following steps: 1) determining key points 90 between the weld material 16 and the second metal piece 14 at each of two points on the periphery of the weld 10; 2) drawing a first line 92 connecting the two key points 90; and 3) finding the deepest point on the weld boundary between the weld material 16 and the second metal piece 14 in a direction perpendicular to the first line 92 and calculating the distance between the deepest point and the first line 92, as indicated by a second line 94.

[0043] 7 illustrates a flowchart of steps in a second method 100 for automated inspection of a weld 10 in accordance with the present disclosure. The second method 100 may be performed by one or more controllers, such as first processor 32 and / or second processor 64. As can be understood in light of the present disclosure, the order of operations in the method is not limited to sequential execution as shown in FIG. 7, but may be performed in one or more varying orders, as appropriate, in accordance with the present disclosure.

[0044] The second method 100 includes, at step 110, acquiring an image of a cross section of the weld 10.

[0045] The second method 100 also includes automatically processing the acquired images to determine measurements of properties of the weld 10 at step 112 .

[0046] The second method 100 proceeds to present the measurement results to the user at step 114. The presented results may include a pass or fail indication for one or more of the measured properties of the weld 10. In some embodiments, the presented results may include a pass or fail indication for the entire weld, which may be determined based on comparing the results of the measurement to predetermined tolerances for one or more different properties of the weld 10.

[0047] The second method 100 also includes adjusting the position in step 116. For example, the automated inspection system 20 can prompt the user to provide one or more additional cross sections of the weld 10 to allow the automated inspection system 20 to measure properties of the weld 10 not shown or otherwise determined in the previous image. The second method 100 can loop through steps 110-116 over several iterations until all properties of all welds 10 for a given test are completed. The operator can adjust the actual measurements within a single sample. The process can be repeated for several samples in a set. The set can include welds from the same product being tested.

[0048] In some embodiments, steps 110, 114, and 116 of second method 100 may be performed, at least in part, by a human operator using user interface 30. In some embodiments, step 112 of second method 100 may be performed using an automated process. However, one or more of steps 110-116 of second method 100 may be performed in other ways in accordance with the present disclosure. In some embodiments, step 112 of second method 100 may be performed, in part or in whole, using an automated process.

[0049] The second method 100 also includes, in step 118, storing all weld property measurements for a given test in a database and generating a report of the measurements. In some embodiments, step 118 can be performed, at least in part, by a human operator using the user interface 30. For example, the user interface 30 can present a list of options for saving property measurements, such as selecting, naming, and annotating the measurements to be saved. Alternatively or additionally, step 118 can be performed, in whole or in part, using an automated process.

[0050] FIG. 8 shows a flowchart of substeps for performing step 112 of automatic image processing in second method 100 of FIG. 7. Step 112 includes segmenting the image into n bodies in substep 120, where n is an integer corresponding to a predetermined number of bodies or components that comprise weld 10. Substep 120 can be performed using one or more convolutional neural networks (CNNs). In some embodiments, substep 120 can use n CNNs, each configured to determine a corresponding body from the image. Examples of substep 120 are shown in the progression from FIG. 3 to FIG. 4A-4C.

[0051] Step 112 also includes reassembling the bodies to form an assembled image in sub-step 122. An example of an assembled image is shown in FIG.

[0052] Step 112 also includes identifying key points in the assembled image in sub-step 124. For example, with reference to Figure 6, sub-step 124 may include determining key points 90 between the weld material 16 and the second metal piece 14 at each of two points on the periphery of the weld 10.

[0053] Step 112 also includes measuring values ​​of properties of the weld using the key points in sub-step 126. For example, with reference to FIG. 6, sub-step 124 may include calculating the distance between the deepest point and first line 92, as indicated by second line 94. Sub-step 126 may further include converting the measured distance from pixels to a linear distance, such as millimeters. Additionally or alternatively, sub-step 126 may include determining an angle measurement, a two-dimensional area, a radius, a perimeter, or a number of features, such as inclusions or pores, within a given area. The number of features within a given area may indicate, for example, the porosity of the weld.

[0054] In some embodiments, the systems and methods of the present disclosure can also include identifying characteristics indicative of defects within a weld based on the body segment. For example, the system can be configured to identify the presence of porosity and / or one or more inclusions, such as slag or silicates, within the weld. Additionally or alternatively, the characteristics may be indicative of other types of defects, such as cracks, burn-through, distortion, undercuts, etc.

[0055] FIG. 9 shows a series of training images used to train an artificial intelligence (AI) model according to the present disclosure.

[0056] To train a segmentation model, a "ground truth" must be defined by annotating training images. This "ground truth" acts as a target for the AI ​​model. After each training epoch (loop), the error against the ground truth is measured and the neural net weights are adjusted accordingly. Test sets are used to measure the quality and accuracy of the model and algorithm. These test sets are used to tune and optimize software parameters.

[0057] FIG. 10 shows a cross-sectional image of a weld illustrating features of the semantic segmentation method according to the present disclosure.

[0058] FIG. 11 shows a cross-sectional image of a first weld between portions M1, M2 disposed generally parallel to one another, illustrating weld properties measured by the method and system of the present disclosure.

[0059] FIG. 12 shows a cross-sectional image of a second weld between portions M1, M2 disposed generally perpendicular to one another, illustrating weld properties measured by the method and system of the present disclosure.

[0060] Table 1 below lists properties of welds that can be measured by the methods and systems of the present disclosure.

[0061] [Table 1]

[0062] Figure 13A shows a cross-sectional image of a weld joining two pieces of metal, with lines indicating measurements identified by a manual inspection process, and Figure 13B shows a cross-sectional image of the weld of Figure 13A, with lines indicating measurements identified by an automated system and method according to the present disclosure.

[0063] FIG. 14 shows an architecture diagram of a system for automated cross-sectional analysis of joints, such as welds, in accordance with the present disclosure.

[0064] The present disclosure provides a method for analyzing a cross-section of a joint between two workpieces, the method including capturing an image of the cross-section of the joint using a camera, analyzing the image of the cross-section of the joint using machine learning techniques to classify two or more body segments that make up the joint, separating the body segments of the joint using machine learning techniques, reassembling the body segments to form an assembled image of the joint, identifying keypoints in the assembled image of the joint, and using the keypoints to measure values ​​of properties of the joint.

[0065] In some embodiments, the joint comprises a weld.

[0066] In some embodiments, the joint comprises one of laser welding, resistance welding, or gas metal arc welding (GMAW).

[0067] In some embodiments, the joint comprises a mechanical fastener.

[0068] In some embodiments, the mechanical fastener comprises a self-piercing rivet (SPR).

[0069] In some embodiments, the machine learning techniques include one or more neural networks.

[0070] In some embodiments, the one or more neural networks include an artificial neural network (ANN).

[0071] In some embodiments, the one or more neural networks include a convolutional neural network (CNN).

[0072] In some embodiments, the one or more neural networks include a convolutional neural network (CNN) associated with each body segment of the two or more body segments.

[0073] In some embodiments, analyzing the image of the cross section of the joint and classifying the two or more body segments that make up the joint includes locating and distinguishing a boundary between the two or more body segments of the joint.

[0074] In some embodiments, measuring the value of the property of the junction further comprises determining one of a linear distance, an angular measurement, a two-dimensional area, a radius, or a number of features within a given area.

[0075] In some embodiments, measuring the value of the property of the junction further includes determining a number of pixels between points associated with the property of the junction, and calculating a linear distance between the points associated with the property based on the number of pixels between the points associated with the property of the junction.

[0076] In some embodiments, the method further includes training the one or more neural networks to identify the two or more body segments of the junction and evaluating the performance of the one or more neural networks to identify the two or more body segments of the junction.

[0077] In some embodiments, the method further includes comparing the value of the property of the junction to one or more predetermined thresholds to determine whether the property of the junction is within acceptable limits.

[0078] In some embodiments, the method further includes storing the assembled image of the joint and the values ​​of the properties of the joint in a database.

[0079] In some embodiments, the property of the junction is one of a plurality of properties of the junction, and the method further comprises using the keypoint to measure a value for each of the plurality of properties of the junction.

[0080] In some embodiments, the method further includes generating a report regarding the joint by an automated process, the report including a value for each of a plurality of characteristics of the joint.

[0081] In some embodiments, generating the report includes generating a graphical representation associated with a plurality of characteristics of the joint.

[0082] The present disclosure also provides a method for analyzing a cross-section of a joint between two workpieces, the method including capturing an image of the cross-section of the joint using a camera, analyzing the image of the cross-section of the joint using machine learning techniques to classify two or more body segments that make up the joint, separating the body segments of the joint using machine learning techniques, reassembling the body segments to form an assembled image of the joint, and identifying characteristics indicative of defects in the joint based on the body segments.

[0083] The present disclosure also provides a system for automated cross-sectional analysis of a joint between two workpieces. The system includes a processor and a memory containing instructions that, when executed by the processor, cause the processor to analyze an image of the cross-section of the joint using machine learning techniques, classify two or more body segments that make up the joint, separate the body segments of the joint using machine learning techniques, reassemble the body segments to form an assembled image of the joint, identify keypoints in the assembled image of the joint, and use the keypoints to measure values ​​of properties of the joint.

[0084] The above-described systems, methods, and / or processes, and steps thereof, can be implemented in hardware, software, or any combination of hardware and software suitable for a particular application. The hardware can include general-purpose computers and / or dedicated computing devices, or specific computing devices, or particular aspects or components of specific computing devices. The processes can be implemented in one or more microprocessors, microcontrollers, embedded microcontrollers, programmable digital signal processors, or other programmable devices, along with internal and / or external memory. Additionally or alternatively, the processes can be embodied in application-specific integrated circuits, programmable gate arrays, programmable array logic, or any other device or combination of devices that can be configured to process electronic signals. It will further be understood that one or more of the processes can be implemented as computer-executable code that can be executed on a machine-readable medium.

[0085] Computer executable code can be written using a structured programming language such as C, an object-oriented programming language such as C++, or any other high-level or low-level programming language (including assembly language, hardware description languages, and database programming languages ​​and techniques), and can be stored, compiled, or interpreted for execution on one of the above devices, as well as on heterogeneous combinations of processors, processor architectures, or combinations of different hardware and software, or any other machines capable of executing program instructions.

[0086] Thus, in one aspect, each method and combination thereof described above can be embodied in computer-executable code that, when executed on one or more computing devices, performs its steps. In another aspect, the methods can be embodied in a system that performs its steps, can be distributed in multiple ways across multiple devices, or all of the functionality can be integrated into a dedicated standalone device or other hardware. In another aspect, the means for performing the steps associated with the processes described above can include any of the hardware and / or software described above. All such permutations and combinations are intended to be within the scope of the present disclosure.

[0087] The above description is not intended to be exhaustive or to limit the present disclosure. Individual elements or features of particular embodiments are generally not limited to particular embodiments, but may be interchangeable where appropriate and may be used in selected embodiments even if not specifically shown or described. They may also be modified in many ways. Such modifications are not considered a departure from the present disclosure, and all such variations are intended to be within the scope of the present disclosure.

Claims

1. 1. A method for analyzing a cross section of a joint between two workpieces, comprising: capturing an image of the cross section of the joint with a camera; using machine learning techniques to analyze the image of the cross section of the joint and classify two or more body segments that make up the joint; using the machine learning technique to separate the body segments at the joint; reassembling the body segments to form an assembled image of the joint; identifying key points in the assembled image of the joint; measuring values ​​of properties of the joint using the key points; A method comprising:

2. The method of claim 1 , wherein the joint comprises one of a weld or a mechanical fastener.

3. The method of claim 1 , wherein the joint comprises one of a laser weld, a resistance weld, or a gas metal arc weld (GMAW).

4. The method of claim 1 , wherein the machine learning techniques include one or more neural networks.

5. The method of claim 4 , wherein the one or more neural networks comprise an artificial neural network (ANN).

6. The method of claim 4 , wherein the one or more neural networks comprise a convolutional neural network (CNN).

7. 5. The method of claim 4, wherein the one or more neural networks include a convolutional neural network (CNN) associated with each body segment of the two or more body segments.

8. 5. The method of claim 4, further comprising training the one or more neural networks to identify the two or more body segments of the junction and evaluating the performance of the one or more neural networks to identify the two or more body segments of the junction.

9. 2. The method of claim 1, wherein analyzing the image of the cross-section of the joint and classifying two or more body segments that make up the joint includes locating and distinguishing a boundary between the two or more body segments of the joint.

10. The method of claim 1 , wherein measuring the value of the property of the junction further comprises determining one of a linear distance, an angular measurement, a two-dimensional area, a radius, or a number of features within a given area.

11. 2. The method of claim 1, wherein measuring a value of the property of the junction further comprises: determining a number of pixels between points associated with the property of the junction; and calculating a linear distance between the points associated with the property based on the number of pixels between the points associated with the property of the junction.

12. 2. The method of claim 1, wherein the characteristic of the junction is one of a plurality of characteristics of the junction, the method further comprising using the key points to measure a value for each of the plurality of characteristics of the junction.

13. The method of claim 12 , further comprising generating a report regarding the joint by an automated process, the report including a value for each of the plurality of characteristics of the joint.

14. 1. A method for analyzing a cross section of a joint between two workpieces, comprising: capturing an image of the cross section of the joint with a camera; using machine learning techniques to analyze the image of the cross section of the joint and classify two or more body segments that make up the joint; using the machine learning technique to separate the body segments at the joint; reassembling the body segments to form an assembled image of the joint; Identifying characteristics indicative of defects in the joint based on the body segments; A method comprising:

15. 1. A system for automated cross-sectional analysis of a joint between two workpieces, comprising: a processor; A memory containing instructions that, when executed by the processor, cause the processor to: using machine learning techniques to analyze an image of a cross section of the joint and classify two or more body segments that make up the joint; using the machine learning technique to separate the body segments at the joint; reassembling the body segments to form an assembled image of the joint; identifying key points in the assembled image of the joint; using the key points to measure values ​​of properties of the joint; Memory and A system comprising:

Images