Method for detecting gluing uniformity of copper foil adhesive tape based on image detection

Abstract

The invention relates to the technical field of image recognition, and discloses a copper foil adhesive tape gluing uniformity detection method based on image detection, which comprises the following steps: carrying out graying processing on a continuous image covering the surface of an adhesive tape, constructing a reference basis through an adhesive-free area, and eliminating the influence of micro light-transmitting scattering in a pixel neighborhood to obtain a purified grey-scale map; the purification gray level is compared with the reference standard to generate a gluing area mask, the glue layer thickness is only calculated for the pixels of the gluing area, local averaging is carried out, and the edge and micropore interference is eliminated through residual error correction and extreme value filtering to obtain the accurate thickness; performing neighborhood statistics on the accurate thickness, identifying a thickness out-of-tolerance pixel, and analyzing a continuous multi-frame thickness change trend to find a dynamic anomaly; the thickness distribution is counted according to regional blocks, a long-term uniformity trend chart is generated, a trend mask is generated according to the trend chart, the trend mask is fused with the thickness out-of-tolerance and dynamic change results, a comprehensive abnormal index is obtained, and an alarm is triggered at an abnormal pixel.

Description

Technical Field

[0001] This invention relates to the field of image recognition technology, specifically to a method for detecting the uniformity of adhesive coating on copper foil tape based on image detection. Background Technology

[0002] Copper foil tape, as an important industrial material, is widely used in electronics, electrical engineering, new energy, and communication equipment. Its performance is highly dependent on the quality of the adhesive coating on the back. The uniformity of the adhesive coating directly affects the tape's bonding strength, conductivity, long-term reliability, and product consistency. Defects such as localized areas of excessive thinness, excessive thickness, gaps, or dynamic fluctuations in the adhesive coating may lead to product bonding failure, unstable signal transmission, or premature aging, posing significant quality risks to downstream applications.

[0003] Traditional methods for inspecting adhesive coating uniformity primarily rely on manual visual sampling or point measurements using contact thickness gauges. Manual inspection is highly subjective, inefficient, and prone to missed inspections due to fatigue, making it difficult to meet the full inspection requirements of high-speed continuous production lines. Contact measurements, on the other hand, are slow, may damage the adhesive surface, and can only obtain discrete point data, failing to achieve full-area, continuous quality monitoring. In recent years, machine vision-based inspection technology has been developed due to its advantages such as non-contact operation, high speed, and rich information. However, when applied to materials with high reflectivity and complex surface characteristics, such as copper foil tape, it still faces many challenges: First, the high reflectivity of the copper foil substrate and the semi-transparent nature of the adhesive layer itself can lead to strong specular reflection and light scattering interference in the imaging, which seriously affects the stability and accuracy of the mapping relationship between grayscale values ​​and actual adhesive thickness. Second, micro-vibrations during the production process, changes in ambient light, and micropores and edge effects that may exist inside the adhesive layer can all introduce noise into the image, making it easy for traditional single-frame grayscale analysis to produce misjudgments. Furthermore, existing methods mostly focus on static uniformity assessment and lack continuous monitoring and long-term statistical analysis of the dynamic trends of the coating process, making it difficult to predict gradual process drift or occasional transient anomalies.

[0004] Therefore, there is an urgent need to develop an online detection method for coating uniformity that can effectively overcome the interference of substrate reflection and light scattering, achieve accurate extraction of adhesive layer areas, and integrate dynamic and long-term statistical analysis, so as to achieve high-precision, high-robustness, and fully automated coating quality monitoring and early warning, and ensure the stability and reliability of copper foil tape products. Summary of the Invention

[0005] This invention provides an image detection-based method for detecting the uniformity of adhesive coating on copper foil tape, which helps to solve the problems mentioned in the background art.

[0006] This invention provides the following technical solution: a method for detecting the uniformity of adhesive coating on copper foil tape based on image detection, comprising: Continuous images of the surface covered by adhesive tape are acquired and converted to grayscale. A reference baseline is constructed using images of areas without adhesive. A neighborhood window is set for each pixel. The micro-transmittance scattering offset is calculated, and the purified grayscale image is obtained by removing the micro-transmittance scattering offset from the original image. The purified grayscale is compared with the reference grayscale to generate a mask for the coated area, and the thickness and uniformity analysis is performed only on the pixels in the coated area. The initial thickness of the adhesive layer is calculated based on the grayscale difference and local averaging is performed. The residual scattering correction is used and extreme value filtering is applied to eliminate edge and micropore interference, thereby obtaining the accurate adhesive layer thickness. Perform neighborhood statistics on the precise thickness, calculate thickness uniformity, and identify pixels with excessive thickness based on the allowable range; Acquire thickness data from multiple consecutive frames and calculate the trend of change to identify dynamic anomalies; The thickness distribution is statistically analyzed by region and a long-term uniformity trend map is generated. A trend mask is generated based on the trend chart, and then filtered and integrated with the thickness deviation and dynamic change results to obtain a comprehensive anomaly index and output an anomaly alarm.

[0007] Optionally, the step of acquiring and converting continuous images of the adhesive tape surface to grayscale, constructing a reference baseline using images of the adhesive-free areas, setting a neighborhood window for each pixel, calculating the micro-transmittance scattering offset, and removing pixels from the original image to obtain a purified grayscale image includes: The surface of the tape is continuously captured across the entire frame using an industrial camera, and the captured images are converted into grayscale format. A reference image is pre-selected from a glue-free area to serve as the true reflection reference of the copper foil body. For each pixel in the entire image, a neighborhood range is set, and the grayscale fluctuation within this range is calculated to characterize the error caused by illumination, reflection and micro-transmittance scattering. This error is then removed from the original image to obtain a clean grayscale image that characterizes the true optical properties of the coating. The system processes multiple consecutive frames to form a purified grayscale image sequence after eliminating environmental optical interference.

[0008] Optionally, the step of comparing the purified grayscale with the reference grayscale and generating a mask for the coated area, and performing thickness and uniformity analysis only on the pixels of the coated area, includes: The actual adhesive coating area is identified based on the difference between the purified grayscale image and the reference image. A binary mask is constructed to completely exclude pixels in the non-adhesive coating area, so that subsequent analysis is carried out only on the adhesive layer area. By establishing the mapping relationship between grayscale and thickness through experimental calibration, the thickness information corresponding to the glue-coated pixels can be initially quantified. The average thickness is calculated within a local area to mitigate the effects of random noise and slight illumination fluctuations, thereby obtaining a local thickness map that reflects the basic thickness trend.

[0009] Optionally, the step of calculating the initial thickness of the adhesive layer based on the grayscale difference and performing local averaging, using residual scattering correction and extreme value filtering to eliminate edge and micropore interference, and obtaining the accurate adhesive layer thickness includes: The theoretical optical performance is estimated by using the local thickness mean, and the actual optical performance is compared with the theoretical results to obtain residual information representing local optical anomalies; The residual information is smoothed by the neighborhood mean method to correct the inherent optical scattering error generated during the imaging process, so that the thickness estimate is closer to the actual physical thickness. Perform local extremum filtering to remove extreme high and low values.

[0010] Optionally, performing neighborhood statistics on the precise thickness, calculating thickness uniformity, and identifying pixels with excessive thickness based on allowable intervals includes: A neighborhood analysis region is set for the final thickness map, and the uniformity evaluation index of the region is obtained by calculating the average thickness and the degree of thickness fluctuation. The thickness is judged based on the allowable thickness range of the process. When the local thickness deviates from the normal range, the pixel is marked as a static abnormal point.

[0011] Optionally, the step of acquiring thickness data from multiple consecutive frames and calculating the trend of change to identify dynamic anomalies includes: Collect multiple frames of thickness data, establish a dynamic thickness change model in the time dimension, and determine the local thickness increase or decrease trend by comparing the thickness differences between previous and subsequent frames. The rate of change per unit time is calculated by combining time intervals, and the overall change characteristics of the neighborhood are statistically analyzed. If the rate of change exceeds the set range, it is considered a dynamic anomaly.

[0012] Optionally, the step of statistically analyzing the thickness distribution by region and generating a long-term uniformity trend map includes: Obtain the overall thickness level of the image and divide it into fixed region units. Statistically analyze the thickness level and uniformity of each region. By accumulating statistics across multiple consecutive frames, a long-term thickness fluctuation trend is plotted, and a visualized uniformity trend map is generated, in which high fluctuation areas correspond to potential long-term process inhomogeneity issues.

[0013] Optionally, the step of generating a trend mask based on the trend chart and filtering and fusing it with the thickness deviation and dynamic change results to obtain a comprehensive anomaly index and output an anomaly alarm includes: Trend masks are generated by extracting regions with long-term non-uniformity based on long-term uniformity trend maps. The mask is fused and filtered with both static thickness deviation results and dynamic change anomaly results. Finally, only the abnormal locations that simultaneously exhibit long-term non-uniformity characteristics and current defect trends are retained, and alarm commands are issued to them.

[0014] The present invention has the following beneficial effects: 1. This method for detecting the uniformity of adhesive coating on copper foil tape based on image detection acquires continuous images covering the entire tape surface and constructs a baseline grayscale value using the adhesive-free area. Neighborhood analysis is then used to eliminate micro-transmittance scattering offsets, resulting in a purified grayscale image from the original image. This lays the foundation for accurate identification of the coated area. The beneficial effect of this method is that it effectively eliminates optical interference caused by the high reflectivity of the copper foil and the semi-transparent nature of the adhesive layer, allowing the image grayscale to more accurately reflect the adhesive layer thickness. This not only improves the stability and accuracy of the detection but also enables full-area, continuous monitoring. Compared with traditional manual inspection and point measurement methods, it significantly improves detection efficiency and is suitable for high-speed continuous production line applications. Simultaneously, by analyzing local grayscale fluctuations through neighborhood windows, the influence of local noise on the judgment can be eliminated at the image level, thereby ensuring the accuracy of adhesive area identification and providing a reliable data foundation for subsequent thickness calculation and uniformity evaluation.

[0015] 2. This method for detecting the uniformity of adhesive coating on copper foil tape based on image detection involves differentiating a purified grayscale image with a reference grayscale image to generate a mask for the coated area. The initial thickness of the adhesive layer is then calculated using the grayscale difference. Local averaging and residual correction are then applied, along with extreme value filtering to eliminate edge and micro-hole interference, ultimately yielding an accurate adhesive layer thickness. The advantages of this design are that it not only quantifies the adhesive layer thickness, reducing errors caused by image noise and scattering, but also eliminates the influence of local micro-holes or edge anomalies on thickness measurement, making the thickness data closer to the actual physical thickness of the adhesive layer. Through local averaging and residual correction, the thickness calculation retains the true variation characteristics of the adhesive layer while smoothing out the influence of optical interference, thus achieving high-precision thickness measurement. Extreme value filtering further ensures the reliability of the thickness data, enabling the system to accurately determine the uniformity of the adhesive layer and local defects even under complex surface conditions. It is particularly suitable for copper foil tapes with high reflectivity on metal substrates and significant adhesive layer transparency, improving the feasibility and accuracy of thickness assessment and defect identification, and providing a solid data foundation for subsequent dynamic monitoring and quality alarms.

[0016] 3. This image-based detection method for copper foil tape coating uniformity detection, based on accurate thickness calculation, sets a neighborhood analysis window for each pixel, statistically analyzes the local average thickness and calculates thickness uniformity, and simultaneously acquires multiple consecutive frames of thickness data and analyzes thickness change trends to identify local thickness anomalies and dynamic changes. The beneficial effect of this design is that it can simultaneously achieve static and dynamic coating defect monitoring, detecting immediate defects such as excessive thickness, insufficient thickness, or blank areas, and providing early warnings of trend anomalies during the coating process, such as process drift or occasional thickness fluctuations. Through neighborhood statistics, local thickness fluctuations can be quantified, reducing the impact of single-point measurement errors on overall judgment; through multi-frame thickness trend analysis, continuously changing anomalies can be captured, enabling early warning of gradual deviations in the production process. This measure not only improves the reliability of coating quality control but also provides enterprises with traceable quality data, helping to optimize process parameters and improve product consistency, thereby significantly reducing downstream application risks caused by coating defects and meeting the full-area, high-precision monitoring requirements of high-speed continuous production lines.

[0017] 4. This image-based detection method for copper foil tape coating uniformity detection divides the full-width thickness distribution into several statistical units. For each unit, the average thickness and uniformity are calculated, and a long-term uniformity trend chart is generated. Simultaneously, a trend mask is used to filter and fuse local thickness anomalies and dynamic changes, generating a comprehensive anomaly index and triggering an alarm. The beneficial effect of this design is that it achieves long-term, macroscopic, and microscopic joint analysis of coating quality. It can not only capture instantaneous local defects but also reflect cumulative non-uniformity problems during production, achieving trend-based anomaly early warning. Through trend mask filtering, the method can distinguish between occasional noise and long-term anomalies, improving the accuracy and reliability of alarms and avoiding false alarms. The output of the comprehensive anomaly index makes the detection results intuitive and usable, providing a scientific basis for production management and process improvement. This function is particularly suitable for continuous production lines where continuous monitoring of the full-width uniformity and dynamic stability of the product is required, thereby effectively improving the overall quality and consistency of copper foil tape and reducing the risk of product failure. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1, see Figure 1 A method for detecting the uniformity of adhesive coating on copper foil tape based on image detection, comprising: Continuous images of the surface covered by adhesive tape are acquired and converted to grayscale. A reference baseline is constructed using images of areas without adhesive. A neighborhood window is set for each pixel. The micro-transmittance scattering offset is calculated, and the purified grayscale image is obtained by removing the micro-transmittance scattering offset from the original image. The purified grayscale is compared with the reference grayscale to generate a mask for the coated area, and the thickness and uniformity analysis is performed only on the pixels in the coated area. The initial thickness of the adhesive layer is calculated based on the grayscale difference and local averaging is performed. The residual scattering correction is used and extreme value filtering is applied to eliminate edge and micropore interference, thereby obtaining the accurate adhesive layer thickness. Perform neighborhood statistics on the precise thickness, calculate thickness uniformity, and identify pixels with excessive thickness based on the allowable range; Acquire thickness data from multiple consecutive frames and calculate the trend of change to identify dynamic anomalies; The thickness distribution is statistically analyzed by region and a long-term uniformity trend map is generated. A trend mask is generated based on the trend chart, and then filtered and integrated with the thickness deviation and dynamic change results to obtain a comprehensive anomaly index and output an anomaly alarm.

[0021] The process involves acquiring and converting continuous images of the adhesive tape surface to grayscale, constructing a reference baseline using images of the adhesive-free areas, setting a neighborhood window for each pixel, calculating the micro-transmittance scattering offset, and removing unwanted pixels from the original image to obtain a cleaned grayscale image. The surface of the tape is continuously captured across the entire frame using an industrial camera, and the captured images are converted into grayscale format. A reference image is pre-selected from a glue-free area to serve as the true reflection reference of the copper foil body. For each pixel in the entire image, a neighborhood range is set, and the grayscale fluctuation within this range is calculated to characterize the error caused by illumination, reflection and micro-transmittance scattering. This error is then removed from the original image to obtain a clean grayscale image that characterizes the true optical properties of the coating. The system processes multiple consecutive frames to form a purified grayscale image sequence after eliminating environmental optical interference.

[0022] The process of comparing the purified grayscale with the reference grayscale and generating a mask for the coated area, and performing thickness and uniformity analysis only on the pixels in the coated area, includes: The actual adhesive coating area is identified based on the difference between the purified grayscale image and the reference image. A binary mask is constructed to completely exclude pixels in the non-adhesive coating area, so that subsequent analysis is carried out only on the adhesive layer area. By establishing the mapping relationship between grayscale and thickness through experimental calibration, the thickness information corresponding to the glue-coated pixels can be initially quantified. The average thickness is calculated within a local area to mitigate the effects of random noise and slight illumination fluctuations, thereby obtaining a local thickness map that reflects the basic thickness trend.

[0023] The process of calculating the initial thickness of the adhesive layer based on grayscale difference and performing local averaging, using residual scattering correction and extreme value filtering to eliminate edge and micropore interference, and obtaining the accurate adhesive layer thickness includes: The theoretical optical performance is estimated by using the local thickness mean, and the actual optical performance is compared with the theoretical results to obtain residual information representing local optical anomalies; The residual information is smoothed by the neighborhood mean method to correct the inherent optical scattering error generated during the imaging process, so that the thickness estimate is closer to the actual physical thickness. Perform local extremum filtering to remove extreme high and low values.

[0024] The process of performing neighborhood statistics on the precise thickness, calculating thickness uniformity, and identifying pixels with excessive thickness based on allowable intervals includes: A neighborhood analysis region is set for the final thickness map, and the uniformity evaluation index of the region is obtained by calculating the average thickness and the degree of thickness fluctuation. The thickness is judged based on the allowable thickness range of the process. When the local thickness deviates from the normal range, the pixel is marked as a static abnormal point.

[0025] The process of acquiring thickness data from multiple consecutive frames and calculating the trend of change to identify dynamic anomalies includes: Collect multiple frames of thickness data, establish a dynamic thickness change model in the time dimension, and determine the local thickness increase or decrease trend by comparing the thickness differences between previous and subsequent frames. The rate of change per unit time is calculated by combining time intervals, and the overall change characteristics of the neighborhood are statistically analyzed. If the rate of change exceeds the set range, it is considered a dynamic anomaly.

[0026] The step of statistically analyzing thickness distribution by region and generating a long-term uniformity trend map includes: Obtain the overall thickness level of the image and divide it into fixed region units. Statistically analyze the thickness level and uniformity of each region. By accumulating statistics across multiple consecutive frames, a long-term thickness fluctuation trend is plotted, and a visualized uniformity trend map is generated, in which high fluctuation areas correspond to potential long-term process inhomogeneity issues.

[0027] The process of generating a trend mask based on the trend chart, filtering and integrating it with the results of thickness deviation and dynamic changes to obtain a comprehensive anomaly index and output an anomaly alarm includes: Trend masks are generated by extracting regions with long-term non-uniformity based on long-term uniformity trend maps. The mask is fused and filtered with both static thickness deviation results and dynamic change anomaly results. Finally, only the abnormal locations that simultaneously exhibit long-term non-uniformity characteristics and current defect trends are retained, and alarm commands are issued to them.

[0028] Example 2: A method for detecting the uniformity of adhesive coating on copper foil tape based on image detection, comprising: Continuous images of the surface covered by adhesive tape are acquired and converted to grayscale. A reference baseline is constructed using images of areas without adhesive. A neighborhood window is set for each pixel. The micro-transmittance scattering offset is calculated, and the purified grayscale image is obtained by removing the micro-transmittance scattering offset from the original image. The purified grayscale is compared with the reference grayscale to generate a mask for the coated area, and the thickness and uniformity analysis is performed only on the pixels in the coated area. The initial thickness of the adhesive layer is calculated based on the grayscale difference and local averaging is performed. The residual scattering correction is used and extreme value filtering is applied to eliminate edge and micropore interference, thereby obtaining the accurate adhesive layer thickness. Perform neighborhood statistics on the precise thickness, calculate thickness uniformity, and identify pixels with excessive thickness based on the allowable range; Acquire thickness data from multiple consecutive frames and calculate the trend of change to identify dynamic anomalies; The thickness distribution is statistically analyzed by region and a long-term uniformity trend map is generated. A trend mask is generated based on the trend chart, and then filtered and integrated with the thickness deviation and dynamic change results to obtain a comprehensive anomaly index and output an anomaly alarm.

[0029] The process involves acquiring and converting continuous images of the adhesive tape surface to grayscale, constructing a reference baseline using images of the adhesive-free areas, setting a neighborhood window for each pixel, calculating the micro-transmittance scattering offset, and removing unwanted pixels from the original image to obtain a cleaned grayscale image. The sampling area is set to include the entire surface of the tape to ensure that the uniformity of the adhesive coating across the entire width can be completely covered; Use the camera at the set frame rate Images are continuously acquired and converted into grayscale images, which are then recorded as a matrix. ,in, Represents the pixel coordinates of the image. Indicates a time sequence number; Images were acquired on a separate, glue-free area. ,Will The grayscale value is used as a reference standard, denoted as . ; in, This indicates the grayscale value reflected by the copper foil itself. For each pixel Set Neighborhood Window ; Calculate each pixel in the neighborhood window Local grayscale fluctuations: ; in: For pixels The micro-transmittance scattering offset, For neighborhood windows Average grayscale value of internal pixels The total number of pixels in the window. These are the pixel coordinates within the neighborhood window; Remove the micro-transmittance scattering offset from the original image: ; in, The grayscale image after removing the micro-transmittance scattering offset; Obtain the image sequence after removing the micro-transmittance scattering offset from consecutive frames: .

[0030] By acquiring continuous images covering the entire adhesive tape surface and constructing a baseline grayscale value using the adhesive-free area, combined with neighborhood analysis to eliminate micro-transmittance scattering offsets, a purified grayscale image is obtained from the original image, laying the foundation for accurate identification of the adhesive-coated area. The beneficial effect of this method is that it effectively eliminates optical interference caused by the high reflectivity of the copper foil and the semi-transparent nature of the adhesive layer, allowing the image grayscale to more accurately reflect the adhesive layer thickness. This not only improves the stability and accuracy of the detection but also enables full-area, continuous monitoring. Compared to traditional manual inspection and point measurement methods, it significantly improves detection efficiency and is suitable for high-speed continuous production line applications. Simultaneously, by analyzing local grayscale fluctuations through neighborhood windows, the influence of local noise on judgment can be eliminated at the image level, thereby ensuring the accuracy of adhesive-coated area identification and providing a reliable data foundation for subsequent thickness calculation and uniformity evaluation.

[0031] The process of comparing the purified grayscale with the reference grayscale and generating a mask for the coated area, and performing thickness and uniformity analysis only on the pixels in the coated area, includes: calculate and The difference between the grayscale values: ; like This indicates that there is an adhesive layer under the current pixel; otherwise, it indicates that there is no adhesive area under the current pixel. Set minimum effective grayscale difference It can be set according to the grayscale value of the glue-free area; Based on the minimum effective grayscale difference Determine the area to be glued: ; in, For the glued pixel set; Output the initial glue application area mask : ; when If the value is 0, then delete the pixel. The adhesive-coated area mask is used to identify the actual adhesive-coated pixels, serving as a condition for subsequent thickness calculations, uniformity analysis, and defect identification. It is only used when... Only when the pixel enters the thickness and uniformity calculation process does it avoid interference from the background and reflective areas, thereby improving the detection stability and accuracy. Define a local grayscale-thickness mapping function: ; in, For adhesive layer thickness, This is an experimental calibration constant, which can be obtained through sample measurement; For each pixel in the adhesive coating area Calculate the initial adhesive layer thickness: ; Calculate the local average thickness of each pixel within the adhesive coating area. : ; Output local average thickness This can reduce the impact of local residual noise while preserving the characteristics of thickness variation.

[0032] The process of calculating the initial thickness of the adhesive layer based on grayscale difference and performing local averaging, using residual scattering correction and extreme value filtering to eliminate edge and micropore interference, and obtaining the accurate adhesive layer thickness includes: Through local average thickness Back-calculation of the theoretical grayscale value of the corresponding colloidal region : ; Will and The difference in grayscale values ​​between them is used as the residual grayscale value. : ; For each glued pixel Define the local scattering correction function: ; in, The average value of pixel residual scattering is used to smooth local optical non-uniformity; right Perform residual correction: ; in, To achieve the precise thickness of the final colloid; After removing residual scattering, the thickness value is closer to the actual physical thickness of the adhesive layer. Further eliminate abnormal thickness at the edges of the adhesive coating and in localized microporous areas, for Perform local extremum filtering: ; in, The median is the thickness of the colloid after local extremum filtering. By eliminating extreme high and low values, the influence of micropores and edge anomalies on thickness estimation can be eliminated.

[0033] The purified grayscale image is differentially divided with the reference grayscale image to generate a mask for the adhesive-coated area. The initial thickness of the adhesive layer is then calculated using the grayscale difference. Local averaging and residual correction are then applied, along with extreme value filtering to eliminate edge and micro-pore interference, ultimately yielding an accurate adhesive layer thickness. The advantages of this design are that it not only quantifies the adhesive layer thickness, reducing errors caused by image noise and scattering, but also eliminates the influence of local micro-pores or edge anomalies on thickness measurement, making the thickness data closer to the actual physical thickness of the adhesive layer. Through local averaging and residual correction, the thickness calculation retains the true variation characteristics of the adhesive layer while smoothing out the influence of optical interference, thus achieving high-precision thickness measurement. Extreme value filtering further ensures the reliability of the thickness data, enabling the system to accurately determine the uniformity of the adhesive layer and local defects under complex surface conditions. This is particularly suitable for copper foil tapes with high reflectivity on metal substrates and significant adhesive layer transparency, improving the feasibility and accuracy of thickness assessment and defect identification, and providing a solid data foundation for subsequent dynamic monitoring and quality alarms.

[0034] The process of performing neighborhood statistics on the precise thickness, calculating thickness uniformity, and identifying pixels with excessive thickness based on allowable intervals includes: For each pixel Set neighborhood analysis window It can be set to 11×11 pixels; Neighborhood analysis window The set of pixels contained within is denoted as : ; Calculate the local average thickness for each pixel: ; in, Neighborhood analysis window The total number of pixels contained within. Indicates the local average thickness; Calculate the thickness uniformity of the neighborhood analysis window : ; The smaller the value, the more uniform the adhesive coating is in the neighborhood analysis window; Based on process requirements and experience calibration, a minimum allowable thickness is set. and maximum allowable thickness ; For each pixel within the adhesive application area Make a judgment: ; in, This indicates that the pixel has a local defect.

[0035] The process of acquiring thickness data from multiple consecutive frames and calculating the trend of change to identify dynamic anomalies includes: Obtain multiple consecutive frames And form a thickness sequence: ; in, This is the initial time sequence number. This is the nth time sequence number; Calculate the thickness change rate between adjacent frames to reflect local adhesive application trends: ; like If so, it is determined to be a local increase in thickness; like If so, it is determined to be a localized reduction in thickness; Convert the thickness change rate to a change rate in units of time. : ; in, The frame interval can be directly determined by the acquisition frame rate. calculate: ; according to Calculate the local average rate of change for each pixel. : ; Set the maximum change threshold according to the process requirements. and minimum change threshold Used to identify abnormal pixels: ; in, This indicates an abnormal trend in the local pixel thickness variation.

[0036] Based on accurate thickness calculation, this method sets a neighborhood analysis window for each pixel, statistically analyzes the local average thickness and calculates thickness uniformity, and simultaneously acquires multiple consecutive frames of thickness data to analyze thickness change trends in order to identify local thickness anomalies and dynamic changes. The beneficial effect of this design is that it can simultaneously achieve static and dynamic coating defect monitoring. It can detect immediate defects such as excessive thickness, insufficient thickness, or blank areas, and also provide early warnings of trend anomalies during the coating process, such as process drift or occasional thickness fluctuations. Through neighborhood statistics, local thickness fluctuations can be quantified, reducing the impact of single-point measurement errors on overall judgment; through multi-frame thickness trend analysis, continuously changing anomalies can be captured, enabling early warning of gradual deviations in the production process. This measure not only improves the reliability of coating quality control but also provides enterprises with traceable quality data, helping to optimize process parameters and improve product consistency, thereby significantly reducing downstream application risks caused by coating defects and meeting the full-area, high-precision monitoring requirements of high-speed continuous production lines.

[0037] The step of statistically analyzing thickness distribution by region and generating a long-term uniformity trend map includes: according to Calculate the full-frame average thickness of each captured image frame. : ; in, The number of pixels for the image width and height; Calculate the full-frame thickness fluctuation of the image : ; The adhesive application area is divided into Each statistical unit contains several pixels; For each unit Calculate the average thickness and standard deviation : ; ; in, This refers to the number of pixels within a unit. Output local uniformity matrix ; For each frame of image Plotting time series curves: ; Local uniformity matrix By overlaying the data with the time series curve over time, a two-dimensional heat map is generated. ; The red area indicates large fluctuations in uniformity, while the blue area indicates low uniformity.

[0038] The process of generating a trend mask based on the trend chart, filtering and integrating it with the results of thickness deviation and dynamic changes to obtain a comprehensive anomaly index and output an anomaly alarm includes: Set threshold ,when It was believed at the time that the statistical unit exhibited long-term unevenness; Generate a binary mask from statistically non-uniform units over a long period of time. : ; use right and Perform the screening; ; ; in, and After filtering and ; Calculate the comprehensive anomaly index for each frame of the image: ; in, This is the normalization coefficient, which can be directly set to 1; when When this happens, the pixel is determined to be abnormal, triggering a local alarm.

[0039] The full-width thickness distribution is divided into several statistical units. For each unit, the average thickness and uniformity are calculated, and a long-term uniformity trend chart is generated. Simultaneously, a trend mask is used to filter and merge local thickness anomalies and dynamic changes, generating a comprehensive anomaly index and triggering an alarm. The beneficial effect of this design is that it enables long-term, macroscopic, and microscopic joint analysis of adhesive coating quality. It can not only capture instantaneous local defects but also reflect cumulative non-uniformity problems during production, achieving trend-based anomaly early warning. Through trend mask filtering, the method can distinguish between occasional noise and long-term anomalies, improving the accuracy and reliability of alarms and avoiding false alarms. The output of the comprehensive anomaly index makes the detection results intuitive and usable, providing a scientific basis for production management and process improvement. This function is particularly suitable for continuous production lines where continuous monitoring of the full-width uniformity and dynamic stability of the product is required, thereby effectively improving the overall quality and consistency of copper foil tape and reducing the risk of product failure.

[0040] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0041] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for detecting the uniformity of adhesive coating on copper foil tape based on image detection, characterized in that, include: Continuous images of the surface covered by adhesive tape are acquired and converted to grayscale. A reference baseline is constructed using images of areas without adhesive. A neighborhood window is set for each pixel. The micro-transmittance scattering offset is calculated, and the purified grayscale image is obtained by removing the micro-transmittance scattering offset from the original image. The purified grayscale is compared with the reference grayscale to generate a mask for the coated area, and the thickness and uniformity analysis is performed only on the pixels in the coated area. The initial thickness of the adhesive layer is calculated based on the grayscale difference and local averaging is performed. The residual scattering correction is used and extreme value filtering is applied to eliminate edge and micropore interference, thereby obtaining the accurate adhesive layer thickness. Perform neighborhood statistics on the precise thickness, calculate thickness uniformity, and identify pixels with excessive thickness based on the allowable range; Acquire thickness data from multiple consecutive frames and calculate the trend of change to identify dynamic anomalies; The thickness distribution is statistically analyzed by region and a long-term uniformity trend map is generated. A trend mask is generated based on the trend chart, and then filtered and integrated with the thickness deviation and dynamic change results to obtain a comprehensive anomaly index and output an anomaly alarm.

2. The method for detecting the uniformity of adhesive coating on copper foil tape based on image detection according to claim 1, characterized in that: The process involves acquiring and converting continuous images of the adhesive tape surface to grayscale, constructing a reference baseline using images of the adhesive-free areas, setting a neighborhood window for each pixel, calculating the micro-transmittance scattering offset, and removing unwanted pixels from the original image to obtain a cleaned grayscale image. The surface of the tape is continuously captured across the entire frame using an industrial camera, and the captured images are converted into grayscale format. A reference image is pre-selected from a glue-free area to serve as the true reflection reference of the copper foil body. For each pixel in the entire image, a neighborhood range is set, and the grayscale fluctuation within this range is calculated to characterize the error caused by illumination, reflection and micro-transmittance scattering. This error is then removed from the original image to obtain a clean grayscale image that characterizes the true optical properties of the coating. The system processes multiple consecutive frames to form a purified grayscale image sequence after eliminating environmental optical interference.

3. The method for detecting the uniformity of adhesive coating on copper foil tape based on image detection according to claim 1, characterized in that: The process of comparing the purified grayscale with the reference grayscale and generating a mask for the coated area, and performing thickness and uniformity analysis only on the pixels in the coated area, includes: The actual adhesive coating area is identified based on the difference between the purified grayscale image and the reference image. A binary mask is constructed to completely exclude pixels in the non-adhesive coating area, so that subsequent analysis is carried out only on the adhesive layer area. By establishing the mapping relationship between grayscale and thickness through experimental calibration, the thickness information corresponding to the glue-coated pixels can be initially quantified. The average thickness is calculated within a local area to mitigate the effects of random noise and slight illumination fluctuations, thereby obtaining a local thickness map that reflects the basic thickness trend.

4. The method for detecting the uniformity of adhesive coating on copper foil tape based on image detection according to claim 1, characterized in that: The process of calculating the initial thickness of the adhesive layer based on grayscale difference and performing local averaging, using residual scattering correction and extreme value filtering to eliminate edge and micropore interference, and obtaining the accurate adhesive layer thickness includes: The theoretical optical performance is estimated by using the local thickness mean, and the actual optical performance is compared with the theoretical results to obtain residual information representing local optical anomalies; The residual information is smoothed by the neighborhood mean method to correct the inherent optical scattering error generated during the imaging process, so that the thickness estimate is closer to the actual physical thickness. Perform local extremum filtering to remove extreme high and low values.

5. The method for detecting the uniformity of adhesive coating on copper foil tape based on image detection according to claim 1, characterized in that: The process of performing neighborhood statistics on the precise thickness, calculating thickness uniformity, and identifying pixels with excessive thickness based on allowable intervals includes: A neighborhood analysis region is set for the final thickness map, and the uniformity evaluation index of the region is obtained by calculating the average thickness and the degree of thickness fluctuation. The thickness is judged based on the allowable thickness range of the process. When the local thickness deviates from the normal range, the pixel is marked as a static abnormal point.

6. The method for detecting the uniformity of adhesive coating on copper foil tape based on image detection according to claim 1, characterized in that: The process of acquiring thickness data from multiple consecutive frames and calculating the trend of change to identify dynamic anomalies includes: Collect multiple frames of thickness data, establish a dynamic thickness change model in the time dimension, and determine the local thickness increase or decrease trend by comparing the thickness differences between previous and subsequent frames. The rate of change per unit time is calculated by combining time intervals, and the overall change characteristics of the neighborhood are statistically analyzed. If the rate of change exceeds the set range, it is considered a dynamic anomaly.

7. The method for detecting the uniformity of adhesive coating on copper foil tape based on image detection according to claim 1, characterized in that: The step of statistically analyzing thickness distribution by region and generating a long-term uniformity trend map includes: Obtain the overall thickness level of the image and divide it into fixed region units. Statistically analyze the thickness level and uniformity of each region. By accumulating statistics across multiple consecutive frames, a long-term thickness fluctuation trend is plotted, and a visualized uniformity trend map is generated, in which high fluctuation areas correspond to potential long-term process inhomogeneity issues.

8. The method for detecting the uniformity of adhesive coating on copper foil tape based on image detection according to claim 1, characterized in that: The process of generating a trend mask based on the trend chart, filtering and integrating it with the results of thickness deviation and dynamic changes to obtain a comprehensive anomaly index and output an anomaly alarm includes: Trend masks are generated by extracting regions with long-term non-uniformity based on long-term uniformity trend maps. The mask is fused and filtered with both static thickness deviation results and dynamic change anomaly results. Finally, only the abnormal locations that simultaneously exhibit long-term non-uniformity characteristics and current defect trends are retained, and alarm commands are issued to them.

Images