Large aperture kdp crystal micro-defect rapid detection and positioning method based on variable magnification ring multiple transformation

Abstract

The application relates to a large-aperture KDP crystal micro-defect rapid detection and positioning method based on variable magnification ring magnification conversion, and relates to the technical field of optical engineering, and aims to solve the problems of low scanning efficiency, low image imaging quality and easy image missing of the existing method. The application adopts a feature point matching method based on a simulated CMT tracking algorithm to determine the conversion coefficients under each variable magnification ring magnification, and further calculates the actual field of view range of the scanning microscope under different magnifications; through edge-seeking motion, the edge image of the element is extracted to determine the element center coordinate value, and the crystal global coordinate conversion is realized; based on the low-magnification scanning microscope, a grid format reciprocating scanning strategy is adopted to scan the micro-defects on the surface of the whole crystal element, and the micro-defects on the surface of the element are positioned based on the medium-magnification scanning microscope; during the scanning and positioning of the micro-defects on the surface of the crystal element, an image processing method based on a gradient algorithm is adopted, and finally the position, shape and size information of each micro-defect are obtained quickly and accurately.

Description

Technical Field

[0001] This invention relates to the field of optical engineering technology, and more specifically, to a rapid detection and location method for micro-defects in large-aperture KDP crystals based on variable magnification ring multiplication. Background Technology

[0002] With the rapid development of human society and the depletion of fossil fuels such as oil, the world is facing an energy crisis caused by the shortage of fossil fuels, making the development of new clean energy sources imperative. Nuclear fusion energy, due to its advantages such as high energy production, abundant reactor raw material reserves, and low hazard of radioactive waste, is currently recognized as the most ideal future energy source. Inertial confinement fusion, as the main method for achieving controlled nuclear fusion, has been listed as a key research area by various countries, such as the National Ignition Facility in the United States and the Shenguang Facility in China. In laser fusion engineering, tens of thousands of large-aperture, high-precision, and high-surface-quality optical elements are required to process the input laser. Among them, KDP crystals, due to their unique optical properties, have become an irreplaceable core component in current laser fusion devices. However, KDP crystal elements obtained through ultra-precision machining in engineering are highly susceptible to laser damage (damage points are typically on the order of tens to hundreds of micrometers, hereinafter referred to as micro-defects) under high-intensity laser conditions. If these damage points are not addressed after they occur, they will rapidly expand under subsequent high-intensity laser irradiation, severely affecting their optical performance and lifespan, thus limiting the widespread application of KDP crystals in laser nuclear fusion engineering. Currently, using ball-end milling cutters to repair surface defects in KDP crystal elements is one of the most effective methods to suppress damage point growth and improve the laser load capacity of crystal elements.

[0003] When repairing surface damage and micro-defects in large-diameter KDP crystal elements, it is first necessary to quickly and accurately detect these micro-defects to obtain information such as the location, shape, and size of all micro-defect points to be repaired on the element's surface. However, the characteristics of KDP crystal elements, such as large diameter, small defect size, uneven distribution, and huge number, make it extremely difficult to achieve rapid and accurate detection and location.

[0004] Currently, for the detection of micro-defects on the surface of full-aperture KDP crystal elements, with a crystal size of 410mm × 410mm, a sub-aperture field of view of 3.566mm × 3.156mm, and a movement speed of 5mm / s, the task takes approximately 4 hours. During the scanning process, a 10ms interval is used to check if the difference between the current position and the image position is less than 0.2mm (if less than 0.2mm, the image capture continues). Analysis of current methods for detecting and locating micro-defects on the surface of large-aperture KDP crystal elements based on sub-aperture stitching reveals the following problems:

[0005] (1) The single image scanning sub-aperture is too small, and the full aperture defect scanning efficiency is low, which does not meet the actual engineering needs (the full aperture KDP crystal element surface micro-defect detection process time is within 1.5h).

[0006] (2) The image acquisition is completed while the scanning motion system is in motion, resulting in low image quality and limiting the improvement of scanning motion speed.

[0007] (3) The timer-based position detection method is prone to missing images of the entire column (the timer may be unstable), resulting in incomplete image acquisition;

[0008] (4) The scanning speed is generally selected to be around 5 mm / s, and the timer is refreshed every 10 ms. Under this condition, the actual position of the defect and the detected defect coordinate position have a position error of not less than 150 μm, which greatly reduces the accuracy of defect detection and positioning (the repeatability of the scanning motion axis itself is 3 μm). Summary of the Invention

[0009] The technical problem to be solved by this invention is:

[0010] Existing methods for detecting and locating micro-defects in large-aperture KDP crystals use a constant aperture field of view for scanning, resulting in low scanning efficiency, low image quality, and a tendency to miss images. Furthermore, they cannot eliminate the positional error between the detected defect location and the actual defect location.

[0011] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows:

[0012] This invention provides a rapid detection and location method for micro-defects in large-aperture KDP crystals based on variable magnification ring gamut transformation, comprising the following steps:

[0013] Step 1. Install the optical crystal element to be tested, assemble and adjust the scanning microscope system, automatically return each axis of the crystal element moving mechanism to zero, and establish an absolute coordinate system;

[0014] Step 2. Adjust the magnification of the scanning microscope's zoom rings, use a feature point matching method based on a simulated CMT tracking algorithm to determine the downconversion coefficients for each zoom ring magnification, complete pixel calibration, and establish an intuitive one-click calibration module for the scanning microscope for users;

[0015] Step 3. Based on the conversion coefficients of different zoom ring magnifications of the scanning microscope obtained in Step 2 and the resolution of the scanning microscope, calculate the actual field of view of the scanning microscope at different magnifications;

[0016] Step 4. Control the scanning microscope to perform edge-finding motion and extract the edge image of the component. Determine the coordinate values ​​of the component edge in the absolute coordinate system, further determine the coordinate values ​​of the component center, and establish the workpiece coordinate system of the large-diameter KDP crystal component based on the component center coordinate values ​​to achieve global coordinate transformation of the crystal.

[0017] Step 5. Adjust the magnification of the scanning microscope's zoom ring to medium or low magnification. Select the scanning step distance based on the actual field of view obtained in Step 3. Use a grid-based reciprocating scanning strategy to scan and acquire images of micro-defects on the surface of the entire crystal element. Simultaneously, call the motion controller to control the motion and detect the motion state during the scanning process. Use an image processing method based on gradient algorithms to process the acquired images and determine the position coordinates of the defects in the absolute coordinate system.

[0018] Step 6. Adjust the magnification of the scanning microscope to medium magnification. Based on the position coordinates of the micro-defects on the surface of the component obtained in Step 5, move the defect points into the field of view of the scanning microscope and acquire images of the micro-defects on the surface of the crystal component.

[0019] Step 7. Based on the micro-defect images of the crystal element surface acquired in Step 6, extract the micro-defect feature information using an image processing method based on gradient algorithm to obtain the location, shape, and size information of each micro-defect.

[0020] Furthermore, the scanning microscope system described in step 1 is used to detect defects on the surface of the optical crystal. The scanning microscope uses a JAI BM-500GE CCD with a resolution of 2456×2058, a maximum frame rate of 15fps, a pixel size of 3.45μm×3.45μm, and a magnification range of 0.58X to 7X.

[0021] Furthermore, step 2 includes the following process:

[0022] Adjust the magnification ring of the scanning microscope to the lowest magnification k1, and adjust the field of view of the scanning microscope to focus on the tip of the cutting tool. Adjust the brightness of the ring light source or the exposure time of the scanning microscope to obtain an image that is as clear as possible while preserving rich detail. Move the cutting tool tip to point A and acquire an image using the scanning microscope. Then move the cutting tool tip a distance d along the X2 axis to point B and acquire an image. Calculate the pixel distance Δ of the cutting tool tip's movement in the image and the actual movement distance d of the cutting tool tip using the simulated CMT algorithm, and calculate the conversion coefficient c.

[0023] c = d / Δ

[0024] The above method was used to perform pixel calibration at different zoom ring magnifications of a scanning microscope.

[0025] Furthermore, in step 3, the actual field of view of the scanning microscope at different magnifications is obtained by multiplying the different magnification down-conversion coefficients by the resolution of the scanning microscope.

[0026] Furthermore, when extracting the component edge image in step 4, the crystal edge image is first converted into a grayscale image, then converted into a high-contrast image and Gaussian noise reduction is performed, and then converted into a binarized image in CV_THRESH_BINARY mode to realize the processing of the crystal edge image. Finally, the specific coordinates of the crystal edge in the scanning motion system are calculated based on the pixel calibration value under the current zoom ring magnification and the coordinate value of the scanning motion system.

[0027] Furthermore, in step 5, the scanning step distance and the actual number of feed steps required in the X1 and Y1 directions are determined based on the actual scanning microscope field of view. During motion control, the ComACRsrvr function library provided by the ACR motion controller is used to realize motion control and motion status detection. When the motion stops, the motion status detection is that there is no motion, and the image is acquired. After the image acquisition is completed, the motion command is fed back to the motion axis. This cycle is repeated to complete the rapid detection of the full aperture of the crystal.

[0028] Furthermore, the gradient-based image processing method described in steps 5 and 7 specifically employs the Canny operator and includes the following steps:

[0029] Step 1: Smooth the image using a Gaussian filter;

[0030] Step 2: Calculate the gradient intensity and direction of each pixel in the image using the finite difference of the first-order partial derivative;

[0031] Step 3: Perform non-maximum suppression on the gradient magnitude to eliminate stray effects caused by edge detection;

[0032] Step 4: Use a dual threshold algorithm to determine real and potential edges, suppressing isolated weak edges in the process;

[0033] The dual threshold algorithm sets an upper threshold and a lower threshold. When a pixel value is greater than the upper threshold, it is considered to be a boundary; when it is less than the lower threshold, it is considered not to be a boundary; and when it is in between, it is considered a candidate.

[0034] Compared with the prior art, the beneficial effects of the present invention are:

[0035] This invention presents a rapid detection and location method for micro-defects in large-aperture KDP crystals based on magnification ring scaling. It utilizes scanning microscopes at different magnifications to rapidly detect and locate surface micro-defects in large-aperture KDP crystal elements, and performs pixel calibration on the scanning microscopes at different magnifications to achieve real-time calibration of the microscopic imaging, significantly improving detection efficiency and stability. This invention uses low magnification ring scaling for micro-defect scanning detection, providing a large scanning field of view and improving scanning efficiency while ensuring effective defect identification, thus achieving rapid detection. Using medium magnification ring scaling for defect localization ensures sufficient field of view and location accuracy, enabling precise location of micro-defects.

[0036] This invention employs a grid-based reciprocating scanning trajectory and uses a motion controller to control the motion and detect the motion state during the scanning process, greatly improving the efficiency of micro-defect detection. It also uses an image processing algorithm based on gradient operators to process the micro-defect image on the surface of the crystal element, removing weak edges from the detected defect image and determining the true and potential edges of the detected defects. This avoids filtering out defects because the defect edges are difficult to separate from the background, thereby greatly improving the accuracy and precision of micro-defect detection for large-aperture KDP crystal elements. Attached Figure Description

[0037] Figure 1 This is a schematic diagram of a rapid detection and location device for micro-defects in large-aperture KDP crystals in an embodiment of the present invention;

[0038] Figure 2 The images shown are of the tool positions during automatic calibration of the scanning microscope in this embodiment of the invention, wherein a) is the image at the scanning calibration (2,2), b) is the image at the positioning calibration (1,1), c) is the image at the scanning calibration (-2,2), and d) is the image at the positioning calibration (-1,1).

[0039] Figure 3 This is the one-click calibration module for scanning microscopes in this embodiment of the invention;

[0040] Figure 4 This is a schematic diagram of the raster scan path in an embodiment of the present invention;

[0041] Figure 5 This is a schematic diagram of the edge-finding trajectory of the KDP crystal in an embodiment of the present invention;

[0042] Figure 6 This is an image showing the effect of determining the crystal edge through image processing in an embodiment of the present invention.

[0043] Figure 7 This is a schematic diagram illustrating the principle of the Canny dual threshold algorithm in an embodiment of the present invention;

[0044] Figure 8 This is a comparison chart of the defect detection effects of various methods in the embodiments of the present invention;

[0045] Figure 9 This is the minimum circumcircle image of the component defect point location and defect contour obtained based on the Canny operator in this embodiment of the invention. Detailed Implementation

[0046] In the description of this invention, it should be noted that the terms "first," "second," and "third" mentioned in the embodiments of this invention are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first," "second," and "third" may explicitly or implicitly include one or more of that feature.

[0047] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0048] Specific Implementation Method 1: This invention provides a rapid detection and location method for micro-defects in large-aperture KDP crystals based on variable magnification ring gamut transformation, including the following steps:

[0049] Step 1. Install the optical crystal element to be tested, assemble and adjust the scanning microscope system, automatically return each axis of the crystal element moving mechanism to zero, and establish an absolute coordinate system;

[0050] Step 2. Adjust the magnification of the scanning microscope's zoom rings, use a feature point matching method based on a simulated CMT tracking algorithm to determine the downconversion coefficients for each zoom ring magnification, complete pixel calibration, and establish an intuitive one-click calibration module for the scanning microscope for users;

[0051] Step 3. Based on the conversion coefficients of different zoom ring magnifications of the scanning microscope obtained in Step 2 and the resolution of the scanning microscope, calculate the actual field of view of the scanning microscope at different magnifications;

[0052] Step 4. Control the scanning microscope to perform edge-finding motion and extract the edge image of the component. Determine the coordinate values ​​of the component edge in the absolute coordinate system, further determine the coordinate values ​​of the component center, and establish the workpiece coordinate system of the large-diameter KDP crystal component based on the component center coordinate values ​​to achieve global coordinate transformation of the crystal.

[0053] Step 5. Adjust the magnification of the scanning microscope's zoom ring to medium or low magnification. Select the scanning step distance based on the actual field of view obtained in Step 3. Use a grid-based reciprocating scanning strategy to scan and acquire images of micro-defects on the surface of the entire crystal element. Simultaneously, call the motion controller to control the motion and detect the motion state during the scanning process. Use a gradient-based image processing method to process the acquired images and determine the position coordinates of the defects in the absolute coordinate system.

[0054] Step 6. Adjust the magnification of the scanning microscope to medium magnification. Based on the position coordinates of the micro-defects on the surface of the component obtained in Step 5, move the defect points into the field of view of the scanning microscope and acquire images of the micro-defects on the surface of the crystal component.

[0055] Step 7. Based on the micro-defect images of the crystal element surface acquired in Step 6, extract the micro-defect feature information using an image processing method based on gradient algorithm to obtain the location, shape, and size information of each micro-defect.

[0056] like Figure 1 As shown, the crystal repair machine tool mentioned in step 1 is a self-developed device for rapid searching and micro-milling repair of micro-defects on the surface of a large-diameter KDP crystal (application number: 201310744691.1). The repair machine tool includes a scanning microscope system (X1, Y1, Z1 axes) for rapid searching of crystal surface defect points and acquisition of real-time image information of the crystal tool surface during the repair process; a defect repair system (X2, Y2, Z2 axes) for three-axis linkage machining removal of defect points; and a repair microscope system (X3, Y3, Z3 axes) for acquisition of tool setting image information and calculation of the distance between the tool and the crystal tool surface.

[0057] The scanning microscope system is used to detect defects on the surface of optical crystals. The scanning microscope uses a JAI BM-500GE CCD with a resolution of 2456×2058, a maximum frame rate of 15fps, a pixel size of 3.45μm×3.45μm, and a magnification range of 0.58X to 7X.

[0058] Step 2 includes the following process:

[0059] Adjust the magnification ring of the scanning microscope to the lowest setting, k1 (0.58X). Adjust the field of view of the scanning microscope to focus on the tip of the cutting tool. By adjusting the brightness of the ring light source or the exposure time of the scanning microscope, the acquired image should be as clear as possible while preserving rich details. Move the cutting tool tip to point A (2mm, 2mm) and acquire an image using the scanning microscope. The image is as follows. Figure 2 As shown in a), move the tool tip along the X2 axis by d to point B (-2mm, 2mm), and acquire an image. The image is as follows. Figure 2 As shown in c), the actual movement distance of the tool tip is 4mm. Based on the simulated CMT algorithm, the pixel distance Δ of the tool tip's movement in the image and the actual movement distance d of the tool tip are calculated, and the conversion coefficient c is calculated.

[0060] c = d / Δ

[0061] Calculate the optical magnification of the scanning microscope lens based on the zoom ring magnification k:

[0062] K = 0.75 × 2 × k

[0063] Based on the above method, pixel calibration was performed at different zoom ring magnifications of the scanning microscope, and the parameters are shown in Table 1.

[0064] Table 1

[0065]

[0066] The final user-friendly, intuitive one-click calibration interface for scanning microscopes is as follows: Figure 3 As shown, when the scanning microscope is at different zoom ring magnifications, one-click calibration can automatically realize the calibration movement of the tool and image processing, thereby obtaining the conversion coefficients at different zoom ring magnifications.

[0067] In step 3, the actual field of view of the scanning microscope at different magnifications is obtained by multiplying the conversion factor of the different magnifications by the resolution of the scanning microscope. The scanning microscope sensor parameters are shown in Table 2.

[0068] Table 2

[0069]

[0070] Calculate the actual field of view of the scanning microscope at different magnifications, i.e.:

[0071] (2456×2058)×3.966=(9740×8162)

[0072] (2456×2058)×1.533=(3765×3155)

[0073] When extracting the component edge image in step 4, the crystal edge will appear as a strip of bright light in the microscope field of view due to strong scattering. First, the crystal edge image is converted into a grayscale image, then into a high-contrast image and Gaussian noise reduction is performed. Then, it is converted into a binarized image in CV_THRESH_BINARY mode to process the crystal edge image. The search pixels are all rows or columns of foreground pixels, so as to obtain the position of the crystal edge in the image. Finally, the specific coordinates of the crystal edge in the scanning motion system are calculated based on the pixel calibration value under the current zoom ring magnification and the coordinate value of the scanning motion system.

[0074] The method in step 4 is used to determine the crystal boundary because there is an installation error (approximately 3 mm) during the clamping of large-aperture KDP crystals. Therefore, it is impossible to directly determine the position of the KDP crystal center in the absolute coordinate system. Thus, it is necessary to determine the boundary position of the KDP crystal element through the calibration of the scanning microscope, thereby ultimately determining the coordinates of the center of the large-aperture KDP crystal element in the absolute coordinate system. According to the field of view of 9.740 mm × 8.162 mm obtained in step 3, it can completely cover the crystal installation error. Therefore, the position of the crystal element center in the absolute coordinate system is determined by the edge position of the crystal element.

[0075] like Figure 4 As shown, in step 5, the scanning step distance and the actual number of feed steps required in the X1 and Y1 directions are determined based on the actual field of view of the scanning microscope. During the motion control process, the ComACRsrvr function library provided by the ACR motion controller is used to realize motion control and motion status detection. When the motion stops, the motion status detection is that there is no motion, and the image is acquired. After the image acquisition is completed, the motion command is fed back to the motion axis. This cycle is repeated to complete the rapid detection of the full aperture of the crystal.

[0076] This invention utilizes a magnification variation method, employing a low magnification ring (0.58X) during the scanning process to achieve a large scanning field of view. This ensures effective identification of defect points while improving scanning efficiency and enabling rapid detection. A medium magnification ring (1.5X) is used for defect localization. The medium magnification ring ensures sufficient field of view and accuracy for point finding; therefore, it is used for the positioning function. The scanning platform achieves a repeatability accuracy better than 10μm on the X1 axis and better than 3μm on the Y1 axis, achieving high precision during both scanning and positioning.

[0077] The gradient-based image processing method described in steps 5 and 7 specifically employs the Canny operator and includes the following steps:

[0078] Step 1: Smooth the image using a Gaussian filter;

[0079] Step 2: Calculate the gradient intensity and direction of each pixel in the image using the finite difference of the first-order partial derivative;

[0080] Step 3: Perform non-maximum suppression on the gradient magnitude to eliminate stray effects caused by edge detection;

[0081] Step 4: Use a dual threshold algorithm to determine real and potential edges, suppressing isolated weak edges in the process.

[0082] The dual threshold algorithm sets an upper threshold and a lower threshold. When a pixel value is greater than the set upper threshold, it is considered to be a boundary; when it is less than the set lower threshold, it is considered not to be a boundary; and when it is in between, it is considered a candidate.

[0083] The method of the present invention was verified as follows:

[0084] Since the rapid detection and location method for micro-defects in large-aperture KDP crystals of this invention involves multiple process flows, in order to ensure the feasibility of the scheme and the defect detection effect, it is necessary to verify the determination of the crystal element edge position and the defect image processing method based on gradient operators in the above steps with examples:

[0085] (1) Determining the edge position of the crystal element

[0086] The movement process of the work platform during the calibration process is as follows: Figure 5 As shown, and before calibration, it must be ensured that the scanning motion platform has returned to zero. The specific steps are as follows:

[0087] 1-1 Move to the left edge of the crystal, pause the movement, and record the coordinates of the left edge;

[0088] 1-2 Move to the right edge of the crystal, pause the movement, and record the coordinates of the left edge;

[0089] 1-3 Mechanical return to zero;

[0090] 1-4 Move to the lower edge of the crystal, pause the movement, and record the coordinates of the lower edge;

[0091] 1-5 Move to the upper edge of the crystal, pause the movement, and record the coordinates of the upper edge;

[0092] 1-6 Mechanical return to zero, assembly calibration completed.

[0093] When the scanning microscope moves to the edge of a crystal, strong scattering occurs at the crystal edge, resulting in bright stripes in the microscope's field of view. Precise processing of the crystal edge image can be achieved by first converting it to a grayscale image, then to a high-contrast image and performing Gaussian noise reduction, and finally converting it to a binarized image in CV_THRESH_BINARY mode. Figure 6 As shown in a), the position of the crystal edge in the image is obtained by searching for entire rows or columns where all pixels are foreground pixels, as shown in a diagram. Figure 6 As shown in b). Then, based on the pixel calibration value at the current zoom level and the coordinate value of the scanning motion system, the specific coordinate value of the crystal edge in the scanning motion system is calculated.

[0094] (2) Development of a Defect Image Processing Algorithm Based on Gradient Method

[0095] Current image processing methods for micro-defects on the surface of KDP crystals mainly employ a direct binarization scheme with a single threshold. This involves calculating the average pixel value of the acquired grayscale image of the micro-defect source on the crystal element surface and using this average pixel value as an empirical threshold for direct binarization. This scheme filters out pixels at the edges of the acquired defect points that are not significantly different from the background, making it difficult to accurately separate the defect image edges from the background. Consequently, the size information of the detected defect points is smaller than the actual defect size. Furthermore, the empirical threshold in the direct binarization scheme changes with variations in the lighting environment, affecting the accuracy of defect detection.

[0096] Therefore, developing gradient-based image processing algorithms requires selecting appropriate gradient operators. There are three types of gradient operators: operators that approximate the derivative of a function using the finite difference method, operators that identify the zero-crossing points of the second derivative of a function, and operators that attempt to match the parametric model of the image with that of the edges. These are the Robert operator, Sobel operator, Laplcian operator, and Canny operator.

[0097] 1. Roberts Operator

[0098] The Robert operator, also known as the cross-differential operator, is a gradient algorithm based on cross-difference. It uses the difference between two adjacent pixels along the diagonal to approximate the gradient magnitude to detect edges, and its edge detection performance is better in the vertical direction than in other directions. For the original image f(x,y), the Robert operator's edge detection output image is g(x,y). The Robert edge detection of an image can be expressed by the following formula:

[0099]

[0100] The Robert operator template can be obtained as follows:

[0101]

[0102] 2Sobel operator

[0103] The Sobel operator is a first-derivative edge detection operator that calculates the gradient vector and norm of image pixels, and then uses convolution to calculate the horizontal and vertical edges. For a 3×3 Sobel kernel, the horizontal kernel G of the image... x and vertical core G y It can be expressed as the following formula:

[0104]

[0105] The gradient magnitude G at a point in the convolutional image can be approximated as:

[0106]

[0107] The Sobel operator has high detection efficiency, but it is not accurate enough for arrangements that are not along the x and y axes. In this case, the Scharr filter can be used to correct the computation kernel. The filter factors are as follows:

[0108]

[0109] 3Laplacian operator

[0110] The Laplacian operator is an edge detection operator based on its second derivative. Based on the expression for the discrete Laplacian, the following templates can be obtained: G1 is the template for the discrete Laplacian operator, and G2 is the extended template:

[0111]

[0112] 4. Canny Operator

[0113] The Canny operator is an optimized operator that integrates filtering, enhancement, and detection. Its goal is to find an optimal edge detection method or to locate the position with the strongest gray-level intensity change in an image. Optimal edge detection is mainly evaluated based on three criteria: low error rate, high localization rate, and minimum response. The implementation steps of the Canny operator are as follows:

[0114] Step 1: Smooth the image using a Gaussian filter;

[0115] Step 2: Use the finite difference of the first-order partial derivative to calculate the gradient intensity and direction of each pixel in the image;

[0116] Step 3: Perform non-maximum suppression on the gradient magnitude to eliminate stray effects caused by edge detection;

[0117] Step 4: Use a double threshold algorithm to detect and determine real and potential edges, while suppressing isolated weak edges in the process.

[0118] The dual-threshold algorithm described in Step 4 is key to the Canny algorithm. By setting an upper threshold and a lower threshold, a pixel value greater than the upper threshold is considered a boundary, less than the lower threshold is considered not a boundary, and values ​​in between are considered candidate values. Figure 7 As shown, A is a strong edge, B and C are candidate options, and D is a non-edge. Furthermore, since B is connected to A, but C is not connected to A, B is considered a weak edge, while C is not an edge.

[0119] By using different gradient operators for defect detection, we can obtain results such as... Figure 8 The test results shown are as follows: Figure 8a) The acquired image of the crystal surface shows four defect points within the field of view. Direct binarization image processing can detect these defect points relatively well, but its effectiveness depends on an empirical threshold. For example, changes in the light source will alter the empirical threshold. It also filters pixels near the defect point edges that are close to the background. Figure 8 As shown in b), the Roberts operator has a strong filtering effect, which can lead to the loss of small defects during processing. Larger defects also have a certain corrosive effect, easily resulting in issues such as... Figure 8 The defect segmentation phenomenon observed in c) indicates that after the Sobel operator processes the image, binarization is required to detect defects, and its effect is similar to direct binarization, such as... Figure 8 As shown in d), the Laplacian operator is susceptible to noise during processing, which can lead to inaccurate defect information, such as... Figure 8 As shown in e), the Canny operator eliminates weak edges by using non-maximum suppression of gradient magnitude and a double thresholding method to determine real and potential edges. Figure 8 As shown in f), the image processing algorithm for detecting micro-defects on the surface of large-aperture KDP crystals based on the Canny operator effectively improves the accuracy of defect detection.

[0120] The fast detection and point finding method proposed in this invention uses a designed grid scanning strategy and a defect detection algorithm based on the Canny operator to successfully detect defects in a full-diameter KDP of 410mm×410mm, and the results have been verified experimentally.

[0121] pass Figure 9 As can be seen, the Canny operator accurately detects all defect points in the acquired image, calculates the minimum circumcircle of the defect contour, and obtains the position coordinates of the defect in the image and the size of the minimum circumcircle radius of the defect as shown in Table 4.

[0122] Table 4

[0123]

[0124] Experimental results show that, while maintaining detection accuracy, the scanning speed can be increased to 10 mm / s. Due to the acceleration / deceleration of the scanning motion system and the image capture dwell time, the average speed throughout the scanning process is 6.81 m / s. The detection of a full-aperture KDP crystal can be completed in just 43 minutes, significantly improving efficiency compared to the 4 hours required by current detection methods. Furthermore, grid movement can be controlled through programming, avoiding conflicts with image acquisition and ensuring the accuracy and integrity of the images. This reduces the image positioning error to approximately 10 μm, substantially minimizing the positioning error.

[0125] While the present invention has been disclosed above, its scope of protection is not limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention, and all such changes and modifications will fall within the scope of protection of the present invention.

Claims

1. A rapid detection and location method for micro-defects in large-aperture KDP crystals based on variable magnification ring multiplication, characterized in that, Includes the following steps: Step 1. Install the optical crystal element to be tested, assemble and adjust the scanning microscope system, automatically return each axis of the crystal element moving mechanism to zero, and establish an absolute coordinate system; Step 2. Adjust the magnification of the scanning microscope's zoom rings, use a feature point matching method based on a simulated CMT tracking algorithm to determine the downconversion coefficients for each zoom ring magnification, complete pixel calibration, and establish an intuitive one-click calibration module for the scanning microscope for users; Step 3. Based on the conversion coefficients of different zoom ring magnifications of the scanning microscope obtained in Step 2 and the resolution of the scanning microscope, calculate the actual field of view of the scanning microscope at different magnifications; Step 4. Control the scanning microscope to perform edge-finding motion and extract the edge image of the component, determine the coordinate value of the component edge in the absolute coordinate system, further determine the coordinate value of the component center, establish the workpiece coordinate system of the large-diameter KDP crystal component based on the coordinate value of the component center, and realize the global coordinate transformation of the crystal. Step 5. Adjust the magnification of the scanning microscope's zoom ring to low magnification. Select the scanning step distance based on the actual field of view obtained in Step 3. Use a grid-based reciprocating scanning strategy to scan and acquire images of micro-defects on the surface of the entire crystal element. Simultaneously, call the motion controller to control the motion and detect the motion state during the scanning process. Use a gradient-based image processing method to process the acquired images and determine the position coordinates of the defects in the absolute coordinate system. Step 6. Adjust the magnification of the scanning microscope to medium magnification. Based on the position coordinates of the micro-defects on the surface of the component obtained in Step 5, move the defect points into the field of view of the scanning microscope and acquire images of the micro-defects on the surface of the crystal component. Step 7. Based on the micro-defect images of the crystal element surface acquired in Step 6, extract the micro-defect feature information using an image processing method based on gradient algorithm to obtain the location, shape, and size information of each micro-defect; Step 2 includes the following process: Adjust the magnification ring of the scanning microscope to the lowest magnification k1, and adjust the field of view of the scanning microscope to focus on the tip of the cutting tool. Adjust the brightness of the ring light source or the exposure time of the scanning microscope to obtain an image that is as clear as possible while preserving rich detail. Move the cutting tool tip to point A and acquire an image using the scanning microscope. Then move the cutting tool tip a distance d along the X2 axis to point B and acquire an image. Calculate the pixel distance of the cutting tool tip's movement in the image using a CMT-like algorithm. And the actual movement distance d of the tool tip, calculate the conversion factor c: The above method is used to calibrate pixels at different zoom ring magnifications of a scanning microscope.

2. The rapid detection and location method for micro-defects in large-aperture KDP crystals based on variable magnification ring multiplication as described in claim 1, characterized in that, The scanning microscope system described in step 1 is used to detect defects on the surface of optical crystals. The scanning microscope uses a JAI BM-500GE CCD with a resolution of 2456×2058, a maximum frame rate of 15fps, a pixel size of 3.45μm×3.45μm, and a magnification range of 0.58X~7X.

3. The rapid detection and location method for micro-defects in large-aperture KDP crystals based on variable magnification ring multiplication as described in claim 2, is characterized in that, In step 3, the actual field of view of the scanning microscope at different magnifications is obtained by multiplying the down-conversion factor of the different magnifications by the resolution of the scanning microscope.

4. The rapid detection and location method for micro-defects in large-aperture KDP crystals based on variable magnification ring multiplication as described in claim 3, is characterized in that... When extracting the component edge image in step 4, the crystal edge image is first converted into a grayscale image, then converted into a high-contrast image and Gaussian noise is removed, and then converted into a binarized image in CV_THRESH_BINARY mode to process the crystal edge image. Finally, the specific coordinates of the crystal edge in the scanning motion system are calculated based on the pixel calibration value under the current zoom ring and the coordinate value of the scanning motion system.

5. The rapid detection and location method for micro-defects in large-aperture KDP crystals based on variable magnification ring multiplication as described in claim 4, characterized in that, In step 5, the scanning step distance and the actual number of feed steps required in the X1 and Y1 directions are determined based on the actual field of view of the scanning microscope. During motion control, the ComACRsrvr function library provided by the ACR motion controller is used to realize motion control and motion status detection. When the motion stops, the motion status detection is that there is no motion, and the image is acquired. After the image acquisition is completed, the motion command is fed back to the motion axis. This cycle is repeated to complete the rapid detection of the full aperture of the crystal.

6. The rapid detection and location method for micro-defects in large-aperture KDP crystals based on variable magnification ring multiplication as described in claim 5, is characterized in that, The gradient-based image processing method described in steps 5 and 7 specifically employs the Canny operator and includes the following steps: Step 1: Smooth the image using a Gaussian filter; Step 2: Calculate the gradient intensity and direction of each pixel in the image using the finite difference of the first-order partial derivative; Step 3: Perform non-maximum suppression on the gradient magnitude to eliminate stray effects caused by edge detection; Step 4: Use a dual threshold algorithm to determine real and potential edges, suppressing isolated weak edges in the process; The dual threshold algorithm sets an upper threshold and a lower threshold. When a pixel value is greater than the upper threshold, it is considered to be a boundary; when it is less than the lower threshold, it is considered not to be a boundary; and when it is in between, it is considered a candidate.

Images