An optical film surface defect detection system

Abstract

The application provides an optical film surface defect detection system, especially suitable for optical film, wafer and sensor detection. The system integrates six modules of illumination, confocal positioning, differential interference microscopy, phase shift, imaging and data processing. The application adopts a double optical path imaging mode: a differential interference microscopic light path recorded by a CCD and a microscopic dark field imaging light path. The differential interference microscopic light path uses interference and phase shift technology to obtain high sensitivity of the microstructure and phase distribution information of the film surface. The microscopic dark field imaging light path highlights the film surface defects, particles and micro-nano structure characteristics by enhancing the surface scattering signal. The application has the characteristics of simple structure, large single measurement range, high resolution and high precision, effectively overcomes the defects of three-dimensional information loss, low efficiency, narrow application range and positioning difficulty of traditional technology, and significantly improves the detection efficiency and applicability.

Description

Technical Field

[0001] This invention relates to the field of precision optical inspection technology, and in particular to an optical thin film surface defect detection system. Background Technology

[0002] Optical thin films refer to one or more thin layers of material with special optical properties deposited or coated on optical elements or independent substrates. The production and preparation of optical thin films are affected by various factors, and their surfaces are prone to various defects that affect the surface flatness and original optical properties. To ensure the quality of optical thin films, it is often necessary to acquire images and perform defect detection on the prepared films.

[0003] In current industrial production processes, the acquisition methods for optical thin film images primarily rely on traditional camera acquisition methods, which involve flat-laying the optical thin film for imaging. This method suffers from inconsistent accuracy, affecting the subsequent identification of film defects. Furthermore, it can only acquire two-dimensional image information of the thin film, making it difficult to obtain three-dimensional information about the film surface through a single inspection. While contact methods such as surface profilometers and atomic force microscopes can achieve three-dimensional surface information detection of optical thin films, they are prone to causing secondary damage to the surface.

[0004] Chinese patent CN 113740034 B, while also disclosing a detection film based on optical interference, only uses the most basic optical system for interference imaging. It lacks the precise imaging capabilities of a microscope objective, and therefore lacks a step to determine the focusing state of the sample, making high-precision image acquisition impossible. Furthermore, it relies too heavily on the training accuracy of the detection model, performing subsequent detection only based on the acquired two-dimensional image; and its use of a basic, improved convolutional neural network results in a lack of real-time detection capabilities. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides an optical thin film surface defect detection system. It can simultaneously acquire two-dimensional and three-dimensional information about the thin film, improving the overall system reliability and reducing the workload of measurement personnel.

[0006] To achieve the objectives of this invention, the technical solution provided by this invention is as follows:

[0007] An optical thin film surface defect detection system includes: an illumination module for forming a collimated beam, wherein the collimated beam is circularly polarized after passing through a polarizer and a first quarter-wave plate;

[0008] The confocal positioning module splits the circularly polarized light into two linearly polarized lights through the first polarizing beam splitter inside it. One beam passes through the imaging lens inside it and reaches the photodetector PD inside it, while the other beam enters the differential micro-interference module.

[0009] The differential micro-interference module includes a second polarizing beam splitter, which splits the incoming light beam into two beams: the upper beam enters the phase-shifting module, and the lower beam, after passing through a relay lens, illuminates the surface of the optical axis crystal. The optical axis crystal differentially shears this beam into two linearly polarized beams with mutually perpendicular vibration directions. These two linearly polarized beams converge at the surface of the optical thin film through a microscope objective, forming a reflected beam. The reflected beam then passes through the microscope objective and the optical axis crystal again, where it is combined into a single reflected beam that returns to the second polarizing beam splitter. One reflected beam enters the phase-shifting module, while the other reflects beam returns to the confocal positioning module.

[0010] The phase-shifting module is equipped with a second quarter-wave plate and a polarizer, which is used to perform phase-shifting processing on the upper beam and one reflected beam. The processed beam then enters the imaging module.

[0011] The imaging module includes a tube imaging lens and a CCD. The tube imaging lens is used to receive a phase-shifted reflected beam and an upper beam to form interference fringes; the CCD is used to acquire the interference fringes.

[0012] The data processing module includes a PC connected to a photodetector PD and a CCD, for receiving and processing signals from the photodetector PD and the CCD to obtain information about the surface of the optical thin film.

[0013] Furthermore, the lighting module includes a light source, a tunable attenuator, and a 4f beam shaping system located on the same horizontal axis. The light source is used to generate a beam, which is then collimated by the tunable attenuator and the 4f beam shaping system.

[0014] Furthermore, the first polarization beam splitter, imaging lens, and photodetector PD of the confocal positioning module are arranged sequentially from bottom to top within the detection light combining device and are located on the same horizontal axis.

[0015] Furthermore, the second polarizing beam splitter, relay lens, optical axis crystal, and microscope objective of the differential micro-interference module are arranged in a top-to-bottom order and are located on the same vertical axis.

[0016] Furthermore, the microscope objective includes an annular aperture and a condenser lens. The annular aperture is used to reduce the two linearly polarized beams into a circle and obliquely illuminate the sample surface through the periphery of the condenser lens.

[0017] The signal of the photodetector PD is: the light intensity signal of one beam is converted into an electrical signal, which is transmitted to the PC for digital processing and spectral calibration to determine the focusing state of the optical thin film; the signal of the CCD is: the optical interference image formed by the optical interference fringes acquired by the CCD is converted into an electrical signal, which is transmitted to the PC for subsequent processing.

[0018] Furthermore, the data processing module also includes measurement software, which is used to detect defects in the optical interference images acquired by the CCD.

[0019] Furthermore, the measurement software is used to detect defects in the optical interference image acquired by the CCD, specifically by using the YOLOv10 network to detect defects in the optical interference image.

[0020] Furthermore, the data processing module also includes a driver rotator for adjusting measurement software parameters to control a three-dimensional displacement platform located below the microscope objective.

[0021] Compared with the prior art, the beneficial effects of the present invention are:

[0022] This invention innovatively integrates a dual-optical-path imaging mode by combining reflective Kola illumination, forming a complementary surface detection scheme. The differential interferometric microscopy optical path utilizes interference and phase-shifting techniques to acquire high-sensitivity information on the microscopic morphology and phase distribution of the thin film surface; the microscopic dark-field imaging optical path effectively highlights surface defects, particles, and micro / nanostructure features by enhancing surface scattering signals. The two imaging optical paths can work independently or in combination, enabling multi-dimensional optical property analysis from the overall morphology of the thin film to local defects. A confocal module is also included to improve the accuracy of image acquisition. Specifically, this is reflected in the following aspects:

[0023] 1) By using polarized light and beam expanding and collimating devices, the beam focusing characteristics are improved, the quality of the detection light is enhanced, and high-quality optical input conditions are provided for subsequent interferometric analysis.

[0024] 2) Dark-field microscopy enhances the contrast of scattered light, helping to reveal minute defects and particle information on the sample surface, providing more details on the surface quality of thin films.

[0025] 3) Combining three-dimensional topography data with two-dimensional scattering data provides a more comprehensive analysis of thin film quality, ensuring a full assessment of surface defects and optical properties; two-dimensional image detection uses an improved YOLOv10 network to maximize the real-time performance and accuracy of detection.

[0026] 4) By converting the beam intensity signal into an electrical signal through the confocal positioning module for digital processing and spectral calibration, it is possible not only to determine the focusing state of the sample, but also to improve the depth resolution and data accuracy of the system, thus significantly improving the accuracy of image acquisition. Attached Figure Description

[0027] Figure 1 A schematic diagram of the implementation structure of the optical thin film surface imaging system based on differential interferometry imaging provided by the present invention.

[0028] Figure 2 This is a flowchart of the defect identification process for optical thin film images in this invention.

[0029] Figure 3 This is a diagram of the improved YOLOv10 network structure used for two-dimensional image defect detection in this invention.

[0030] Figure 4 This is a schematic diagram of the optical thin film defects to be detected in this invention. Different colors represent different types of defects.

[0031] Figure 1 In the diagram: 1-Illumination module, 2-Confocal positioning module, 3-Differential micro-interference module, 4-Phase shift module, 5-Imaging module, 6-Data processing module, 7-Light source, 8-Tunable attenuator, 9-4f beam shaping system, 10-Polarizer, 11-First quarter-wave plate, 12-First polarizing beam splitter, 13-Imaging lens, 14-Photodetector (PD), 15-Second polarizing beam splitter, 16-Relay lens, 17-Optical axis crystal, 18-Microscopic objective, 19-Three-dimensional displacement platform, 20-Second quarter-wave plate, 21-Analyzer, 22-Tube imaging lens, 23-CCD, 24-PC, 25-Driver rotator, 26-Measurement software. Detailed Implementation

[0032] The system and detection method of the present invention will now be described in detail with reference to the accompanying drawings.

[0033] See Figure 1The optical thin film surface defect detection system mainly includes an illumination module 1, a polarizer 10, a first quarter-wave plate 11, a confocal positioning module 2, a differential micro-interference module 3, a phase-shifting module 4, an imaging module 5, and a data processing module 6. The illumination module 1 includes a light source 7, a tunable attenuator 8, and a 4f beam shaping system 9. The light emitted from the light source passes through the above components sequentially, then through the polarizer 10 and the first quarter-wave plate 11. The axes of the light source 7, tunable attenuator 8, 4f beam shaping system 9, polarizer 10, and first quarter-wave plate 11 are all on the same straight line. The confocal positioning module 2 includes a first polarizing beam splitter 12, an imaging lens 13, and a photodetector PD14. These are arranged in a top-to-bottom order within the detection light synthesis device, with the axes of all components on the same straight line. The light beam enters the first polarizing beam splitter 12 after passing through the first quarter-wave plate 11. The photodetector PD14 and the data processing module 6... PC24 is connected for information transmission. The differential interference module 3 consists of a second polarizing beam splitter 15, a relay lens 16, an optical axis crystal 17, a microscope objective 18, and a three-dimensional displacement platform 19 arranged in a bottom-up order within the detection light receiving device, with their axes all on the same axis. Light enters the second quarter-wave plate 20 of the phase shift module 4 through the second polarizing beam splitter 15. The phase shift module 4 consists of the second quarter-wave plate 20 and an analyzer 21. After passing through the second quarter-wave plate 20 and the analyzer 21, the light enters the Tube imaging lens 22 of the imaging module 5. The imaging module 5 includes a CCD 23 and a Tube imaging lens 22. The CCD 23 transmits information with PC24 of the data processing module 6. The data processing module 6 includes PC24, a driver rotator 25, and measurement software 26. PC24 interacts with the driver rotator 25 and transmits information with the measurement software 26.

[0034] The detection method of the optical thin film surface defect detection system provided by the present invention includes the following steps:

[0035] The optical thin film to be tested is placed on a three-dimensional motion platform 19. The rotator 25 drives the measurement software 26, which is connected to the PC 24, to control the three-dimensional motion platform 19 to ensure that the surface of the thin film is within the optimal focal plane of the optical system. A white light beam is emitted by the light source 7. After the incident beam is adjusted by the attenuator 8, it enters the beam expanding, collimating and filtering system to optimize the beam quality and form a collimated beam. This collimating and filtering system is a 4f beam shaping system 9, specifically a system composed of two convex lenses.

[0036] Angle of passage of collimated incident beam The polarizer 10 generates linearly polarized light, which is then converted into circularly polarized light by the first quarter-wave plate 11 at an angle α = 45°. The phase delay is expressed by the following formula:

[0037]

[0038] In the formula, Δn is the refractive index difference, d is the propagation distance of light in the waveplate, and λ is the wavelength of light;

[0039] The phase difference of the first quarter wave plate here is The beam is converted into circularly polarized light. Circularly polarized light is equivalent to the superposition of two linearly polarized lights of equal intensity and orthogonal to each other. After the beam is converted into circularly polarized light, the first polarizing beam splitter 12 can separate the beam more accurately, laying the foundation for subsequent interferometric imaging.

[0040] After the incident circularly polarized light passes through the first polarizing beam splitter 12 with an angle of θ = 45°, it is split into two linearly polarized beams with mutually perpendicular polarization directions. One incident beam passes through the imaging lens 13 and reaches the photodetector PD14 inside it; the other incident beam enters the second polarizing beam splitter 15.

[0041] The second polarizing beam splitter 15 splits the linearly polarized light into an upper and lower beam. The upper incident beam enters the phase shift module 4, is phase-delayed by the second quarter-wave plate 20 at an angle α = 45°, and is converted into circularly polarized light; the lower incident beam is irradiated onto the surface of the optical axis crystal by the relay lens 13. The optical axis crystal 17 differentially shears the linearly polarized light into two beams of linearly polarized light with mutually perpendicular vibration directions, the shearing direction forming a fixed angle with the surface of the optical thin film;

[0042] A sheared beam is converged onto the surface of the optical thin film by a microscope objective 18. The microscope objective 18 includes an annular aperture and a condenser lens, with the annular aperture located above the condenser lens. The beam is reduced to a circle by the annular aperture, passes through the periphery of the condenser lens, and obliquely illuminates the sample, forming a dark-field microscopic image. The reflective Kola illumination method reduces stray light and improves the uniformity of illumination. The reflected beam ultimately carries information about the microstructure of the thin film sample.

[0043] The reflected beam passes through the microscope objective 18 and the optical axis crystal 17 again, and is then combined into a single beam before returning. It is then split into two reflected beams by the second polarizing beam splitter 15. One reflected beam enters the confocal positioning module 2 for optical processing. This beam is reflected by the first polarizing beam splitter 12 and passes through the imaging lens 13. The corresponding light intensity signal enters the photodetector PD14, is converted into an electrical signal, and transmitted to the PC24 for digital processing and spectral calibration to determine the sample's focusing state, thereby improving the system's depth resolution and data accuracy.

[0044] The other reflected beam enters the phase-shifting module 4 after passing through the second polarizing beam splitter 15. After reflection, the beam's polarization direction changes; therefore, the reflected beam is converted into new linearly polarized light after passing through the second quarter-wave plate 20, and its phase changes simultaneously. Compared to the incident beam, it is now phase-lagging. Therefore, the two beams passing through phase shift module 4 have a constant phase difference.

[0045] The analyzer 21 ensures that the two beams (the upper incident beam and the other reflected beam) vibrate in the same direction, satisfying the coherence condition and becoming coherent beams. This generates an interference field, allowing the longitudinal thickness of the measured surface to be represented by intensity contrast. The intensity formula for the generated interference image is as follows:

[0046] I = I0cos 2 (Δφ)

[0047] In the formula, I is the light intensity of the interference pattern, I0 is the light intensity of the beam entering the analyzer (i.e., the light intensity of the upper incident beam and the other reflected beam), and Δφ is the phase difference between the two beams.

[0048] The CCD23 records the interference field generated by two coherent beams, displaying in real time the relief-like measurement results with the microscopic morphology and characteristics of the thin film. The microscopic dark-field imaging optical path in the system further enhances the imaging contrast of surface defects, particles, and micro / nano structures. The final generated two-dimensional scattering image obtains the scattering characteristics of the sample surface and records the shape, size, and distribution information of defects.

[0049] CCDs can simultaneously acquire two-dimensional interferometric images and two-dimensional non-interferometric ordinary images. The acquired interferometric images are converted into electrical signals, digitized by the image acquisition card, and transmitted to a PC for further processing. Interferometric images carry phase differences, interference fringes, light intensity, and minute information about the object. By calculating the phase distribution of the interferometric images, the three-dimensional data information displayed in the thin-film image can be obtained.

[0050] For those based on having The principle of interference imaging using two coherent beams with a phase difference is employed, and the phase distribution of the interference fringes is calculated using the phase extraction method. Two images with a phase difference of... The interference images have the following interference light intensities:

[0051] I1(x,y)=I0(x,y)+I m (x,y)cos[φ(x,y)]

[0052]

[0053] In the formula, I1(x,y) and I2(x,y) are the interference light intensities of the two images, respectively, and I0(x,y) is the background light intensity, obtained by capturing an interference-free image. m (x,y) is the amplitude of the interference fringes, and φ(x,y) is the phase distribution to be determined.

[0054] Using trigonometric identities, eliminate amplitude I m The formula for calculating the phase is as follows, and the phase distribution of the interference fringes can be calculated using this formula:

[0055]

[0056] In actual measurements, the phase value is limited to 0 ≤ φ ≤ 2π. If the phase changes by more than 2π, it will "jump," leading to inaccurate measurement results. To accurately reconstruct the three-dimensional height of the thin film surface, phase unwrapping is performed before calculating the height information to obtain a continuous phase map and correct for any jumping phase information.

[0057] By utilizing a global unwrapping model, the phase structure of the overall image is optimized by minimizing the phase gradient. This ensures the smoothness of the overall image by progressively correcting the phase of each pixel. The specific optimization function is as follows:

[0058]

[0059] in, These represent the rates of change of the phase field in the x and y directions, respectively;

[0060] During the optimization process, the gradient of the objective function E(φ) with respect to the phase field is calculated and gradually minimized to find the optimal solution. Simultaneously, the phase is continuously adjusted to ultimately obtain the optimized continuous phase map φ. unwrapped ;

[0061] After obtaining the continuous phase diagram, the three-dimensional height information of the sample surface can be calculated using a formula. The relationship between the interference fringe phase and the surface height h(x,y) can be expressed as:

[0062]

[0063] In the formula, λ is the wavelength of light, and n eff is the effective refractive index, and h(x,y) is the height of the sample surface relative to the reference surface;

[0064] Using the above formula, the height of the sample surface relative to the reference surface can be calculated, which allows for three-dimensional visualization of thin film surface defects and ultimately obtains the three-dimensional height information of the sample surface.

[0065] The two-dimensional non-interference ordinary image acquired in the CCD is simultaneously transmitted to the measurement software (26) used for defect detection for further processing. The specific defect identification process is as follows: Figure 2 As shown, a modified YOLOv10 network is used to detect defects in thin films. The quality of the dataset determines the effectiveness of the network in training the defect detection model.

[0066] First, image augmentation techniques were used to expand the image dataset. Cropping, rotation, and brightening methods were employed to improve the quality of the dataset for subsequent model training.

[0067] For complex and diverse thin film defects, the original YOLOv10 network does not perform well when used directly. To address the diverse types and complex appearances of thin film defects, the YOLOv10 network is improved. The improved YOLOv10 network structure is shown in the diagram below. Figure 3 As shown, the area within the red box represents the improved structure;

[0068] Specifically, the improvements include replacing the Conv layers of the backbone network with RefConv convolutions, introducing the CARAFE sampling method and BiFPN pyramid network structure into the neck network, and adding the STA global attention mechanism. These improvements enhance the model's ability to extract features from both small and long-range defects, enabling high-precision and real-time defect detection of thin film images.

[0069] Data processing module 6 integrates the 3D topography data obtained by processing 2D interferometric images and the 2D defect identification results obtained by processing 2D non-interferometric ordinary images. First, spatial mapping is performed to align the 3D and 2D information to the same reference frame for analysis. The specific mapping process is as follows:

[0070] First, key points (such as edge points and corner points) are extracted from the 3D topography data, and key information of defect areas (such as the center point of the rectangle and the four corner points) is extracted from the 2D YOLO detection result image for feature point matching. Then, affine transformation is used to align the coordinates between the 3D data and the 2D image. The RANSAC random sampling consensus algorithm is used to find the transformation relationship between the matching points, and an affine transformation matrix is ​​calculated. This process includes translation, scaling, and rotation. The formula is as follows:

[0071]

[0072] Where a, b, c, d are the parameters of the affine transformation, t x , t y It is the translation amount;

[0073] Finally, the coordinates of the 2D YOLO detection result area are mapped to the coordinate system of the 3D topography image through affine transformation. The defect information detected by YOLO is cross-validated with the calculated results of the 3D topography data, integrating the height distribution and local defect conditions of the sample surface. A threshold for cross-validation is set based on the actual defect type and characteristics of the measured component, and the acceptability of both 3D and 2D information is judged to determine whether the region of the component is defective. Ultimately, a comprehensive detection result for the thin film is generated, achieving efficient and accurate detection of surface defects in optical thin films and similar optical components using this system.

[0074] Figure 4 This diagram illustrates common defect types in optical thin films. These include both long-distance, large-sized defects (such as crystal lines and cracks in the image) and small-sized defects (such as shallow spots and depressions in the image). Overall, they are characterized by their diverse types and sizes, varied shapes, and ease of confusion with the thin film background. By combining this information with other defect types encountered in actual production, this invention utilizes three-dimensional and two-dimensional information processing on the acquired images to efficiently and accurately identify and detect these complex and diverse optical thin film defects.

[0075] It should be noted that the connection between PC24, driver rotator 25 and measurement software 26 in data processing module 6 is not fixed. Driver rotator 25 and measurement software 26 can be installed in PC24, or they can be set up independently. The three-dimensional displacement platform 19 can also be directly controlled by operating PC24.

[0076] It should be noted that the differential interference microscopy optical path of the present invention is the optical path formed by the upper incident beam entering the Tube imaging lens 22 in the imaging module 5 and the other reflected beam. This differential interference microscopy optical path utilizes interference and phase shifting techniques to obtain the microscopic morphology and phase distribution information of the thin film surface with high sensitivity.

[0077] The optical path for dark-field microscopy imaging is as follows: linearly polarized light is split downwards by the second polarizing beam splitter 15 and then passes sequentially through the relay lens 16, the optical axis crystal 17, and the microscope objective 18. This dark-field microscopy imaging path uses a ring aperture and a condenser lens on the microscope objective 18 to ensure that the incident light illuminates the sample from the side at an angle, preventing it from directly entering the condenser lens. Only the scattered light has its propagation direction altered and enters the condenser lens for imaging, resulting in a completely black background while the sample details are brightly visible, highlighting surface defects, particles, and micro / nanostructure features of the thin film.

[0078] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. An optical thin film surface defect detection system, characterized in that, include: An illumination module (1) is used to form a collimated beam, which is then passed through a polarizer (10) and a first quarter-wave plate (11) to form circularly polarized light; The confocal positioning module (2) splits the circularly polarized light into two linearly polarized lights through the first polarizing beam splitter (12) inside it. One beam passes through the imaging lens (13) inside it to reach the photodetector PD (14) inside it, and the other beam enters the differential micro-interference module (3). The differential micro-interference module (3) is equipped with a second polarizing beam splitter (15) to split the incoming beam into two beams, the upper beam and the lower beam. The lower beam is irradiated on the surface of the optical axis crystal (17) by a relay lens (16). The optical axis crystal (17) differentially shears the beam into two linearly polarized beams with mutually perpendicular vibration directions. The two linearly polarized beams are converged on the surface of the optical thin film by a microscope objective (18) to form a reflected beam. The reflected beam is then combined into a single reflected beam after passing through the microscope objective (18) and the optical axis crystal (17) again. It is then split into two reflected beams again at the second polarizing beam splitter (15). One reflected beam enters the phase shift module (4) and the other reflected beam returns to the confocal positioning module (2). The phase-shifting module (4) is equipped with a second 1 / 4 wave plate (20) and a polarizer (21) for performing phase-shifting processing on the upper beam and one reflected beam. The processed beam enters the imaging module (5). The imaging module (5) includes a Tube imaging lens (22) and a CCD (23). The Tube imaging lens (22) is used to receive a phase-shifted reflected beam and an upper beam to form interference fringes; the CCD (23) is used to acquire the interference fringes. The data processing module (6) includes a PC (24) connected to a photodetector PD (14) and a CCD (23) for receiving and processing signals from the photodetector PD (14) and the CCD (23) to obtain information about the surface of the optical thin film.

2. The optical thin film surface defect detection system according to claim 1, characterized in that, The lighting module (1) includes a light source (7) located on the same horizontal axis, a tunable attenuator (8) and a 4f beam shaping system (9), the light source being used to generate a beam, which is then collimated by the tunable attenuator (8) and the 4f beam shaping system (9).

3. The optical thin film surface defect detection system according to claim 1, characterized in that, The first polarization beam splitter (12), imaging lens (13) and photodetector PD (14) of the confocal positioning module (2) are arranged in the detection light synthesis device in a bottom-up order and are located on the same horizontal axis.

4. The optical thin film surface defect detection system according to claim 1, characterized in that, The second polarizing beam splitter (15), relay lens (16), optical axis crystal (17) and microscope objective (18) of the differential micro-interference module (3) are arranged in a top-to-bottom order and are located on the same vertical axis.

5. The optical thin film surface defect detection system according to claim 1, characterized in that, The microscope objective (18) includes an annular aperture and a condenser lens. The annular aperture is used to reduce the two linearly polarized beams into a circle and obliquely irradiate the sample surface through the periphery of the condenser lens.

6. The optical thin film surface defect detection system according to claim 1, characterized in that, The signal of the photodetector PD (14) is: the light intensity signal of one beam is converted into an electrical signal, which is transmitted to the PC (24) for digital processing and spectral calibration to determine the focusing state of the optical thin film; the signal of the CCD (23) is: the optical interference image formed by the optical interference fringes collected by the CCD (23) is converted into an electrical signal, which is transmitted to the PC for subsequent processing.

7. The optical thin film surface defect detection system according to claim 1, characterized in that, The data processing module also includes measurement software (26), which is used to detect defects in the optical interference image acquired by the CCD (23).

8. The optical thin film surface defect detection system according to claim 7, characterized in that, The measurement software (26) is used to detect defects in the optical interference image acquired by the CCD (23), specifically by using the YOLOv10 network to detect defects in the optical interference image.

9. The optical thin film surface defect detection system according to claim 8, characterized in that, The data processing module (6) further includes a driver rotator (25), which is used to adjust the parameters of the measurement software (26) to control the three-dimensional displacement platform (19), which is located below the microscope objective (18).

Images