A water guide laser control method and device, and readable storage medium

Abstract

This invention relates to the field of water-guided laser technology, and discloses a control method and device for water-guided lasers, as well as a readable storage medium. The control method for water-guided lasers includes the following steps: using a coaxial vision sensor to acquire a coaxial image of the nozzle inlet end face and the laser focused spot through a first optical axis, where the first optical axis is shared by the beam splitter and the water-guided laser optical path; and performing dark field correction processing on the coaxial image. This water-guided laser control method, by acquiring a coaxial image of the nozzle inlet end face and the laser focused spot through a coaxial vision sensor, changes the ratio of meridional rays to oblique rays incident on the water jet, transforming the energy distribution of the water jet cross-section from a centrally concentrated Gaussian-like distribution to a uniform distribution with sufficient energy at the edges. This effectively solves the problem of uneven processing caused by insufficient edge energy in traditional water-guided laser processing, improving processing accuracy and surface quality.

Description

Technical Field

[0001] This invention relates to the field of water-guided laser technology, specifically to a water-guided laser control method, a water-guided laser control device, and a readable storage medium. Background Technology

[0002] Water-guided laser processing technology is a special processing technology that uses water jets as the laser transmission medium. It utilizes the total internal reflection effect at the water-air interface to confine laser energy within the water jet and transmit it to the workpiece surface for processing. It is widely used in aerospace composite material processing, semiconductor wafer dicing, and precision electronic component manufacturing.

[0003] However, the inherent Gaussian intensity distribution of laser beams results in high energy density at the center and low energy density at the edges. When the laser is coaxially coupled into a water jet, the energy distribution across the water jet cross-section inherits this Gaussian characteristic, exhibiting a non-uniform state with concentrated energy at the center and severely insufficient energy at the edges. This uneven energy distribution directly leads to differences in material removal rates at different locations on the workpiece surface during processing, causing problems such as inconsistent groove depths, poor edge roughness, and insufficient processing accuracy. This severely restricts the application of water-guided laser processing technology in high-precision, high-quality processing scenarios.

[0004] Currently, to address the issue of uneven energy distribution in water-guided laser processing, existing technologies mainly employ two solutions: First, adding beam shaping devices to the laser path to reshape the Gaussian beam into a flat-top beam, thereby improving the energy distribution of the water beam. However, this approach requires a complex optical system, significantly increasing equipment costs and system complexity. Furthermore, the beam shaping effect is greatly affected by fluctuations in laser parameters, limiting its adaptability to different processing conditions. Second, using multiple scans or increasing path overlap to compensate for insufficient energy in edge regions through energy superposition. However, this approach significantly reduces processing efficiency, extends the production cycle, and multiple scans can easily lead to heat accumulation on the workpiece, negatively impacting processing quality. Summary of the Invention

[0005] The purpose of this invention is to provide a solution to the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for controlling a water-guided laser, comprising the following steps: Using a coaxial vision sensor, a coaxial image of the nozzle inlet end face and the laser focused spot is acquired through the first optical axis. The first optical axis is the shared optical axis of the beam splitter and the water-guided laser optical path. Perform dark field correction processing on coaxial images; Median filtering is applied to the coaxial image after dark field correction. Adaptive histogram equalization preprocessing is performed on the coaxial image after median filtering to obtain the first processed image; Based on the first image, obtain the first center coordinates of the laser focused spot and the second center coordinates of the nozzle inlet end face; Input the second center coordinates and the first center coordinates into the offset-light field distribution mapping model, and calculate the target offset vector of the light spot relative to the second center coordinates; Based on the target offset vector and calibration mapping matrix, the angle control value of the beam pointing adjustment mechanism is obtained; Based on the angle control quantity, the optical field adjustment controller drives the beam pointing adjustment mechanism to move, and presets the center of the laser focusing spot to the target position that is off-center from the nozzle inlet end face.

[0007] Optionally, the step of obtaining the first center coordinates of the laser focused spot based on the first image includes: Based on the first image, the first center coordinates of the laser focused spot are obtained using the gray-scale centroid method.

[0008] Optionally, the step of obtaining the first center coordinates of the laser focused spot based on the first image using the gray-scale centroid method includes: The Otsu method was used to determine the segmentation threshold for the laser focused spot in the first image; Based on the first image and the segmentation threshold, extract the spot region; Within the light spot area, the first center coordinates are calculated using pixel grayscale values ​​as weights; A sliding median filter is applied to the first center coordinates of the first images captured in multiple consecutive frames.

[0009] Optionally, the step of obtaining the second center coordinates of the nozzle inlet end face based on the first image includes: Based on the first image, the Hough transform circle detection algorithm is used to obtain the second center coordinates of the nozzle inlet end face.

[0010] Optionally, the step of obtaining the second center coordinates of the nozzle inlet end face using the Hough transform circle detection algorithm based on the first image includes: The edge of the nozzle inlet end face region in the first image is extracted using the Cannibal operator; Based on the edge, the point set of the edge is obtained, and the Hough transform circle detection algorithm is applied to the point set of the edge to detect the initial circle center and radius; Based on the initial center and radius, the coordinates of the second center are obtained by least-squares circle fitting.

[0011] Optionally, the calibration process for the offset-light field distribution mapping model is as follows: Using the second center coordinate as the origin, select multiple different offsets within the nozzle inlet end face; Based on multiple different offsets, the laser focused spot is adjusted to multiple offset positions corresponding to the multiple different offsets; At multiple offset locations, the two-dimensional energy distribution of the water jet outlet cross section was measured using a beam analyzer. Based on calculations of multiple two-dimensional energy distributions, multiple uniformity indices are obtained; Establish mapping relationships between multiple different offsets and multiple uniformity indices; Based on the mapping relationship, the target offset vector is obtained through surface fitting; The uniformity index is the ratio of the standard deviation to the mean of the energy distribution within the effective area of ​​the water jet cross section.

[0012] Alternatively, the calibration mapping matrix is ​​obtained in the following way: Multiple sampling points are selected within the travel range of the beam pointing adjustment mechanism; A preset control value is applied to each of the multiple sampling points, and the calibration mapping matrix is ​​obtained by using the least squares method based on the first center coordinate.

[0013] Optionally, the control device for the water-guided laser includes: The acquisition unit is used to acquire a coaxial image of the nozzle inlet end face and the laser focused spot through a first optical axis using a coaxial vision sensor. The first optical axis is a shared optical axis between the beam splitter and the water-guided laser optical path. The first image processing unit is used to perform dark field correction processing on the coaxial image; The second image processing unit is used to perform median filtering on the coaxial image after dark field correction. The third image processing unit is used to perform adaptive histogram equalization preprocessing on the coaxial image after median filtering to obtain the processed first image. The coordinate acquisition unit is used to acquire the first center coordinates of the laser focused spot and the second center coordinates of the nozzle inlet end face based on the first image; The calculation unit is used to input the second center coordinates and the first center coordinates into the offset-light field distribution mapping model, and calculate the target offset vector of the light spot with energy distribution relative to the second center coordinates; The control quantity acquisition unit is used to acquire the angle control quantity of the beam pointing adjustment mechanism based on the target offset vector and the calibration mapping matrix; The control unit is used to control the light field adjustment controller to drive the beam pointing adjustment mechanism to move based on the angle control amount, so as to preset the center of the laser focusing spot to the target position that is offset from the center of the nozzle inlet end face.

[0014] Optionally, the control device for the water-guided laser also includes a memory and a processor. The memory stores programs or instructions that can be run on the processor. When the program or instructions are executed by the processor, they implement the steps of the vision-guided water-guided laser processing offset preset method as described in any of the above technical solutions.

[0015] Optionally, a program or instructions are stored on a readable storage medium, which, when executed by a processor, implements the steps of the vision-guided water-guided laser processing offset preset method as described in any of the above technical solutions.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: the control method and apparatus for water-guided lasers, and the readable storage medium, have the following advantages: This invention acquires coaxial images of the nozzle inlet face and the laser focusing spot using a coaxial vision sensor, extracts the coordinates of the spot center and the nozzle center, calculates the target offset that optimizes the energy distribution based on a pre-calibrated offset-light field distribution mapping model, and then drives the beam pointing adjustment mechanism to preset the spot to a target position off-center from the nozzle. This changes the ratio of meridional rays to oblique rays incident on the water jet, transforming the energy distribution of the water jet cross-section from a centrally concentrated Gaussian-like distribution to a uniform distribution with sufficient energy at the edges. This effectively solves the problem of uneven processing caused by insufficient edge energy in traditional water-guided laser processing, improving processing accuracy and surface quality.

[0017] This invention does not require the introduction of complex beam shaping devices. It only integrates a coaxial vision acquisition module and a beam pointing adjustment mechanism into the existing water-guided laser system. It uses the combination of visual positioning and pre-calibration model to achieve energy distribution control. The system modification is small, the hardware cost is low, and it is easy to upgrade existing equipment. At the same time, the offset-light field distribution mapping model can be recalibrated for nozzles of different specifications, which can adapt to a variety of processing conditions.

[0018] This invention can achieve uniform control of water jet energy distribution through a single offset preset, without the need for multiple scans or increasing path overlap rate for energy compensation. Furthermore, by calibrating the mapping matrix, the target offset vector is converted into a control quantity for the beam pointing adjustment mechanism, driving the adjustment mechanism to quickly complete the spot position adjustment. At the same time, it fundamentally avoids the problem of workpiece heat accumulation caused by multiple scans, ensuring the stability of processing quality. Attached Figure Description

[0019] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0020] Figure 1 This is a flowchart illustrating a water-guided laser control method according to an embodiment of the present invention; Figure 2 This is a block diagram of a control device for a water-guided laser according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the calibration results of the offset-light field distribution mapping model. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Please see Figure 1 The technical solution provided by this invention is a control method for a water-guided laser, characterized by the following steps: Step 101: Using a coaxial vision sensor, a coaxial image of the nozzle inlet end face and the laser focused spot is acquired through a first optical axis, wherein the first optical axis is a shared optical axis of the beam splitter and the water-guided laser optical path; Step 102: Dark field correction processing is performed on the coaxial image; Step 103: Median filtering processing is performed on the coaxial image after dark field correction processing; Step 104: Adaptive histogram equalization preprocessing is performed on the coaxial image after median filtering processing to obtain the processed first image; Step 105: Based on... First image: Obtain the first center coordinates of the laser focused spot and the second center coordinates of the nozzle inlet end face; Step 106: Input the second center coordinates and the first center coordinates into the offset-light field distribution mapping model, and calculate the target offset vector of the energy distribution spot relative to the second center coordinates; Step 107: Based on the target offset vector and the calibration mapping matrix, obtain the angle control amount of the beam pointing adjustment mechanism; Step 108: Based on the angle control amount, control the light field adjustment controller to drive the beam pointing adjustment mechanism to move, and preset the center of the laser focused spot to the target position that is offset from the center of the nozzle inlet end face.

[0023] In the specific implementation process, it is worth noting that the eight steps of this method form a complete closed-loop control logic of "image acquisition, image preprocessing, location extraction, target calculation and control execution". The output of each step is the only input of the next step. The steps are strictly executed in sequence, and there is no parallel or reverse execution, which ensures the determinism and reproducibility of the control process. Step 101: This invention employs a beam splitter to achieve a shared optical path between the first optical axis and the water-guided laser, fundamentally solving the inherent defects of existing non-coaxial vision systems. In existing non-coaxial vision systems, the camera and laser optical path are installed at a fixed angle, and the position of the light spot captured by the camera is a projection of the actual laser illumination position, resulting in an unavoidable parallax error. This error changes with the working distance between the nozzle and the camera, failing to meet the high-precision offset control requirements of this invention. In this invention, however, the beam splitter is fixedly installed at a 45° angle in the focusing optical path of the water-guided laser. The high-energy laser beam passes through the beam splitter and is focused by the focusing lens onto the nozzle inlet face. Simultaneously, the reflected light from the nozzle inlet face and the focused laser spot returns along the original optical path, is reflected by the beam splitter, and enters the coaxial vision sensor. This ensures that the imaging optical axis of the coaxial vision sensor is completely aligned with the transmission optical axis of the laser, guaranteeing accurate camera imaging. The position of each pixel in the acquired image corresponds one-to-one with the actual irradiation position of the laser on the nozzle inlet face, with no viewing angle deviation. The coaxial vision sensor includes a CCD camera and an adjustable attenuator assembly. The adjustable attenuator assembly consists of multiple attenuators with different transmittance and a stepper motor-driven rotating wheel. By switching different combinations of attenuators through the stepper motor-driven rotating wheel, a wide range of laser intensity attenuation can be achieved. This is because the laser energy used in water-guided laser processing is high, and direct irradiation of the CCD camera would cause severe overexposure and saturation of the image, with all pixel gray values ​​reaching their maximum values, making it impossible to identify the outline details of the light spot and nozzle. The adjustable attenuator assembly can automatically adjust the attenuation level according to the current laser power, ensuring that the gray values ​​of the light spot and nozzle in the acquired coaxial image are within the linear response range of the CCD camera, providing high-quality raw data for subsequent image preprocessing and center coordinate extraction.

[0024] Step 102, dark field correction processing, is a preprocessing step specifically designed to address the inherent dark current noise of CCD cameras. The imaging principle of a CCD camera is to convert light signals into electrical signals through photodiodes. Even in the absence of any light input, photodiodes will generate a weak current due to thermal excitation, i.e., dark current. This current will form a fixed background noise in the image, and the noise intensity will increase with the extension of camera working time and the increase of ambient temperature. The specific implementation of dark field correction is as follows: During the device power-on initialization phase, the laser light source is turned off and the camera lens is blocked, and a completely black dark field image is acquired. The gray value of each pixel in this image is the dark current noise value at the corresponding position. When acquiring coaxial images subsequently, the gray value of each pixel in the acquired original coaxial image is subtracted from the gray value of the corresponding pixel in the dark field image, which can completely eliminate the fixed dark current noise of the CCD camera. If the dark field correction step is omitted, the dark current noise will be superimposed on the image of the light spot and nozzle, resulting in distorted gray distribution of the light spot and blurred nozzle edges, ultimately significantly increasing the center coordinate extraction error and failing to meet the sub-pixel level positioning requirements.

[0025] Step 103, median filtering, is a filtering step specifically designed for salt-and-pepper noise in industrial environments. Water-guided laser processing environments involve a large amount of metal dust, cutting fluid splashes, and electromagnetic interference. These factors cause randomly distributed bright and dark spots, i.e., salt-and-pepper noise, to appear in the acquired coaxial image. The specific implementation of median filtering is as follows: a sliding window is used to traverse the entire image, and the gray values ​​of the pixels within the window are sorted. The median value is taken as the new gray value of the center pixel of the window. Compared with commonly used mean filtering, median filtering has two significant advantages: first, it can effectively remove salt-and-pepper noise while completely preserving the edge information of the image, preventing blurring of the edges of the laser spot and nozzle; second, it has a strong ability to suppress isolated noise points. Even if a certain proportion of salt-and-pepper noise points exist in the image, a good filtering effect can still be obtained. If mean filtering is used, the gray values ​​of the noise points will be averaged to the surrounding pixels, causing the edges of the laser spot and nozzle to become blurred, seriously affecting the subsequent edge extraction and center positioning accuracy.

[0026] Step 104, adaptive histogram equalization preprocessing, aims to address the problem of insufficient contrast in coaxial images. During water-guided laser processing, interference such as reflections and water films on the nozzle surface causes the acquired image to be generally dark, with small grayscale differences between the spot area and the background area, and between the nozzle edge and the background area, making them difficult to distinguish. This invention employs a contrast-limited adaptive histogram equalization algorithm. Specifically, the entire image is divided into multiple equally sized sub-blocks, and histogram equalization is performed on each sub-block separately. Simultaneously, the contrast enhancement amplitude of each sub-block is limited to avoid excessive enhancement of background noise. Compared to ordinary global histogram equalization, this algorithm can adaptively enhance the contrast of local image areas while effectively suppressing noise amplification, resulting in a clearer grayscale distribution of the spot and sharper nozzle edges. The first image obtained after this step shows significantly improved distinction between the spot area and the background area, and between the nozzle edge and the background area, providing high-quality input data for subsequent center coordinate extraction.

[0027] Step 105, obtaining the first and second center coordinates, is the core foundational step of the entire control process. The accuracy of coordinate extraction directly determines the accuracy of subsequent target offset vector calculation, thus affecting the control effect of water jet energy distribution. This invention employs two completely different algorithms for center positioning based on the different geometric features of the laser focusing spot and the nozzle inlet face: for laser focusing spots without clear edges, the gray-scale centroid method is used for center positioning; for nozzle inlet faces with standard circular contours, the Hough transform circle detection algorithm is used for center positioning. This dual-algorithm combination fully leverages the advantages of each algorithm, enabling sub-pixel-level positioning accuracy for both the spot and nozzle center simultaneously. If a single algorithm is used to extract both centers simultaneously, the positioning error of one center will be too large, failing to meet the accuracy requirements of this invention.

[0028] Step 106: The offset-light field distribution mapping model is the core mathematical model for achieving uniform control of water jet energy distribution in this invention. This model establishes a quantitative mapping relationship between the spot offset and the uniformity of energy distribution across the water jet outlet cross section through pre-tested calibration. Due to different nozzle specifications, water jet pressures, and laser parameters, the transmission characteristics of the laser in the water jet will change, thus altering the optimal offset. Therefore, this model needs to be recalibrated after the equipment is used for the first time, after nozzle replacement, or after adjusting water jet pressure or laser parameters. The specific process for calculating the target offset vector is as follows: the difference between the currently acquired first center coordinates and second center coordinates is used to obtain the current offset vector of the spot relative to the nozzle center; the current offset vector is input into the pre-calibrated offset-light field distribution mapping model to calculate the water jet energy uniformity corresponding to the current offset position; then, the position that minimizes the uniformity index is searched in the model using numerical optimization methods. The offset vector corresponding to this position is the target offset vector. This step addresses the problem of uneven energy distribution by starting from the coupling link between the laser and the water jet, without introducing complex beam shaping devices or sacrificing processing efficiency.

[0029] Step 107: The calibration mapping matrix is ​​a key parameter for establishing a linear transformation relationship between the image pixel coordinate system and the beam pointing adjustment mechanism angle coordinate system. In this invention, the target offset vector is calculated based on the image pixel coordinates, and its unit is pixels. The beam pointing adjustment mechanism controls the position of the light spot by deflection angle, and its input control quantity is an angle value. Therefore, the pixel offset must be converted into an angle control quantity through the calibration mapping matrix. The calibration process of the calibration mapping matrix is ​​as follows: multiple sampling points are uniformly selected within the entire stroke range of the beam pointing adjustment mechanism; a known preset angle control quantity is applied to each sampling point to drive the beam pointing adjustment mechanism to rotate, and then the light spot image is acquired through a coaxial vision sensor to calculate the coordinates of the light spot in the image pixel coordinate system; the preset angle control quantities of all sampling points and the corresponding pixel coordinates are combined to form a calibration dataset, and the dataset is solved using the least squares method to obtain the optimal calibration mapping matrix. The least squares method can ensure that the conversion accuracy from pixel offset to angle control quantity meets the requirements of high-precision offset preset throughout the entire adjustment range by minimizing the sum of squared conversion errors of all sampling points.

[0030] Step 108: In this invention, a two-dimensional fast reflector driven by piezoelectric ceramics is used as the beam pointing adjustment mechanism. This mechanism has extremely high angular resolution and response speed. After receiving the angle control quantity, the light field adjustment controller converts it into a corresponding analog voltage signal and outputs it to the piezoelectric ceramic driver of the fast reflector. Under the action of voltage, the piezoelectric ceramics undergo micron-level deformation, driving the reflector to produce precise two-dimensional deflection, thereby changing the transmission direction of the laser beam and causing the focused spot to translate at the nozzle inlet end face to the target offset position. When the spot moves to the target offset position, the laser beam is no longer coaxially coupled. Instead of being directly coupled into the water jet, the light beams are eccentrically coupled into the water jet, which reduces the proportion of meridional rays and increases the proportion of oblique rays incident into the water jet. The energy distribution of the water jet cross section changes from a centrally concentrated Gaussian distribution to a uniform distribution with sufficient energy at the edges. This fundamentally solves the problem of uneven processing caused by insufficient edge energy in traditional water-guided laser processing. The entire control process, from image acquisition to spot offset preset, takes very little time and can be automatically completed during workpiece material change intervals or processing path jump intervals in water-guided laser processing. It will not occupy normal processing time and will not affect the overall processing cycle.

[0031] Furthermore, the step of obtaining the first center coordinates of the laser focused spot based on the first image includes: obtaining the first center coordinates of the laser focused spot based on the first image using the gray-scale centroid method.

[0032] In the specific implementation process, it is worth noting that the gray-scale centroid method is chosen as the algorithm for locating the center of the laser focusing spot in this step. This is determined by the inherent optical characteristics of the laser focusing spot. The gray-scale distribution of the laser focusing spot exhibits a typical Gaussian-like distribution, that is, the gray-scale value is the highest in the central area and gradually decreases towards the edge area, without a clear and sharp edge contour. If an edge detection-based center positioning algorithm is used, it will be difficult to accurately extract the effective edge of the spot, resulting in a significant increase in center positioning error, which cannot meet the high-precision control requirements of this invention. The gray-scale centroid method makes full use of all the gray-scale information of the spot, using the gray-scale value of each pixel in the spot area as a weight to calculate the centroid of the entire spot area. This can effectively avoid the positioning error caused by edge blurring, and at the same time, it has the characteristics of low computational complexity and good real-time performance, which can meet the real-time requirements of the control process of this invention.

[0033] Furthermore, the first image obtained after the aforementioned dark field correction, median filtering, and adaptive histogram equalization preprocessing has effectively eliminated the dark current noise of the CCD camera and the salt-and-pepper noise of the industrial site, and significantly enhanced the contrast between the spot area and the background area, further improving the positioning accuracy of the gray-scale centroid method, so that the positioning accuracy of the spot center coordinates can reach the sub-pixel level.

[0034] Furthermore, the step of obtaining the first center coordinates of the laser focused spot based on the first image using the gray-scale centroid method includes: determining the segmentation threshold of the laser focused spot in the first image using the Otsu method; extracting the spot region based on the first image and the segmentation threshold; calculating the first center coordinates within the spot region using pixel gray-scale values ​​as weights; and performing sliding median filtering on the first center coordinates of the first center coordinates of multiple consecutively captured first images.

[0035] In the specific implementation process, it is worth noting that this step breaks down the gray-scale centroid method into four sequentially executed sub-steps, forming a complete spot center localization process of "determining the segmentation threshold, extracting the spot area, calculating the first center coordinates with pixel gray values ​​as weights, and performing sliding median filtering on the first center coordinates of multiple consecutive frames". Each sub-step is designed to address specific technical issues in the spot localization process, ensuring the accuracy and stability of the localization results.

[0036] The sub-step of determining the segmentation threshold using the Otsu method is chosen as the threshold segmentation algorithm for the spot region in this invention because the Otsu method is an adaptive threshold segmentation algorithm based on maximizing inter-class variance. It can automatically calculate the optimal segmentation threshold according to the gray-level distribution of the image, without the need for manually setting a fixed threshold in advance. In the actual application of water-guided laser processing, the laser power will be adjusted according to different processing materials and processes, and the ambient lighting conditions will also change over time, causing large fluctuations in the overall gray-level value of the spot image. If a manually set fixed threshold is used, when the laser power or ambient lighting changes, the spot region may be incompletely extracted or the extracted region may contain a lot of background noise, which seriously affects the accuracy of subsequent centroid calculation. The Otsu method can automatically adapt to these changes and always accurately separate the spot region from the background region, ensuring the consistency and reliability of spot region extraction under different working conditions.

[0037] The sub-step of extracting the spot region involves determining the segmentation threshold, then identifying pixels in the first image with gray values ​​greater than or equal to the threshold as pixels in the spot region, and pixels with gray values ​​less than the threshold as pixels in the background region. This allows for the complete extraction of the spot region from the first image. This step only performs subsequent centroid calculations within the extracted spot region, completely eliminating the interference of noise points in the background region on the centroid calculation results. If the centroid is calculated directly on the entire image, noise points in the background region will be used as weights in the calculation, causing the calculated centroid position to deviate from the true energy center of the spot, resulting in a large positioning error.

[0038] The sub-step of calculating the first center coordinates using pixel grayscale values ​​as weights uses grayscale values ​​as weights to calculate the centroid, rather than simply calculating the geometric center of the spot area. This is determined by the energy distribution characteristics of the laser focused spot. The grayscale value of the laser focused spot is positively correlated with the laser energy density at that location. The higher the grayscale value, the greater the laser energy density, and the greater the contribution to the energy center of the spot. The centroid calculated using pixel grayscale values ​​as weights is essentially the energy center of the spot, not the geometric center. The core objective of this invention is to achieve uniform processing by controlling the distribution of laser energy within the water jet. Therefore, only by accurately locating the energy center of the spot can the accuracy of subsequent offset calculations be guaranteed, thereby achieving the expected energy control effect.

[0039] The sub-step of performing sliding median filtering on the first center coordinates of multiple consecutive frames is to eliminate centroid coordinate jitter caused by random fluctuations in laser power. In actual operation, the laser output power will have slight random fluctuations, which will cause subtle changes in the grayscale distribution of the light spot, thus causing random jitter in the calculated first center coordinates. Sliding median filtering effectively filters out such random fluctuations by sorting the first center coordinates of multiple consecutively captured images and taking the median value after sorting as the output coordinate of the current frame, ensuring that the output first center coordinates are stable and reliable. If this filtering step is omitted, the unstable centroid coordinates will cause the subsequently calculated target offset vector to change frequently, causing the beam pointing adjustment mechanism to produce unnecessary frequent movements. This will not only reduce the service life of the mechanism, but also cause the water beam energy distribution to be unstable, affecting the processing quality.

[0040] Furthermore, the step of obtaining the second center coordinates of the nozzle inlet end face based on the first image includes: obtaining the second center coordinates of the nozzle inlet end face based on the first image using the Hough transform circle detection algorithm.

[0041] In the specific implementation process, it is worth noting that the Hough transform circle detection algorithm is selected as the core algorithm for nozzle inlet end face center positioning in this step. This algorithm perfectly matches the inherent geometric characteristics of the nozzle inlet end face and the actual industrial working conditions of water-guided laser processing. The nozzle inlet end face is a standard circular mechanical structure with high processing accuracy and a clear, continuous circular edge contour. This is the premise for using the Hough transform circle detection algorithm. Compared with other circle detection algorithms, the Hough transform circle detection algorithm has irreplaceable anti-interference advantages. There are many interference factors in the water-guided laser processing site, such as metal dust adhesion, cutting fluid splash residue, nozzle surface reflection, and slight wear and scratches caused by long-term use. These factors can cause local breaks, gray-scale abrupt changes, or false edges in the collected nozzle edges. Ordinary least squares circle fitting algorithms have extremely high requirements for edge integrity. Even if there are a few missing or interference points on the edge, a large fitting error will occur. The Hough transform circle detection algorithm, based on the parameter space voting principle, maps each edge point in the image space to the parameter space for cumulative statistics. The parameter with the highest vote is the most likely circle parameter, which can effectively tolerate the effects of edge breaks, local noise, and false edges. It can still stably detect the circular outline of the nozzle in complex industrial environments. At the same time, the first image obtained after the aforementioned dark field correction, median filtering, and adaptive histogram equalization preprocessing has effectively eliminated various noise interferences and significantly enhanced the contrast between the nozzle edge and the background area, providing high-quality input data for the Hough transform circle detection algorithm and further improving the accuracy and reliability of the detection results. The second center coordinates obtained in this step are the reference origin of the entire offset preset process. The subsequent calculation of the target offset vector and the control of the beam pointing adjustment mechanism are all based on this coordinate. Therefore, its positioning accuracy directly determines the energy regulation effect of the entire system.

[0042] Furthermore, the step of obtaining the second center coordinates of the nozzle inlet end face based on the first image using the Hough transform circle detection algorithm includes: using the Cannibal operator to extract the edge of the nozzle inlet end face region in the first image; obtaining the point set of the edge based on the edge, performing the Hough transform circle detection algorithm on the point set of the edge to detect the initial circle center and radius; and obtaining the second center coordinates based on the initial circle center and radius using least squares circle fitting.

[0043] In the specific implementation process, it is worth noting that this step decomposes the Hough transform circle detection algorithm into three progressive sub-steps, forming a three-level nozzle center positioning process: "extracting the edge of the nozzle inlet end face region in the first image using the Canni operator, detecting the initial circle center radius, and obtaining the second center coordinates using least squares circle fitting". This process retains the strong anti-interference characteristics of the Hough transform algorithm and solves the problem of insufficient positioning accuracy when used alone, achieving sub-pixel level accurate positioning of the nozzle center in complex industrial environments.

[0044] The sub-step of extracting the edge of the nozzle inlet end face region in the first image using the Canni operator is chosen because it is currently the optimal edge detection algorithm in the field of computer vision that can simultaneously meet the three core requirements of low false detection rate, high positioning accuracy, and single edge response. The Canni operator can extract real edges that are single-pixel wide, continuous, and accurately positioned, effectively filtering out noise points, reflection points, and false edges in the image, providing a high-quality edge point set for subsequent Hough transform circle detection. If other edge detection algorithms, such as the Sobel operator and the Robert operator, are used, the extracted edges often have problems such as being multi-pixel wide, discontinuous, and having large positioning deviations, which will lead to a large number of false circles in subsequent Hough transform circle detection, making it impossible to accurately identify the true contour of the nozzle.

[0045] For the point set obtained based on edge acquisition, the Hough transform circle detection algorithm is applied to the point set of the edge to detect the initial circle center and radius. This step achieves coarse localization of the nozzle's circular contour through Hough transform circle detection. The core principle of the Hough transform circle detection algorithm is to map each edge point in the image space to a conical surface in the parameter space, and then accumulate votes in the parameter space. The parameter point with the highest vote is the most likely initial circle center and radius. This algorithm has a strong tolerance for edge breaks, local occlusion, and noise points. Even if there is a small amount of wear, scratches, or impurities attached to the nozzle edge, it can accurately detect the approximate circle center and radius of the nozzle. However, the initial circle center and radius obtained by Hough transform circle detection can only achieve pixel-level accuracy, which cannot meet the high-precision requirements of the offset preset in this invention. Therefore, subsequent least squares circle fitting is required to improve the accuracy.

[0046] For the sub-step of obtaining the second center coordinates based on the initial circle center and radius using least squares circle fitting, this step achieves sub-pixel-level precise positioning of the nozzle center through least squares circle fitting, building upon the coarse positioning using Hough transform. Specifically, using the initial circle center obtained from the Hough transform as the center and the initial radius as the reference, effective edge points within a certain range are selected. Then, by minimizing the sum of the squared distances from all effective edge points to the fitted circle, the optimal circle center coordinates and radius are obtained. The least squares circle fitting algorithm can fully utilize the information of all effective edge points, improving the positioning accuracy of the circle center coordinates from the pixel level to the sub-pixel level. The second center coordinates obtained in this step are the reference origin for the entire offset preset process. Subsequent calculations of the target offset vector and control actions of the beam pointing adjustment mechanism are all based on this coordinate as the absolute reference. Its sub-pixel-level positioning accuracy is a necessary prerequisite for achieving precise control of water jet energy distribution.

[0047] Furthermore, such as Figure 3As shown, the calibration process of the offset-light field distribution mapping model is as follows: taking the second center coordinate as the origin, select multiple different offsets within the nozzle inlet end face; based on the multiple different offsets, adjust the laser focusing spot to multiple offset positions corresponding to the multiple different offsets; at the multiple offset positions, use a beam analyzer to measure multiple two-dimensional energy distributions of the water jet outlet cross section; based on the calculation of the multiple two-dimensional energy distributions, obtain multiple uniformity indices; establish the mapping relationship between the multiple different offsets and the multiple uniformity indices; based on the mapping relationship, obtain the target offset vector through surface fitting; the uniformity index is the ratio of the standard deviation to the mean of the energy distribution within the effective area of ​​the water jet cross section.

[0048] In the specific implementation process, it is particularly worth noting that this step fully defines the calibration process of the offset-light field distribution mapping model. This model is the core mathematical foundation for the active control of water beam energy distribution in this invention, and its calibration accuracy directly determines the final effect of subsequent energy homogenization. The entire calibration process follows the logical sequence of "selecting multiple different offsets, corresponding offset positions, measuring the two-dimensional energy distribution of the water beam outlet cross section, calculating multiple uniformity indices, establishing the mapping relationship between offsets and uniformity indices, and obtaining the target offset vector through surface fitting." Each step provides the necessary input for subsequent steps, ensuring that the model can accurately reflect the quantitative relationship between the spot offset and the water beam energy distribution.

[0049] The present invention selects multiple offsets with the second center coordinate as the origin. The second center coordinate (i.e., the true center of the nozzle inlet end face) obtained in the previous steps is used as the origin of the calibration coordinates. This fundamentally ensures that all offsets are defined relative to the true center of the nozzle, avoiding the overall model offset caused by the reference deviation. When selecting offsets in the nozzle inlet end face, a polar coordinate grid is uniformly distributed. Sampling points are uniformly selected at different polar angles and polar diameters with the origin as the center, covering the effective coupling area of ​​the entire nozzle inlet. This selection method can ensure that the sampling points are uniformly distributed in the entire plane and will not miss any area where the optimal offset may exist. This provides comprehensive and reliable basic data for subsequent surface fitting. It should be noted that this model is not universal and immutable. When the equipment is changed to different specifications of nozzles, the water jet pressure is adjusted, or the laser parameters are changed, the transmission characteristics of the laser in the water jet will change, and the corresponding optimal offset will also change. Therefore, this calibration process must be repeated to establish a new mapping model.

[0050] For the sub-step of adjusting the laser focusing spot to the corresponding offset position based on multiple different offsets, this step achieves precise control of the spot position through a beam pointing adjustment mechanism. In specific implementation, each offset to be calibrated is input into a pre-calibrated calibration mapping matrix, which is converted into an angle control quantity of the beam pointing adjustment mechanism. Then, the light field adjustment controller drives the adjustment mechanism to move the spot precisely to the target offset position. This process ensures that the spot position of each calibration sampling point is completely consistent with the preset offset, eliminating the influence of position error on the calibration result.

[0051] For the sub-step of measuring the two-dimensional energy distribution of the water jet outlet cross section using a beam analyzer, this step chooses to measure the energy distribution at the water jet outlet cross section rather than at the nozzle inlet. This is because what we need to control is the energy distribution of the water jet that ultimately acts on the workpiece surface. The beam analyzer can accurately collect the energy density data of each point on the water jet outlet cross section to obtain a complete two-dimensional energy distribution cloud map, providing raw data support for the subsequent calculation of uniformity index.

[0052] For the sub-steps of calculating multiple uniformity indices, this invention uses the ratio of the standard deviation to the mean of the energy distribution within the effective area of ​​the water jet cross-section as the uniformity index. This is an internationally recognized quantitative evaluation method for energy distribution uniformity. The standard deviation reflects the dispersion of the energy distribution, while the mean reflects the overall energy level. The ratio of the two can objectively and accurately measure the uniformity of the energy distribution within the water jet cross-section: the smaller the value, the more uniform the energy distribution; when the value is 0, it indicates that the energy distribution is completely uniform. By calculating the uniformity index corresponding to each offset position, we can transform the abstract energy distribution state into a quantifiable and comparable value.

[0053] The sub-step of establishing the mapping relationship between offset and uniformity index uses the offset of all sampling points as the independent variable and the corresponding uniformity index as the dependent variable to establish a discrete point set mapping relationship. This point set completely records the water jet energy uniformity corresponding to different offset positions in the nozzle inlet plane, which is the basic data for subsequent surface fitting.

[0054] For the sub-step of obtaining the target offset vector through surface fitting, this invention uses a two-dimensional Gaussian surface fitting algorithm to fit the above discrete mapping point set to obtain a continuous surface function. This function can describe the water jet energy uniformity corresponding to the offset at any position in the nozzle inlet plane. Since the smaller the uniformity index, the more uniform the energy distribution, the offset corresponding to the lowest point of this surface function is the target offset vector that optimizes the water jet energy distribution. After fitting, the surface function and the corresponding target offset vector are stored in the system memory and can be directly called in subsequent processing without repeating the calibration process.

[0055] The entire calibration process is essentially about systematically finding the spot position that achieves the optimal ratio of meridional and oblique rays incident on the water jet, as shown in the instruction manual. Figure 3 As shown, when the laser spot offset relative to the nozzle center is 0 μm, the laser beam coaxially couples into the water jet, with a very high proportion of meridional rays. The energy distribution of the water jet outlet cross section exhibits a regular concentric circle-like Gaussian distribution with highly concentrated energy at the center and extremely low energy at the edges. As the laser spot offset increases to 10 μm, the proportion of meridional rays decreases, while the proportion of oblique rays increases significantly. The energy distribution begins to shift from being concentrated at the center to being ring-shaped, and the high-energy region at the center gradually disperses. When the offset increases to 20 μm, the proportion of oblique rays further increases, and the energy distribution of the water jet cross section exhibits a relatively uniform ring-shaped distribution, with significant replenishment of energy at the edges. When the offset continues to increase to 30 μm, the overall energy distribution tends to be flat, but an excessively large offset will cause some laser energy to fail to couple into the water jet, resulting in a decrease in energy utilization. The optimal offset found through calibration is precisely the light ratio balance point that ensures the most uniform energy distribution of the water jet cross section while maintaining energy utilization.

[0056] Furthermore, the calibration mapping matrix is ​​obtained by selecting multiple sampling points within the stroke range of the beam pointing adjustment mechanism; applying a preset control value to each of the multiple sampling points; and solving the calibration mapping matrix using the least squares method based on the first center coordinates.

[0057] In the specific implementation process, it is worth noting that the calibration mapping matrix is ​​the core bridge connecting the visual perception system and the motion execution system. Its core function is to establish a quantitative linear transformation relationship between the image pixel coordinate system and the beam pointing adjustment mechanism angle coordinate system. In this invention, the target offset vector is calculated based on the image acquired by the coaxial vision sensor, and its unit is pixels. The beam pointing adjustment mechanism drives the light spot to produce physical displacement by receiving angle control quantities. The two belong to completely different coordinate systems. The unit and dimension conversion must be completed through the calibration mapping matrix in order to achieve precise control of the light spot position.

[0058] In the sub-step of selecting multiple sampling points within the travel range of the beam pointing adjustment mechanism, this invention adopts a uniformly distributed sampling method to cover the entire effective adjustment travel of the beam pointing adjustment mechanism. This sampling method can ensure that the conversion accuracy of pixel offset to angle control is consistent throughout the entire adjustment range, avoiding the problem of high accuracy in the central area and low accuracy in the edge area. If the sampling points are unevenly distributed or the number of sampling points is insufficient, the conversion error in some areas will increase significantly, making it impossible to achieve accurate spot offset preset. At the same time, the selection of sampling points must avoid the mechanical blind zone and nonlinear region of the adjustment mechanism to ensure the validity and reliability of the sampling data.

[0059] The sub-step of applying a preset control value to each of the multiple sampling points and collecting data based on the first center coordinates constructs a calibration dataset through the correspondence between known inputs and measured outputs. In specific implementation, a preset standard angle control value is applied to each sampling point to drive the beam pointing adjustment mechanism to generate a corresponding deflection, so that the laser focusing spot moves to a specific position on the nozzle inlet end face. Then, the current spot image is acquired through a coaxial vision sensor, and the first center coordinates corresponding to the position are calculated using the aforementioned gray-scale centroid method. By traversing all sampling points, multiple sets of corresponding data pairs of "preset angle control value and measured pixel coordinates" can be obtained. These data pairs constitute the original dataset for solving the calibration mapping matrix. This step strictly uses the first center coordinates as the measured output, ensuring that the calibration data is completely consistent with the coordinate system used in the subsequent processing, and eliminating systematic errors caused by coordinate system differences.

[0060] The sub-step of obtaining the calibration mapping matrix using the least squares method is chosen because it can obtain the globally optimal linear transformation matrix by minimizing the sum of squared transformation errors of all sampling points. The calibration mapping matrix is ​​a linear matrix, corresponding to the transformation coefficients in the X and Y axes respectively. The least squares method can make full use of the information of all sampling points, balance the transformation errors in each region, and minimize the average transformation error over the entire adjustment range. After the solution is obtained, the calibration mapping matrix is ​​stored in the system memory and can be directly called in subsequent processing without repeating the calibration process.

[0061] It is important to note that the calibration mapping matrix is ​​not permanently valid. When the equipment undergoes optical system adjustments, beam pointing adjustment mechanism replacement, maintenance, or relocation, the relative position of the optical system may change slightly, causing the original calibration mapping matrix to become invalid. In this case, the calibration process must be repeated. Furthermore, the calibration accuracy of the calibration mapping matrix directly determines the final accuracy of the beam offset preset, which in turn affects the control effect of the water beam energy distribution. Therefore, the calibration process must be carried out under conditions of stable equipment operation and minimal environmental interference to ensure the accuracy of the calibration results.

[0062] Furthermore, such as Figure 2As shown, the control device 200 for a water-guided laser includes: a data acquisition unit 201, used to acquire a coaxial image of the nozzle inlet end face and the laser focusing spot via a first optical axis using a coaxial vision sensor, wherein the first optical axis is shared by the beam splitter and the water-guided laser optical path; a first image processing unit 202, used to perform dark field correction processing on the coaxial image; a second image processing unit 203, used to perform median filtering processing on the coaxial image after dark field correction processing; a third image processing unit 204, used to perform adaptive histogram equalization preprocessing on the coaxial image after median filtering processing to obtain the processed first image; and a coordinate acquisition unit 205, used for... In the first image, the first center coordinates of the laser focused spot and the second center coordinates of the nozzle inlet end face are obtained; the calculation unit 206 is used to input the second center coordinates and the first center coordinates into the offset-light field distribution mapping model, and calculate the target offset vector of the energy distribution spot relative to the second center coordinates; the control quantity acquisition unit 207 is used to obtain the angle control quantity of the beam pointing adjustment mechanism based on the target offset vector and the calibration mapping matrix; the control unit 208 is used to control the light field adjustment controller to drive the beam pointing adjustment mechanism to move based on the angle control quantity, and preset the center of the laser focused spot to the target position that is offset from the center of the nozzle inlet end face.

[0063] In the specific implementation process, it is worth noting that this control device adopts a modular and hierarchical architecture design, which decomposes the entire water-guided laser offset preset control process into multiple functionally independent and clearly defined units. Each unit strictly corresponds to one or more steps in the aforementioned method. The units communicate with each other through standardized data interfaces, and the data flow is completely consistent with the method execution logic, ensuring the determinism and reliability of the device operation. The entire device forms a complete closed loop of "visual perception, image preprocessing, coordinate positioning, parameter calculation and motion execution", which can automatically complete the entire control process from image acquisition to spot offset preset.

[0064] Acquisition Unit 201: Corresponding to the original sensing unit used in the aforementioned method to acquire the coaxial image of the nozzle inlet end face and the laser focused spot, it is the only data input port of the entire system and is responsible for acquiring high-quality original images that meet the requirements of subsequent processing. Its function corresponds to the content of step 101 in the aforementioned method. This unit achieves the sharing of the first optical axis with the water-guided laser optical path through a beam splitter, thereby eliminating the inherent parallax error of the non-coaxial vision system from the root and ensuring that the image position corresponds one-to-one with the actual laser irradiation position. At the same time, it integrates a CCD camera and an adjustable attenuation plate group, which can automatically adjust the incident light intensity according to the current laser power to avoid image overexposure and saturation caused by high-energy laser, and provide clear and accurate original image data for all subsequent processing steps.

[0065] First image processing unit 202: Corresponding to the first-level preprocessing unit used for dark field correction of coaxial images in the aforementioned method, it is the starting point of the image preprocessing process and is responsible for eliminating the inherent dark current noise of the CCD camera. Its function corresponds to the content of step 102 in the aforementioned method. This unit performs pixel-by-pixel gray value subtraction correction on the real-time acquired coaxial image using the dark field image pre-acquired during the device power-on initialization stage, removes the fixed background noise generated by the thermal excitation of the camera photodiode, avoids noise superposition leading to distortion of the gray distribution of the light spot and blurring of the nozzle edge, and ensures the accuracy of subsequent image processing results.

[0066] The second image processing unit 203 corresponds to the secondary preprocessing unit used in the aforementioned method to perform median filtering on the coaxial image after dark field correction. It is an intermediate step in the image preprocessing process and is responsible for eliminating salt-and-pepper noise generated by the industrial environment. Its function corresponds to the content of step 103 in the aforementioned method. This unit adopts a sliding window median filtering algorithm, which can effectively remove isolated bright spots and dark spots caused by metal dust adhesion, cutting fluid splashing and electromagnetic interference, while completely preserving the edge information of the light spot and nozzle, without causing edge blurring, thus laying the foundation for subsequent edge extraction and center positioning.

[0067] The third image processing unit 204 corresponds to the three-level preprocessing unit used in the aforementioned method for adaptive histogram equalization preprocessing of the coaxial image after median filtering. It is the final step in the image preprocessing process and is responsible for solving the problem of insufficient image contrast. Its function corresponds to the content of step 104 in the aforementioned method. This unit adopts a contrast-limited adaptive histogram equalization algorithm, which can adaptively enhance the contrast of local areas of the image while suppressing the excessive amplification of background noise. This significantly improves the distinction between the spot area and the background area, and between the nozzle edge and the background area, and finally outputs a high-quality first image that meets the center positioning requirements.

[0068] Coordinate acquisition unit 205: Corresponding to the core positioning unit used in the aforementioned method to acquire the first center coordinates of the laser focusing spot and the second center coordinates of the nozzle inlet end face, it is the reference generation module of the entire system. It is responsible for outputting high-precision coordinates of the two centers as the reference for all subsequent calculations. Its function corresponds to the content of step 105 in the aforementioned method. This unit adopts a dual-algorithm combination strategy for different geometric features of the spot and the nozzle: for laser focusing spots without clear sharp edges and with a Gaussian-like gray distribution, the gray centroid method is used to locate the energy center; for nozzle inlet end faces with standard circular contours, the Hough transform circle detection algorithm is used to locate the geometric center. At the same time, it achieves sub-pixel-level positioning accuracy for the two centers, providing reliable reference data for subsequent offset calculations.

[0069] Solution Unit 206: Corresponding to the core decision unit used to solve the target offset vector in the aforementioned method, it is the core module for achieving uniform control of water jet energy distribution in this invention. It is responsible for calculating the target offset that optimizes the water jet energy distribution, and its function corresponds to step 106 in the aforementioned method. This unit stores a pre-calibrated offset-light field distribution mapping model. After receiving the first and second center coordinates output by the coordinate acquisition unit, it calculates the water jet energy uniformity corresponding to the current offset position and searches for the target offset vector that minimizes the uniformity index through numerical optimization. This unit starts from the coupling link between the laser and the water jet, and changes the ratio of meridional rays to oblique rays incident on the water jet by adjusting the spot offset, thereby achieving active control of the water jet energy distribution.

[0070] Control quantity acquisition unit 207: Corresponding to the coordinate transformation unit used in the aforementioned method to acquire the angle control quantity of the beam pointing adjustment mechanism, it is the core bridge connecting the visual perception system and the motion execution system. It is responsible for converting the offset in the pixel coordinate system into an angle control quantity that the actuator can recognize. Its function corresponds to the content of step 107 in the aforementioned method. This unit stores a pre-calibrated calibration mapping matrix. After receiving the target offset vector output by the solution unit, it converts the pixel-unit offset into an angle-unit control quantity through linear transformation, thereby achieving accurate mapping between the visual coordinate system and the actuator coordinate system.

[0071] Control Unit 208: Corresponding to the final execution unit used to drive the beam pointing adjustment mechanism in the aforementioned method, it is the only output port of the entire system and is responsible for accurately presetting the beam spot position. Its function corresponds to the content of step 108 in the aforementioned method. After receiving the angle control quantity output by the control quantity acquisition unit, this unit converts it into a corresponding drive signal and outputs it to the light field adjustment controller, thereby driving the beam pointing adjustment mechanism to produce a precise deflection, causing the laser focusing beam spot to move to the target position deviating from the center of the nozzle, and finally realizing the transformation of the energy distribution of the water jet cross-section from a centrally concentrated Gaussian distribution to a uniform distribution with sufficient energy at the edges.

[0072] Furthermore, the water-guided laser control device 200 also includes a memory 210 and a processor 209. The memory 210 stores programs or instructions that can be executed on the processor 209. When the program or instructions are executed by the processor, they implement the steps of the vision-guided water-guided laser processing offset preset method as described in any of the above embodiments.

[0073] In the specific implementation process, it is worth noting that this embodiment provides a general hardware implementation of the control device. Through the collaborative work of the processor and memory, it fully carries the execution logic of all the aforementioned method steps, so that the technical solution of the present invention can be applied in the form of a standardized hardware platform.

[0074] Processor 209: Corresponding to the core computing unit mentioned above that executes all computation and control logic, it is the command center of the entire control device, responsible for scheduling all functional units to complete the complete offset preset control process. Its functions cover all computation and control operations from step 101 to step 108 in the aforementioned method, including driving the acquisition unit to acquire coaxial images, scheduling three image processing units to complete image preprocessing in sequence, controlling the coordinate acquisition unit to extract two center coordinates, the instruction solving unit to solve the target offset vector, the control quantity acquisition unit to generate angle control quantity, and outputting drive signals to the control unit to complete the spot offset preset. The processor has sufficient computing power to complete image preprocessing, sub-pixel level coordinate extraction, and numerical optimization calculation in real time, ensuring that the entire control process can be completed quickly within the processing interval without affecting the normal processing cycle.

[0075] Memory 210: Corresponding to the aforementioned storage unit used to store all programs and data, it is the data warehouse of the control device. Its storage contents are divided into three categories: The first category is the programs or instructions that implement the method of the present invention, including the execution code of all core algorithms such as image preprocessing algorithm, gray centroid method, Hough transform circle detection algorithm, and offset calculation algorithm; the second category is the pre-calibration data required for system operation, including offset-light field distribution mapping model and calibration mapping matrix. These data are the core foundation for realizing energy distribution regulation and can be updated according to changes in equipment operating conditions; the third category is temporary data generated during system operation, including acquired coaxial images, intermediate calculation results, operation logs, etc. The memory usually contains two parts: non-volatile storage and volatile storage. Non-volatile storage is used to store programs and pre-calibration data for a long time, while volatile storage is used to temporarily store data during operation.

[0076] When the processor 209 executes the program or instructions stored in the memory 210, it can fully implement all the steps of the vision-guided water-guided laser processing offset preset method of any of the above embodiments. Therefore, the water-guided laser control device 200 has all the beneficial effects of the vision-guided water-guided laser processing offset preset method of any of the above embodiments.

[0077] Furthermore, a program or instructions are stored on a readable storage medium, which, when executed by a processor, implement the steps of the vision-guided water-guided laser processing offset preset method as described in any of the above embodiments.

[0078] In specific implementation, it is particularly noteworthy that this embodiment provides a computer-readable storage medium implementation of the technical solution of the present invention. By embedding the program or instructions for implementing the water-guided laser processing offset preset method in a readable storage medium, the technical solution of the present invention can exist, be distributed, and deployed independently in the form of a software product. A readable storage medium refers to various non-temporary computer storage media capable of storing computer programs or instructions, including but not limited to hard disks, solid-state drives, USB flash drives, flash memory cards, read-only memories, optical discs, and all other physical media that can be read and stored by a computer, excluding temporary transmission media such as carrier waves. The program or instructions stored on the readable storage medium completely encompass all execution logic of the vision-guided water-guided laser processing offset preset method of any of the above embodiments, covering coaxial image acquisition. The complete code for three-level image preprocessing, dual-algorithm center coordinate extraction, target offset vector calculation, angle control quantity conversion, and beam offset preset execution, implemented using a readable storage medium, offers significant advantages: First, it facilitates standardized software distribution and batch deployment. Users do not need to write their own code; they only need to load the program or instructions from the storage medium into the control device of the water-guided laser equipment. Second, it facilitates software upgrades and maintenance. When it is necessary to optimize algorithm accuracy, adapt to new nozzles, or add new functions, only the program or instructions in the storage medium need to be updated, without any changes to the hardware structure of the equipment, greatly reducing the system's upgrade costs and maintenance difficulty. Third, it improves software portability. The same program or instructions can run on different models and configurations of water-guided laser control devices, demonstrating broad applicability.

[0079] When the program or instructions stored on the readable storage medium are executed by the processor, all the steps of the vision-guided water-guided laser processing offset preset method of any of the above embodiments can be fully implemented. Therefore, the readable storage medium has all the beneficial effects of the vision-guided water-guided laser processing offset preset method of any of the above embodiments.

[0080] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for controlling a water-guided laser, characterized in that, Includes the following steps: Using a coaxial vision sensor, a coaxial image of the nozzle inlet end face and the laser focused spot is acquired through the first optical axis. The first optical axis is the shared optical axis of the beam splitter and the water-guided laser optical path. The coaxial image is subjected to dark field correction processing; The coaxial image after dark field correction is then subjected to median filtering. The coaxial image after median filtering is preprocessed with adaptive histogram equalization to obtain the first processed image. Based on the first image, obtain the first center coordinates of the laser focused spot and the second center coordinates of the nozzle inlet end face; Input the second center coordinates and the first center coordinates into the offset-light field distribution mapping model, and calculate the target offset vector of the light spot relative to the second center coordinates; Based on the target offset vector and calibration mapping matrix, the angle control value of the beam pointing adjustment mechanism is obtained; Based on the angle control amount, the light field adjustment controller drives the beam pointing adjustment mechanism to move, and presets the center of the laser focusing spot to a target position that is offset from the center of the nozzle inlet end face.

2. The control method for water-guided lasers according to claim 1, characterized in that, The step of obtaining the first center coordinates of the laser focused spot based on the first image includes: Based on the first image, the first center coordinates of the laser focused spot are obtained using the gray-scale centroid method.

3. The control method for water-guided lasers according to claim 2, characterized in that, The step of obtaining the first center coordinates of the laser focused spot based on the first image using the gray-scale centroid method includes: The Otsu method is used to determine the segmentation threshold for the laser focused spot in the first image; Based on the first image and the segmentation threshold, extract the spot region; Within the light spot area, the first center coordinates are calculated using pixel grayscale values ​​as weights; A sliding median filter is applied to the first center coordinates of multiple consecutively captured frames of the first image.

4. The control method for water-guided lasers according to claim 1, characterized in that, The step of obtaining the second center coordinates of the nozzle inlet end face based on the first image includes: Based on the first image, the second center coordinates of the nozzle inlet end face are obtained by using the Hough transform circle detection algorithm.

5. The control method for water-guided lasers according to claim 4, characterized in that, The step of obtaining the second center coordinates of the nozzle inlet end face using the Hough transform circle detection algorithm based on the first image includes: The edge of the nozzle inlet end face region in the first image is extracted using the Cannibal operator; Based on the edge, obtain the point set of the edge, and perform the Hough transform circle detection algorithm on the point set of the edge to detect the initial circle center and radius; Based on the initial center and the radius, the second center coordinates are obtained by least-squares circle fitting.

6. The control method for water-guided lasers according to claim 1, characterized in that, The calibration process of the offset-light field distribution mapping model is as follows: Using the second center coordinate as the origin, select multiple different offsets within the nozzle inlet end face; Based on multiple different offsets, the laser focused spot is adjusted to multiple offset positions corresponding to the multiple different offsets; At multiple offset positions, a beam analyzer was used to measure multiple two-dimensional energy distributions at the cross-section of the water jet outlet. Based on the calculation of multiple two-dimensional energy distributions, multiple uniformity indices are obtained; Establish a mapping relationship between multiple different offsets and multiple uniformity indices; Based on the mapping relationship, the target offset vector is obtained through surface fitting; The uniformity index is the ratio of the standard deviation to the mean of the energy distribution within the effective area of ​​the water jet cross section.

7. The control method for water-guided lasers according to claim 1, characterized in that, The calibration mapping matrix is ​​obtained in the following way: Multiple sampling points are selected within the travel range of the beam pointing adjustment mechanism; A preset control value is applied to each of the plurality of sampling points, and the calibration mapping matrix is ​​obtained by using the least squares method based on the first center coordinates.

8. A control device for a water-guided laser, characterized in that, include: The acquisition unit is used to acquire a coaxial image of the nozzle inlet end face and the laser focused spot through a first optical axis using a coaxial vision sensor. The first optical axis is a shared optical axis between the beam splitter and the water-guided laser optical path. The first image processing unit is used to perform dark field correction processing on the coaxial image; The second image processing unit is used to perform median filtering on the coaxial image after dark field correction processing; The third image processing unit is used to perform adaptive histogram equalization preprocessing on the coaxial image after median filtering to obtain the processed first image. The coordinate acquisition unit is used to acquire the first center coordinates of the laser focusing spot and the second center coordinates of the nozzle inlet end face based on the first image; The calculation unit is used to input the second center coordinates and the first center coordinates into the offset-light field distribution mapping model, and calculate the target offset vector of the light spot with energy distribution relative to the second center coordinates; The control quantity acquisition unit is used to acquire the angle control quantity of the beam pointing adjustment mechanism based on the target offset vector and the calibration mapping matrix. The control unit is used to control the light field adjustment controller to drive the beam pointing adjustment mechanism to move based on the angle control amount, so as to preset the center of the laser focusing spot to a target position that is offset from the center of the nozzle inlet end face.

9. A control device for a water-guided laser, characterized in that, Also includes: A memory and a processor, the memory storing a program or instructions executable on the processor, the program or instructions, when executed by the processor, implementing the steps of the vision-guided water-guided laser processing offset preset method as described in any one of claims 1 to 7.

10. A readable storage medium having a program or instructions stored thereon, characterized in that, When the program or the instructions are executed by the processor, they implement the steps of the vision-guided water-guided laser processing offset preset method as described in any one of claims 1 to 7.

Images