Laser processing molten pool surface image acquisition method and device

Abstract

The invention relates to the technical field of laser processing, and particularly provides a laser processing molten pool surface image acquisition method which comprises the following steps: acquiring a laser energy distribution type; obtaining the area of the oxide layer region according to the laser energy distribution type; according to the area of the oxide layer region, calculating an average gray value of the region in a single-frame image; and according to the single-frame average gray value, carrying out statistical averaging on the single-frame average gray value of the multi-frame image to obtain a multi-frame average gray value of the molten pool oxide layer. According to the method, laser energy distribution types are distinguished firstly, then the area of the oxide layer region is obtained in a targeted manner, and the gray value is calculated; according to the method, accurate acquisition and analysis of surface images of molten pools with different laser distribution types are realized, uniform energy characteristics of flat-topped distribution laser are adapted, gradient energy characteristics of Gaussian distribution laser are also considered, and finally, through single-frame and multi-frame gray value analysis, the accuracy of molten pool analysis in the laser processing process is improved.

Description

Technical Field

[0001] This invention relates to the field of laser processing technology, and specifically provides a method and apparatus for acquiring images of the surface of a laser-processed molten pool. Background Technology

[0002] Aluminum alloys play an irreplaceable role in laser processing in the automotive, aerospace, electronics and precision manufacturing industries. Their high strength-to-weight ratio and excellent thermal conductivity make aluminum alloy parts widely used in modern manufacturing.

[0003] However, aluminum and its alloys have high laser reflectivity, making absorptivity a key process parameter affecting energy coupling, molten pool formation, and the final microstructure / defect evolution. Recent studies on the absorption behavior of aluminum alloys have shown that, in thermal conductivity mode, absorptivity is the result of the interaction between the oxide layer floating or forming on the molten pool surface and the liquid metal directly irradiated by the laser. This composite surface significantly alters the laser reflection-absorption behavior, thus affecting absorptivity and molten pool stability. In thermally conductive molten pools, the presence of the oxide layer not only optically enhances local absorption but also further influences the distribution of the liquid metal and the molten pool morphology by altering surface tension and Marangoni flow.

[0004] Accordingly, there is a need in the field for a new method and apparatus for acquiring images of the surface of a laser-processed molten pool to solve the above-mentioned technical problems. Summary of the Invention

[0005] The present invention aims to solve the above-mentioned technical problems, namely, to solve the problem of poor stability of existing data acquisition methods.

[0006] In a first aspect, the present invention provides a method for acquiring images of the surface of a laser-processed molten pool, the method comprising the following steps: Obtain the laser energy distribution type; The area of ​​the oxide layer region is obtained based on the laser energy distribution type. Calculate the average gray value of the region in a single frame image based on the area of ​​the oxide layer region; Based on the single-frame average gray value, the single-frame average gray values ​​of multiple frames of images are statistically averaged to obtain the multi-frame average gray value of the molten pool oxide layer.

[0007] Based on the above settings, this method first distinguishes the laser energy distribution type, then specifically acquires the area of ​​the oxide layer region and calculates the grayscale value, achieving accurate acquisition and analysis of molten pool surface images for different laser distribution types. The overall process is logically coherent, adapting to both the uniform energy characteristics of flat-top laser distribution and the gradient energy characteristics of Gaussian laser distribution. Finally, through single-frame and multi-frame grayscale value analysis, the accuracy of molten pool analysis during laser processing is improved.

[0008] In the preferred embodiment of the above-mentioned laser processing molten pool surface image acquisition method, the step of obtaining the area of ​​the oxide layer region based on the laser energy distribution type includes: If the laser energy distribution is a flat-top laser, acquire a coaxial camera image of the flat-top laser and determine the laser irradiation area through image processing; Based on the laser irradiation area, the red light component of the original image is extracted, and after processing, the visible molten pool area within the laser irradiation range of the flat-top distribution is obtained; Perform feature analysis on the visible molten pool region and calculate the area of ​​the liquid metal region within the region; The area of ​​the oxide layer region within the laser irradiation range is calculated by combining the total area of ​​the actual molten pool measured by a metal microscope with the area of ​​the liquid metal region.

[0009] Based on the above settings, and considering the characteristics of flat-top laser distribution, a refined division of the molten pool region under flat-top laser processing is achieved by stepwise determining the laser irradiation area, extracting the visible molten pool region, calculating the liquid metal area, and deriving the oxide layer area by combining the actual total molten pool area. This process fully utilizes the uniform energy of flat-top lasers, reducing region identification errors caused by uneven energy distribution, improving the relevance and accuracy of oxide layer area calculation, and providing reliable data for molten pool condition assessment in flat-top laser processing scenarios.

[0010] In the preferred embodiment of the above-mentioned laser processing molten pool surface image acquisition method, the step of obtaining the area of ​​the oxide layer region based on the laser energy distribution type includes: If the laser energy distribution is a Gaussian distribution laser, set specific process parameters to acquire images of the stable stage and locate the coordinates of the laser spot center; Based on the center coordinates, the boundary of the laser irradiation area is determined, and the acquired image is cropped and denoised. Based on the preprocessed acquired image, a corresponding method is selected according to the laser power and powder supply status, and the image is binarized. The longest perimeter contour is extracted based on the binarized image, and contour artifacts are identified and processed through positional relationship and roundness calculation. Based on the corrected contour features, the area of ​​the liquid metal region and the area of ​​the oxide layer region are calculated.

[0011] Based on the above settings, and considering the characteristics of Gaussian laser energy distribution—high energy center and low energy edge—the problem of molten pool images being easily affected by energy gradient interference under Gaussian lasers is effectively solved through steps such as spot center localization, image preprocessing, binarization, and contour artifact removal. In particular, contour artifact recognition and correction avoids misjudgment of regions caused by uneven energy distribution, ensuring the accuracy of liquid metal and oxide layer area calculations, and adapting to the complex image features of Gaussian laser processing scenarios.

[0012] In the preferred embodiment of the above-mentioned laser processing molten pool surface image acquisition method, if the laser energy distribution is a flat-top laser, acquiring a coaxial camera image of the flat-top laser and determining the laser irradiation area through image processing includes: The acquired flat-top distributed laser coaxial camera images are converted into grayscale images; The grayscale image is denoised using a 5×5 pixel Gaussian filter kernel; The denoised grayscale image is processed using a binarization algorithm to obtain the laser irradiation area.

[0013] Based on the above settings, the irradiation area of ​​the flat-top distributed laser is determined through grayscale conversion, 5×5 pixel Gaussian filtering for noise reduction, and binarization algorithms. This effectively filters out image noise and accurately extracts the physical boundary of the actual laser irradiation. This step defines a clear spatial reference for subsequent molten pool region analysis, reduces interference from background areas outside the laser irradiation range, and improves the basic accuracy of molten pool region identification under flat-top laser.

[0014] In the preferred embodiment of the above-mentioned laser processing molten pool surface image acquisition method, the step of extracting the red light component of the original image based on the laser irradiation area, and processing it to obtain the visible molten pool area within the flat-top distributed laser irradiation range; performing feature analysis on the visible molten pool area and calculating the area of ​​the liquid metal region within the area includes: Extract the red light component from the original image; The red light component image is denoised using a 5×5 pixel Gaussian filter. Calculate the Otsu binarization threshold. If the Otsu binarization threshold exceeds the range, a constant threshold binarization is used; otherwise, the Otsu binarization threshold is used. The minimum threshold is 21 and the maximum threshold is 255 under the powder-free processing state. The minimum threshold is gmin=36 and the maximum threshold is 65. The visible molten pool region is obtained through binarization. The intersection of the visible molten pool region and the laser irradiation region is calculated to obtain the visible molten pool region within the laser irradiation range; The visible molten pool region is subjected to adaptive threshold binarization with a size of 15×15 pixels to obtain an adaptive threshold binarized region. Binarize the region with a gray value of 255 in the visible molten pool region to obtain the binarized region with the highest gray value; The visible oxide layer region of the laser irradiation area is calculated as: the adaptive threshold binarization processing + the binarized region with the highest gray value. The liquid metal area is calculated as follows: the visible molten pool area within the laser irradiation range minus the visible oxide layer area within the laser irradiation area. If the laser power is less than 600W, the liquid metal area is set to 0. Otherwise, the liquid metal area is equal to the liquid metal area directly exposed to the irradiation area, and the negative value generated after the subtraction is set to 0.

[0015] Based on the above settings, ambient light interference is reduced by extracting the red light component. Adaptive and constant threshold binarization are used to process the visible molten pool region, and dual binarization is employed to distinguish between the oxide layer and liquid metal, achieving precise segmentation of the visible molten pool region under a flat-top laser. Simultaneously, the liquid metal area is corrected based on the laser power, further ensuring the data's validity in low-power scenarios and improving the reliability of liquid metal and oxide layer area calculations.

[0016] In the preferred embodiment of the above-mentioned laser processing molten pool surface image acquisition method, if the laser energy distribution is a Gaussian distribution laser, setting specific process parameters to acquire images during the stable phase and locating the laser spot center coordinates includes: The laser process parameters were set to 600W power, 0m / s laser speed, and 10s processing time. Coaxial camera video data was collected during the processing. The last 7 seconds of the video data were selected as the image of the stabilization stage. The Otsu binarization method was used to obtain the annular contour of the oxide layer region in the frame image. The gray value of the oxide layer region was set to 255, and the gray value of the liquid metal region inside Avox was manually set to 255. Calculate the average center coordinates of a single frame Calculate the average center coordinates of the video sequence. : ; in: Here, n represents the average center coordinates of a single frame, and n255 represents the number of pixels with a grayscale value of 255. (Ic, Jc) represents the average center coordinates of the video sequence, Nstart refers to the ID number of the first image in the coaxial camera data, and Nend refers to the ID number of the last image in the coaxial camera data.

[0017] Based on the above settings, images of the stable phase were acquired by setting specific process parameters, and the center of the Gaussian laser spot was located based on the average coordinates of the high grayscale region, ensuring the stability and accuracy of the spot center coordinates. This center serves as the spatial reference for subsequent image analysis, effectively reducing regional analysis errors caused by laser drift or jitter, and providing a reliable reference for the spatial positioning of the molten pool image under Gaussian laser.

[0018] In the preferred embodiment of the above-mentioned laser processing molten pool surface image acquisition method, the step of determining the laser irradiation area boundary based on the center coordinates and performing cropping and noise reduction preprocessing on the acquired image; based on the preprocessed acquired image, selecting a corresponding method according to the laser power and powder supply status, and performing binarization processing, including: The laser irradiation area is set as the center of the light spot, and a circular area with the center coordinates (Ic, Jc) of the light spot and a radius of rl is defined, where rl is the distance from the center of the light spot to the edge of the laser. The acquired single-frame image is cropped into a 200×200 pixel image. The cropped image is then denoised using a 9×9 pixel Gaussian filter kernel. In the case of powderless laser processing, Otsu binarization is used when the laser power is in the range of 300W-600W, and fixed threshold binarization is used when the laser power is <300W to generate a binarized image. For other process parameters, adaptive threshold binarization with a size of 15×15 pixels is used to generate a binarized image. The binarized image is subjected to two opening operations using a kernel function with a size of 5×5 pixels to eliminate white spot noise in the image.

[0019] Based on the above settings, the boundary of the laser irradiation area is determined based on the center of the light spot. Image preprocessing is achieved by combining 200×200 pixel cropping and 9×9 pixel Gaussian filtering, which can focus the core area and reduce noise interference. For different power and powder supply conditions, an appropriate binarization method is selected, and white point noise is eliminated by two opening operations, which significantly improves the contrast and clarity of the Gaussian laser image, laying a high-quality image foundation for subsequent contour extraction and artifact recognition.

[0020] In the preferred embodiment of the above-mentioned laser processing molten pool surface image acquisition method, the step of extracting the longest perimeter contour based on the binarized image and identifying and processing contour artifacts through positional relationship and roundness calculation includes: Extract the longest perimeter contour from the binarized image; determine whether the center of the laser spot is within the area enclosed by the longest perimeter contour; If not present, it is determined to be either a no-fake or a Type I fake, and the fake ID is set to 0; If it is, then calculate the contour roundness φ according to the formula φ=4πS / C², where S is the area of ​​the region enclosed by the contour and C is the perimeter of the contour; If φ < 0.65, it is determined to be a Type II or Type V artifact; if φ ≥ 0.65, it is determined to be a Type III artifact when the laser power is ≤ 1200W, and a Type IV artifact when the laser power is > 1800W.

[0021] Based on the above settings, by extracting the longest perimeter contour and combining the positional relationship between the laser spot center and the contour with roundness calculations to identify Type IV contour artifacts, it is possible to accurately distinguish and mark artifacts such as oxide layer adhesion and spatter interference caused by Gaussian laser energy distribution. This processing effectively avoids the influence of artifacts on the area calculation, ensuring the authenticity and accuracy of subsequent liquid metal and oxide layer area calculations.

[0022] In the preferred embodiment of the above-mentioned laser processing molten pool surface image acquisition method, if the artifact ID=0, the liquid metal radius = the shortest distance from the laser spot center (Ic,Jc) to the longest perimeter contour; if the artifact ID=3, the liquid metal radius = the shortest distance from the laser spot center (Ic,Jc) to the second longest perimeter contour; calculate the molten pool half-width: ; Where rom is the half-width of the molten pool in the multi-channel metallographic results; Liquid metal area = Liquid metal area within the laser irradiation zone; The area of ​​the oxide layer region = Amp - Alm, where Amp is the visible area of ​​the molten pool; Calculate the intersection of Aox_l=Aox and the laser-irradiated area to obtain the area of ​​the oxide layer region within the laser irradiation range.

[0023] Based on the above settings, the radius of the liquid metal is dynamically determined according to different contour artifact types, and the area of ​​the liquid metal and oxide layer is calculated in conjunction with the half-width of the molten pool, thus achieving adaptive processing of complex molten pool contours under Gaussian lasers. This method can flexibly handle contour features under different processing conditions, ensuring the consistency between the area calculation results and the actual molten pool state, and improving the reliability of the data.

[0024] In a second aspect, the present invention also provides a laser processing molten pool surface image acquisition device for the laser processing molten pool surface image acquisition method described in any of the above claims. The acquisition device includes a laser processing device and an image acquisition and processing device. The laser processing device includes a continuous fiber laser, a laser focusing device, a powder feeding device, and a stepper motor. The continuous fiber laser is used to emit a flat-top distributed laser or a Gaussian distributed laser. The laser focusing device is used to focus the laser. The powder feeding device is used to direct the laser towards the substrate through a coaxial powder feeding head. The powder feeding device is also used to deliver metal powder to the substrate. The stepper motor is used to drive the continuous fiber laser and the laser focusing device to move synchronously to achieve scanning processing. The image acquisition and processing device includes a coaxial camera and a computer. The coaxial camera is used to acquire dynamic images of the molten pool within the laser irradiation range. The computer is electrically connected to the coaxial camera via a network cable and is used to receive image data and execute acquisition methods and calculation steps. Based on the above setup, the device, through the coordinated operation of laser processing equipment and image acquisition and processing equipment, can both emit flat-top or Gaussian distributed lasers for scanning processing and acquire molten pool images in real time and perform analysis methods. The components of the device are highly adaptable, and the computer can efficiently process image data and set camera parameters, ensuring the synchronization and accuracy of laser processing and image acquisition and analysis. Attached Figure Description

[0025] The preferred embodiments of the present invention are described below with reference to the accompanying drawings, in which: Figure 1 A flowchart of the data acquisition method of the present invention is shown; Figure 2 A flowchart of the method for calculating the oxide layer area of ​​a flat-top distributed laser according to the present invention is shown; Figure 3 A flowchart of the method for calculating the area of ​​the oxide layer region of Gaussian distributed laser light according to the present invention is shown; Figure 4 A schematic diagram of the data acquisition device of the present invention is shown; Figure 5 The processing of the Gaussian distributed laser of the present invention is shown, and the maximum perimeter contour artifact type after binarization processing of the molten pool image is illustrated. Figure 6 A flowchart illustrating the steps of an embodiment of the present invention is shown.

[0026] Figure label: 1. Computer; 2. Laser processing equipment; 3. Coaxial camera; 4. Powder cylinder; 5. Substrate. Detailed Implementation

[0027] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the invention and are not intended to limit the scope of protection of the invention. Those skilled in the art can make adjustments as needed to adapt to specific applications.

[0028] It should be noted that in the description of this invention, the terms "center," "upper," "lower," "left," "right," "inner," and "outer," which indicate directional or positional relationships, are based on the directional or positional relationships shown in the accompanying drawings. These are merely for ease of description and do not indicate or imply that the structure must have a specific orientation, or be constructed and operated in a specific orientation; therefore, they should not be construed as limitations on this invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0029] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "connected," "connected," and "linked" should be interpreted broadly. For example, they can refer to a fixed connection or a detachable connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0030] In a first aspect, the present invention provides a method for acquiring images of the surface of a laser-processed molten pool, the method comprising the following steps: Obtain the laser energy distribution type; The area of ​​the oxide layer region is obtained based on the laser energy distribution type; Calculate the average gray value of the region in a single frame image based on the area of ​​the oxide layer region; Based on the average gray value of a single frame, the average gray values ​​of multiple frames of images are statistically averaged to obtain the average gray value of the molten pool oxide layer across multiple frames. This method first distinguishes the laser energy distribution type, then specifically acquires the area of ​​the oxide layer region and calculates the grayscale value, achieving accurate acquisition and analysis of molten pool surface images for different laser distribution types. The overall process is logically coherent, adapting to both the uniform energy characteristics of flat-top laser distribution and the gradient energy characteristics of Gaussian laser distribution. Finally, through single-frame and multi-frame grayscale value analysis, the accuracy of molten pool analysis during laser processing is improved.

[0031] Furthermore, based on the laser energy distribution type, the area of ​​the oxide layer region is obtained, including: If the laser energy distribution is a flat-top laser, acquire coaxial camera images of the flat-top laser and determine the laser irradiation area through image processing; Based on the laser irradiation area, the red light component of the original image is extracted, and after processing, the visible molten pool area within the laser irradiation range with a flat top distribution is obtained. Perform feature analysis on the visible molten pool region and calculate the area of ​​the liquid metal region within that region; The area of ​​the oxide layer within the laser irradiation range is calculated by combining the total area of ​​the actual molten pool measured by a metal microscope and the area of ​​the liquid metal region.

[0032] Taking advantage of the characteristics of flat-top laser distribution, this paper proposes a method to achieve refined division of the molten pool region under flat-top laser conditions. This method involves stepwise determination of the laser irradiation area, extraction of the visible molten pool region, calculation of the liquid metal area, and derivation of the oxide layer area by combining the actual total molten pool area. This process fully utilizes the uniform energy of flat-top lasers, reducing region identification errors caused by uneven energy distribution, and improving the relevance and accuracy of oxide layer area calculation. This provides reliable data for assessing the molten pool condition in flat-top laser processing scenarios.

[0033] Furthermore, if the laser energy distribution is a flat-top laser, a coaxial camera image of the flat-top laser is acquired, and the laser irradiation area is determined through image processing, including: The acquired flat-top distributed laser coaxial camera images are converted into grayscale images; The grayscale image is denoised using a 5×5 pixel Gaussian filter kernel. A binarization algorithm is used to process the denoised grayscale image to obtain the laser irradiation area.

[0034] The illumination area of ​​the flat-top distributed laser is determined by grayscale conversion, 5×5 pixel Gaussian filtering for noise reduction, and binarization algorithm. This effectively filters out image noise and accurately extracts the physical boundary of the actual laser illumination. This step defines a clear spatial reference for subsequent molten pool region analysis, reduces interference from background areas outside the laser illumination range, and improves the basic accuracy of molten pool region identification under flat-top laser illumination.

[0035] Furthermore, based on the laser irradiation area, the red light component of the original image is extracted and processed to obtain the visible molten pool region within the flat-topped laser irradiation range; feature analysis is performed on the visible molten pool region to calculate the area of ​​the liquid metal region within this area, including: Extract the red light component from the original image; The red light component image was denoised using a 5×5 pixel Gaussian filter. Calculate the Otsu binarization threshold. If the Otsu binarization threshold exceeds the range, use constant threshold binarization; otherwise, use the Otsu binarization threshold. The minimum threshold is 21 and the maximum threshold is 255 under the powder-free processing state. The minimum threshold is gmin=36 and the maximum threshold is 65. The visible molten pool region is obtained through binarization. The intersection of the visible molten pool region and the laser irradiation region is calculated to obtain the visible molten pool region within the laser irradiation range; An adaptive threshold binarization process with a size of 15×15 pixels is applied to the visible molten pool region to obtain the adaptive threshold binarized region. Binarize the region with a gray value of 255 in the visible molten pool area to obtain the binarized region with the highest gray value. The visible oxide layer region in the laser-irradiated area is calculated as: adaptive threshold binarization + binarized region with the highest grayscale value. The liquid metal area is calculated as follows: visible molten pool area within the laser irradiation range - visible oxide layer area within the laser irradiation area. If the laser power is less than 600W, the liquid metal area is set to 0. Otherwise, the liquid metal area is equal to the liquid metal area directly exposed to the irradiation area, and the negative value generated after the subtraction is set to 0.

[0036] By extracting the red light component to reduce ambient light interference, and combining adaptive and constant threshold binarization to process the visible molten pool region, and further distinguishing between the oxide layer and liquid metal through dual binarization, fine segmentation of the visible molten pool region under a flat-top laser is achieved. Simultaneously, the liquid metal area is corrected based on the laser power, further ensuring the rationality of the data in low-power scenarios and improving the reliability of the liquid metal and oxide layer area calculations.

[0037] Furthermore, based on the laser energy distribution type, the area of ​​the oxide layer region is obtained, including: If the laser energy distribution is Gaussian, set specific process parameters to acquire images during the stable phase and locate the coordinates of the laser spot center. Based on the center coordinates, the boundary of the laser irradiation area is determined, and the acquired image is cropped and denoised. Based on the preprocessed acquired image, the corresponding method is selected according to the laser power and powder supply status, and the image is binarized. The longest perimeter contour is extracted based on the binarized image, and contour artifacts are identified and processed by calculating the positional relationship and roundness. Based on the corrected contour features, the area of ​​the liquid metal region and the area of ​​the oxide layer region are calculated.

[0038] To address the characteristics of Gaussian laser energy distribution—high energy center and low energy edge—this paper effectively solves the problem of energy gradient interference in molten pool images under Gaussian lasers by employing steps such as spot center localization, image preprocessing, binarization, and contour artifact removal. In particular, contour artifact recognition and correction avoids misjudgment of regions caused by uneven energy distribution, ensuring the accuracy of liquid metal and oxide layer area calculations, and adapting to the complex image features of Gaussian laser processing scenarios.

[0039] Furthermore, if the laser energy distribution is Gaussian, specific process parameters are set to acquire images during the stable phase, and the coordinates of the laser spot center are located, including: The laser process parameters were set to 600W power, 0m / s laser speed, and 10s processing time. Coaxial camera video data was collected during the processing. The last 7 seconds of the video data were selected as the image of the stabilization stage. The Otsu binarization method was used to obtain the ring contour of the oxide layer region in the frame image. The gray value of the oxide layer region was set to 255, and the gray value of the liquid metal region inside Avox was manually set to 255. Calculate the average center coordinates of a single frame Calculate the average center coordinates (Ic, Jc) of the video sequence: ; in Here, n represents the average center coordinates of a single frame, and n255 represents the number of pixels with a grayscale value of 255. (Ic, Jc) represents the average center coordinates of the video sequence, Nstart refers to the ID number of the first image in the coaxial camera data, and Nend refers to the ID number of the last image in the coaxial camera data.

[0040] By setting specific process parameters to acquire images during the stable phase, and locating the center of the Gaussian laser spot based on the average coordinates of high grayscale regions, the stability and accuracy of the spot center coordinates were ensured. This center serves as the spatial reference for subsequent image analysis, effectively reducing regional analysis errors caused by laser drift or jitter, and providing a reliable reference for spatial positioning of molten pool images under Gaussian laser illumination.

[0041] Furthermore, based on the center coordinates, the boundary of the laser irradiation area is determined, and the acquired image is preprocessed by cropping and denoising. Based on the preprocessed acquired image, a corresponding method is selected according to the laser power and powder supply status to perform binarization processing, including: The laser irradiation area is set as the center of the spot, and a circular area with the center coordinates (Ic, Jc) of the spot center and a radius of rl is defined, where rl is the distance from the center of the spot to the edge of the laser. The acquired single-frame image is cropped into a 200×200 pixel image. The cropped image is then denoised using a 9×9 pixel Gaussian filter kernel. In the case of powderless laser processing, Otsu binarization is used when the laser power is in the range of 300W-600W, and fixed threshold binarization is used when the laser power is <300W to generate a binarized image. For other process parameters, adaptive threshold binarization with a size of 15×15 pixels is used to generate a binarized image. The binarized image is subjected to two opening operations using a kernel function with a size of 5×5 pixels to eliminate white spot noise in the image.

[0042] The boundary of the laser irradiation area is determined based on the center of the laser spot. Image preprocessing is achieved by combining 200×200 pixel cropping and 9×9 pixel Gaussian filtering, which can focus the core area and reduce noise interference. Adaptive binarization methods are selected for different power and powder supply conditions. White point noise is eliminated by two opening operations, which significantly improves the contrast and clarity of the Gaussian laser image and lays a high-quality image foundation for subsequent contour extraction and artifact recognition.

[0043] Furthermore, based on the binarized image, the longest perimeter contour is extracted, and contour artifacts are identified and processed through positional relationships and roundness calculations, including: Extract the longest perimeter contour from the binarized image; determine whether the center of the laser spot is within the region enclosed by the longest perimeter contour; If not present, it is determined to be either a no-fake or a Type I fake, and the fake ID is set to 0; If it is, then calculate the contour roundness φ according to the formula φ=4πS / C², where S is the area of ​​the region enclosed by the contour and C is the perimeter of the contour; If φ < 0.65, it is determined to be a Type II or Type V artifact; if φ ≥ 0.65, it is determined to be a Type III artifact when the laser power is ≤ 1200W, and a Type IV artifact when the laser power is > 1800W.

[0044] By extracting the longest perimeter contour and combining the positional relationship between the laser spot center and the contour with roundness calculations to identify Type IV contour artifacts, it is possible to accurately distinguish and mark artifacts such as oxide layer adhesion and spatter interference caused by Gaussian laser energy distribution. This processing effectively avoids the impact of artifacts on the area calculation, ensuring the authenticity and accuracy of subsequent liquid metal and oxide layer area calculations.

[0045] Furthermore, if the illusory ID=0, the liquid metal radius = the shortest distance from the laser spot center (Ic,Jc) to the longest perimeter contour; if the illusory ID=3, the liquid metal radius = the shortest distance from the laser spot center (Ic,Jc) to the second longest perimeter contour; calculate the half-width of the molten pool: ; Where rom is the half-width of the molten pool in the multi-channel metallographic results; Liquid metal area = Liquid metal area within the laser irradiation zone; The area of ​​the oxide layer region = Amp - Alm, where Amp is the visible area of ​​the molten pool; Calculate the intersection of Aox_l=Aox and the laser-irradiated area to obtain the area of ​​the oxide layer region within the laser irradiation range.

[0046] By dynamically determining the radius of the liquid metal based on different contour artifact types and combining this with the half-width of the molten pool to calculate the area of ​​the liquid metal and oxide layer, adaptive processing of complex molten pool contours under Gaussian lasers is achieved. This method can flexibly handle contour features under different processing conditions, ensuring consistency between the area calculation results and the actual molten pool state, thus improving data reliability.

[0047] This invention also provides a laser processing molten pool surface image acquisition device for the laser processing molten pool surface image acquisition method described in any of the above claims. The acquisition device includes a laser processing equipment and an image acquisition and processing equipment. The laser processing equipment includes a continuous fiber laser, a laser focusing device, a powder feeding device, and a stepper motor. The continuous fiber laser is used to emit a flat-top distributed laser or a Gaussian distributed laser. The laser focusing device is used to focus the laser. The powder feeding device is used to direct the laser towards the substrate through a coaxial powder feeding head. The powder feeding device is also used to deliver metal powder to the substrate. The stepper motor is used to drive the continuous fiber laser and the laser focusing device to move synchronously to achieve scanning processing. The image acquisition and processing equipment includes a coaxial camera and a computer. The coaxial camera is used to acquire dynamic images of the molten pool within the laser irradiation range. The computer is electrically connected to the coaxial camera via a network cable and is used to receive image data and execute acquisition methods and calculation steps. Based on the above setup, the device, through the coordinated operation of laser processing equipment and image acquisition and processing equipment, can both emit flat-top or Gaussian distributed lasers for scanning processing and acquire molten pool images in real time and perform analysis methods. The components of the device are highly adaptable, and the computer can efficiently process image data and set camera parameters, ensuring the synchronization and accuracy of laser processing and image acquisition and analysis.

[0048] The specific steps of the data acquisition method of the present invention include the following steps. Example 1: Image acquisition of molten pool surface for flat-top distributed lasers S1: Obtain laser energy distribution type The energy distribution of the laser output beam is detected by a laser power meter to confirm that the current laser is a flat-top distribution laser (energy uniformity > 90%), and the acquisition and recognition logic corresponding to the flat-top distribution is selected.

[0049] S2: Determine the laser irradiation area (Al) S2.1: Start the laser, set the power to 800W and the scanning speed to 10mm / s, and turn on the coaxial camera to acquire the coaxial image of the pilot laser (exposure time 50μs, gain 10dB).

[0050] S2.2: After receiving the image, the computer converts it into an 8-bit grayscale image (pixel value 0-255).

[0051] S2.3: A Gaussian filter kernel (σ=1.0) with a size of 5×5 pixels is used to denoise the grayscale image and eliminate high-frequency noise.

[0052] S2.4: Apply the Otsu binarization algorithm (maximum inter-class variance method) to automatically determine the threshold, and segment the image into the laser-illuminated area (foreground, pixel value 255) and the non-illuminated area (background, pixel value 0), to obtain the set of pixel coordinates of the laser-illuminated area A1.

[0053] S3: Extract the visible molten pool region (Avmp_l) within the laser irradiation range. S3.1: Extract the red light component (R channel) from the original color image (RGB format), and ignore the G and B channels to reduce the interference of plasma blue and green light.

[0054] S3.2: Use a 5×5 pixel Gaussian filter (σ=1.0) to denoise the red light component image.

[0055] S3.3: Calculate the Otsu binarization threshold gOtsu. In this embodiment, gOtsu=45 under the powder supply processing state (within the range of 36-65). Therefore, Otsu binarization is used to obtain the visible molten pool area Avmp (the foreground is the molten pool, with a pixel value of 255).

[0056] S3.4: Calculate the overlapping area of ​​Avmp and Al through pixel coordinate intersection operation to obtain the visible molten pool area Avmp_l within the laser irradiation range.

[0057] S4: Calculate the area of ​​the liquid metal region (Alm_l) S4.1: Apply adaptive threshold binarization of Avmp with a size of 15×15 pixels (block size 15, constant C=2) to obtain the adaptive threshold binarization region AAdaptive_threshold (oxide candidate region).

[0058] S4.2: Binarize the pixels with a gray value of 255 in Avmp to obtain the highest gray value binarized region A255_threshold (high-gloss oxide layer region).

[0059] S4.3: Calculate the visible oxide layer region (Pixel-level union operation).

[0060] S4.4: Calculate the liquid metal area Alm = number of pixels of Avmp_l - number of pixels of Avox_l (unit: pixel²). Since the current laser power is 800W > 600W, Alm_l = Alm (negative values ​​are set to 0).

[0061] S4.5: Converts pixel area to actual area (μm²) through pixel size calibration (1 pixel = 5.2μm).

[0062] S5: Calculate the area of ​​the oxide layer region (Aox_l) S5.1: The cross-section of the processed substrate was photographed using an Olympus GX51 metal microscope, and the actual total area of ​​the molten pool, Aom, was measured to be 120000μm².

[0063] S5.2: Calculate the intersection area of ​​Aom and Al complement (i.e., the area of ​​the molten pool outside the laser irradiation range) = 15000μm².

[0064] S5.3: The area of ​​the oxide layer region within the laser irradiation range is obtained by using the formula Aox_l=Aom-Alm_l-15000μm².

[0065] Example 2: Image acquisition of molten pool surface for Gaussian distributed lasers S1: Obtain laser energy distribution type The laser power meter was used to detect and confirm that the current laser is of Gaussian distribution (the energy is distributed normally with high energy in the center and low energy at the edges). The acquisition and recognition logic corresponding to the Gaussian distribution was then selected.

[0066] S2: Locate the center coordinates of the laser spot (Ic, Jc) S2.1: Set the laser process parameters: power 600W, scanning speed 0m / s (static irradiation), processing time 10s, and turn on the camera to collect video data (frame rate 10fps).

[0067] S2.2: Extract the last 7 seconds of the video (70 frames in total) as the stable phase image.

[0068] S2.3: Apply Otsu binarization to each frame of the image, extract the annular contour of the oxide layer region Avox, and manually set the pixel values ​​of Avox and the internal liquid metal region to 255.

[0069] S2.4: Calculate the average center of a single frame (iˉ,jˉ): Count the pixel coordinates (i,j) of grayscale value = 255 in each frame, and calculate iˉ = ∑i / n255 and jˉ = ∑j / n255 (n255 is the total number of high grayscale pixels in this frame).

[0070] S2.5: Calculate the average center coordinates (Ic, Jc) = (685, 520) pixels (image resolution 1280×1024) over 70 frames, and use them as the center of the laser spot.

[0071] S3: Image preprocessing and laser irradiation area determination S3.1: Crop a single frame image into a 200×200 pixel region centered at (Ic, Jc) (to reduce invalid background).

[0072] S3.2: Use a 9×9 pixel Gaussian filter (σ=1.5) to denoise the cropped image and smooth the image edges.

[0073] S3.3: Define the laser irradiation area Al as a circular area with (Ic, Jc) as the center and a radius rl = 50 pixels (actually 260μm) (rl is determined by measuring the pixel distance at the edge of the light spot).

[0074] S4: Image Binarization Processing S4.1: This embodiment is for powder supply processing, so 15×15 pixel adaptive threshold binarization is used (block size 15, constant C=3).

[0075] S4.2: Perform two opening operations (erosion followed by dilation) on the binarized image using a 5×5 pixel rectangular kernel function to eliminate white spot noise (splashing interference) with a diameter < 5 pixels.

[0076] S5: Contour Artifact Detection and Processing S5.1: Use OpenCV's findContours function to extract all contours in the binarized image and select the contour C1 with the longest perimeter.

[0077] S5.2: Determine whether the center of the light spot (Ic, Jc) is inside C1 (using the pointPolygonTest function). In this embodiment, the center is inside C1.

[0078] S5.3: Calculate the roundness of C1 φ=4πS / C², where S=3140 pixels² (outline area) and C=198 pixels (outline perimeter), and get φ=0.63<0.65, which is determined to be a Type II artifact (ID=2).

[0079] S6: Calculate the area of ​​liquid metal and oxide layer S6.1: Since the Type II artifact does not affect the calculation of the liquid metal radius, the closest distance from the laser center to C1 is taken as rlm = 30 pixels (actually 156μm).

[0080] S6.2: Calculate the liquid metal area Alm = πrlm² = 2826 pixels² (actual 75000μm²).

[0081] S6.3: The visible area of ​​the molten pool is measured to be Amp = 4500 pixels², therefore the area of ​​the oxide layer region is Aox = 4500 - 2826 = 1674 pixels².

[0082] S6.4: Calculate the intersection of Aox and Al to obtain the area of ​​the oxide layer region within the laser irradiation range, Aox_l = 1674 pixels² (because Aox is completely within Al).

[0083] The technical solutions of the present invention have been described in conjunction with the optional embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A laser processing molten pool surface image acquisition method, characterized in that, The acquisition method steps include: acquiring a laser energy distribution type; obtaining an oxide layer area according to the laser energy distribution type; calculating an average gray value of the area in a single frame image according to the oxide layer area; statistically averaging single frame average gray values of multiple frames of images to obtain a multiple frame average gray value of the oxide layer of the molten pool.

2. The laser-processed molten pool surface image acquisition method according to claim 1, characterized by, The oxide layer area is obtained according to the laser energy distribution type, including: if the laser energy distribution is flat-top distribution laser, collecting coaxial camera images of the flat-top distribution laser, and determining a laser irradiation area through image processing; extracting a red light component of an original image according to the laser irradiation area, and obtaining a visible molten pool area within the laser irradiation range of the flat-top distribution laser after processing; performing feature analysis on the visible molten pool area, and calculating a liquid metal area within the area; combining an actual molten pool total area measured by a metal microscope and the liquid metal area to calculate an oxide layer area within the laser irradiation range.

3. The laser-processed molten pool surface image acquisition method according to claim 1, characterized by, The oxide layer area is obtained according to the laser energy distribution type, including: if the laser energy distribution is Gaussian distribution laser, setting specific process parameters to collect stable stage images, and locating a laser spot center coordinate; determining a laser irradiation area boundary based on the center coordinate, and performing cropping and denoising preprocessing on the collected images; based on the preprocessed collected images, selecting a corresponding mode according to laser power and powder supply state, and performing binarization processing; based on the binarized image, extracting a longest perimeter contour, and identifying and processing contour artifacts through position relationship and roundness calculation; based on the corrected contour features, calculating a liquid metal area and an oxide layer area.

4. The laser-processed molten pool surface image acquisition method according to claim 2, characterized by, If the laser energy distribution is flat-top distribution laser, the coaxial camera images of the flat-top distribution laser are collected, and the laser irradiation area is determined through image processing, including: converting the collected coaxial camera images of the flat-top distribution laser into gray scale images; performing denoising processing on the gray scale images using a 5*5 pixel size Gaussian filter kernel; performing processing on the denoised gray scale images using a binarization algorithm to obtain the laser irradiation area.

5. The laser-processed molten pool surface image acquisition method according to claim 4, characterized by, The red light component of the original image is extracted according to the laser irradiation area, and the visible molten pool area within the laser irradiation range of the flat-top distribution laser is obtained after processing; performing feature analysis on the visible molten pool area, and calculating a liquid metal area within the area, including: extracting a red light component from an original image; performing denoising processing on the red light component image using a 5*5 pixel size Gaussian filter; calculating an Otsu binarization threshold value, if the Otsu binarization threshold value exceeds a range, using a constant threshold binarization, otherwise using the Otsu binarization threshold value, wherein the minimum threshold value is 21 and the maximum threshold value is equal to 255 in a powder-free processing state, the minimum threshold value is gmin=36 and the maximum threshold value is 65; obtaining a visible molten pool area through binarization processing; calculating an intersection of the visible molten pool area and the laser irradiation area to obtain a visible molten pool area within the laser irradiation range; Adaptive threshold binarization processing is performed on the visible molten pool area with a 15*15 pixel size to obtain an adaptive threshold binarization area; Binarization processing is performed on the area with a gray value of 255 in the visible molten pool area to obtain a highest gray value binarization area; A visible oxide layer area of the laser irradiation area is calculated = the adaptive threshold binarization processing + the highest gray value binarization area; A liquid metal area = the visible molten pool area in the laser irradiation range - the visible oxide layer area of the laser irradiation area; if the laser power < 600 W, the liquid metal area is set to 0; otherwise, the liquid metal area = the liquid metal area directly exposed to the irradiation area, and the negative value generated after the subtraction operation is set to 0.

6. The laser-processed molten pool surface image acquisition method according to claim 3, characterized by, If the laser energy distribution is a Gaussian distribution laser, specific process parameter acquisition stable stage images are set, and the laser spot center coordinates are located, including: The laser process parameters are set to 600 W power, 0 m / s laser speed, and 10 s processing time, and coaxial camera video data in the processing process is collected; the last 7 seconds of frame images in the video data are selected as the stable stage images; an Otsu binarization method is used to obtain the annular contour of the oxide layer area in the frame image, the gray value of the oxide layer area is set to 255, and the gray value of the liquid metal area in the visible area of the molten pool oxide layer is manually set to 255; computing a single frame average center coordinate computing an average center coordinate for a video sequence : ； wherein (Ic, Jc) is the average center coordinate of the video sequence, Nstart denotes the start image ID number of the coaxial camera data, and Nend denotes the last image ID number of the coaxial camera data.

7. The laser-processed molten pool surface image acquisition method according to claim 6, characterized by, Based on the center coordinates, the laser irradiation area boundary is determined, and the collected images are cropped and denoised for preprocessing; Based on the preprocessed collected images, corresponding methods are selected according to the laser power and powder supply state, and binarization processing is performed, including: The laser irradiation area is set as the spot center, and the spot center coordinates (Ic, Jc) are taken as the center of a circular area with a radius rl, where rl is the distance from the spot center to the laser edge; the collected single frame image is cropped to a 200*200 pixel image; the cropped image is denoised using a 9*9 pixel size Gaussian filter kernel; If it is a powder-free laser processing collected video sequence, Otsu binarization is used when the laser power is in the range of 300 W-600 W, and fixed threshold binarization is used when the laser power < 300 W to generate a binarization image; If it is the rest of the process parameters, adaptive threshold binarization with a 15*15 pixel size is used to generate a binarization image; The binarization image is subjected to 2 times of opening operation using a 5*5 pixel size kernel function to eliminate white point noise in the image.

8. The laser-processed molten pool surface image acquisition method according to claim 7, characterized by, Based on the binarization image, the longest perimeter contour is extracted, and false contours are recognized and processed through position relationship and roundness calculation, including: The longest perimeter contour in the binarization image is extracted; it is judged whether the laser spot center is in the area surrounded by the longest perimeter contour; If not, it is determined to be no false contour or I-type false contour, and the false contour ID is set to 0; If yes, then the profile roundness φ is calculated according to the formula where S is the area of the region enclosed by the profile and C is the circumference of the profile. If φ < 0.65, it is determined to be II-type or V-type false contour; if φ ≥ 0.65, it is determined to be III-type false contour when the laser power ≤ 1200 W, and IV-type false contour when the laser power > 1800 W.

9. The laser-processed molten-pool surface image acquisition method according to claim 8, characterized by, If artifact ID = 0, the liquid metal radius = the closest distance from the laser spot center (Ic, Jc) to the longest perimeter contour; if artifact ID = 3, the liquid metal radius = the closest distance from the laser spot center (Ic, Jc) to the second longest perimeter contour; calculate the molten pool half-width: ; wherein romis the molten pool half-width in the multi-pass metallographic result; Liquid metal area = area of liquid metal in the laser irradiation region; Oxidation layer area = Amp - Alm, where Amp is the area of the visible range of the molten pool; Calculate the intersection of Aox_l = Aox and the laser irradiation area to obtain the oxidation layer area in the laser irradiation range.

10. A laser processing molten pool surface image acquisition device, characterized in that, The laser processing molten pool surface image acquisition method of any one of claims 1-9, wherein the acquisition device comprises a laser processing device and an image acquisition processing device; the laser processing device comprises a continuous fiber laser, a laser focusing device, a powder feeding device, and a stepping motor, the continuous fiber laser is used to emit a flat-top distributed laser or a Gaussian distributed laser; the laser focusing device is used to focus the laser, the powder feeding device is used to direct the laser to the substrate through a coaxial powder feeding head, the powder feeding device is also used to deliver metal powder to the substrate, and the stepping motor is used to drive the continuous fiber laser and the laser focusing device to move synchronously to realize scanning processing; the image acquisition processing device comprises a coaxial camera and a computer, the coaxial camera is used to acquire the dynamic image of the molten pool in the laser irradiation range, the computer is electrically connected to the coaxial camera through a network cable, and the computer is used to receive image data and execute the acquisition method and the calculation step.

Images