Defect detection method for film coating on surface of plastic substrate based on lmc process

Abstract

The application relates to the technical field of coating defect visual detection, in particular to a plastic substrate surface coating defect detection method based on an LMC process, which comprises the following steps: acquiring a film layer morphology gray scale image of a plastic substrate surface coating, constructing a pit direction difference degree of each pixel point, obtaining a pit edge crack characteristic value of each pixel point, then acquiring a pit characteristic value of each pixel point, extracting a suspected pit feature point, combining a position relationship between each pixel point and the suspected pit feature point to obtain a pit dense saliency of each pixel point, using a four-element Fourier saliency detection algorithm to acquire a film layer pit saliency map, and obtaining a region of the coating pit defect through image segmentation. The application can improve the detection precision of the coating defect.

Description

Technical Field

[0001] This application relates to the field of visual inspection technology for coating defects, specifically to a method for detecting coating defects on the surface of plastic substrates based on LMC technology. Background Technology

[0002] LMC (Low Motion Coating) is a method for coating non-metallic surfaces using atomic-level electromagnetic shock. LMC-based coatings offer significant advantages in smoothness, fully revealing the details of the plastic substrate surface while maintaining its original high-gloss finish. However, the coating quality on plastic substrates using LMC is greatly affected by the operating current, easily leading to defects such as pits, cracks, and holes, affecting the uniformity and density of the coating. Therefore, defect detection of the coating on plastic substrate surfaces is necessary.

[0003] Currently, most methods rely on industrial visual inspection to analyze the morphological features of the film layer on the surface of plastic substrates through film morphology images, thereby enabling defect detection of the coating on the plastic substrate surface. However, during the coating process of plastic substrates using the LMC process, small-scale pit defects may form due to the impact of metal atoms and metal ions. Since the characteristics of these pit defects on the surface of plastic substrates are often relatively small, existing technologies cannot accurately identify these pit defects in the film morphology images. This leads to false positives and false negatives when performing defect detection on the surface of plastic substrates after coating, affecting the accuracy of defect detection for plastic substrate coatings. Summary of the Invention

[0004] To address the aforementioned technical problems, this application provides a method for detecting surface coating defects on plastic substrates based on the LMC process, thereby resolving the existing issues.

[0005] The method for detecting surface coating defects of plastic substrates based on LMC process in this application adopts the following technical solution:

[0006] One embodiment of this application provides a method for detecting coating defects on the surface of a plastic substrate based on LMC technology, comprising the following steps:

[0007] Obtain grayscale images of the film morphology on the surface of a plastic substrate;

[0008] By analyzing the distribution and variation of grayscale texture in the local neighborhood of each pixel in the grayscale image of the film morphology, the difference in the direction of the pits of each pixel is obtained.

[0009] Based on the gradient changes in the local neighborhood of each pixel and the difference in the pit direction of each pixel, the pit edge crack feature value of each pixel is obtained.

[0010] By using the pit direction difference degree and pit edge crack feature value, the pit feature value of each pixel is obtained, so as to extract the suspected pit feature points of each pixel, and combine the positional relationship between each pixel and each suspected pit feature point to obtain the pit density salience of each pixel.

[0011] The quaternary Fourier saliency detection algorithm is used to combine the gray value, gradient value, pit edge crack feature value and pit density saliency of each pixel to obtain the saliency map of the film pit, and the region of coating pit defect is obtained by image segmentation.

[0012] Preferably, the method for obtaining the difference in pit direction of each pixel is as follows:

[0013] G j =(Hg j -Hd j )×(Hg j +Hd j );

[0014] In the formula, G j Hg represents the difference in pit direction at the j-th pixel. j and Hd j Let be the horizontal texture complexity and the vertical texture complexity of the window region of the j-th pixel, respectively.

[0015] Preferably, a window of a preset size is used as the region window for each pixel, with each pixel as the center.

[0016] Preferably, the method for obtaining the horizontal texture complexity and the vertical texture complexity is as follows:

[0017] Obtain the grayscale row vectors and grayscale column vectors within the pixel region window. Use the average of the permutation entropy of all grayscale row vectors within the region window of each pixel as the horizontal texture complexity of the region window of each pixel. Use the average of the permutation entropy of all grayscale column vectors within the region window of each pixel as the vertical texture complexity of the region window of each pixel.

[0018] Preferably, the method for obtaining the pit edge crack feature values ​​of each pixel is as follows:

[0019]

[0020] In the formula, D j Let F be the feature value of the pit edge crack of the j-th pixel, n be the number of neighboring pixels in the region window of the j-th pixel, and F be the feature value of the pit edge crack of the j-th pixel. j and G j F represents the gradient magnitude and pit direction difference of the j-th pixel, respectively. j,g and G j,gThese represent the gradient magnitude and the difference in pit direction of the g-th pixel within the region window of the j-th pixel, respectively; where the gradient magnitude of the pixel is obtained through an edge detection algorithm.

[0021] Preferably, the pit feature value of each pixel is the sum of the pit direction difference degree of each pixel and the pit edge crack feature value.

[0022] Preferably, the step of extracting suspected concave feature points of each pixel further includes:

[0023] The average value of the pit feature value of all pixels is used as the screening threshold, and pixels with pit feature values ​​greater than the screening threshold are all designated as target pixels. Centered on each pixel, the K target pixels that are closest to each pixel are selected as the K suspected pit feature points of each pixel.

[0024] Preferably, the method for calculating the saliency of the pit density of each pixel is as follows:

[0025]

[0026] In the formula, V j Let be the saliency of the pit density at the j-th pixel, exp() be an exponential function with the natural constant as the base, and ds j,k DS is the Euclidean distance between the j-th pixel and its k-th suspected concave feature point. j,k Let K be the average of the pit feature values ​​of the j-th pixel and its k-th suspected pit feature point, where K is the number of suspected pit feature points.

[0027] Preferably, the method for obtaining the saliency map of the film pits is as follows:

[0028] Edge detection is performed on the grayscale image of the film morphology to obtain the gradient map of the film morphology;

[0029] The gray values ​​of each pixel in the grayscale image of the film morphology are replaced with the pit edge crack feature value and the pit density saliency, respectively. The images after replacement are denoted as pit edge crack feature map and pit density feature map, respectively.

[0030] The grayscale image of the film morphology, the gradient map of the film morphology, the feature map of pit edge cracks, and the feature map of pit density are used as four feature maps in the phase spectrum model PQRT of quaternion Fourier transform. The saliency map of the film pit is obtained by using the quaternion Fourier saliency detection algorithm.

[0031] Preferably, the method for obtaining the region of the coating pit defect is as follows: the watershed segmentation algorithm is used to segment the saliency map of the film pit to obtain the segmented film pit defect region.

[0032] This application has at least the following beneficial effects:

[0033] This application analyzes the pit defect characteristics after LMC coating process, and more accurately measures the edge crack characteristics and pit density characteristics of pit defects on the film morphology, making the pit defect characteristics on the plastic substrate surface more accurate, and avoiding the problems of false detection and missed detection when detecting pit defects on the plastic substrate surface after coating.

[0034] This application combines the edge crack characteristics and pit density characteristics of pit defects on the film morphology, and uses the quaternary Fourier saliency detection algorithm to extract the saliency map of the film pit, making the pit defects on the film morphology more obvious and improving the saliency of the pit defects on the film morphology. Thus, the image segmentation algorithm can more accurately identify the pit defect area on the film surface and improve the accuracy of defect detection on the coating surface of plastic substrate.

[0035] This application solves the problem of poor saliency of pit defect features on the surface of plastic substrates, enabling existing segmentation algorithms to more accurately identify pit defect features on the film surface, thereby accurately segmenting the pit defect area on the film surface, thus solving the problem of false detection and missed detection when performing defect detection on the surface of plastic substrates after coating. Attached Figure Description

[0036] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 A flowchart illustrating the steps of the LMC-based method for detecting surface coating defects in plastic substrates provided in this application;

[0038] Figure 2 This is a schematic diagram illustrating the process of obtaining the saliency map of the film pits provided in this application. Detailed Implementation

[0039] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the method for detecting surface coating defects of plastic substrates based on LMC process proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0040] Unless otherwise defined, terms such as “comprising,” “including,” or any other variations thereof are intended to cover a non-exclusive inclusion, such that a circuit structure, article, or device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such an article or device. Without further limitation, an element defined by the phrase “comprising one…” does not exclude the presence of other identical elements in the article or device that includes said element. Furthermore, the term “and / or” as used herein includes any and all combinations of one or more of the associated listed items. All technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0041] The following description, in conjunction with the accompanying drawings, details the specific scheme of the LMC-based plastic substrate surface coating defect detection method provided in this application.

[0042] This application provides an embodiment of a method for detecting coating defects on the surface of a plastic substrate based on the LMC process. For details, please refer to [link to relevant documentation]. Figure 1 This includes the following steps:

[0043] Step 1: Obtain a grayscale image of the film morphology on the surface of the plastic substrate.

[0044] In this embodiment, a defect detection method for the LMC coating surface is implemented through industrial visual inspection. First, a plastic substrate coated using the LMC process is acquired. Then, an image sensor is used to obtain an image of the coating morphology on the surface of the plastic substrate through visual inspection. In this embodiment, the image sensor is mainly divided into CCD image sensors and CMOS image sensors. CCD image sensors have the characteristics of high sensitivity and low noise, while CMOS image sensors have the characteristics of low power consumption and high integration. In this embodiment, a CCD image sensor is selected to acquire the image of the coating morphology.

[0045] To avoid interference from external noise and improve the accuracy of images acquired by intelligent sensors in industrial visual inspection, the film morphology image is converted to grayscale. The grayscale-converted film morphology image is then input into a filtering algorithm. The filtering algorithm can be bilateral filtering, Gaussian filtering, or mean filtering. This embodiment does not specify any particular algorithm. In this embodiment, bilateral filtering is selected for denoising. The denoised image is denoised as the film morphology grayscale image. Bilateral filtering is a well-known technique, and the specific process will not be described in detail.

[0046] Step 2: By analyzing the distribution and variation of grayscale texture in the local neighborhood of each pixel in the grayscale image of the film morphology, the difference in the direction of the pits of each pixel is obtained.

[0047] In the LMC process, the instability of the operating current during atomic-level electromagnetic shock affects the compactness of the film layer, resulting in pits of varying sizes and irregular shapes. Since the changes during atomic-level electromagnetic shock are relatively small, the significance of these pit defects detected by smart sensors is often minimal. Consequently, existing technologies cannot accurately identify pit defect features on the film surface from film morphology images, leading to false positives and false negatives when performing defect detection on plastic substrates after coating. Therefore, to improve the accuracy of defect detection on plastic substrate coatings, it is necessary to more accurately identify pit defect features in the grayscale images of the film morphology acquired by smart sensors.

[0048] Furthermore, to accurately analyze the pit defects on the film morphology identified by the smart sensor, a 9×9 region window will be set centered on each pixel in the grayscale image of the film morphology. The implementer can choose the size of the region window according to the actual situation. At the same time, since the pit defects on the film layer are of varying sizes and irregular shapes, the directional difference between the horizontal and vertical texture changes near the pit defect locations in the grayscale image of the film morphology will be relatively large.

[0049] Therefore, by using the gray values ​​within the region window of each pixel in the grayscale image of the film morphology, we can obtain the grayscale row vector and grayscale column vector within the region window. The grayscale row vector and grayscale column vector can respectively reflect the horizontal and vertical grayscale texture changes near each pixel in the grayscale image of the film morphology. The greater the difference in complexity between the horizontal and vertical grayscale textures near a certain pixel in the grayscale image of the film morphology, and the higher the complexity of the horizontal and vertical grayscale texture changes, the more it can reflect the directional difference characteristics at the location of the pit defects in the grayscale image of the film morphology.

[0050] Furthermore, the mean of the permutation entropy of all grayscale row vectors within the region window of each pixel is denoted as the horizontal texture complexity of the region window of each pixel, and the mean of the permutation entropy of all grayscale column vectors within the region window of each pixel is denoted as the vertical texture complexity of the region window of each pixel. The horizontal texture complexity and the vertical texture complexity reflect the complexity of the horizontal texture and the vertical texture within the region window, respectively. The calculation of the permutation entropy is a well-known technique, and the specific process will not be elaborated here.

[0051] Based on the above analysis, the difference in pit orientation of each pixel in the grayscale image of the film morphology is calculated:

[0052] G j =(Hg j -Hd j )×(Hg j +Hd j);

[0053] In the formula, G j Hg represents the difference in pit direction at the j-th pixel. j and Hd j Let be the horizontal texture complexity and the vertical texture complexity of the window region of the j-th pixel, respectively.

[0054] Because the pit defects formed during the atomic-level electromagnetic shock process of LMC have varying sizes and irregular shapes, the complex changes between the horizontal and vertical textures of the pit defect locations in the grayscale image of the film morphology identified by the smart sensor will show significant differences. This can effectively enhance the pit defect features in the grayscale image of the film morphology, which is beneficial for more accurately identifying pit defects on the surface coating of plastic substrates.

[0055] Step 3: Based on the gradient changes in the local neighborhood of each pixel and the difference in the pit direction of each pixel, obtain the pit edge crack feature value of each pixel.

[0056] Typically, the instability during the atomic-level electromagnetic shock process of LMC technology can easily lead to cracks appearing at the edges of the pits in the film morphology identified by smart sensors, causing significant and chaotic changes in the orientation of the pits at the edges.

[0057] Furthermore, the grayscale image of the film morphology is input into the edge detection algorithm. The edge detection algorithm is used to obtain the gradient magnitude of all pixels in the grayscale image of the film morphology, and the image obtained after edge detection is used as the gradient map of the film morphology. The edge detection algorithm can be the Sobel edge detection algorithm or the Canny edge detection algorithm. This embodiment does not make a specific limitation. In this embodiment, the Canny edge detection algorithm is selected to obtain the gradient magnitude of all pixels in the grayscale image of the film morphology and the gradient map of the film morphology. The Canny edge detection algorithm is a well-known technology, and the specific process will not be described in detail.

[0058] If the gradient change within the region window of a certain pixel is greater and the variation in the pit direction is more inconsistent, it can reflect the greater the edge cracking feature of the pit defect to a certain extent. Conversely, when the pixel is not located at the pit defect position, the gradient change within the region window is small and the variation in the pit direction is not obvious. At this time, it does not have the edge cracking feature of the pit defect.

[0059] Based on the above analysis, the pit and crack feature values ​​of each pixel in the grayscale image of the film morphology are calculated:

[0060]

[0061] In the formula, D jLet F be the feature value of the pit edge crack of the j-th pixel, n be the number of neighboring pixels in the region window of the j-th pixel, and F be the feature value of the pit edge crack of the j-th pixel. j and G j F represents the gradient magnitude and pit direction difference of the j-th pixel, respectively. j,g and G j,g These represent the gradient magnitude and the difference in pit direction for the g-th pixel within the region window of the j-th pixel, respectively.

[0062] Because the atomic-level electromagnetic shock force of the LMC process is too great, pit defects are formed. The edges of the pit locations are often accompanied by edge cracks. Therefore, the edge crack feature value of the pit reflects the edge crack features on the pit defects in the grayscale image of the film morphology identified by the smart sensor. The larger the edge crack feature on the pit defect, the more clearly the coating defect features based on the LMC process can be highlighted, thereby achieving more accurate identification of pit defect features on the film morphology. This avoids the problems of false detection and missed detection when performing defect detection on the surface of plastic substrates after coating.

[0063] Furthermore, in this embodiment, for ease of understanding and description, the gray values ​​of each pixel in the grayscale image of the film morphology are replaced with the corresponding pit and crack feature values. All pixels in the grayscale image of the film morphology are traversed for replacement, and the replaced feature map is recorded as the pit and crack feature map of the film morphology.

[0064] Step 4: Obtain the pit feature value of each pixel by using the pit direction difference degree and pit edge crack feature value, so as to extract the suspected pit feature points of each pixel, and combine the positional relationship between each pixel and each suspected pit feature point to obtain the pit density salience of each pixel.

[0065] Generally, pits are formed due to the unstable operating current of LMC process. When pits are formed, multiple pits are often formed in a local area, rather than just one pit. Moreover, the distance between different pits in the local area is relatively close. This reflects the characteristic phenomenon of dense pits in the film morphology image identified by smart sensors.

[0066] Meanwhile, the more obvious the difference in pit orientation and the pit edge crack features at the pixel positions in the grayscale image of the film morphology, the more prominent the pit defects in the film morphology will be. Therefore, the sum of the difference in pit orientation and the pit edge crack feature value of each pixel in the grayscale image of the film morphology is recorded as the pit feature value of each pixel in the grayscale image of the film morphology. The pit feature value reflects the characteristics of pit defects in the grayscale image of the film morphology identified by the intelligent sensor. The larger the pit feature value, the more likely the pixel is to be located in a pit position.

[0067] Furthermore, in order to analyze the dense pitting feature in the film morphology image identified by the intelligent sensor, and in combination with the coating characteristics of the LMC process, to more accurately identify the pitting defect features on the film surface, the average value of the pit feature values ​​of all pixels in the film morphology grayscale image is used as the screening threshold, and pixels with pit feature values ​​greater than the screening threshold are designated as target pixels. In this embodiment, with each pixel as the center, the K target pixels closest to the center pixel are selected and denoted as the K suspected pitting feature points of each pixel in the film morphology grayscale image. In this embodiment, K is set to 20, but the implementer can choose the value according to the actual situation. Meanwhile, the calculation of the distance between pixels is a prior art and will not be described in detail in this embodiment.

[0068] Based on the above analysis, the pit density saliency of each pixel in the grayscale image of the film morphology is calculated:

[0069]

[0070] In the formula, V j Let be the saliency of the pit density at the j-th pixel, exp() be an exponential function with the natural constant as the base, and ds j,k DS is the Euclidean distance between the j-th pixel and its k-th suspected concave feature point. j,k Let K be the average of the pit feature values ​​of the j-th pixel and its k-th suspected pit feature point, where K is the number of suspected pit feature points.

[0071] The pit density significance reflects the characteristic phenomenon of dense pits in the film morphology identified by the smart sensor after LMC coating. The greater the pit density significance, the more it can highlight the dense characteristics of pit defects after LMC coating. In combination with the coating characteristics of LMC process, it is beneficial to more accurately identify the pit defect characteristics on the film surface and improve the accuracy of defect detection of coating on plastic substrate surface.

[0072] Furthermore, in this embodiment, for ease of understanding and description, the gray values ​​of each pixel in the grayscale image of the film morphology are replaced with the corresponding pit density saliency. All pixels in the grayscale image of the film morphology are traversed for replacement, and the replaced feature map is recorded as the pit density feature map of the film morphology.

[0073] Step 5: Using the quaternary Fourier saliency detection algorithm, the gray value, gradient value, pit edge crack feature value and pit density saliency of each pixel are combined to obtain the saliency map of the film pit, and the region of coating pit defect is obtained by image segmentation.

[0074] During the coating process on the surface of plastic substrates using LMC technology, small-scale pit defects may be formed due to the impact of metal atoms and metal ions. The pit defect features on the surface of plastic substrates identified by smart sensors are often relatively small.

[0075] Therefore, this embodiment utilizes the quaternary Fourier saliency detection algorithm, combined with the pit features after LMC coating, to improve the saliency of pits in the film morphology. The grayscale image of the film morphology, the gradient map of the film morphology, the pit edge crack feature map, and the pit density feature map are used as four feature maps in the phase spectrum model PQRT of the quaternion Fourier transform, and input into the quaternary Fourier saliency detection algorithm to obtain the film pit saliency map. It should be noted that the quaternary Fourier saliency detection algorithm is a known technology, and the specific process will not be elaborated further. The schematic diagram of the process for obtaining the film pit saliency map is shown below. Figure 2 As shown.

[0076] Finally, an image segmentation algorithm is used to obtain the region of coating defects on the surface of the plastic substrate. The image segmentation algorithm can be a region growing segmentation algorithm, an Otsu threshold segmentation algorithm, or a watershed segmentation algorithm. In this embodiment, the watershed segmentation algorithm is used to segment the saliency map of the film pits to segment the region image of the film pit defects, thereby realizing the defect detection of the coating on the surface of the plastic substrate. The watershed segmentation algorithm is a well-known technology, and the specific process will not be described in detail.

[0077] It is understood that references to "one embodiment" or "some embodiments" in this specification mean that one or more embodiments of this application include the specific features, structures, or characteristics described in connection with that embodiment. Therefore, the appearance of phrases such as "in one embodiment," "in some embodiments," "in other embodiments," or "in still other embodiments" in different parts of this specification does not necessarily refer to the same embodiment, but rather means "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0078] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous. Moreover, the sequence numbers of the steps in the embodiments do not imply a specific order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments in this specification.

[0079] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for detecting surface coating defects on plastic substrates based on LMC technology, characterized in that, Includes the following steps: Obtain grayscale images of the film morphology on the surface of a plastic substrate; By analyzing the distribution and variation of grayscale texture in the local neighborhood of each pixel in the grayscale image of the film morphology, the difference in the direction of the pits of each pixel is obtained. Based on the gradient changes in the local neighborhood of each pixel and the difference in the pit direction of each pixel, the pit edge crack feature value of each pixel is obtained. By using the pit direction difference degree and pit edge crack feature value, the pit feature value of each pixel is obtained, so as to extract the suspected pit feature points of each pixel, and combine the positional relationship between each pixel and each suspected pit feature point to obtain the pit density salience of each pixel. The quaternary Fourier saliency detection algorithm is used to combine the gray value, gradient value, pit edge crack feature value and pit density saliency of each pixel to obtain the saliency map of the film pit, and the region of coating pit defect is obtained by image segmentation. The method for obtaining the difference in pit direction of each pixel is as follows: ； In the formula, Let the pit direction difference of the j-th pixel be , and These represent the horizontal and vertical texture complexity of the window region at the j-th pixel, respectively. The methods for obtaining the horizontal and vertical texture complexity are as follows: Obtain each grayscale row vector and each grayscale column vector within the pixel region window. Use the average of the permutation entropy of all grayscale row vectors within the region window of each pixel as the horizontal texture complexity of the region window of each pixel. Use the average of the permutation entropy of all grayscale column vectors within the region window of each pixel as the vertical texture complexity of the region window of each pixel. The method for obtaining the pit edge crack feature values ​​of each pixel is as follows: ； In the formula, Let the feature value of the pit edge crack at the j-th pixel be . Let be the number of neighboring pixels within the window of the j-th pixel region. and These represent the gradient magnitude and the difference in pit direction for the j-th pixel, respectively. and These represent the gradient magnitude and the pit direction difference of the g-th pixel within the j-th pixel region window, respectively; where the gradient magnitude of the pixel is obtained through an edge detection algorithm. The method for calculating the saliency of the pit density of each pixel is as follows: ； In the formula, Let be the saliency of the pit density at the j-th pixel. It is an exponential function with the natural constant as its base. Let be the Euclidean distance between the j-th pixel and its k-th suspected concave feature point. Let K be the average of the pit feature values ​​of the j-th pixel and its k-th suspected pit feature point, where K is the number of suspected pit feature points.

2. The method for detecting surface coating defects of plastic substrates based on LMC process as described in claim 1, characterized in that, Centered on each pixel, a window of a preset size is used as the region window for each pixel.

3. The method for detecting surface coating defects of plastic substrates based on LMC process as described in claim 1, characterized in that, The pit feature value of each pixel is the sum of the pit direction difference degree and the pit edge crack feature value of each pixel.

4. The method for detecting surface coating defects of plastic substrates based on LMC process as described in claim 1, characterized in that, The step of extracting suspected concave feature points of each pixel further includes: The average value of the pit feature value of all pixels is used as the screening threshold, and pixels with pit feature values ​​greater than the screening threshold are all designated as target pixels. Centered on each pixel, the K target pixels that are closest to each pixel are selected as the K suspected pit feature points of each pixel.

5. The method for detecting surface coating defects of plastic substrates based on LMC process as described in claim 1, characterized in that, The method for obtaining the saliency map of the film pits is as follows: Edge detection is performed on the grayscale image of the film morphology to obtain the gradient map of the film morphology; The gray values ​​of each pixel in the grayscale image of the film morphology are replaced with the pit edge crack feature value and the pit density saliency, respectively. The images after replacement are denoted as pit edge crack feature map and pit density feature map, respectively. The grayscale image of the film morphology, the gradient map of the film morphology, the feature map of pit edge cracks, and the feature map of pit density are used as four feature maps in the phase spectrum model PQRT of quaternion Fourier transform. The saliency map of the film pit is obtained by using the quaternion Fourier saliency detection algorithm.

6. The method for detecting surface coating defects of plastic substrates based on LMC process as described in claim 1, characterized in that, The method for obtaining the region of the coating pit defect is as follows: the watershed segmentation algorithm is used to segment the saliency map of the film pit to obtain the segmented film pit defect region.

Images