A visual real-time monitoring system and method for welding process based on DMD technology

Abstract

The present invention discloses a real-time visual monitoring system and method for welding processes based on DMD technology, belonging to the field of welding monitoring. The system comprises: a visual acquisition module for real-time acquisition of images of the welding area; a DMD fill light control module for dynamically adjusting the light projected onto the welding area based on the acquired images; an image processing and analysis module for preprocessing the acquired images and extracting and analyzing features of the welding area; a welding process control and planning module for extracting the 3D contour of the welding area and planning welding process parameters based on the morphological changes of the spatially structured light projected by the DMD fill light control module; and a control and feedback module for real-time adjustment of welding parameters based on the analysis results of the image processing and analysis module and issuing an alarm when an anomaly is detected. The present invention uses DMD technology to precisely control fill light, effectively suppressing overexposure caused by strong welding light, enabling the camera to clearly capture detailed information about the welding area, and improving the accuracy of welding detection.

Description

Technical Field

[0001] The present invention relates to the field of welding monitoring, and in particular to a visual real-time monitoring system and method for a welding process based on DMD technology. Background Art

[0002] During the welding process, accurate monitoring of the welding status is crucial to ensuring welding quality. Traditional welding monitoring methods have many shortcomings, such as:

[0003] 1. The strong light generated during welding can easily cause the camera to overexpose the image, making it difficult to clearly capture the details of the welding area, and thus unable to accurately judge key information such as the quality of the weld and the state of the molten pool. Figure 1 The imaging results of a common active molten pool camera are shown. From the figure, it can be observed that there is a lot of overexposure in the smooth molten pool area at the front end of the welding wire. This is because the molten pool surface in this area is smooth, which easily forms light beam reflection. For the active lighting molten pool monitoring solution, this area will be overexposed.

[0004] 2. Complex changes in welding ambient light, such as differences in indoor and outdoor light, interference from other light sources in the workshop, etc., will affect the stability and accuracy of the monitoring system. Existing monitoring systems are difficult to adapt to these ambient light changes.

[0005] 3. For some minor welding defects, such as fine pores and cracks, ordinary visual monitoring methods have insufficient resolution and contrast, making it difficult to effectively identify them. Summary of the Invention

[0006] The purpose of the present invention is to overcome the problems of overexposure, poor adaptability to ambient light, and difficulty in identifying small defects in existing welding process monitoring, and to provide a welding process visual real-time monitoring system and method based on DMD technology to improve the accuracy and reliability of welding process monitoring.

[0007] The object of the present invention is achieved through the following technical solutions:

[0008] In a first aspect, a visual real-time monitoring system for a welding process based on DMD technology is provided, comprising:

[0009] Visual acquisition module: used to collect images of the welding area in real time;

[0010] DMD fill light control module: used to dynamically adjust the light projected onto the welding area according to the collected image;

[0011] Image processing and analysis module: used to pre-process the collected images and extract and analyze the features of the welding area;

[0012] Welding process control and planning module: extracts the 3D contour of the welding area and plans welding process parameters based on the changes in the spatial structured light morphology projected by the DMD fill light control module;

[0013] Control and feedback module: adjusts welding process parameters in real time based on the analysis results of the image processing and analysis module, and issues an alarm when an abnormality is detected.

[0014] In some embodiments, the visual acquisition module includes a CMOS camera.

[0015] In some embodiments, the DMD fill light control module includes a DMD module controller and an array light source, a DMD spatial light modulator, and an output light lens group connected in sequence; the DMD module controller is controlled and connected to the array light source and the DMD spatial light modulator respectively.

[0016] In a second aspect, a method for visual real-time monitoring of a welding process based on DMD technology is provided, which is used in the visual real-time monitoring system for a welding process based on DMD technology described in the first aspect, comprising the following steps:

[0017] S1. Initialize the visual acquisition module, image processing and analysis module, welding process control and planning module, and control and feedback module;

[0018] S2. Real-time acquisition of images of the welding area through the visual acquisition module;

[0019] S3. Based on the captured image, the DMD fill light control module dynamically adjusts the light projected onto the welding area; and triggers the visual acquisition module to perform image exposure, capturing an image of the welding area at this time;

[0020] S4. The image collected in step S3 is preprocessed by the image processing and analysis module, and the features of the welding area are extracted and analyzed;

[0021] S5. Based on the changes in the spatial structured light morphology projected by the DMD fill light control module, the welding process control and planning module extracts the 3D contour of the welding area and plans the welding process parameters;

[0022] S6. Based on the analysis results of step S4, the control and feedback module adjusts the welding process parameters in real time and issues an alarm when an abnormality is detected.

[0023] In some embodiments, the steps further include:

[0024] S7. After the welding process is completed, the images and analysis data collected during the monitoring process are saved.

[0025] In some embodiments, dynamically adjusting the light projected onto the welding area by the DMD fill light control module includes:

[0026] The image is scanned and the pixel positions of the overexposed areas in the image are recorded. The pixel positions of the overexposed areas are input into the DMD module controller. The DMD module controller performs scale scaling and spatial position offset based on the pixel positions, and then outputs the control signal to turn off the fill light for the corresponding DMD unit in the DMD spatial light modulator.

[0027] In some embodiments, the pre-processing of the image collected in step S3 by the image processing and analysis module includes:

[0028] The image is denoised and enhanced using image processing algorithms.

[0029] In some embodiments, extracting and analyzing the characteristics of the welding area includes:

[0030] The welding area is segmented and extracted based on the region segmentation algorithm of FCN fully convolutional neural network.

[0031] In some embodiments, extracting the 3D contour of the welding area by the welding process control and planning module includes:

[0032] The sinusoidal grating image projected by the DMD fill light control module is used to calculate the phase principal value using the four-step phase shift method;

[0033] The 3D contour of the welding area is reconstructed according to the phase main value.

[0034] In some embodiments, it further includes:

[0035] Monocular weld tracking is performed based on the reconstructed 3D contour of the weld area.

[0036] It should be further explained that the technical features corresponding to the above embodiments can be combined or replaced with each other to form a new technical solution if there is no conflict.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] 1. Effectively suppress overexposure: Through DMD technology, the fill light is precisely controlled, and the light intensity of the welding area can be adjusted in real time, effectively suppressing the overexposure caused by strong welding light, allowing the camera to clearly capture detailed information of the welding area and improve the accuracy of weld quality inspection.

[0039] 2. Adapt to changes in ambient light: The system can automatically adjust the DMD fill light parameters according to different welding ambient light conditions to ensure that the welding area always has appropriate lighting, improving the adaptability and stability of the monitoring system in complex environments.

[0040] 3. Improve defect recognition capabilities: The multi-segment coded modulated light formed by DMD technology enhances the detailed features of the welding area. Combined with image processing and analysis algorithms, it can more accurately identify tiny welding defects such as fine pores and cracks, which helps to improve welding quality.

[0041] 4. Achieve automatic navigation and tracking for 3D welding: Stripe structured light obtained through DMD spatial light modulation can extract the 3D dimensions of the entire welding area. The 3D topography obtained through structured light reconstruction can clearly determine the direction of the weld bead at the front end of the molten pool. It provides navigation function for welding curved welds, allowing the welding robot to perform monocular weld tracking. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a schematic diagram of the imaging results of an existing common active melt pool camera;

[0043] Figure 2 Schematic diagram of a visual real-time monitoring system for welding process based on DMD technology of the present invention;

[0044] Figure 3 This is a working diagram of the monitoring system of the present invention;

[0045] Figure 4 Schematic diagram of the structure of the DMD fill light control module of the present invention;

[0046] Figure 5 This is a flow chart of the monitoring method of the present invention;

[0047] Figure 6 This is a schematic diagram of overexposure control of the present invention;

[0048] Figure 7 Schematic diagram of the region segmentation algorithm structure of the present invention;

[0049] Figure 8 This is a schematic diagram of the region segmentation result of the present invention;

[0050] Figure 9 This is a flow chart of structured light multi-frame data acquisition in the present invention;

[0051] Figure 10 This is a schematic diagram of the 3D morphology reconstruction result of the present invention;

[0052] Figure 11 Schematic diagram of weld navigation according to the present invention. DETAILED DESCRIPTION

[0053] The technical solutions of the present invention are described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present application generally described and shown in the drawings herein can be arranged and designed in various different configurations. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0054] It should be noted that the defects existing in the solutions in the above-mentioned prior art are the results obtained by the inventor after practice and careful research. Therefore, the discovery process of the above-mentioned problems and the solutions proposed in the embodiments of this application below for the above-mentioned problems should be the contributions made by the inventor to this application in the process of invention and creation, and should not be understood as technical contents known to technical personnel in this field.

[0055] In response to the technical problems pointed out in the background technology, the embodiments provided by the present invention are as follows:

[0056] Reference Figure 2 In an exemplary embodiment, a visual real-time monitoring system for a welding process based on DMD technology includes:

[0057] Visual acquisition module: used to collect images of the welding area in real time;

[0058] DMD fill light control module: used to dynamically adjust the light projected onto the welding area according to the collected image;

[0059] Image processing and analysis module: used to pre-process the collected images and extract and analyze the features of the welding area;

[0060] Welding process control and planning module: extracts the 3D contour of the welding area and plans welding process parameters based on the changes in the spatial structured light morphology projected by the DMD fill light control module;

[0061] Control and feedback module: adjusts welding process parameters in real time based on the analysis results of the image processing and analysis module, and issues an alarm when an abnormality is detected.

[0062] Specifically, if Figure 3-Figure 4As shown, the visual acquisition module includes a CMOS camera for real-time capture of the weld pool image. Camera parameters, such as resolution and frame rate, are selected based on the requirements of the welding scenario to ensure clear capture of dynamic changes in the weld area. The DMD fill light control module includes a DMD module controller (not shown), along with a sequentially connected array light source, a DMD spatial light modulator, and an output light lens assembly. The DMD module controller is controlled and connected to the array light source and the DMD spatial light modulator, respectively. The core component of the DMD fill light control module is the digital micromirror device (DMD). The DMD consists of a large number of tiny mirrors. By controlling the angles of these mirrors, the intensity, angle, and distribution of light projected onto the weld area can be precisely adjusted. The DMD fill light control module works in conjunction with the visual acquisition module, adjusting the fill light parameters in real time based on the captured image information.

[0063] The image processing and analysis module processes and analyzes the images acquired by the visual acquisition module. First, image processing algorithms perform preprocessing operations such as denoising and enhancement to improve image quality. Then, image recognition technology is used to extract and analyze features of the weld area, such as weld shape and position, weld pool size and shape, and wire position and length.

[0064] The welding process control and planning module projects the spatial structured light formed by the DMD fill light control module onto the welding area. The target contours of the weld seam, molten pool, etc. in the welding area are extracted through the morphological changes of the projected light, and the welding parameters are planned and adjusted according to the size of the area to be filled.

[0065] The control and feedback module provides control and feedback on the welding process based on the results of the image processing and analysis module. If any abnormalities are detected during the welding process, such as weld deviations or defects, the control and feedback module sends a signal to the welding equipment, adjusts the welding parameters, and issues an alert to the operator.

[0066] Furthermore, the system is constructed as follows:

[0067] 1. Choose a suitable camera, such as an industrial camera with high resolution and high frame rate, to meet the needs of monitoring the rapid dynamic changes in the welding process. Install the camera in a position where it can clearly capture the welding area and ensure that its field of view covers the entire welding area. The camera FOV parameters are calculated as follows:

[0068]

[0069] Where HFOV is the horizontal field of view of the camera, WFOV is the vertical field of view of the camera, DFOV is the diagonal field of view of the camera, W is the physical width of the sensor CMOS, H is the physical length of the sensor CMOS, and f is the focal length of the lens.

[0070] 2. Select a high-performance DMD fill light control module and determine the DMD model and parameters based on the size of the welding scene and the lighting requirements. Synchronize the DMD fill light control module with the camera to ensure that the fill light and image acquisition time match.

[0071] 3. Build an image processing and analysis module. You can use a computer or a dedicated image processing chip, install the corresponding image processing and analysis software, and realize real-time processing and analysis of the collected images.

[0072] 4. Establish a control and feedback module, connect the output signal of the image processing and analysis module to the control interface of the welding equipment, and realize real-time adjustment and feedback of welding process parameters. Specifically, the control module is connected to the actuator of the welding equipment through the relevant motor control communication protocol and is responsible for the specific execution logic. The feedback module realizes the feedback closed loop of the actuator position, torque, etc. through the real-time encoder data and the current of the actuator motor, thus achieving more precise welding control. At the same time, the input information of the control module is provided by the image processing and analysis module, including the offset information of the welding wire and the weld bead, the position information between the welding wire and the molten pool, etc., which realizes the planning and execution functions of the control feedback module.

[0073] Furthermore, the system is debugged and optimized, including:

[0074] 1. After the system is set up, perform initialization and debugging. Set the camera's exposure time, gain, and other parameters to capture clear images under various lighting conditions. Also, initialize the DMD fill light control module and determine the initial fill light mode and parameters.

[0075] 2. Conduct welding experiments, capturing images of the welding process under different welding process conditions and observing the effectiveness of the DMD fill light control module. Based on the captured image quality and analysis results, optimize the DMD fill light parameters, such as adjusting the reflector angle and changing the light intensity and distribution, to achieve the best fill light effect.

[0076] 3. Optimize the image processing and analysis algorithm. By comparing the accuracy and efficiency of different algorithms in welding image analysis, select the most appropriate algorithm and adjust its parameters to improve the recognition accuracy of welding area features.

[0077] 4. Test the response speed and accuracy of the control and feedback system to ensure that when welding anomalies are detected, welding parameters can be adjusted promptly and accurately, and an alarm can be issued to the operator.

[0078] In another exemplary embodiment, a method for visually monitoring a welding process in real time based on DMD technology is provided, which is used in the visually monitoring system for a welding process in real time based on DMD technology, referring to Figure 5 , including the following steps:

[0079] S1. Initialize the visual acquisition module, image processing and analysis module, welding process control and planning module, and control and feedback module; set camera parameters such as exposure time and gain; initialize the DMD fill light mode and parameters; and load the image processing and analysis algorithm.

[0080] S2. The visual acquisition module acquires images of the welding area in real time at a set frame rate;

[0081] S3. Based on the captured image, the DMD fill light control module dynamically adjusts the light projected onto the welding area; and triggers the visual acquisition module to perform image exposure, capturing an image of the welding area at this time;

[0082] S4. Preprocess the image collected in step S3 through the image processing and analysis module, and extract and analyze the features of the welding area through algorithms such as edge detection and feature extraction, such as calculating the center line of the weld, the area and shape of the molten pool, etc.

[0083] S5. Based on the changes in the spatial structured light morphology projected by the DMD fill light control module, the welding process control and planning module extracts the 3D contour of the welding area and plans the welding process parameters. Specifically, through the geometric position relationship between the DMD projector and the camera, the weld width, the volume of the area to be filled, and the internal morphology of the weld bead at the current welding position are calculated, thereby outputting control parameters such as the welding gun height, welding speed, and wire feeding speed corresponding to the current welding position.

[0084] S6. Based on the analysis results from step S4, the control and feedback module adjusts welding process parameters in real time and issues an alarm if an anomaly is detected. Specifically, the image processing and analysis module determines whether the welding process is normal. If a weld deviation exceeds the allowable range or a defect is detected, the control and feedback module sends a control signal to the welding equipment to adjust parameters such as welding current, voltage, and welding speed, and simultaneously issues an alarm to the operator to notify them of any problems during the welding process.

[0085] Furthermore, the method further comprises the steps of:

[0086] S7. After the welding process is completed, the system stops working and saves the images and analysis data collected during the monitoring process to facilitate subsequent evaluation and tracing of welding quality.

[0087] During fill light, the DMD fill light control module adjusts the state of the projection reflective surface. When the adjustment is completed, it generates a DMDReadySignal signal. After receiving this signal, the DMD module controller starts to trigger the fill light source to generate a pulse beam. After a certain delay, it triggers the visual acquisition module to expose the image, thereby obtaining a real-time monitoring image of the welding under the corresponding modulated light.

[0088] When the DMD adjusts the overexposed area of ​​the image, the DMD fill light control module dynamically adjusts the light projected onto the welding area, including:

[0089] By capturing the melt pool image, scanning the image and recording the pixel position of the overexposed area in the image, the pixel position of the overexposed area is input into the DMD module controller. The DMD module controller performs scale scaling and spatial position offset based on the pixel position, and then outputs the control signal to turn off the fill light for the corresponding DMD unit in the DMD spatial light modulator. Figure 6 As shown in the figure, the DMD units in the DMD spatial light modulator corresponding to the pixel positions in the overexposed area of ​​the image, when a certain position is overexposed, the DMD unit on the right turns off the fill light, thereby avoiding the overexposure of the pixels in the image space.

[0090] Furthermore, the image processing and analysis module pre-processes the image collected in step S3, including:

[0091] The image is denoised and enhanced through image processing algorithms (grayscale transformation, filtering, etc.) to improve the image clarity and contrast.

[0092] Furthermore, if Figure 7 As shown in the figure, the regional segmentation algorithm based on the FCN fully convolutional neural network is used to segment and extract the welding area, which can effectively obtain the data of each part. The fully convolutional neural network can extract small targets, and the features of the small objects are derived from the features provided by the original input image. For areas of interest such as larger grooves, the algorithm restores the segmented image through downsampling pooling and upsampling, thereby realizing regional segmentation. Figure 8 This is the result of the FCN network segmenting the molten pool image, where red represents the welding wire, blue represents the groove, green represents the weld bead, yellow represents the molten pool, and cyan represents the nozzle.

[0093] Furthermore, the extraction of the 3D contour of the welding area by the welding process control and planning module includes:

[0094] The sinusoidal grating image projected by the DMD fill light control module is used to calculate the phase principal value using the four-step phase shift method;

[0095] The 3D contour of the welding area is reconstructed according to the phase main value.

[0096] Specifically, the structured light multi-frame data acquisition process is as follows: Figure 9 As shown, after a certain delay, the visual acquisition module is triggered to perform image exposure. The delay time threshold is set to 5ms. As long as the delay time is within 5ms, a sinusoidal grating image is projected through the DMD device every 1ms. In this embodiment, four sinusoidal grating images are obtained (i.e., i is an integer from 1 to 4), and their light intensity distribution can be expressed as:

[0097] ,in is the light intensity of the image at (x,y), is the background light intensity, is the intensity of the stripe light modulation, is the phase information. Through the four-step phase shift method, the phase shift value is The intervals between them change four times, that is, the projector projects four images with a phase difference of Phase image of Coded pictures ( ), its encoding formula is:

[0098]

[0099] Through these four images, the main phase value can be solved using the formula:

[0100]

[0101] The phase main value is unique to each pixel, thereby marking the relative position on the projected image.

[0102] Furthermore, based on the relevant areas of image segmentation, including information such as the parent material surface, welding wire position, molten pool position, and weld bead position, the weld bead line can be obtained based on the intersection position of the parent material surface and the weld bead. By obtaining the weld bead lines on both sides of the groove, the center of the welding process can be tracked. Specifically, the internal parameters of the welding pool camera can be calibrated first to determine its principal point position, imaging system focal length and other parameters. The imaging system formula is as follows:

[0103]

[0104] After obtaining the camera calibration results, the scaling factor between physical space and pixel space is determined. This is then combined with the pixel spacing between the wire position and the two groove lines to convert the resulting physical position into a physical position. This allows for control of wire swing and weld edge, wire length, and weld pool shape. When the wire position is closer to the left groove line, the welding gun can be controlled to the right, and vice versa. By identifying the welding wire, the actual wire length can be calculated using an appropriate scaling factor, enabling dry-out tracking data during the welding process. The geometry and size of the weld pool can also be calculated to control the welding process. When wire penetration is detected in the weld pool image area during the reverse welding process, the unified wire feed speed is controlled to reduce the heat input and molten iron filling volume during the welding process, allowing time for the weld pool to solidify and ensuring that the wire can re-strike on the solidified metal substrate. Furthermore, through visual monitoring of the welding process, if the weld hole is small, the swing width can be appropriately increased to reduce the risk of wire penetration and ensure that the welding result meets the required quality. The control of arc stability during the welding process can be achieved by effectively controlling the length of the welding wire for automatic control. When the welding material bulges, the welding wire will be melted shorter. After visual detection of the welding wire shortening, the wire extension can be adjusted to achieve a stable length of the welding wire, thereby maintaining the arc state.

[0105] Furthermore, the method further comprises:

[0106] Based on the reconstructed 3D contour of the welding area, monocular weld tracking is performed. Specifically, the welding robot can perform monocular weld tracking, thereby providing navigation for welding curved welds. The 3D topography obtained by structured light reconstruction can clearly obtain the direction of the weld at the front end of the molten pool, as shown in Figure 10. By calculating the angle between the contour line of the feature point on the weld and the movement of the welding device, the current heading angle deviation of the welding mechanism is obtained. At the same time, the center position offset of the welding mechanism and the weld can be determined by the center position of the current feature point. ,pass , The welding mechanism tracking parameters can realize the navigation function. Figure 11 Based on the results of 3D reconstruction, the contour information of the current weld can be known. The volume of the area to be filled can be calculated by integration, and the welding power supply wire feeding speed, welding current and voltage, and welding robot walking speed can be set to ensure the appropriate filling volume:

[0107]

[0108] in, is the wire radius, is the wire feeding speed, For cladding efficiency, is the cross-sectional area of ​​the groove, is the wire speed, is the metal filling volume per unit time.

[0109] Taking car body welding in automobile manufacturing as an example, the visual real-time monitoring system of the present invention is installed in the car body welding production line. During the welding process, the system captures real-time images of the welding area. The DMD fill-light control module effectively suppresses overexposure caused by the intense welding light, enabling the camera to clearly capture the shape of the weld and the state of the weld pool. The image processing and analysis module analyzes the captured images, accurately identifying subtle defects in the weld and providing timely feedback to the control and feedback system. The control and feedback system adjusts welding parameters and optimizes the welding process, improving the quality and production efficiency of the car body welding.

[0110] The above specific implementation methods are detailed descriptions of the present invention. It cannot be considered that the specific implementation methods of the present invention are limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, they can make several simple deductions and substitutions without departing from the concept of the present invention, which should be regarded as falling within the scope of protection of the present invention.

Claims

1. A visual real-time monitoring system for welding process based on DMD technology, characterized in that: include: Visual acquisition module: used to collect images of the welding area in real time; DMD fill light control module: used to dynamically adjust the light projected onto the welding area according to the collected image; the DMD fill light control module includes a DMD module controller and an array light source, a DMD spatial light modulator, and an output light lens group connected in sequence; the DMD module controller is controlled and connected to the array light source and the DMD spatial light modulator respectively; Image processing and analysis module: used to pre-process the collected images and extract and analyze the features of the welding area; Welding process control and planning module: extracts the 3D contour of the welding area and plans welding process parameters based on the changes in the spatial structured light morphology projected by the DMD fill light control module; Control and feedback module: adjusts welding process parameters in real time based on the analysis results of the image processing and analysis module, and issues an alarm when an abnormality is detected.

2. The DMD-based real-time visual monitoring system for welding process according to claim 1, characterized in that: The visual acquisition module includes a CMOS camera.

3. A method for visual real-time monitoring of a welding process based on DMD technology, used in a visual real-time monitoring system for a welding process based on DMD technology as claimed in any one of claims 1 to 2, characterized in that: The following steps are involved: S1. Initialize the visual acquisition module, image processing and analysis module, welding process control and planning module, and control and feedback module; S2. Real-time acquisition of images of the welding area through the visual acquisition module; S3. Based on the captured image, the DMD fill light control module dynamically adjusts the light projected onto the welding area; and triggers the visual acquisition module to perform image exposure, capturing an image of the welding area at this time; S4. The image collected in step S3 is preprocessed by the image processing and analysis module, and the features of the welding area are extracted and analyzed; S5. Based on the changes in the spatial structured light morphology projected by the DMD fill light control module, the welding process control and planning module extracts the 3D contour of the welding area and plans the welding process parameters; S6. Based on the analysis results of step S4, the control and feedback module adjusts the welding process parameters in real time and issues an alarm when an abnormality is detected.

4. The method for visual real-time monitoring of welding process based on DMD technology according to claim 3, characterized in that: Also includes the steps: S7. After the welding process is completed, the images and analysis data collected during the monitoring process are saved.

5. The method for visual real-time monitoring of welding process based on DMD technology according to claim 3, characterized in that: The method of dynamically adjusting the light projected onto the welding area by using the DMD fill light control module includes: The image is scanned and the pixel positions of the overexposed areas in the image are recorded. The pixel positions of the overexposed areas are input into the DMD module controller. The DMD module controller performs scale scaling and spatial position offset based on the pixel positions, and then outputs the control signal to turn off the fill light for the corresponding DMD unit in the DMD spatial light modulator.

6. The method for visual real-time monitoring of welding process based on DMD technology according to claim 3, characterized in that: The image processing and analysis module pre-processes the image collected in step S3, including: The image is denoised and enhanced using image processing algorithms.

7. The method for visual real-time monitoring of welding process based on DMD technology according to claim 3, characterized in that: The extraction and analysis of the characteristics of the welding area includes: The welding area is segmented and extracted based on the region segmentation algorithm of FCN fully convolutional neural network.

8. The method for visual real-time monitoring of welding process based on DMD technology according to claim 3, characterized in that: The extraction of the 3D contour of the welding area by the welding process control and planning module includes: The sinusoidal grating image projected by the DMD fill light control module is used to calculate the phase principal value using the four-step phase shift method; The 3D contour of the welding area is reconstructed according to the phase main value.

9. The method for visual real-time monitoring of welding process based on DMD technology according to claim 8, characterized in that: Also includes: Monocular weld tracking is performed based on the reconstructed 3D contour of the weld area.

Images