A defect detection and identification method and device based on image detection

Abstract

The present application discloses a defect detection and identification method based on image detection, which relates to the field of image processing technology and comprises the following steps: S1: an industrial camera cooperates with a light source to capture image data of a plate in actual production; S2: a processor receives the plate data image captured by the industrial camera and pre-processes the plate data image; S3: an image processing algorithm is used to extract features of pits and shadows, and corresponding features are extracted for each image captured by the industrial camera, and feature matching is performed; S4: based on the extracted feature matching results, images captured by multiple industrial cameras are aligned to the same coordinate system using an image registration algorithm; S5: the registered image is verified, and finally the detection result of the pits on the plate is output; through the cooperation of the industrial camera and the light source, as well as the steps of image pre-processing and feature extraction, defects such as pits on the surface of the plate can be accurately captured and identified, and the possibility of missed detection and false detection can be reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a defect detection and identification method and device based on image detection. Background Art

[0002] With the efficient application of Internet technology in smart factories, more and more factories are moving towards intelligent processing. In the board industry, the traditional classification of board defects mainly relies on the experience of workers, which is highly subjective and inefficient. Therefore, it is necessary to introduce automated means to achieve accurate classification of board defects. At present, for the detection of board defects, images are generally used for detection, and images are undoubtedly the most efficient and easiest to obtain information. However, the high similarity between rotten knots and dead knots in the board defect category, the high similarity between tree cores, wood cracks, and straight textures, and the high similarity between tree knots and mountain textures make it difficult to identify the above board defects using algorithms for image detection. In addition, the convolutional neural network model based on transfer learning in the prior art can automatically extract high-order features and achieve accurate classification of high-similarity images; however, the convolutional neural network has a lot of parameters. If it is trained from scratch, its training time is long, the computing power consumption is large, and a large amount of labeled data is required, which brings limitations to the implementation of the transfer learning model.

[0003] In the prior art, a Chinese invention patent with announcement number CN113610831B discloses a plate defect detection method based on computer image technology and transfer learning. First, the data is preprocessed; then a multi-layer convolutional neural network transfer learning model is constructed; then the preprocessed data is trained using the constructed transfer learning model; secondly, the trained data is processed to obtain a batch of new image data; then the new image data is input into the transfer learning model for training to obtain a critical value; then the new image data is searched for defect features, a computer image technology model is constructed, and then the defect features are quantified, and finally the critical value is combined with the threshold of the defect classification, thereby obtaining the final plate defect detection model to be detected, and plate defect detection is performed according to the plate defect detection model.

[0004] However, some tiny pits may appear on the plate, which may leave obvious marks on the surface of the plate, affecting the appearance and strength of the plate, and will have a serious impact on the load-bearing capacity of the plate. If there are too many pits, the processing efficiency of the plate will be reduced, and the scrap rate will be increased. In addition, the pits may not be presented by only photographing the plate with an industrial camera, which reduces the accuracy of plate defect detection. Summary of the invention

[0005] The embodiments of the present application provide a defect detection and identification method and device based on image detection, thereby solving the problem in the prior art that pits on a plate cannot be presented, thereby affecting the accuracy of plate defect detection.

[0006] The present invention provides a method for defect detection and recognition based on image detection, comprising the following steps:

[0007] S1: The industrial camera and the light source cooperate to capture the image data of the plate in actual production, and multiple industrial cameras transmit the captured plate image data to the processor;

[0008] S2: The processor receives the plate data image taken by the industrial camera and pre-processes the plate data image, including image background removal, image enhancement, image denoising, contrast enhancement, and grayscale processing, so as to improve the image quality and highlight the features of pits and shadows;

[0009] S3: Use image processing algorithms to extract features of pits and shadows, where pit features include shape and depth, and shadow features include brightness, boundary, shape and size. Image processing algorithms include threshold segmentation, morphological processing, edge detection and contour extraction. For each image taken by an industrial camera, extract corresponding features and perform feature matching.

[0010] S4: Based on the extracted feature matching results, the images taken by multiple industrial cameras are aligned to the same coordinate system using an image registration algorithm;

[0011] S5: Verify the registered image to check whether the pits and shadow features are accurately aligned, use visual inspection to evaluate, and adjust and optimize the light source position, industrial camera parameters and image processing algorithm based on the verification results to improve the registration accuracy and pit detection accuracy, and finally output the detection results of the plate pits.

[0012] Furthermore, step S3 further includes the following steps:

[0013] S301: preprocessing the acquired image, including background removal, contrast enhancement, noise removal, etc., to improve image quality and highlight pit features;

[0014] S302: applying threshold segmentation technology to convert the image into a binary image, presetting a threshold, and distinguishing the pit area from the background area;

[0015] S303: performing morphological processing on the binary image, including erosion and dilation operations, wherein the erosion operation can remove small noise points and make the edge of the pit clearer; and the dilation operation can enhance the internal area of ​​the pit;

[0016] S304: using the Canny edge detection algorithm to perform edge detection on the image after morphological processing, identifying the area with obvious grayscale changes in the image, and obtaining preliminary information on the pit contour through edge detection;

[0017] S305: Based on the edge detection, the preliminary information of the pit contour is extracted to obtain the shape, depth and size characteristics of the pit.

[0018] Furthermore, the threshold segmentation in step S3 can separate the shadow from the background in the shadow detection and simplify the image processing, and obtain a binary image after the threshold segmentation, which can be used as the input of the subsequent morphological processing, feature extraction and other steps. The formula is: S(x,y) = {255, if I(x,y) <T0,otherwise},

[0019] Where T is the threshold, S(x,y) represents the binary image after segmentation, and I(x,y) represents the grayscale value of the original grayscale image at the pixel position (x,y).

[0020] Furthermore, the morphological processing in step S3 can reduce the foreground area through the corrosion operation, which helps to remove small noise and interference, making the pit area clearer. The corrosion operation can make the edge of the pit more obvious, and the dilation operation can enhance the internal area of ​​the pit.

[0021] By performing a dilation operation first and then an erosion operation through a closing operation, small holes and broken connections in the foreground area can be filled, thereby obtaining a more complete pit shape.

[0022] By performing an erosion operation first and then a dilation operation through an open operation, small protrusions can be removed and the boundaries can be smoothed, making the shape of the pit smoother.

[0023] Corrosion: E(x,y)=min{S(x+i,y+j)|(i,j)∈K}

[0024] Dilation: D(x,y)=max{S(xi,yj)|(i,j)∈K}

[0025] Among them, E(x, y) and D(x, y) represent the images after corrosion and expansion respectively, S(x, y) represents the input binary image, and K represents the structural element.

[0026] Furthermore, step S4 further includes the following steps:

[0027] S401: Establish the same or similar feature points in the image information taken by different industrial cameras to find the corresponding relationship between these image information;

[0028] S402: Fusing image information from different perspectives to form a comprehensive and accurate representation of pit and shadow features.

[0029] Furthermore, the number of industrial cameras is 3-5, and the shooting angle of the industrial cameras is 45 degrees to 90 degrees, and the industrial cameras are evenly distributed on both sides above the plate;

[0030] Select two plate images taken from different perspectives and name them Image1 and Image2 respectively;

[0031] Detect and describe feature points in two plate images respectively;

[0032] By comparing the descriptors of these feature points, 100 pairs of matching feature points are found;

[0033] According to the correspondence between the two plate images, the feature points corresponding to the same physical position in the two images are matched;

[0034] Calculate the transformation matrix based on the matched feature points and align Image2 with Image1;

[0035] Transform Image2 into the same coordinate system as Image1 to ensure spatial consistency between the two images.

[0036] Furthermore, step S5 further includes the following steps:

[0037] S501: Select a certain number of sample images from the registered images for verification, where the sample images contain pits and shadow features of different sizes, shapes and positions to cover various possible situations;

[0038] S502: Determine an index for evaluating the alignment accuracy of pits and shadow features, such as overlap rate, center point distance, and edge alignment;

[0039] S503: Performing manual visual inspection on the selected verification samples to verify the alignment of the pit and shadow features;

[0040] S504: Use an automated algorithm to detect pit and shadow features on the verification sample; compare the algorithm detection result with the manual visual inspection result, and calculate the value of the verification index; if the algorithm detection result is consistent with the manual visual inspection result, and the value of the verification index is within an acceptable range, it can be considered that the feature alignment is accurate;

[0041] S505: Perform a detailed analysis on the completed verification results, determine the samples with inaccurate alignment of pits and shadow features, identify the common problems in these samples, and determine the light source parameters that need to be adjusted based on the analysis results, including the height and illumination angle of the light source.

[0042] Furthermore, the light source is an LED lamp, the irradiation angle of the light source is 30 degrees to 60 degrees, and the height of the light source is 10 cm to 50 cm above the plate.

[0043] Furthermore, step S5 further includes the following steps:

[0044] S506: extracting images of pits on the plate whose depth exceeds 10% of the plate thickness;

[0045] S507: Connect two pits with the farthest distance between them;

[0046] S508: The structural strength of the plate without pits is set to 100%, and the structural strength of the plate with two pits is set to x%. Each pit will cause the structural strength of the plate to decrease by y%.

[0047] S509: setting the number of pits on the connecting line to n, and calculating the remaining structural strength of the plate; when the remaining structural strength of the plate is greater than 50%, determining that the plate is easy to break;

[0048] Among them, y% is the coefficient of reduction in the structural strength of the plate caused by each pit, x% is the structural strength of the plate with two pits, and n is the number of pits.

[0049] A defect detection and identification device based on image detection is used to execute a defect detection and identification method based on image detection, comprising an image input module, an image detection module, an output module and a verification module.

[0050] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0051] Through the cooperation of industrial cameras and light sources, as well as steps such as image preprocessing and feature extraction, defects such as pits on the surface of the plate can be accurately captured and identified, reducing the possibility of missed detection and false detection. Multiple industrial cameras simultaneously capture images at different angles, which can shorten the detection time and improve detection efficiency. The automated image processing and analysis process also reduces the need for manual intervention and further improves efficiency. Through accurate detection and identification of defects such as pits in the plate, problems in the production process can be discovered and dealt with in a timely manner, thereby improving the level of product quality control. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is a flow chart of Example 1 of the present application;

[0053] Figure 2 This is a schematic diagram of the plate testing for this application;

[0054] Figure 3 This is a flow chart of the second embodiment of the present application;

[0055] Figure 4 This is a flow chart of the third embodiment of the present application;

[0056] Figure 5 This is a flow chart of the fourth embodiment of the present application;

[0057] Figure 6 This is a flow chart of Embodiment 5 of the present application;

[0058] Figure 7 This is a schematic diagram of plate detection in Example 5 of the present application;

[0059] Figure 8 A schematic diagram of the structure of a defect detection and identification device based on image detection provided in this application. DETAILED DESCRIPTION

[0060] To facilitate the understanding of the present invention, the present application will be described more comprehensively below with reference to the relevant drawings; the drawings show preferred embodiments of the present invention, but the present invention can be implemented in many different forms and is not limited to the embodiments described herein; on the contrary, the purpose of providing these embodiments is to enable a more thorough and comprehensive understanding of the disclosed content of the present invention.

[0061] It should be noted that the terms “vertical”, “horizontal”, “up”, “down”, “left”, “right” and similar expressions used in this document are only for illustrative purposes and do not represent the only implementation method.

[0062] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which the present invention belongs; the terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention; the term "and / or" used herein includes any and all combinations of one or more related listed items.

[0063] Embodiment 1: To solve the problem that the pits may not be shown when photographing the plate only with an industrial camera, thus reducing the accuracy of plate defect detection:

[0064] like Figure 1-Figure 2 As shown, the embodiment of the present application provides a defect detection and identification method based on image detection, comprising the following steps:

[0065] S1: The industrial camera and the light source cooperate to capture the image data of the plate in actual production, and multiple industrial cameras transmit the captured plate image data to the processor;

[0066] S2: The processor receives the plate data image taken by the industrial camera and pre-processes the plate data image, including image background removal, image enhancement, image denoising, contrast enhancement, and grayscale processing, so as to improve the image quality and highlight the features of pits and shadows;

[0067] S3: Use image processing algorithms to extract features of pits and shadows, where pit features include shape and depth, and shadow features include brightness, boundary, shape and size. Image processing algorithms include threshold segmentation, morphological processing, edge detection and contour extraction. For each image taken by an industrial camera, extract corresponding features and perform feature matching.

[0068] S4: Based on the extracted feature matching results, the images taken by multiple industrial cameras are aligned to the same coordinate system using an image registration algorithm;

[0069] S5: Verify the registered image to check whether the pits and shadow features are accurately aligned, use visual inspection to evaluate, and adjust and optimize the light source position, industrial camera parameters and image processing algorithm based on the verification results to improve the registration accuracy and pit detection accuracy, and finally output the detection results of the plate pits.

[0070] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0071] Through the cooperation of industrial cameras and light sources, as well as steps such as image preprocessing and feature extraction, defects such as pits on the surface of the plate can be accurately captured and identified, reducing the possibility of missed detection and false detection. Multiple industrial cameras simultaneously capture images at different angles, which can shorten the detection time and improve detection efficiency. The automated image processing and analysis process also reduces the need for manual intervention and further improves efficiency. Through accurate detection and identification of defects such as pits in the plate, problems in the production process can be discovered and dealt with in a timely manner, thereby improving the level of product quality control.

[0072] Example 2: In order to more accurately extract the features of pits and shadows on the plate, Figure 3 As shown, step S3 also includes:

[0073] S301: preprocessing the acquired image, including background removal, contrast enhancement, noise removal, etc., to improve image quality and highlight pit features;

[0074] S302: applying threshold segmentation technology to convert the image into a binary image, presetting a threshold, and distinguishing the pit area from the background area;

[0075] S303: performing morphological processing on the binary image, including erosion and dilation operations, wherein the erosion operation can remove small noise points and make the edge of the pit clearer; and the dilation operation can enhance the internal area of ​​the pit;

[0076] S304: using the Canny edge detection algorithm to perform edge detection on the image after morphological processing, identifying the area with obvious grayscale changes in the image, and obtaining preliminary information on the pit contour through edge detection;

[0077] S305: Based on the edge detection, contour extraction is performed to obtain the shape, depth and size characteristics of the pit.

[0078] The threshold segmentation in step S3 can separate the shadow from the background in shadow detection and simplify image processing, and obtain a binary image after threshold segmentation, which can be used as input for subsequent morphological processing, feature extraction and other steps. The formula is: S(x,y)={255,if I(x,y) <T0,otherwise},

[0079] Where T is the threshold, S(x,y) represents the binary image after segmentation, and I(x,y) represents the grayscale value of the original grayscale image at the pixel position (x,y);

[0080] The morphological processing in step S3 can reduce the foreground area through the corrosion operation, which helps to remove small noise and interference, making the pit area clearer. The corrosion operation can make the edge of the pit more obvious, and the dilation operation can enhance the internal area of ​​the pit.

[0081] By performing a dilation operation first and then an erosion operation through a closing operation, small holes and broken connections in the foreground area can be filled, thereby obtaining a more complete pit shape.

[0082] By performing an erosion operation first and then a dilation operation through an open operation, small protrusions can be removed and the boundaries can be smoothed, making the shape of the pit smoother.

[0083] Corrosion: E(x,y)=min{S(x+i,y+j)|(i,j)∈K}

[0084] Dilation: D(x,y)=max{S(xi,yj)|(i,j)∈K}

[0085] Among them, E(x, y) and D(x, y) represent the images after corrosion and expansion respectively, S(x, y) represents the input binary image, and K represents the structural element.

[0086] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0087] Through the above image processing steps, the features of the pits and shadows on the plate can be accurately extracted, the image processing process is simplified, and the processing efficiency is improved.

[0088] Embodiment 3: In order to further improve the detection efficiency and strengthen the image acquisition of the plate, Figure 4 As shown, step S4 also includes:

[0089] S401: Establish the same or similar feature points in the image information taken by different industrial cameras to find the corresponding relationship between these image information;

[0090] S402: Fusing image information from different perspectives to form a comprehensive and accurate representation of pit and shadow features;

[0091] The number of industrial cameras is 3-5, and the shooting angle of the industrial cameras is 45 degrees to 90 degrees. The industrial cameras are evenly distributed on both sides above the plate.

[0092] Select two plate images taken from different perspectives and name them Image1 and Image2 respectively;

[0093] Detect and describe feature points in two plate images respectively;

[0094] By comparing the descriptors of these feature points, 100 pairs of matching feature points are found;

[0095] According to the correspondence in the two plate images, the matched feature points correspond to the same physical positions in the two images;

[0096] Calculate the transformation matrix based on the matched feature points and align Image2 with Image1;

[0097] Transform Image2 into the same coordinate system as Image1 to ensure spatial consistency between the two images.

[0098] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0099] By using multiple industrial cameras, comprehensive coverage of the plate surface can be achieved, reducing the possibility of missed detection. Different shooting angles can capture light reflection and shadow information in different directions, helping to improve the accuracy and reliability of defect detection.

[0100] Example 4: In order to show the characteristics of the pit shadow on the plate as much as possible, Figure 5 As shown, step S5 also includes:

[0101] S501: Select a certain number of sample images from the registered images for verification, where the sample images contain pits and shadow features of different sizes, shapes and positions to cover various possible situations;

[0102] S502: Determine an index for evaluating the alignment accuracy of pits and shadow features, such as overlap rate, center point distance, and edge alignment;

[0103] S503: Performing manual visual inspection on the selected verification samples to verify the alignment of the pit and shadow features;

[0104] S504: Use an automated algorithm to detect pit and shadow features on the verification sample; compare the algorithm detection result with the manual visual inspection result, and calculate the value of the verification index; if the algorithm detection result is consistent with the manual visual inspection result, and the value of the verification index is within an acceptable range, it can be considered that the feature alignment is accurate;

[0105] S505: Perform a detailed analysis on the completed verification results, determine the samples with inaccurate alignment of pits and shadow features, identify the common problems in these samples, and determine the light source parameters that need to be adjusted based on the analysis results, including the height and illumination angle of the light source.

[0106] The light source is an LED lamp, the irradiation angle of the light source is 30 degrees to 60 degrees, and the height of the light source is 10 cm to 50 cm above the plate.

[0107] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0108] By adjusting the position of the light source, the distribution of light on the surface of the plate and the projection of shadows can be changed, making pits and shadow features more obvious and easier to identify, which helps to improve the accuracy of feature extraction and reduce the possibility of missed detection and false detection. Appropriate light source parameter settings can improve the brightness and contrast of the image, reduce reflection and shadow interference, and make the image clearer and more detailed. Real-time monitoring and adjustment of light source parameters can ensure that the detection system is always in the best condition.

[0109] Example 5: When the depth of the pit on the plate exceeds 10% of the plate thickness, it may affect the structural strength of the plate, and when the deeper pits on multiple plates approach a straight line, the plate may be easily broken. Based on the above problems, Figure 6-Figure 7 As shown, step S5 also includes:

[0110] S506: extracting images of pits on the plate whose depth exceeds 10% of the plate thickness;

[0111] S507: Connect two pits with the farthest distance between them;

[0112] S508: The structural strength of the plate without pits is set to 100%, and the structural strength of the plate with two pits is set to x%. Each pit will cause the structural strength of the plate to decrease by y%.

[0113] S509: setting the number of pits on the connecting line to n, calculating the remaining structural strength of the plate; when the remaining structural strength of the plate is greater than 50%, determining that the plate is easy to break.

[0114] Among them, y% is the coefficient of reduction in the structural strength of the plate caused by each pit, x% is the structural strength of the plate with two pits, and n is the number of pits.

[0115] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0116] By connecting the extracted pits and setting the structural strength of the pit-free plate to 100%, it is possible to quantitatively evaluate the impact of the pits on the structural strength of the plate, intuitively understand the extent to which the pits weaken the performance of the plate, and through continuous monitoring and analysis of the impact of pits on the structural strength of the plate, optimize the production process and reduce the adverse factors that may cause pits.

[0117] The present application also provides a defect detection and identification device based on image detection, such as Figure 8 As shown, a defect detection and identification method based on image detection is performed, including an image input module, an image detection module, an output module and a verification module. The image input module can take a plate image, the image detection module can receive a plate image and perform detection calculation processing, the output module outputs the detection result of the image detection module, and the verification module is used to verify the accuracy of the detection.

[0118] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A defect detection and identification method based on image detection, characterized in that: The following steps are involved: S1: The industrial camera and the light source cooperate to capture the image data of the plate in actual production, and multiple industrial cameras transmit the captured plate image data to the processor; S2: The processor receives the plate data image taken by the industrial camera and pre-processes the plate data image, including image background removal, image enhancement, image denoising, contrast enhancement, and grayscale processing, so as to improve the image quality and highlight the features of pits and shadows; S3: Use image processing algorithms to extract features of pits and shadows, where pit features include shape and depth, and shadow features include brightness, boundary, shape and size. Image processing algorithms include threshold segmentation, morphological processing, edge detection and contour extraction. For each image taken by an industrial camera, extract corresponding features and perform feature matching. S4: Based on the extracted feature matching results, the images taken by multiple industrial cameras are aligned to the same coordinate system using an image registration algorithm; S5: Verify the registered image to check whether the pits and shadow features are accurately aligned, use visual inspection to evaluate, and adjust and optimize the light source position, industrial camera parameters and image processing algorithm based on the verification results to improve the registration accuracy and pit detection accuracy, and finally output the detection results of the plate pits; Step S5 also includes the following steps: S501: Select a certain number of sample images from the registered images for verification, where the sample images contain pits and shadow features of different sizes, shapes and positions to cover various possible situations; S502: Determine an indicator for evaluating the alignment accuracy of pits and shadow features; S503: Performing manual visual inspection on the selected verification samples to verify the alignment of the pit and shadow features; S504: Use an automated algorithm to detect pit and shadow features on the verification sample; compare the algorithm detection result with the manual visual inspection result, and calculate the value of the verification index; if the algorithm detection result is consistent with the manual visual inspection result, and the value of the verification index is within an acceptable range, it is considered that the feature alignment is accurate; S505: Analyze the completed verification results in detail, determine the samples with inaccurate alignment of pits and shadow features, identify the common problems in these samples, and determine the light source parameters that need to be adjusted based on the analysis results, including the height and irradiation angle of the light source; S506: extracting images of pits on the plate whose depth exceeds 10% of the plate thickness; S507: Connect two pits with the farthest distance between them; S508: The structural strength of the plate without pits is set to 100%, and the structural strength of the plate with two pits is set to x%. Each pit will cause the structural strength of the plate to decrease by y%. S509: setting the number of pits on the connecting line to n, and calculating the remaining structural strength of the plate; when the remaining structural strength of the plate is less than 50%, determining that the plate is easy to break; Among them, y% is the coefficient of reduction in the structural strength of the plate caused by each pit, x% is the structural strength of the plate with two pits, and n is the number of pits.

2. The defect detection and identification method based on image detection according to claim 1, characterized in that: Step S3 also includes the following steps: S301: preprocessing the acquired image to improve image quality and highlight pit features; S302: applying threshold segmentation technology to convert the image into a binary image, presetting a threshold, and distinguishing the pit area from the background area; S303: performing morphological processing on the binary image, including erosion and dilation operations, wherein the erosion operation can remove small noise points and make the edge of the pit clearer; and the dilation operation can enhance the internal area of ​​the pit; S304: using the Canny edge detection algorithm to perform edge detection on the image after morphological processing, identifying the area with obvious grayscale changes in the image, and obtaining preliminary information on the pit contour through edge detection; S305: Based on the edge detection, the preliminary information of the pit contour is extracted to obtain the shape, depth and size characteristics of the pit.

3. The defect detection and identification method based on image detection according to claim 2 is characterized in that: The threshold segmentation in step S3 can separate the shadow from the background in shadow detection and simplify image processing, and obtain a binary image after threshold segmentation, which can be used as input for subsequent morphological processing and feature extraction. The formula is: Where T is the threshold, S(x,y) represents the binary image after segmentation, and I(x,y) represents the grayscale value of the original grayscale image at the pixel position (x,y).

4. The defect detection and identification method based on image detection according to claim 3 is characterized in that: The morphological processing in step S3 can reduce the foreground area through the corrosion operation, which helps to remove small noise and interference, making the pit area clearer. The corrosion operation can make the edge of the pit more obvious, and the dilation operation can enhance the internal area of ​​the pit. By performing a dilation operation first and then an erosion operation through a closing operation, small holes and broken connections in the foreground area can be filled, thereby obtaining a more complete pit shape. By performing an erosion operation first and then a dilation operation through an open operation, small protrusions can be removed and the boundaries can be smoothed, making the shape of the pit smoother. Corrosion: E(x,y)=min{S(x+i,y+j)|(i,j)∈K} Dilation: D(x,y)=max{S(xi,yj)|(i,j)∈K} Among them, E(x, y) and D(x, y) represent the images after corrosion and expansion respectively, S(x, y) represents the input binary image, and K represents the structural element.

5. The defect detection and identification method based on image detection according to claim 1, characterized in that: Step S4 also includes the following steps: S401: Establish the same or similar feature points in the image information taken by different industrial cameras to find the corresponding relationship between these image information; S402: Fusing image information from different perspectives to form a comprehensive and accurate representation of pit and shadow features.

6. The defect detection and identification method based on image detection according to claim 5 is characterized in that: The number of industrial cameras is 3-5, and the shooting angle of the industrial cameras is 45 degrees to 90 degrees. The industrial cameras are evenly distributed on both sides above the plate; Select two plate images taken from different perspectives and name them Image1 and Image2 respectively; Detect and describe feature points in two plate images respectively; By comparing the descriptors of these feature points, 100 pairs of matching feature points are found; According to the correspondence between the two plate images, the feature points corresponding to the same physical position in the two images are matched; Calculate the transformation matrix based on the matched feature points and align Image2 with Image1; Transform Image2 into the same coordinate system as Image1 to ensure spatial consistency between the two images.

7. The defect detection and identification method based on image detection according to claim 1, characterized in that: The light source is an LED lamp, the irradiation angle of the light source is 30 degrees to 60 degrees, and the height of the light source is 10 cm to 50 cm above the plate.

8. A defect detection and identification device based on image detection, characterized in that: Used to execute a defect detection and identification method based on image detection as described in any one of claims 1 to 7, comprising an image input module, an image detection module, an output module and a verification module.

Images