A surface defect detection device and method based on polarized optical dark field imaging

Abstract

The application discloses a kind of surface defect detection device and method based on polarized optical dark field imaging, belong to industrial defect imaging detection technical field, the device includes: coaxial arrangement polarized camera, imaging lens, polarized ring LED light source and sample to be detected;Wherein, polarized camera and imaging lens form a polarized imaging system, polarized ring LED light source emits linearly polarized light with dip angle, linearly polarized light occurs reflection and scattering on the sample to be measured;Linearly polarized light is incident to the surface of sample to be detected, and a small amount of reflection and most of scattered light of sample to be detected are collected by imaging lens and polarized camera through the circular hole in the center of polarized ring LED light source, and finally imaging is realized.The defect depth can be calculated from the defect width and the defect dip angle, so as to realize the low-frequency linear polarization degree inversion defect depth, and solve the problem that the defect depth information cannot be effectively extracted in the prior art.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of industrial defect imaging detection, and particularly relates to a surface defect detection device and method based on polarized optical dark field imaging. BACKGROUND

[0002] In the field of industrial manufacturing and product quality detection, the detection of sample surface flaws (such as scratches, dents, etc.) is crucial to ensure product performance. For materials such as metals, the depth of defects directly affects key indicators such as mechanical strength and corrosion resistance, so the depth of defects is an important basis for determining product quality. Currently, surface flaw detection techniques based on optical imaging (such as traditional dark field imaging, asymmetric light illumination differential method, etc.) can quickly locate the defect position and clearly image, but their imaging mechanism relying on light intensity contrast cannot effectively extract the depth information of the defect, nor can it distinguish flaws of different severity. The obtained image only provides the "planar" distribution of scratches, and the lack of depth information seriously restricts the quantitative analysis and grading of defects. On the other hand, some techniques that can quantify depth (such as light section method, white light interference and confocal microscopic imaging, etc.) obtain the three-dimensional topography of the sample surface through optical tomography or point-by-point scanning, but these methods require complex optical system design, precise mechanical control or high-cost instruments and equipment, resulting in large and complex equipment, high maintenance costs, weak system robustness and low detection efficiency, which is difficult to adapt to the high-speed, large-batch and low-cost detection requirements of industrial production lines.

[0003] Therefore, how to realize rapid and high-sensitivity analysis of surface defect depth through optical imaging means without complex optical components and precise scanning has become a technical problem to be solved in the field of industrial detection. SUMMARY

[0004] To solve the above technical problems, the application adopts the following technical solutions:

[0005] A surface defect detection device based on polarized optical dark field imaging, characterized in that it comprises a polarized camera, an imaging lens, a polarized ring LED light source and a sample to be detected arranged coaxially.

[0006] The polarization camera and the imaging lens form a polarization imaging system, the polarization annular LED light source emits linearly polarized light with a fixed inclination angle, and is incident on the surface of the sample to be detected in a reflective dark-field illumination mode; the linearly polarized light is incident on the surface of the sample to be detected, and a small amount of specularly reflected light and most of the scattered light of the sample to be detected pass through the circular hole at the center of the polarization annular LED light source and are collected by the imaging lens and the polarization camera to form a polarization image for defect depth inversion; the polarization camera supports synchronous acquisition of images in four polarization directions in a single exposure, which is used to calculate a low-frequency linear polarization degree image; and the surface defect detection device based on the polarization optical dark-field imaging is used to invert the defect depth based on the low-frequency linear polarization degree image and the geometric relationship of the defect.

[0007] A surface defect detection method based on polarization optical dark-field imaging, used in the surface defect detection device based on polarization optical dark-field imaging, comprising:

[0008] Step 1, calculating a linear polarization degree image, specifically comprising: placing the surface defect detection device based on polarization optical dark-field imaging coaxially, adjusting the polarization annular LED light source to a suitable brightness, and synchronously shooting original polarization images in four polarization directions by using the polarization camera; calculating Stokes vectors from the original polarization images in the four polarization directions, and generating a low-frequency linear polarization degree image from the Stokes vectors;

[0009] Step 2, performing three-dimensional block matching denoising filtering on the low-frequency linear polarization degree image obtained in step 1 to obtain a denoised low-frequency linear polarization degree image;

[0010] Step 3, calibrating a linear equation of the defect section inclination angle and the low-frequency linear polarization degree according to the denoised low-frequency linear polarization degree image obtained in step 2;

[0011] Step 4, performing depth inversion on the defect according to the linear equation of the defect section inclination angle and the low-frequency linear polarization degree calculated in step 3;

[0012] Step 5, performing zero-frequency filtering on the original polarization images in the four polarization directions shot in step 1 and calculating a denoised high-frequency linear polarization degree image;

[0013] Step 6, performing weighted fusion on the denoised low-frequency linear polarization degree image obtained in step 2 and the high-frequency linear polarization degree image obtained in step 5.

[0014] The present application has the following beneficial effects:

[0015] The present application provides a low-frequency linear polarization degree inversion depth method. The low-frequency linear polarization degree image of the sample can be calculated from the original polarization image taken by the polarization camera, and the contrast difference between the defects and the background is improved compared with the ordinary intensity image. In the low-frequency linear polarization degree image, the width of the defect can be directly measured by the image, and the section angle of the defect (such as a scratch) has a strong linear correlation with the low-frequency linear polarization degree of the defect. By calibrating the linear relationship, the low-frequency linear polarization degree can be used to solve the section angle of the defect. According to the specific geometric relationship of various defects (such as V-shaped scratches), the depth of the defect can be calculated from the width of the defect and the inclination angle of the defect, so as to realize the low-frequency linear polarization degree inversion of the defect depth, and solve the problem that the prior art cannot effectively extract the defect depth information.

[0016] In addition to directly calculating and generating the low-frequency polarization degree image based on the original polarization image, the contrast of the defect scattering signal can be enhanced by performing zero-frequency filtering processing on the original polarization image and then calculating the linear polarization degree. On this basis, the three-dimensional block matching noise reduction algorithm can be applied to effectively suppress the background noise and improve the detection sensitivity. In the detection process, the high-frequency linear polarization degree image and the low-frequency polarization degree image after zero-frequency filtering processing are fused by adjusting the weighting proportion parameter, so as to adjust and control the visualization degree of defects of different grades. The present application can realize flexible adaptation of various detection standards.

[0017] The imaging lens and the polarization camera are directly connected in the present application, and the polarization camera is placed in a straight line and coaxially with the polarization ring LED light source and the sample to be detected. The structure is simple, and the problem of complex system structure in the traditional method is solved.

[0018] The polarization ring LED light source used in the present application is an industrial-grade LED, the imaging lens can be an industrial lens, and the polarization camera can be a commercial polarization camera. There is no consumable and expensive imaging objective, and the overall cost of the device is low. The algorithm module of the present application is embedded in the equipment system, and only the daily cleaning of the equipment is required in the detection. The present application is suitable for low-cost detection devices in industry, and solves the problems of high cost and difficult maintenance in the traditional method, and is not easy to industrialize.

[0019] The present application provides a reflection dark-field imaging condition by using a polarization ring LED light source to irradiate the sample at a large angle. The light source inclination and polarization direction are fixed and do not need to be adjusted again, so that a measurement device without optical calibration and with strong robustness can be realized.

[0020] The imaging lens of the present application can be a lens with a reduction ratio, which can realize large-field imaging of the sample, and can cooperate with a high-frame-rate polarization camera to realize high-speed and real-time detection. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1It is a structure schematic view of the surface defect detection device based on polarized optical dark field imaging of the application, wherein 1-polarized camera, 2-imaging lens, 3-polarized ring LED light source, 4-sample to be detected;

[0022] Figure 2 It is a calibration and measurement result graph of the application, wherein (a) is a scatter plot and linear fitting result graph of the scratch section angle and the low-frequency linear polarization degree, (b) is a low-frequency linear polarization degree distribution histogram of scratches with different section angles, and (c) is a comparison graph of the scratch inversion depth value and the measured depth value;

[0023] Figure 3 It is a surface low-frequency linear polarization degree graph of good product capacitors and substandard capacitors and a statistical scatter plot of scratch depth inversion values thereof, wherein (a) is a surface low-frequency linear polarization degree graph of a substandard capacitor with a deep scratch, (b) is a surface low-frequency linear polarization degree graph of a good product capacitor with a shallow scratch, and (c) is a statistical scatter plot of scratch depth inversion values of a plurality of good product capacitors and substandard capacitors calculated by the low-frequency linear polarization degree inversion depth method;

[0024] Figure 4 It is a linear polarization degree graph of the same capacitor sample under different filtering conditions, wherein (a) is a low-frequency linear polarization degree graph of the same capacitor sample, (b) is a high-low-frequency weighted fusion linear polarization degree graph of the same capacitor sample, and (c) is a high-frequency linear polarization degree graph of the same capacitor sample. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical scheme and advantages of the application clearer and more comprehensible, the application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the application and do not limit the application. In addition, the technical features involved in each embodiment of the application described below can be combined with each other as long as they do not conflict with each other.

[0026] The application provides a surface defect detection device based on polarized optical dark field imaging, as shown in Figure 1 The application provides a surface defect detection device based on polarized optical dark field imaging, as shown in

[0027] The principle of the technical scheme of the present application is that the polarized ring LED light source 3 emits linearly polarized light, the light is incident on the surface of the sample 4 to be detected at a certain angle, a small amount of reflected light and most of the scattered light are collected by the imaging lens 2 and the polarization camera 1 through the circular hole in the center of the polarization ring LED light source 3, and finally imaged.

[0028] Specifically, the light emitted by the polarization ring LED light source 3 is similar to conical light, which can realize large-angle dark-field illumination, and the light is reflected and scattered on the measurement plane of the sample 4 to be detected.

[0029] The polarization ring LED light source 3 is a LED light source with a certain angle, which emits linearly polarized light and is incident on the surface of the sample 4 to be detected at a certain angle. The large-angle incidence provides the condition for dark-field illumination. The reflected light and the scattered light are collected by the polarization camera 1 through the circular hole in the center of the polarization ring LED light source 3 after the imaging lens 2. The angle, wavelength and polarization direction of the polarization ring LED light source 3 are not limited. The inclined white light polarized light used in the detection of the capacitance surface can provide better dark-field illumination conditions while ensuring sufficient light intensity. The scattered light on the surface of the sample 4 to be detected is mainly collected during imaging. If the sample 4 to be detected is of other types, the polarization ring LED light source 3 can be replaced by other angles or wavelengths.

[0030] The imaging lens 2 can completely image the sample 4 to be detected on the detector of the polarization camera 1. Different magnification imaging lenses 2 can be replaced according to the size of the sample 4 to be detected and the detection requirements, which can realize complete imaging of the topography of the sample 4 to be detected in the field of view, avoid edge distortion, and select appropriate magnification lenses to detect multiple samples 4 in one image, further improving the detection efficiency.

[0031] The exposure time of the polarization camera 1 is within 1ms, which supports single-exposure synchronous capture of polarization images in four polarization directions such as 0°, 45°, 90° and 135°, and can eliminate the time difference error caused by traditional mechanical rotation of the polarization plate. If a common non-polarization camera is used instead of the polarization camera 1, a polarization module can be added between the imaging lens 2 and the polarization ring LED light source 3, such as a motor-driven linear polarization plate combination, a liquid crystal polarization modulator and other element combinations, which can also detect defects with the same effect.

[0032] The sample 4 to be detected is not limited to the capacitance sample in the embodiment of the present application, and can also be replaced by other samples with high reflection characteristics.

[0033] Further, the present application provides a surface defect detection method based on polarization optical dark-field imaging, comprising:

[0034] Step 1, calculating the linear polarization degree image, specifically comprising: placing the device according to the present application in accordance with Figure 1 Coaxial arrangement, adjusting the polarization annular LED light source 3 to the appropriate brightness, and synchronously shooting the original polarization images of four polarization directions by using the polarization camera 1. The Stokes vector is calculated from the original polarization images of the four polarization directions, and the low-frequency linear polarization degree image is generated by calculating the Stokes vector.

[0035] Step 2, performing three-dimensional block matching denoising filtering on the low-frequency linear polarization degree image obtained in step 1, specifically comprising: first, performing block matching on the low-frequency linear polarization degree image obtained in step 1, and stacking similar image blocks into a three-dimensional matrix; then, performing collaborative filtering and threshold processing in the transform domain to eliminate noise components; finally, reconstructing the image through inverse transformation to generate the denoised low-frequency linear polarization degree image.

[0036] Step 3, calibrating the linear equation of the defect section angle and the low-frequency linear polarization degree according to the denoised low-frequency linear polarization degree image obtained in step 2, specifically comprising: performing three-dimensional topography scanning (such as step instrument scanning) on the actual sample 4 to be detected to obtain the section angle of each defect, and extracting the linear polarization degree value of the corresponding defect position in the denoised low-frequency linear polarization degree image obtained in step 2, using the section angle of the defect and the low-frequency linear polarization degree value of the defect position to draw a scatter plot, and performing linear fitting on all data points to obtain the linear equation of the defect section angle and the low-frequency linear polarization degree of the sample 4 to be detected.

[0037] Step 4, depth inversion of the defect according to the linear equation calculated in step 3, specifically comprising: according to the denoised low-frequency linear polarization degree image obtained in step 2, the width and low-frequency linear polarization degree of the defect can be extracted, and the low-frequency linear polarization degree is substituted into the linear equation in step 3 to obtain the defect section angle. According to the geometric relationship of the defect section profile (such as V-shaped scratch), the defect depth can be calculated by knowing the defect section angle and the defect width, that is, the low-frequency linear polarization degree is used to invert the defect depth; this method is the low-frequency linear polarization degree inversion depth method.

[0038] Step 5, performing zero-frequency filtering on the original polarization images of the four polarization directions obtained in step 1 and calculating the denoised high-frequency linear polarization degree image, specifically comprising: performing Fourier transform on the original polarization images of the four polarization directions obtained in step 1 respectively, filtering out the zero-frequency component, and then performing inverse Fourier transform, calculating the high-frequency linear polarization degree image by calculating the Stokes vector as described in step 1, and performing three-dimensional block matching denoising filtering as described in step 2, and finally obtaining the denoised high-frequency linear polarization degree image. The high-frequency linear polarization degree image can significantly enhance the contrast between defects and uniform background.

[0039] Step 6: Perform weighted fusion of the noise-reduced low-frequency linear polarization map obtained in Step 2 and the high-frequency linear polarization map obtained in Step 5. Specifically, this includes adjusting the weighting ratio parameter according to the detection requirements to perform weighted fusion of the noise-reduced low-frequency linear polarization map obtained in Step 2 and the noise-reduced high-frequency linear polarization map obtained in Step 5. This allows for the control of the visualization level of different levels of defects and enables flexible adaptation to multiple detection standards.

[0040] This invention performs low-frequency and high-frequency linear polarization imaging detection on scratches of varying depths on the surface of capacitor samples, and uses low-frequency linear polarization to invert the scratch depth. Furthermore, this invention also detects scratches on several good and defective capacitors on a production line, statistically analyzes the difference in scratch depth between good and defective products, and performs high- and low-frequency weighted fusion detection on several samples. This demonstrates that this invention can effectively extract defect depth information and classify and display defects of different degrees, enabling rapid detection of good and defective products on the production line.

[0041] Reference Figure 2 The calibration and measurement results are shown in the figure. The method of this invention is used to image different scratches on the same sample and calculate the low-frequency linear polarization degree. The geometric contours of these scratches are detected by a profilometer to obtain the inclination angle of the cross section. Figure 2 (a) shows the scatter plot and linear fitting results of the scratch cut surface inclination angle and low-frequency linear polarization degree, where the straight line is the fitted straight line, the points are the scratch sampling points, and the linear correlation R between the scratch cut surface inclination angle (as shown on the horizontal axis) and the low-frequency linear polarization degree (as shown on the vertical axis) is... 2 The value is 0.945, indicating a good linear relationship between the low-frequency linear polarization degree and the scratch cut surface inclination angle. This straight line can be used as the calibration result. Figure 2 The inset in (a) shows the step sample scanning result of one of the scratches, with the horizontal axis representing the scan length and the vertical axis representing the scratch depth. The scratch is approximately V-shaped. Different cut angles are extracted ( Figure 2 The low-frequency linear polarization distribution histogram of the scratches with inclination angles of 28°, 32°, 36°, 41°, 46°, 50°, 54°, 58°, and 62° in (b) is shown below. Figure 2 As shown in (b), the horizontal axis represents the low-frequency linear polarization degree, and the vertical axis represents the pixel count. It can be seen that the scratch cut angle α that the method of this invention can distinguish is 4-5°. Therefore, the low-frequency linear polarization degree and the scratch cut angle are linearly related. The scratch cut is approximately V-shaped, and its cut angle is related to the scratch width and scratch depth. The scratch width can be obtained by measuring the pixel length on the low-frequency linear polarization degree image. Based on the geometric relationship of the scratch morphology, the scratch depth can be calculated; this is the low-frequency linear polarization degree inversion depth method. This low-frequency linear polarization degree inversion depth method was used to measure six unknown scratches. The scratch depth values ​​inverted by the low-frequency linear polarization degree were compared with the scratch depth values ​​measured by the profilometer, as shown...Figure 2 As shown in (c), the horizontal axis represents the scratch count, the vertical axis represents the scratch depth, the dots represent the measured depth values, and the squares represent the inverted depth values; the maximum relative error between the inverted depth value and the measured depth value is 0.08. This demonstrates that the method of the present invention can quickly obtain the scratch depth within a certain accuracy range.

[0042] Reference Figure 3 As shown, Figure 3 of (a) Figure 3 (b) uses surface low-frequency linear polarization maps of a defective capacitor and a good capacitor as examples to demonstrate the detection effect after low-frequency linear polarization calculation and 3D block matching noise reduction in this invention. The color mapping bar on the right represents the value of linear polarization. It can be seen that scratches may appear on the surface of both good and defective capacitors; however, the scratch depth of the good capacitor is below the threshold and will not affect product performance or cause potential harm. The scratch depth of several good and defective capacitors was calculated and statistically analyzed using the low-frequency linear polarization inversion depth method, such as... Figure 3 As shown in (c), the horizontal axis represents the scratch count, the vertical axis represents the scratch depth, the circles represent good products, and the squares represent defective products; the scratch depth on defective products is significantly higher than that on good products, and there is a clear boundary between the two. This result indicates that the present invention can effectively detect the scratch depth on the surface of highly reflective planar samples, enabling rapid evaluation and classification of product quality in a production line environment.

[0043] Reference Figure 4 The figure shows the degree of linear polarization of the same capacitor sample under different filtering conditions. Figure 4 of (a) Figure 4 (b) Figure 4 (c) shows the low-frequency linear polarization degree map, the high-low frequency weighted fusion linear polarization degree map, and the high-frequency linear polarization degree map for the same capacitor sample, respectively. All results have undergone three-dimensional block matching noise reduction processing. The color mapping bars on the right represent the values ​​of linear polarization. Through comparison... Figure 4 Three pictures, Figure 4 (c) can highlight finer details of flaws. Figure 4 (a) can only reflect larger and deeper defects on the sample surface. Figure 4 (b) lies between the two. By adjusting the ratio of high and low frequencies during fusion, different levels of detail in the linear polarization diagram can be presented, thereby matching the detection of different quality control standards.

[0044] The raw polarization image captured by the polarization camera 1 can calculate the low-frequency linear polarization degree image of the sample, and the low-frequency linear polarization degree image improves the contrast difference between the defects and the background compared with the ordinary intensity image. In the low-frequency linear polarization degree image, the width of the defect can be directly measured by the image, and the section angle of the defect (such as a scratch) has a strong linear correlation with the low-frequency linear polarization degree of the defect, and by calibrating the linear relationship, the low-frequency linear polarization degree can be used to solve the section angle of the defect. According to the specific geometric relationship of various defects (such as V-shaped scratches), the defect depth can be calculated from the defect width and the defect angle, so as to realize the low-frequency linear polarization degree inversion of the defect depth. The present application solves the problem that the prior art cannot effectively extract the defect depth information.

[0045] In addition to directly calculating and generating the low-frequency polarization degree image based on the raw polarization image, the contrast of the defect scattering signal can be enhanced by performing zero-frequency filtering processing on the raw polarization image and then calculating the linear polarization degree. On this basis, the three-dimensional block matching noise reduction algorithm can effectively suppress the background noise and improve the detection sensitivity. In the detection process, by adjusting the weighted proportion parameter to fuse the high-frequency linear polarization degree image and the low-frequency polarization degree image after zero-frequency filtering processing, the visualization degree of defects of different grades can be adjusted. The present application can realize flexible adaptation of various detection standards.

[0046] The imaging lens 2 and the polarization camera 1 are directly connected, and are placed in a straight line and coaxially with the polarization ring LED light source 3 and the sample to be detected 4, so that the structure is simple, and the problem of complex system structure in the traditional method is solved.

[0047] The polarization ring LED light source 3 used in the present application is an industrial-grade LED, the imaging lens 2 can be an industrial lens, and the polarization camera 1 can be a commercial polarization camera. There is no consumable and expensive imaging objective, and the overall cost of the device is low. The algorithm module of the present application is embedded in the equipment system, and only the daily cleaning of the equipment is required in the detection. The present application is suitable for low-cost detection devices in industry, and solves the problems of high cost and difficult maintenance in traditional methods, and is not easy to industrialize. The present application provides a reflection type dark field imaging condition by using a polarization ring LED light source to irradiate the sample at a large angle. The inclination angle and the polarization direction of the light source are fixed and do not need to be adjusted again, so that a measurement device without optical calibration and with strong robustness can be realized.

[0048] The imaging lens 2 of the present application can be a lens with a reduction ratio, which can realize large field imaging of the sample, and can cooperate with a high frame rate polarization camera to realize high-speed and real-time detection.

[0049] The above description is only an embodiment of the present application, and does not limit the scope of the present application. Any equivalent structure or equivalent process transformation obtained by using the content of the specification and the drawings, or direct or indirect application in other related system fields, is also included in the protection scope of the present application.

[0050] That which is not specifically described in the specification of the present application is considered to be the state of the art by those skilled in the art.

Claims

1. A surface defect detection method based on polarized light optical dark field imaging, characterized by, The application relates to a surface defect detection method based on polarized optical dark field imaging. Step 1: calculating a linear polarization degree image, specifically comprising the following steps: arranging a surface defect detection device based on polarized optical dark field imaging in a coaxial manner, adjusting a polarized ring LED light source to a proper brightness, synchronously shooting original polarization images in four polarization directions by using a polarization camera, calculating a Stokes vector from the original polarization images in the four polarization directions, and generating a low-frequency linear polarization degree image from the Stokes vector; Step 2: performing three-dimensional block matching denoising filtering on the low-frequency linear polarization degree image obtained in step 1 to obtain a denoised low-frequency linear polarization degree image; Step 3: calibrating a linear equation of a defect section angle and a low-frequency linear polarization degree according to the denoised low-frequency linear polarization degree image obtained in step 2; Step 4: performing depth inversion on a defect according to the linear equation of the defect section angle and the low-frequency linear polarization degree obtained in step 3; Step 5: performing zero-frequency filtering on the original polarization images in the four polarization directions shot in step 1 and calculating a denoised high-frequency linear polarization degree image; Step 6: performing weighted fusion on the denoised low-frequency linear polarization degree image obtained in step 2 and the high-frequency linear polarization degree image obtained in step 5. The surface defect detection device based on polarized optical dark field imaging comprises a coaxially arranged polarization camera, an imaging lens, a polarized ring LED light source and a sample to be detected. The polarization camera and the imaging lens form a polarization imaging system, the polarized ring LED light source emits linearly polarized light with a fixed angle and is incident on the surface of the sample to be detected in a reflection type dark field illumination mode; the linearly polarized light is incident on the surface of the sample to be detected, a small amount of specular reflection light and most of the scattered light of the sample to be detected are collected by the imaging lens and the polarization camera through a circular hole in the center of the polarized ring LED light source, and a polarization image for defect depth inversion is formed; the polarization camera supports synchronous acquisition of images in four polarization directions in a single exposure, and is used for calculating a low-frequency linear polarization degree image; the surface defect detection device based on polarized optical dark field imaging is used for inverting the depth of a defect based on the low-frequency linear polarization degree image and the geometric relationship of the defect.

2. The surface defect inspection method based on polarized light optical dark field imaging according to claim 1, characterized in that, The imaging lens completely images the sample to be detected on the detector of the polarization camera; different magnification imaging lenses are replaced according to the size of the sample to be detected and detection requirements, and the topography of the sample to be detected is completely imaged in the field of view.

3. The method of detecting surface defects based on polarized optical dark field imaging according to claim 1, wherein, The exposure time of the polarization camera is within 1ms, and the polarization camera supports synchronous capture of polarization images in four polarization directions in a single exposure.

4. The method of detecting surface defects based on polarized optical dark field imaging according to claim 1, wherein, A common non-polarization camera is used to replace the polarization camera, and a polarization module is added between the imaging lens and the polarized ring LED light source.

5. The method of detecting surface defects based on polarized optical dark field imaging according to claim 1, wherein, Step 2 comprises the following steps: firstly, block matching is performed on the low-frequency linear polarization degree image obtained in step 1, similar image blocks are stacked into a three-dimensional matrix; then, collaborative filtering and threshold processing are performed in a transform domain to eliminate noise components; finally, the image is reconstructed through inverse transformation to generate a denoised low-frequency linear polarization degree image.

6. The method of detecting surface defects based on polarized optical dark field imaging according to claim 1, wherein, Step 3 comprises: performing three-dimensional topography scanning on the actual sample to be detected to obtain the section angle of each defect, and extracting the linear polarization degree value of the corresponding defect position in the denoised low-frequency linear polarization degree map obtained in step 2, and using the section angle of the defect and the low-frequency linear polarization degree value of the defect position to draw a scatter plot, and performing linear fitting on all data points to obtain a linear equation of the defect section angle and the low-frequency linear polarization degree of the actual sample to be detected.

7. The method of detecting surface defects based on polarized optical dark field imaging according to claim 1, wherein, Step 4 comprises: extracting the width and low-frequency linear polarization degree of the defect according to the denoised low-frequency linear polarization degree map obtained in step 2, substituting the low-frequency linear polarization degree into the linear equation of the defect section angle and the low-frequency linear polarization degree in step 3 to obtain the defect section angle; according to the geometric relationship of the defect section profile, the defect depth is calculated according to the known defect section angle and defect width, and the depth inversion of the defect is realized.

8. The method of detecting surface defects based on polarized optical dark field imaging according to claim 1, wherein, Step 5 comprises: performing Fourier transform on the original polarization maps of the four polarization directions obtained in step 1 respectively, filtering out the zero frequency component, and then performing inverse Fourier transform respectively, then calculating the Stokes vector as described in step 1, generating a high-frequency linear polarization degree map from the Stokes vector, and performing the three-dimensional block matching denoising filtering as described in step 2, and finally obtaining a denoised high-frequency linear polarization degree map.

9. The method of detecting surface defects based on polarized optical dark field imaging according to claim 1, wherein, Step 6 comprises: according to the detection requirement, adjusting the weighting proportion parameter to weight and fuse the denoised low-frequency linear polarization degree map obtained in step 2 and the denoised high-frequency linear polarization degree map obtained in step 5.

Images