Snapshot type automatic optical microscopic defect detection system and detection method

Abstract

The invention discloses a snapshot type automatic optical microscopic defect detection system and a detection method. The resolution ratio and accuracy of fine defect identification are improved through the microscopic imaging module, the high-speed coding spatial light modulation module is introduced to carry out spatial modulation on a light field carrying defect information, and a camera module can obtain a compressed measurement image through one-time exposure in cooperation with relative movement of a to-be-detected product and the microscopic imaging module. And a high-resolution defect image is recovered through a calculation reconstruction algorithm, the limitation of the frame rate of the camera is broken through, the mechanical scanning delay is eliminated, and high-speed snapshot type detection is realized. The method is particularly suitable for rapid and nondestructive online detection of defects such as internal bubbles and surface cracks of workpieces such as glass and silicon wafers on an industrial production line.

Description

Technical Field

[0001] This invention relates to the field of automated optical inspection, and more particularly to a snapshot-type automated optical microscopic defect inspection system and method. Background Technology

[0002] Internal defects in transparent objects (such as bubbles, cracks, impurities, and stress concentration areas) are key factors affecting their optical performance, mechanical strength, and product yield. Traditional inspection methods, such as manual visual inspection, are inefficient, subjective, and prone to fatigue. While machine vision-based transmitted light inspection methods have improved upon these methods, their detection capabilities are limited for micron- or submicron-sized defects, especially those with low contrast to the background.

[0003] In existing technologies, high-precision detection typically relies on techniques such as optical coherence tomography (OCT) or confocal microscopy. However, these techniques are mostly based on point or line scanning principles, requiring mechanical scanning to construct two-dimensional or three-dimensional images. This severely limits the imaging speed, making it difficult to meet the demands of high-speed, real-time detection on industrial production lines. Summary of the Invention

[0004] In view of this, the present invention provides a snapshot-type automated optical microscopic defect detection system and method to overcome the above-mentioned problems existing in the prior art.

[0005] The snapshot-type automatic optical microscopic defect detection system provided by the present invention includes a light source module, a support unit, a microscopic imaging module, a spatial light modulation module, a camera module, and a computing module. The light source module illuminates along a first optical axis, and the spatial light modulation module and the camera module are arranged along a second optical axis. The first optical axis and the second optical axis do not coincide.

[0006] The light source module is used to provide a light source for the product under test, the carrier unit is used to carry the product under test, the microscopic imaging module is used to microscopically magnify the test area of ​​the product under test, and the carrier unit can move relative to the microscopic imaging module in the plane where the product carrier surface is located.

[0007] The microscopic imaging module and the camera module are set on different axes;

[0008] The spatial light modulation module is used to spatially modulate at least a portion of the light from the microscopic imaging module and then output it to the camera module; within a unit exposure time of the camera module, the spatial light modulation module can switch the encoding pattern multiple times;

[0009] The camera module is used to acquire a compressed measurement image obtained during the exposure time by the movement of the product under test relative to the microscopic imaging module and the switching of the encoding pattern by the spatial light modulation module.

[0010] The calculation module is used to calculate and output a reconstructed product image based on the compressed measurement image, the coding pattern switching information of the spatial light modulation module, and the movement information of the product under test relative to the microscopic imaging module.

[0011] The snapshot-type automatic optical microscopic defect detection system provided by this invention improves the resolution and accuracy of identifying minute defects by setting up a microscopic imaging module, introducing a high-speed encoded spatial light modulation module to spatially modulate the light field carrying defect information, and coordinating with the relative movement of the product under test, so that the camera module can acquire a compressed image in a single exposure, and then recover a high-resolution defect image through a computational reconstruction algorithm. This invention creatively proposes a novel automatic optical microscopic defect detection method for products under test through a combination of software and hardware, completely breaking through the frame rate limitation of the camera and completely eliminating the time delay caused by mechanical scanning, realizing high-speed "snapshot" detection, while possessing high sensitivity and high flexibility. It is particularly suitable for rapid, non-destructive online detection of defects such as bubbles and cracks inside transparent workpieces such as glass and optical crystals on industrial production lines.

[0012] In some implementations, the product under test is a transparent product, with the light source and the spatial light modulation module located on opposite sides of the stage; or, the product under test is an opaque product, with the light source and the spatial light modulation module located on the same side of the stage. This solution is not only applicable to the microscopic defect detection of transparent objects such as optical glass, optical crystals, and transparent plastics where high precision is required for detecting minute defects (such as bubbles, cracks, impurities, stress concentration areas, etc.), but it is also applicable to the surface microscopic defect detection of non-transparent objects (such as wafers and display array substrates) where high precision is required for surface morphology defects, further broadening the application scope of this solution.

[0013] In some implementations, a beam-splitting component is also included, located between the spatial light modulation module and the camera module. The beam-splitting component is positioned on a second optical axis and is used to vertically direct at least a portion of the light from the microscopic imaging module towards the spatial light modulation module. By setting the beam-splitting prism, the light can be perpendicularly incident on the spatial light modulation module, then perpendicularly exit from the spatial light modulation module and enter the camera module for reception. Thus, compared to the case where the light obliquely enters the spatial light modulation module and then obliquely exits from the spatial light modulation module before entering the camera module for reception, the field of view of the light in the spatial light modulation module is larger, and the area of ​​the spatial light modulation module can be utilized more fully. For image reconstruction of the same product, this obviously significantly reduces the frequency of coded pattern switching and the amount of data computation required for the reconstruction algorithm. Therefore, the microscopic image reconstruction of the entire product can be achieved in a shorter time, resulting in faster and more efficient microscopic defect detection, higher detection accuracy, and more precise resolution.

[0014] In some implementations, the beam-splitting component is a semi-reflective prism or a triangular prism, which can guide the light from the microscopic imaging module vertically to the spatial light modulation module without blocking the further propagation of the light from the spatial light modulation module to the camera module, thereby enabling the compressed measurement image to be successfully obtained and minimizing light loss.

[0015] In some implementations, a relay imaging module is further provided between the spatial light modulation module and the camera module. This relay imaging module enhances the imaging effect of the spatially modulated light on the camera module. In one implementation, the relay imaging module includes a first lens and a second lens, and the spatial light modulation module, the first lens, the second lens, and the camera module constitute a 4F optical system.

[0016] In some implementations, the light source module includes a first light source and a third lens. The focal length of the third lens is f3, and the distance between the third lens and the product under test is between f3 and 10f3. The distance between the third lens and the product under test is neither too far nor too close, allowing the third lens to better focus the light emitted from the light source onto the product under test, thereby ensuring sufficient brightness.

[0017] In some implementations, the microscopic imaging module includes an objective lens and a tube lens, with the objective lens located between the support unit and the tube lens. This achieves the microscopic imaging effect described in this invention using only two components: the objective lens and the tube lens. The structure is simple and cost-effective.

[0018] In some implementations, the spatial light modulation module is a digital micromirror device or a spatial light modulator, and the camera module is a CMOS camera or a CCD camera. The system structure of this invention is simple, with the core encoding component being a mature DMD or SLM, which can be used with common CMOS or CCD cameras. It requires no complex scanning device, and the system exhibits high stability and reliability.

[0019] In some implementations, the spatial light modulation module operates using one or more of the following mask patterns: random mask, complementary mask, and structurally complementary mask. Such mask pattern configurations result in more accurate and reliable image reconstruction and higher precision in identifying microscopic defects.

[0020] In another aspect, the present invention provides a defect detection method applied to the aforementioned snapshot-type automated optical microscopy defect detection system, comprising:

[0021] Turn on the light source module and illuminate the product under test with a uniform beam of light;

[0022] The spatial light modulation module is controlled to load a preset coded pattern, and the relative movement between the carrier unit and the microscopic imaging module is controlled.

[0023] The spatial light modulation module is controlled to operate, and the camera module performs exposure to acquire a compressed measurement image;

[0024] The computing module processes the compressed measurement image using a computational reconstruction algorithm and outputs a reconstructed product image.

[0025] The defect detection method of the snapshot-type automatic optical microscopic defect detection system of the present invention can complete a complete product image reconstruction and microscopic defect measurement with only a few exposures of the camera or even a single exposure, completely eliminating the time delay caused by mechanical scanning, and is particularly suitable for high-speed online inspection on the production line. Attached Figure Description

[0026] Figure 1 A schematic diagram of a snapshot-type automated optical microscopic defect detection system provided in an embodiment of the present invention;

[0027] Figure 2 This is a schematic diagram of a snapshot-type automated optical microscopic defect detection system provided in another embodiment of the present invention;

[0028] Figure 3 This is a schematic diagram of a snapshot-type automated optical microscopic defect detection system provided in another embodiment of the present invention;

[0029] Figure 4 This is a schematic diagram of the optical system architecture of a defect detection system provided in another embodiment of the present invention;

[0030] Figure 5 This is a flowchart of the snapshot-type automatic optical microscopic defect detection method provided in the embodiments of the present invention;

[0031] Figure 6 This is a schematic diagram illustrating the detection results of the detection system and method of this application on internal defects in silicon dioxide;

[0032] Explanation of the labels in the diagram:

[0033] 10 - First light source; 110, 210 - Light source modules; 20 - Third lens; 30, 130, 230 - Product under test; 40, 140, 240 - Carrier unit; 150 - Microscopic imaging module; 50 - Objective lens; 60 - Tube lens; 70, 170, 270 - Spatial light modulation module; 80 - Beam splitter assembly; 190 - Relay imaging module; 901 - First lens; 902 - Second lens; 100, 1000, 2000 - Camera module; 1100, 2100 - Calculation module; C1 - First optical axis; C2 - Second optical axis. Detailed Implementation

[0034] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0035] The core concept of this invention lies in: improving the resolution and accuracy of identifying minute defects through a microscopic imaging module; introducing a high-speed encoded spatial light modulation module to spatially modulate the light field carrying defect information; and coordinating with the relative movement of the product under test and the microscopic imaging module, enabling the camera module to acquire a compressed measurement image in a single exposure. A high-resolution defect image is then recovered through a computational reconstruction algorithm, overcoming camera frame rate limitations, eliminating mechanical scanning delays, and achieving high-speed "snapshot" detection. The system structure, working principle, and detection method of each embodiment are described in detail below with reference to the accompanying drawings.

[0036] Example 1: Basic Inspection System for Transparent Products

[0037] This embodiment corresponds to Figure 1 The snapshot-type automatic optical microscopic defect detection system shown is suitable for detecting minute defects (such as bubbles, cracks, impurities, stress concentration areas, etc.) inside transparent products (such as optical glass, optical crystals, transparent plastics, etc.). In this embodiment, the light source and spatial light modulation module are located on both sides of the stage, and the defect information is obtained by transmission imaging.

[0038] 1.1 Detailed Description of System Structure

[0039] like Figure 1As shown, the snapshot-type automatic optical microscopic defect detection system of this embodiment includes: a light source module 110, a support unit 140, a microscopic imaging module 150, a spatial light modulation module 170, a camera module 1000, and a computing module 1100. The light source module 110, the support unit 140, and the microscopic imaging module 150 are arranged along a first optical axis, while the spatial light modulation module 170 and the camera module 1000 are arranged along a second optical axis. The first optical axis C1 and the second optical axis C2 do not coincide. In this embodiment, it is preferable that the first optical axis C1 and the second optical axis C2 are perpendicular to optimize the optical path layout and reduce component interference.

[0040] The specific structure and parameter configuration of each module are as follows:

[0041] Light source module 110: Used to provide a uniform illumination source for the product under test 130. In one implementation, the light source module 110 includes a first light source and a third lens, wherein the first light source is a white LED light source; the focal length of the third lens is f3, which is selected as 50mm in this embodiment; the distance between the third lens and the product under test 130 is set to 3f3 (i.e., 150mm), which is between f3 and 10f3, and can effectively converge the divergent light emitted by the first light source onto the test area of ​​the product under test 130, ensuring that the test area has sufficient illumination, while avoiding the illumination range being too small due to the distance being too close, or the light intensity attenuating too much due to the distance being too far.

[0042] The carrier unit 140 is used to carry the transparent product 130 to be tested. In this embodiment, the carrier unit 140 adopts an electric translation stage, which is driven by a servo motor. The repeatability of the positioning can reach ±0.1μm, enabling high-precision translation of the product 130 to be tested within the plane of its carrier surface. The carrier unit 140 can move relative to the microscopic imaging module 150. By preset the movement trajectory, it can achieve full coverage detection of the entire test area of ​​the product 130 to be tested. The movement speed can be adjusted according to the detection efficiency requirements.

[0043] Microscopic imaging module 150: Used to magnify the test area of ​​the product under test 130 to improve the resolution of identifying minute defects. In this embodiment, the microscopic imaging module 150 includes an objective lens and a tube lens, with the objective lens located between the support unit 140 and the tube lens. This combination of objective and tube lens structure is simpler, lower in cost, and easier to assemble and debug compared to a complex microscopic imaging system. It should be noted that the microscopic imaging module can also adopt other commonly used implementation methods in the industry, and this application does not limit it to any particular method.

[0044] Spatial light modulation module 170: This module modulates at least a portion of the light from the microscopic imaging module 150 using spatial characteristics before outputting it to the camera module 1000. In this embodiment, the spatial light modulation module 170 uses a digital micromirror device (DMD) with a pixel size of 10.8 μm, a resolution of 1920×1080, and a frame rate of up to 227 Hz, enabling high-speed coded pattern switching. The mask pattern used by the spatial light modulation module 170 combines a random mask and a complementary mask. The random mask is generated using a random number generation algorithm, and the complementary mask is the inverse pattern of the random mask. Alternating between these two masks improves the accuracy and reliability of image reconstruction. Within the unit exposure time of the camera module 1000, the spatial light modulation module 170 can switch coded patterns multiple times. In this embodiment, the unit exposure time is set to 4 ms, during which the DMD switches coded patterns 5 times, achieving the superposition and acquisition of multi-frame coded information.

[0045] It should be noted that in this embodiment, the spatial light modulation module 170 is placed at an angle relative to the direction of the light emitted from the microscopic imaging module 150. That is, the light emitted from the microscopic imaging module 150 is obliquely incident on the working surface of the DMD, and then the DMD performs spatial characteristic modulation before outputting it to the camera module. The angle of oblique incidence can be, for example, 45 degrees, so that the first optical axis C1 is perpendicular to the second optical axis C2.

[0046] Camera module 1000: Used to acquire compressed measurement images obtained during the exposure time by the movement of the product under test 130 relative to the microscopic imaging module 150 and the switching of the encoding pattern by the spatial light modulation module 170. In this embodiment, the camera module 1000 is a CMOS camera with a pixel size of 3.45μm, a resolution of 2592×2048, and a frame rate of up to 14fps. The pixel size of the spatial light modulation module 170 (10.8μm) is larger than that of the camera module 1000 (3.45μm), and the pixel size ratio between the two is approximately 3:1, that is, 1 pixel of the DMD corresponds to 3×3 pixels of the CMOS camera. This configuration can ensure the accurate correspondence between the encoding pattern and the camera pixels, ensure the smooth execution of the image reconstruction algorithm, and avoid the impact of pixel correspondence confusion on reconstruction accuracy.

[0047] It should be noted that in other implementations, the spatial light modulation module and camera module can be selected from commonly used mature products on the market, and there are no requirements regarding their pixel size relationship, thus demonstrating the feasibility of this patented solution. The system structure of this invention is simple, with the core encoding component being a mature DMD or SLM, which can be used with common CMOS or CCD cameras. No complex scanning device is required, resulting in high system stability and reliability.

[0048] It should also be noted that in some implementations, the light-emitting area of ​​the uniform light source is larger than the camera module's area. This ensures that the compressed image data acquired by the camera module is authentic, guaranteeing the reliability of the entire system's operation and the accuracy of the microscopic defect detection results.

[0049] The calculation module 1100 is used to calculate and output a reconstructed product image based on the compressed measurement image, the coded pattern switching information of the spatial light modulation module 170, and the movement information of the product under test 130 relative to the microscopic imaging module 150. In this embodiment, the calculation module 1100 adopts an industrial control computer with powerful data processing capabilities. The calculation reconstruction algorithm running in the calculation module 1100 adopts an iterative optimization algorithm, specifically the Alternating Direction Multiplier Method (ADMM). This algorithm has a fast convergence speed, high reconstruction accuracy, and can complete the reconstruction processing of compressed measurement data in a short time.

[0050] 1.2 System Working Process

[0051] The detection system in this embodiment works as follows:

[0052] 1. System initialization: The calculation module 1100 configures the parameters of each module, including setting the luminous intensity of the light source module 110, the distance between the third lens and the product under test 130; setting the moving trajectory and moving speed of the carrier unit 140; setting the encoding pattern sequence and switching frequency of the spatial light modulation module 170; and setting the exposure time, frame rate, and other parameters of the camera module 1000.

[0053] 2. Place the product to be tested: Place the transparent product to be tested (such as a silicon dioxide optical crystal) on the carrier unit 140, and adjust the position of the carrier unit 140 so that the test area of ​​the product to be tested is aligned with the illumination range of the light source module 110 and the field of view of the microscopic imaging module 150.

[0054] 3. Turn on the light source: Start the light source module 110. The light emitted by the first light source is focused by the third lens and uniformly illuminates the test area of ​​the product under test 130. When the light passes through the product under test 130, its light field distribution changes due to internal defects (such as bubbles and cracks). The light carrying the defect information is emitted out of the product under test 130.

[0055] 4. Microscopic imaging and light modulation: The light carrying the defect information enters the microscopic imaging module 150, and after being magnified by the objective lens and corrected by the tube lens, a clear intermediate image is formed; the light corresponding to the intermediate image is transmitted to the spatial light modulation module 170, and the DMD switches at high speed according to the preset coding pattern sequence to modulate the spatial characteristics of the light and change the amplitude or phase distribution of the light.

[0056] 5. Compressed Image Acquisition: During the 4ms exposure time of the camera module 1000, the carrier unit 140 moves the product under test 130 along a preset trajectory. At the same time, the DMD completes 5 encoding pattern switching. The camera module 1000 acquires the light signal after movement and multiple modulation and superposition to form a compressed measurement image and transmits it to the computing module 1100.

[0057] 6. Image Reconstruction and Defect Recognition: The calculation module 1100 calls the ADMM iterative optimization algorithm, combines the received compressed measurement image, the DMD encoding pattern switching timing information, and the movement position information of the carrier unit 140 (obtained through the servo motor encoder), and performs iterative calculations to reconstruct a high-resolution image of the product under test; then, through a preset defect recognition algorithm (such as threshold segmentation + morphological processing), the reconstructed image is analyzed to identify the location, size, shape, and other information of the defects, and the detection results are output.

[0058] Example 2: Transparent Product Inspection System with Relay Imaging Module

[0059] This embodiment corresponds to Figure 2 The detection system shown, based on Embodiment 1, adds a relay imaging module 190 between the spatial light modulation module 170 and the camera module 1000 to improve the imaging effect of the spatially modulated light on the camera module 1000. It is suitable for precision transparent product detection scenarios with higher imaging quality requirements (such as high-end optical lens defect detection).

[0060] 2.1 Supplementary Explanation of System Structure

[0061] The core difference between this embodiment and Embodiment 1 lies in the addition of a relay imaging module 190. The function of the relay imaging module 190 is to transmit and image the modulated light signal emitted from the spatial light modulation module 170 without distortion, correcting aberrations (such as spherical aberration and chromatic aberration) during light propagation, so that the modulated light signal can be clearly and accurately imaged onto the image sensor of the camera module 1000. This avoids image blurring caused by light propagation divergence or aberrations, thereby improving the quality of the compressed measurement image and providing a more reliable data foundation for subsequent image reconstruction. In one implementation, the relay imaging module 190 includes a first lens 901 and a second lens 902, both of which are achromatic lenses, forming a 4f relay imaging system. It should be noted that the relay imaging module 190 can also employ other implementation methods that can improve imaging effects; this application does not limit this.

[0062] The structure and parameters of other modules in this embodiment are consistent with those in Embodiment 1.

[0063] 2.2 Explanation of differences in system operation process

[0064] The working process of this embodiment is basically the same as that of Embodiment 1, except that a relay imaging stage is added after "microscopic imaging and light modulation": the light modulated by the spatial light modulation module 170 first enters the relay imaging module 190, is collimated by the first lens 901 and focused by the second lens 902 to form a clear modulated light image, which is then transmitted to the camera module 1000 for compressed measurement image acquisition. Due to the correction effect of the relay imaging module 190, the signal-to-noise ratio of the compressed measurement image acquired in this embodiment is improved by about 20% compared with Embodiment 1. The reconstructed image has sharper edges and clearer defect details, and can identify tiny bubble defects with a diameter as low as 300nm.

[0065] Example 3: Opaque Product Surface Inspection System

[0066] This embodiment corresponds to Figure 3 The detection system shown is suitable for detecting minute surface defects (such as scratches, dents, protrusions, and impurity adhesions) on opaque products (such as wafers, display array substrates, and metal sheets). In this embodiment, the light source and the spatial light modulation module are located on the same side of the stage, and a reflective imaging method is used to obtain defect information.

[0067] 3.1 Special Notes on System Structure

[0068] The overall system architecture of this embodiment is similar to that of Embodiment 1, including a light source module 210, a support unit 240, a microscopic imaging module 150, a spatial light modulation module 270, a camera module 2000, and a computing module 2100. The core difference lies in the optical path layout and the configuration of the light source module, as detailed below:

[0069] Optical path layout: The light source module 210 and the spatial light modulation module 270 are located on the same side (above) of the support unit 240. The light source module 210 illuminates the support unit 240 at a certain angle. The light is reflected from the surface of the product under test and enters the microscopic imaging module 250 and the spatial light modulation module 270. After spatial characteristic modulation by the spatial light modulation module 270, the light is output to the camera module 2000. The spatial light modulation module 170 and the camera module 1000 are set along the second optical axis C2. The second optical axis C2 does not coincide with the first optical axis C1 of the light source module 110. The microscopic imaging module 250 and the camera module 2000 are not set on the same axis.

[0070] Light source module 210: Utilizes an oblique-type LED light source array, comprising four uniformly distributed LED light-emitting units with a wavelength of 405nm (violet light). This wavelength is relatively short, resulting in minimal diffraction and enhancing the imaging contrast of minute surface defects. The light source module 210 does not employ a third lens; instead, it adjusts the angle of each LED light-emitting unit to ensure the light illuminates the surface of the product under test 230 at a 15° incident angle. This prevents specular reflections from directly entering the microscopic imaging module 150, reducing background noise.

[0071] Other module adaptations: The spatial light modulation module 270 uses a spatial light modulator (SLM) with a pixel size of 8μm, a resolution of 1920×1080, and a frame rate of 120Hz. The mask pattern is a structurally complementary mask, which is optimized for the grayscale characteristics of surface defects and can enhance the grayscale difference between defects and the background. The camera module 2000 uses a CCD camera with a pixel size of 4.65μm, a resolution of 1628×1236, and a frame rate of 25fps. Its pixel size is larger than that of the SLM, and the ratio between the two is 2:1, ensuring accurate matching between encoding and imaging.

[0072] It should be noted that in other implementations, the spatial light modulation module and camera module can be selected from commonly used mature products on the market, and there are no requirements regarding their pixel size relationship, thus demonstrating the feasibility of this patented solution. The system structure of this invention is simple, with the core encoding component being a mature DMD or SLM, which can be used with common CMOS or CCD cameras. No complex scanning device is required, resulting in high system stability and reliability.

[0073] 3.2 System Working Process

[0074] The working process of this embodiment is similar to that of Embodiment 1, with the main difference being in the lighting and imaging methods:

[0075] After initializing the system, place the opaque product under test (such as a silicon wafer) on the carrier unit 240 and adjust the carrier unit 240 so that the test area is aligned with the light source illumination range and the microscopic imaging field of view.

[0076] When the light source module 210 is turned on, the four LED light-emitting units emit 405nm violet light at a 15° incident angle onto the surface of the product under test 230. After the light is reflected by the surface, the reflected light carrying information about defects such as scratches and impurities enters the microscopic imaging module 150 and is magnified by the objective lens and tube lens to form an intermediate image.

[0077] The intermediate image light propagates to the SLM, and the SLM completes four structural complementary mask switching within the camera exposure time (the exposure time is set to 8ms in this embodiment). At the same time, the carrier unit 240 drives the product under test 230 to move.

[0078] The camera module 2000 acquires the superimposed reflected light signals, forms a compressed measurement image, and transmits it to the computing module 2100. The computing module uses a deep learning reconstruction algorithm (based on U-Net network) to reconstruct a high-resolution surface image by combining modulation information and motion information, and finally completes the defect identification and detection result output.

[0079] Example 4: High-precision detection system with spectrometer component

[0080] This embodiment corresponds to Figure 4 The optical system architecture shown in this embodiment includes an additional beam-splitting component 80 (semi-reflective and semi-transparent prism), located between the spatial light modulation module 70 and the camera module 100, and situated on the second optical axis. This component is used to vertically guide at least a portion of the light from the microscopic imaging module towards the spatial light modulation module 70, thereby further improving the detection efficiency and accuracy of the system.

[0081] 4.1 Detailed Description of System Optical Architecture

[0082] like Figure 4 As shown, the optical system of this embodiment is specifically arranged as follows: the first light source 10, the third lens 20, the carrier unit 40, the objective lens 50, the tube lens 60 and the semi-reflective prism 80 are sequentially arranged on the first optical axis; the spatial light modulation module 70 (digital micromirror device DMD), the semi-reflective prism 80, the relay imaging module (first lens 901 + second lens 902) and the CCD camera 100 are sequentially arranged on the second optical axis, and the first optical axis and the second optical axis are perpendicular.

[0083] Among them, the beam splitter 80 is a semi-reflective and semi-transparent prism made of K9 optical glass with a beam splitting ratio of 1:1. It can vertically reflect the light from the tube lens 60 that is propagating along the first optical axis to the spatial light modulation module 70 on the second optical axis (that is, realize the light from the first optical axis to the second optical axis by 90°). At the same time, it allows the light that is modulated by the spatial light modulation module 70 and propagating along the second optical axis to pass through itself and be directed to the subsequent relay imaging module and camera module 100.

[0084] Compared to optical path designs without beam splitters, the advantages of this embodiment are as follows: after being reflected by the semi-reflective prism 80, the light is perpendicularly incident on the effective working surface of the spatial light modulation module 70. The modulated light then exits perpendicularly and passes through the prism into the camera module 100. This perpendicular incident / exit method maximizes the field of view of the spatial light modulation module 70 (in this embodiment, the field of view is increased to 25°, which is about 30% higher than the method of obliquely incident on the effective working surface of the spatial light modulation module 70), and the effective area utilization rate of the spatial light modulation module 70 is increased from 65% to 95%. For the same 10mm×10mm product under test, the frequency of encoding pattern switching required in this embodiment is reduced from 20 times in Embodiment 1 to 8 times, the data calculation amount of the reconstruction algorithm is reduced by 60%, and the image reconstruction time is shortened from 0.8s to 0.32s, significantly improving the detection efficiency. At the same time, the perpendicular optical path reduces the deflection loss and aberration of the light, improving the detection accuracy by 25% and enabling more precise differentiation of adjacent small defects.

[0085] In this embodiment, the beam splitter 80 can also be replaced with a prism. A right-angle prism is selected, and the light is redirected 90° through the principle of total internal reflection. Its advantage is that the light reflection loss is smaller (less than 1%). Compared with the 5% loss of a semi-reflective and semi-transparent prism, it can further improve the intensity of the light signal and improve the detection effect of weak defect signals.

[0086] It should be noted that the DMD pattern switching frequency can be 50-100 images within a single exposure time. If the product to be tested is too large to be measured in a single exposure, multiple exposures can be performed until the entire product is measured.

[0087] Example 5: Snapshot-type Automated Optical Microscopic Defect Detection Method

[0088] This embodiment corresponds to Figure 5 The flowchart of the detection method shown is applicable to the detection system described in any of the embodiments one to four above, and specifically includes the following steps:

[0089] Step S1: Turn on the light source module and illuminate the product under test with a uniform beam of light.

[0090] The computing module sends control commands to the light source module to activate the light source (LED light source). The light source parameters are adjusted according to the type of product under test (transparent / opaque): for transparent products, the distance between the third lens and the product under test is adjusted to a preset value (e.g., 150mm in Example 1), ensuring that the light is focused and uniformly transmitted to the product after passing through the lens; for opaque products, the emission angle of the light source array is adjusted to ensure that the light is uniformly reflected at a preset incident angle (e.g., 15° in Example 3) to illuminate the surface of the product under test. The light intensity is ensured to remain stable within a preset range (5000-8000 lux) to avoid detection errors caused by uneven illumination.

[0091] Step S2: Control the spatial light modulation module to load the preset coded pattern, and control the relative movement between the carrier unit and the microscopic imaging module.

[0092] Based on the defect type and detection accuracy requirements of the product under test, the calculation module sends an coded pattern loading command to the spatial light modulation module, loading a preset mask pattern sequence (random mask + complementary mask / structural complementary mask), and setting the pattern switching frequency (e.g., 227Hz). Simultaneously, the calculation module sends motion control commands to the servo motor of the carrier unit, controlling the carrier unit to move the product under test along a preset trajectory (e.g., row array scanning). The moving speed is set according to the exposure time and detection efficiency requirements (e.g., 20mm / s), and the moving position information is collected in real time through the motor encoder and synchronously transmitted to the calculation module.

[0093] Step S3: Control the spatial light modulation module to operate, and the camera module to perform exposure and acquire compressed measurement images.

[0094] The computing module sends synchronization control commands to the spatial light modulation module and the camera module, enabling them to work together: within the camera module's unit exposure time (e.g., 4ms / 8ms), the spatial light modulation module switches the encoded pattern at a preset frequency (e.g., 5 times / 4 times), while the carrier unit continuously moves the product under test. The camera module's image sensor continuously acquires the light signal after the product under test moves and undergoes multiple modulation and superposition, converts the light signal into an electrical signal, and forms a digital compressed measurement image after analog-to-digital conversion. This image is then transmitted in real-time to the computing module's SSD for storage via a USB 3.0 interface. In this step, the camera module's exposure time must be precisely matched with the number of pattern switching times of the spatial light modulation module and the moving speed of the carrier unit to ensure that a single compressed measurement image can completely record multiple frames of modulation information and movement information over a distance.

[0095] Step S4: The calculation module uses a computational reconstruction algorithm to process the compressed measurement image and outputs a reconstructed product image.

[0096] The calculation module calls a preset computational reconstruction algorithm (iterative optimization algorithm / deep learning algorithm), reads the stored compressed measurement image, the timing information of the coded pattern switching of the spatial light modulation module (such as the switching time points of each pattern), and the timing information of the movement position of the carrier unit (such as the movement distance at each time point), and performs iterative calculations using these three as inputs. Taking the ADMM iterative optimization algorithm as an example, the specific process is as follows: ① Initialize the reconstructed image; ② Construct a measurement matrix based on the coded pattern and movement information; ③ Perform matching operations between the compressed measurement image and the measurement matrix to obtain the initial reconstruction result; ④ Optimize the reconstruction result through regularization constraints to reduce the reconstruction error; ⑤ Repeat the iteration until the error is less than a preset threshold (such as 10). -4 This outputs the final high-resolution reconstructed product image.

[0097] Finally, the calculation module analyzes the reconstructed image using a defect recognition algorithm, marks the defect area, calculates parameters such as the size, quantity, and location of the defects, generates an inspection report (including text description and defect-marked images), and can transmit it to the monitoring terminal of the industrial production line via a network interface to achieve real-time monitoring and feedback of defects.

[0098] Experimental verification results

[0099] Using the detection system of Example 4 and the detection method of Example 5, a silica sample containing two 3×3 circular hole array defects with diameters of 1000 nm and 500 nm, respectively, was tested. The test results are as follows: Figure 6 As shown, the reconstructed image clearly distinguishes two sizes of circular aperture arrays. The edge sharpness of the 1000nm aperture reaches 95%, and the edge sharpness of the 500nm aperture reaches 90%. The defect location error is less than ±2μm, and the detection time is only 0.35s. Compared with the traditional mechanical scanning optical inspection system (detection time 15s, 500nm defects cannot be clearly distinguished), the detection efficiency of this invention is improved by 42 times, and the defect recognition resolution is improved by 1 time, fully verifying the high-speed and high-precision detection advantages of this invention.

[0100] Finally, it should be noted that the above specific embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to examples, 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, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A snapshot automated optical microscopic defect detection system, characterized by, The system comprises a light source module, a bearing unit, a microscopic imaging module, a spatial light modulation module, a camera module, and a calculation module. The light source module is configured to provide a light source for a product to be tested, the bearing unit is configured to bear the product to be tested, and the microscopic imaging module is configured to perform microscopic amplification on a test area of the product to be tested. The microscopic imaging module is arranged in a non-coaxial manner with the camera module. The spatial light modulation module is configured to modulate at least part of the light from the microscopic imaging module in terms of spatial characteristics and then output the light to the camera module. The camera module is configured to acquire a compressed measurement image obtained by moving the product to be tested relative to the microscopic imaging module and switching the encoding pattern of the spatial light modulation module within an exposure time. The calculation module is configured to calculate a reconstructed product image according to the compressed measurement image, the encoding pattern switching information of the spatial light modulation module, and the movement information of the product to be tested relative to the microscopic imaging module.

2. The snapshot automated optical microscopic defect detection system of claim 1, wherein, The product to be tested is a transparent product, and the light source and the spatial light modulation module are located on two sides of the bearing unit. The product to be tested is an opaque product, and the light source and the spatial light modulation module are located on the same side of the bearing unit.

3. The snapshot automated optical microscopic defect detection system of claim 1, wherein, The system further comprises a light splitting assembly located between the spatial light modulation module and the camera module.

4. The snapshot automated optical microscopic defect detection system of claim 3, wherein, The light splitting assembly is located on a second optical axis and is configured to vertically direct at least part of the light from the microscopic imaging module to the spatial light modulation module.

5. The snapshot automated optical microscopic defect detection system of claim 1, wherein, The light splitting assembly is a half-reflective half-transmissive prism or a triple prism.

6. The snapshot automated optical microscopic defect detection system of claim 1, wherein, A relay imaging module is further arranged between the spatial light modulation module and the camera module.

7. The snapshot automated optical microscopic defect detection system of claim 1, wherein, The light source module comprises a first light source and a third lens.

8. The snapshot automated optical microscopic defect detection system of claim 1, wherein, The microscopic imaging module comprises an objective lens and a tube lens.

9. The snapshot automated optical microscopic defect detection system of claim 1, wherein, The spatial light modulation module is a digital micromirror device or a spatial light modulator.

10. A defect detection method applied to the snapshot automatic optical microscopic defect detection system according to any one of claims 1 to 9, characterized in that, The camera module is a CMOS camera or a CCD camera. The mask pattern operated by the spatial light modulation module is one or more of a random mask, a complementary mask, and a structure complementary mask. The system comprises: Turning on the light source module to irradiate the product to be tested with a uniform light beam. Controlling the spatial light modulation module to load a preset encoding pattern and controlling the relative movement between the bearing unit and the microscopic imaging module. Controlling the spatial light modulation module to operate and the camera module to perform exposure to acquire a compressed measurement image. The calculation module processes the compressed measurement image using a calculation reconstruction algorithm to output a reconstructed product image.

Images