A System and Method for Laser Line Length Control in Microscopic Visual Inspection Based on Parametric Calculation

Abstract

This invention relates to a method for controlling the laser line length in microscopic visual inspection based on parametric calculation, belonging to the field of optical microscopic inspection technology, and includes: inputting the target laser line length L. lim The process involves: parameter calculation; adjusting the position or parameters of the cylindrical lens, convex lens, and objective lens to match the calculation results; optical path transmission: the light source emits a collimated laser, which passes sequentially through the cylindrical lens (focusing in the Y direction), the convex lens (adjusting the divergence angle in the X direction), and the objective lens (focusing on the sample surface); real-time detection: a CMOS sensor detects the actual generated laser line length and spot shape; error judgment: determining whether the actual line length meets the target line length error range (<2%): if yes, the final parameter combination and optical path configuration are output, and the process ends; if not, feedback is sent to the parameter calculation module for dynamic fine-tuning of parameters and readjustment of optical components. This invention can improve the accuracy of laser line length in scenarios such as semiconductor defect detection and biological microscopy imaging.

Description

Technical Field

[0001] This invention relates to a method for controlling the laser line length in microscopic visual inspection based on parametric calculation, belonging to the field of optical microscopic inspection technology. Background Technology

[0002] In existing technologies, microscopic visual inspection systems typically control the laser line length by manually adjusting optical components (such as lenses and apertures). There are generally two technical solutions: mechanical adjustment, which adjusts the line length by moving the light source or lens position, but physical displacement easily introduces vibration interference and has a limited adjustment range; and fixed lens group, which uses a preset lens combination to generate a fixed line length, lacking flexibility and unable to adapt to different inspection scenarios.

[0003] The existing methods have the following drawbacks:

[0004] Highly dependent on experience: It requires manual trial and error to select lens combinations and spacing, which is inefficient and prone to errors. For example, operators need to repeatedly change lenses with different focal lengths or adjust the optical path distance, which is time-consuming and difficult to guarantee accuracy.

[0005] The parameters are complexly coupled: the line length is affected by multiple variables (such as laser diameter, lens focal length, and optical path spacing), lacks a systematic model, and is difficult to quickly match the requirements.

[0006] Insufficient accuracy: Manual adjustment is prone to introducing errors, resulting in poor line length stability and affecting detection repeatability. For example, mechanical vibration or thermal drift may cause optical path deviation, further reducing measurement reliability. Summary of the Invention

[0007] This invention provides a method for controlling the laser line length in microscopic visual inspection based on parametric calculation. Its main purpose is to improve the accuracy of laser line length in scenarios such as semiconductor defect detection and biological microscopic imaging.

[0008] To achieve the above effects, the present invention aims to provide a laser line length control system for microscopic visual inspection based on parametric calculation, comprising a laser source, a cylindrical mirror, a convex lens, an objective lens, and a CMOS sensor arranged sequentially.

[0009] Another objective of this invention is to provide a method for adjusting the laser line length in microscopic visual inspection based on parametric calculation, comprising:

[0010] S1, Input target laser line length L line ;

[0011] S2. Parameter calculation: Based on the formulaic model, calculate the feasible combination of collimated laser diameter Dx, cylindrical lens focal length fcyl, convex lens focal length fconv, distance from convex lens to objective lens d, and objective lens focal length fobj.

[0012] S3. Adjust the position or parameters of the cylindrical mirror, convex lens, and objective lens to match the solution results;

[0013] S4. Optical path transmission: The light source emits collimated laser light, which passes sequentially through a cylindrical mirror (focusing in the Y direction), a convex lens (adjusting the divergence angle in the X direction), and an objective lens (focusing on the sample surface).

[0014] S5. Real-time detection: The CMOS sensor detects the actual length of the generated laser line and the shape of the laser spot.

[0015] S6. Error Judgment: Determine whether the actual line length meets the target line length error range (<2%).

[0016] If so, output the final parameter combination and optical path configuration, and the process ends;

[0017] If not, the feedback is sent to the parameter calculation module to dynamically fine-tune the parameters and readjust the optical components.

[0018] Furthermore, the target laser line length L in S1 line The method for determining the size is accomplished through pixel size conversion.

[0019] Specifically, it includes: S11, obtaining pixel equivalent; S12, laser line length conversion, extracting the outline of the laser line in the image, calculating its pixel length in the image; and combining the pixel equivalent to calculate the actual length.

[0020] Furthermore, in order to increase the target laser line length L line To improve accuracy, sub-pixel precision is used to optimize edge pixels. Specifically, this includes: pixel-level coarse edge localization: first, threshold segmentation is used to find the approximate pixel position of the laser line edge; grayscale information of pixels surrounding the coarsely located edge is extracted to form a grayscale curve; based on the continuity of the grayscale curve, interpolation is used to calculate the precise position of the edge within the pixel; error suppression is achieved by eliminating image noise interference with the grayscale distribution through Gaussian filtering to ensure the accuracy of interpolation.

[0021] Furthermore, Gaussian filtering in error suppression includes the following steps: Gaussian kernel generation, the formula for the two-dimensional Gaussian function is:

[0022] ,

[0023] Where (x, y) are the coordinates relative to the kernel center, and σ is the standard deviation; the size of the Gaussian kernel is determined according to the requirements, the weight value of each position in the kernel is calculated, and the sum of the weights is normalized; for image convolution operation, with the current pixel as the center, neighboring pixels of the same size as the Gaussian kernel are taken, the gray value of the neighboring pixels is multiplied by the weight of the corresponding position in the Gaussian kernel, and then all products are summed to obtain the new gray value of the current speed limit; for image edge pixels, a padding method is used to ensure that the convolution operation can cover all pixels.

[0024] Gaussian filtering can effectively smooth images, suppress noise, and at the same time preserve image edge information relatively well.

[0025] Furthermore, the formula described in step 2 is:

[0026] ,

[0027] Where Dx is the collimated laser diameter, fobj is the objective lens focal length, fconv is the convex lens focal length, and d is the distance from the convex lens to the objective lens.

[0028] Furthermore, the above formula is derived through the following steps:

[0029] A collimated beam diverges after passing through a convex lens. Its divergence angle can be approximated as:

[0030] ;

[0031] The distance from the convex lens to the objective lens is d. During this process, the beam diameter increases with the increase of the propagation distance. The beam diameter when it reaches the objective lens is:

[0032] ;

[0033] The objective lens focuses a diverging beam of light onto the sample surface, and its magnification can be expressed as:

[0034] The object distance is Image distance is ;

[0035] A laser line is formed on the sample surface.

[0036]

[0037] in,

[0038] ,

[0039] Substituting it into the original expression, we get

[0040] ,

[0041] when hour,

[0042]

[0043] The final conclusion is: .

[0044] In step 2, the cylindrical mirror is used to focus the beam in the Y direction and works in conjunction with the convex lens to provide compensation.

[0045] In practice, the target line length is first fixed as a constraint condition, the formula is transformed into a constraint equation, the parameter range is then substituted, the point set that satisfies all constraints is selected, and finally the optimal feasible solution is output according to priority (such as cost).

[0046] In step 3, the lens spacing is adjusted first, and the line length is controlled by combining physical constraints of parameters (such as focal length range) and error grading (such as fine adjustment for small errors and lens replacement for large errors).

[0047] The lens spacing has little impact on the laser line length, resulting in lower sensitivity and easier operation.

[0048] Beneficial effects:

[0049] 1. The present invention employs a parameterized calculation method for adjusting the laser line length in microscopic visual inspection, which enables more accurate testing, avoids excessive reliance on human experience, and increases the adjustment range. At the same time, it avoids the problem of fixed lens group schemes that use preset lens combinations to generate fixed line lengths, which lack flexibility and cannot adapt to different detection scenarios.

[0050] 2. This invention uses pixel size conversion to determine the target laser line length L. line The measurement can eliminate the influence caused by hardware differences. Combined with the parameterized calculation of the laser line length adjustment of microscopic visual inspection, it can more accurately calculate the feasible point set that satisfies all constraints, effectively improving the testing efficiency.

[0051] 3. The present invention, after optimization with sub-pixel precision, can more accurately capture edge details, ensuring the accuracy of laser line length in scenarios such as semiconductor defect detection and biological microscopy imaging.

[0052] 4. The microscopic visual inspection laser line length control system of the present invention, which adopts parametric calculation, has a simple structural design, resulting in fewer hardware influencing factors and making it easy to make precise adjustments. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating a method for adjusting the laser line length in microscopic visual inspection based on parametric calculation, according to an embodiment of the present invention.

[0054] Figure 2 This is a schematic diagram of a laser line length control method for microscopic visual inspection based on parameterized calculation, provided in an embodiment of the present invention.

[0055] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0056] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0057] This application provides a system and method for adjusting the laser line length for microscopic visual inspection based on parametric calculation.

[0058] Example 1:

[0059] Reference Figure 1 The diagram shown illustrates a flowchart of a method for controlling the laser line length in microscopic visual inspection based on parametric calculation, according to an embodiment of the present invention. In this embodiment, the parametric calculation-based method for controlling the laser line length in microscopic visual inspection is applicable to the inspection of microscopic semiconductor devices, particularly suitable for silicon / gallium oxide-based wafers, and for inspection scenarios requiring short laser line lengths and high energy densities. (Refer to...) Figure 2 The diagram shown illustrates the principle of a laser line length control method for microscopic visual inspection based on parametric calculation, according to an embodiment of the present invention. The method includes a laser source, a cylindrical mirror, a convex lens, an objective lens, and a CMOS sensor arranged sequentially. The cylindrical mirror has a focal length of fcyl = 50 mm, and the convex lens has a focal length of fconv = 100 mm. The cylindrical mirror is used to focus the beam in the Y direction and works in conjunction with the convex lens to provide compensation.

[0060] The method is as follows:

[0061] S1, Input target laser line length L line ;

[0062] The target laser line length L is determined by pixel size conversion. line The determination.

[0063] Specifically, it includes: S11, obtaining pixel equivalent; S12, laser line length conversion, extracting the outline of the laser line in the image, calculating its pixel length in the image; and combining the pixel equivalent to calculate the actual length.

[0064] In order to increase the target laser line length L line To improve accuracy, sub-pixel precision is used to optimize edge pixels. Specifically, this includes: pixel-level coarse edge localization: first, threshold segmentation is used to find the approximate pixel position of the laser line edge; grayscale information of pixels surrounding the coarsely located edge is extracted to form a grayscale curve; based on the continuity of the grayscale curve, interpolation is used to calculate the precise position of the edge within the pixel; error suppression is achieved by eliminating image noise interference with the grayscale distribution through Gaussian filtering to ensure the accuracy of interpolation.

[0065] Gaussian filtering in error suppression includes the following steps: Gaussian kernel generation, the formula for the two-dimensional Gaussian function is:

[0066] ,

[0067] Where (x, y) are the coordinates relative to the kernel center, and σ is the standard deviation; the size of the Gaussian kernel is determined according to the requirements, the weight value of each position in the kernel is calculated, and the sum of the weights is normalized.

[0068] In this embodiment, the Gaussian kernel is:

[0069]

[0070] Image convolution operation takes the current pixel as the center and takes the neighboring pixels with the same size as the Gaussian kernel. In this embodiment, it takes the eight pixels around the 3x3 matrix and the center pixel. The gray value of the neighboring pixels is multiplied by the weight of the corresponding position in the Gaussian kernel, and then all the products are summed to obtain the new gray value of the current speed limit.

[0071] Gaussian filtering can effectively smooth images, suppress noise, and at the same time preserve image edge information relatively well.

[0072] S2. Parameter calculation: Based on the formulaic model, calculate the feasible combination of collimated laser diameter Dx, cylindrical lens focal length fcyl, convex lens focal length fconv, distance from convex lens to objective lens d, and objective lens focal length fobj.

[0073] The formula is:

[0074] ,

[0075] Where Dx is the collimated laser diameter, fobj is the objective lens focal length, fconv is the convex lens focal length, and d is the distance from the convex lens to the objective lens.

[0076] First, the target laser line length L line As constraints, the formula is transformed into a constraint equation. Then, the focal length of the convex lens, fconv=100mm, is substituted to select the point set that satisfies all constraints, which includes the collimated laser diameter Dx, the objective lens focal length fobj, and the distance d from the convex lens to the objective lens. Finally, the optimal feasible solution is output according to priority (such as cost).

[0077] S3. Adjust the position or parameters of the cylindrical mirror, convex lens, and objective lens to match the solution results;

[0078] The distance d from the convex lens to the objective lens is selected for adjustment first. Combined with the physical constraints of the collimating laser diameter Dx and the objective lens focal length fobj, and error grading processing, the line length can be controlled.

[0079] If the error is small, make fine adjustments; if the error is large, replace the objective lens.

[0080] The lens spacing has little impact on the laser line length, has low sensitivity, and is easier to operate.

[0081] S4. Optical path transmission: The light source emits collimated laser light, which passes sequentially through a cylindrical mirror (focusing in the Y direction), a convex lens (adjusting the divergence angle in the X direction), and an objective lens (focusing on the sample surface).

[0082] S5. Real-time detection: The CMOS sensor detects the actual length of the generated laser line and the shape of the laser spot.

[0083] S6. Error Judgment: Determine whether the actual line length meets the target line length error range (<2%).

[0084] If so, output the final parameter combination and optical path configuration, and the process ends;

[0085] If not, the feedback is sent to the parameter calculation module to dynamically fine-tune the parameters and readjust the optical components.

[0086] In this embodiment, liquid nitrogen cooling is used to reduce dark current noise in the CMOS sensor.

[0087] Example 2:

[0088] This invention provides a method for rapid scanning and detection of larger-sized samples with an expanded laser line length range.

[0089] The focal length of the cylindrical lens is fcyl=100mm, and the focal length of the convex lens is fconv=200mm.

[0090] Example 3: This invention is suitable for large-area imaging of biological tissues, where a longer laser line length and more uniform energy distribution are required.

[0091] The focal length of the cylindrical lens is fcyl=200mm, and the focal length of the convex lens is fconv=300mm.

[0092] In detail, the microscopic visual inspection laser line length control system based on parameterized calculation described in this embodiment of the invention employs the same methods as described above. Figure 1 The method described herein is the same as the laser line length control method for microscopic visual inspection based on parameterized calculation, and can produce the same technical effect, so it will not be repeated here.

[0093] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for controlling the laser line length in microscopic visual inspection based on parametric calculation, characterized in that, The system to which this method is applied includes a laser source, a cylindrical mirror, a convex lens, an objective lens, and a CMOS sensor arranged in sequence. The method includes the following steps: S1, Input target laser line length L line ; S2. Parameter calculation: Based on the formulaic model, calculate the collimated laser diameter D. x Feasible combinations of cylindrical lens focal length fcyl, convex lens focal length fconv, distance from convex lens to objective lens d, and objective lens focal length fobj; S3. Adjust the position or parameters of the cylindrical mirror, convex lens, and objective lens to match the solution results; S4. Optical path transmission: The light source emits collimated laser light, which is focused by a cylindrical mirror, the divergence angle is adjusted by a convex lens in the direction perpendicular to the focusing direction of the cylindrical mirror, and the objective lens is focused on the sample surface. S5. Real-time detection: The CMOS sensor detects the actual length of the generated laser line and the shape of the laser spot. S6. Error Judgment: Determine whether the actual line length meets the error range of less than 2% of the target line length. If so, output the final parameter combination and optical path configuration, and the process ends; If not, feedback is sent to the parameter calculation module to dynamically fine-tune the parameters and readjust the optical components; The formula mentioned in step 2 is ， Among them, D x denoted as the collimated laser diameter, fobj as the objective lens focal length, fconv as the convex lens focal length, and d as the distance from the convex lens to the objective lens.

2. The method for adjusting the laser line length in microscopic visual inspection based on parameterized calculation as described in claim 1, characterized in that, The formula described in step 2 is derived from the following steps: When a collimated beam is passed through a convex lens, its divergence angle can be approximated as: ； The distance from the convex lens to the objective lens is d. During this process, the beam diameter increases with the increase of the propagation distance. The beam diameter when it reaches the objective lens is: ； The objective lens focuses a diverging beam of light onto the sample surface, and its magnification can be expressed as: ； A laser line is formed on the sample surface. in, ， Substituting it into the original expression, we get ， when hour, The final conclusion is: .

3. The method for adjusting the laser line length in microscopic visual inspection based on parameterized calculation as described in claim 2, characterized in that, In step S1, the target laser line length L line The method for determining the size is accomplished through pixel size conversion.

4. The method for adjusting the laser line length in microscopic visual inspection based on parameterized calculation as described in claim 3, characterized in that, Step S1 includes: S11, obtaining pixel equivalent; S12, laser line length conversion, extracting the outline of the laser line in the image, calculating its pixel length in the image; and calculating the actual length by combining the pixel equivalent.

5. The method for adjusting the laser line length in microscopic visual inspection based on parameterized calculation as described in claim 4, characterized in that, The subpixel precision optimization of edge pixels includes: pixel-level coarse edge localization: first, threshold segmentation is used to find the approximate pixel position of the laser line edge; grayscale information of pixels around the coarsely located edge is extracted to form a grayscale curve; based on the continuity of the grayscale curve, interpolation is used to calculate the precise position of the edge within the pixel; error suppression is achieved by eliminating the interference of image noise on the grayscale distribution through Gaussian filtering to ensure the accuracy of interpolation.

6. The method for adjusting the laser line length in microscopic visual inspection based on parameterized calculation as described in claim 5, characterized in that, Gaussian filtering in error suppression includes the following steps: Gaussian kernel generation, the formula for the two-dimensional Gaussian function is: ， Where (x, y) are the coordinates relative to the kernel center, and σ is the standard deviation; the size of the Gaussian kernel is determined according to the requirements, the weight value of each position in the kernel is calculated, and the sum of the weights is normalized; for image convolution operation, with the current pixel as the center, neighboring pixels of the same size as the Gaussian kernel are taken, the gray value of the neighboring pixels is multiplied by the weight of the corresponding position in the Gaussian kernel, and then all products are summed to obtain the new gray value of the current speed limit; for image edge pixels, a padding method is used to ensure that the convolution operation can cover all pixels.

Images