Intelligent detection system for surface quality of copper foil

Abstract

The invention discloses an intelligent detection system for the surface quality of a copper foil. The intelligent detection system comprises an acquisition module, a construction module and a detection module, the acquisition module is used for acquiring a multi-angle multi-spectral image of the surface of the copper foil and preprocessing the multi-angle multi-spectral image; the construction module is used for carrying out automatic identification and positioning on the surface of the copper foil by utilizing a defect detection algorithm based on deep learning based on the preprocessed multi-angle multispectral image, and constructing a copper foil surface defect identification model; and the detection module is used for comprehensively detecting the surface defects of the copper foil to be detected by using the copper foil surface defect identification model. According to the invention, comprehensive and accurate detection of surface defects of the copper foil can be realized, and the quality control level of the copper foil is improved.

Description

Technical Field

[0001] The invention belongs to the field of information technology, and in particular relates to an intelligent detection system for copper foil surface quality. Background Art

[0002] Surface quality inspection is a key link in the production process of copper foil. Traditional manual inspection methods are inefficient and easily affected by subjective factors, making it difficult to meet the requirements of fast and accurate inspection. In order to achieve online intelligent inspection of copper foil surface quality, many technical difficulties need to be overcome. First, the environment of the copper foil production workshop is harsh, and impurities such as dust and oil are easily attached to the surface of the copper foil, causing serious interference to the quality of image acquisition. How to obtain high-quality images in harsh environments is a major challenge. Secondly, there are various types of surface defects of copper foil, including scratches, pinholes, oxidation spots, etc. The shapes and sizes of different defects vary greatly, and the distribution is random. Traditional methods based on threshold segmentation and morphological operations are difficult to accurately extract defective areas. In addition, the high speed of copper foil production puts extremely high demands on the real-time performance of the detection system. How to improve the detection speed while meeting the accuracy is another problem to be solved urgently. Finally, there are differences in the surface characteristics of copper foils of different production batches and different models. The detection system needs to have adaptive capabilities and be able to automatically adjust relevant parameters according to actual conditions to ensure the stability and reliability of detection. Summary of the invention

[0003] In view of the problems that impurities are easily attached to the surface of copper foil and the surface characteristics of copper foils of different batches and models vary greatly, the present invention proposes an intelligent copper foil surface quality detection system, which can realize comprehensive and accurate detection of copper foil surface defects and improve the quality control level of copper foil.

[0004] To achieve the above object, the present invention provides the following solutions:

[0005] A copper foil surface quality intelligent detection system, comprising: an acquisition module, a construction module and a detection module;

[0006] The acquisition module is used to acquire multi-angle and multi-spectral images of the copper foil surface and perform preprocessing;

[0007] The construction module is used to automatically identify and locate the copper foil surface based on the preprocessed multi-angle multi-spectral image and to construct a copper foil surface defect recognition model by using a deep learning-based defect detection algorithm;

[0008] The detection module is used to use the copper foil surface defect recognition model to achieve comprehensive detection of the surface defects of the copper foil to be detected.

[0009] Preferably, the acquisition module includes: a collection unit, a preprocessing unit and a feature extraction unit;

[0010] The acquisition unit is used to acquire the copper foil surface image;

[0011] The preprocessing unit is used to preprocess the collected copper foil surface image using adaptive image enhancement technology;

[0012] The feature extraction unit is used to design a special light source layout scheme in combination with the surface characteristics of the copper foil, and to obtain the pre-processed image features using multi-angle and multi-spectral imaging technology.

[0013] Preferably, the feature extraction unit includes: a model building subunit, an optimization subunit, a data acquisition subunit and a feature acquisition subunit;

[0014] The model building subunit is used to obtain the reflectivity distribution of the copper foil surface at different incident angles and wavelengths, and to establish a reflection model of the copper foil surface;

[0015] The optimization subunit is used to optimize the light source layout and multi-spectral imaging scheme on the copper foil surface based on the reflection model;

[0016] The data acquisition subunit is used to use a multispectral camera to acquire image data of different bands on the copper foil surface under the optimized light source layout and multispectral imaging scheme;

[0017] The feature acquisition subunit is used to determine the area and position of impurities and defects in the image data through a support vector machine algorithm, and to acquire the surface texture features and shape features of the copper foil.

[0018] Preferably, the construction module includes: a region extraction unit, a defect recognition unit and a recognition model construction unit;

[0019] The region extraction unit is used to extract the background region, normal texture region and defect suspected region on the surface of the copper foil based on the preprocessed multi-angle multi-spectral image by using a semantic segmentation network based on deep learning;

[0020] The defect recognition unit is used to determine whether the suspected defect area is a real defect by combining the area shape and texture features, and classify and identify the defect type to obtain a defect recognition result;

[0021] The recognition model building unit is used to build a copper foil surface defect recognition model based on the defect recognition result.

[0022] Preferably, the semantic segmentation network based on deep learning adopts an improved SegNet network structure, extracts multi-scale features of the image through a convolutional neural network, and then upsamples the feature map through a deconvolution network, and finally obtains a pixel-level segmentation result of the copper foil surface image.

[0023] Preferably, the regional shape features are: area, perimeter, circularity;

[0024] The regional texture features are: gray-level co-occurrence matrix and local binary pattern.

[0025] Preferably, the defect recognition unit comprises: a defect determination subunit, a defect information output subunit and a defect determination subunit;

[0026] The defect determination subunit is used to input the extracted regional shape features and texture features into the defect determination model based on the support vector machine, and by comparing the training sample data, determine whether the region is a real defect with a confidence level of 90%;

[0027] The defect information output subunit is used to output and save the center coordinates, defect type, and confidence information of the area if it is determined to be a real defect area;

[0028] The defect determination subunit is used to traverse all areas and, if any real defect area is detected, determine that the copper foil image has defects.

[0029] Preferably, the detection module comprises: an image acquisition unit, an image convolution unit, an image decomposition unit and a defect detection unit;

[0030] The image acquisition unit is used to acquire the copper foil surface texture image;

[0031] The image convolution unit is used to convolve the copper foil surface texture image using a directional Gabor filter according to the directional characteristics of the copper foil surface texture, so as to obtain a texture image with enhanced directionality;

[0032] The image decomposition unit is used to perform multi-scale decomposition on the directionally enhanced texture image using Gabor filter groups of different scales and directions according to the periodic characteristics of the copper foil surface texture, so as to obtain multi-scale texture features;

[0033] The defect detection unit is used to classify and identify the multi-scale texture features according to the copper foil surface defect recognition model to determine whether there are defects on the copper foil surface; if there are defects, the defect information is output and the defective copper foil is removed.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] In order to solve the problem that impurities are easily attached to the surface of copper foil and the surface characteristics of copper foils of different batches and models vary greatly, the present invention uses adaptive image enhancement technology to pre-process the collected images to improve the image quality; and designs a special multi-angle and multi-spectral light source layout scheme to obtain rich surface texture information. On this basis, a defect detection algorithm based on deep learning is used to automatically identify and locate various defects on the surface of copper foil, and an adaptive parameter adjustment mechanism and incremental learning strategy are introduced to enable the detection system to adaptively adapt to the detection requirements of different copper foil products, and continuously improve the stability and reliability of detection. The present invention can realize comprehensive and accurate detection of copper foil surface defects and improve the quality control level of copper foil. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0037] Figure 1 The present invention is a schematic diagram of the structure of an intelligent copper foil surface quality detection system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0040] Embodiment 1

[0041] like Figure 1 As shown, the present invention provides a copper foil surface quality intelligent detection system, including: an acquisition module, a construction module and a detection module;

[0042] The acquisition module is used to acquire multi-angle and multi-spectral images of the copper foil surface and perform preprocessing;

[0043] The construction module is used to automatically identify and locate the copper foil surface based on the preprocessed multi-angle multi-spectral image and to construct a copper foil surface defect recognition model by using a deep learning-based defect detection algorithm;

[0044] The detection module is used to use the copper foil surface defect recognition model to achieve comprehensive detection of the surface defects of the copper foil to be detected.

[0045] In this embodiment, the acquisition module includes: a collection unit, a preprocessing unit and a feature extraction unit;

[0046] The acquisition unit is used to acquire the copper foil surface image;

[0047] The preprocessing unit is used to preprocess the collected copper foil surface image using adaptive image enhancement technology;

[0048] The feature extraction unit is used to design a special light source layout scheme based on the surface characteristics of the copper foil, and to obtain the pre-processed image features using multi-angle and multi-spectral imaging technology.

[0049] Among them, in order to address the problem that dust, oil and other impurities are easily attached to the surface of copper foil, adaptive image enhancement technology is used to pre-process the collected images. By automatically adjusting image parameters such as contrast and brightness, the image quality is improved and the impact of harsh environment on detection accuracy is reduced.

[0050] Among them, the feature extraction unit includes: a model building subunit, an optimization subunit, a data acquisition subunit and a feature acquisition subunit;

[0051] The model building subunit is used to obtain the reflectivity distribution of the copper foil surface at different incident angles and wavelengths, and to establish a reflection model of the copper foil surface;

[0052] The optimization subunit is used to optimize the light source layout and multi-spectral imaging scheme on the copper foil surface based on the reflection model;

[0053] The data acquisition subunit is used to use a multispectral camera to acquire image data of different bands on the copper foil surface under the optimized light source layout and multispectral imaging scheme;

[0054] The feature acquisition subunit is used to determine the area and position of impurities and defects in the image data through a support vector machine algorithm, and to acquire the surface texture features and shape features of the copper foil.

[0055] Specifically, in order to construct the copper foil surface reflection model, the reflectivity of the copper foil surface at different incident angles (0°-60°) and wavelengths (400nm-1000nm) is first measured by a spectrometer to obtain the reflectivity distribution data. Then the reflectivity data is fitted using the least squares method to obtain the parameters of the reflection model. According to the reflection model and the principle of multispectral imaging, the genetic algorithm is used to optimize the light source layout and the light intensity distribution of each band, where the objective function is the weighted sum of image contrast and signal-to-noise ratio, and the constraint condition is the limit of the number and power of light sources. After 50 generations of iteration, the optimal light source layout scheme is obtained. The copper foil surface is multispectrally imaged using the optimized light source to obtain image data of 10 bands (450nm, 500nm, 550nm, 600nm, 650nm, 700nm, 750nm, 800nm, 850nm, 900nm). The multispectral images are registered using a mutual information-based registration algorithm, and the registration error is controlled within 5 pixels. The wavelet transform denoising algorithm is used to remove Gaussian noise and salt and pepper noise in the image to obtain the corrected image data. The image features are extracted and classified using the support vector machine algorithm. The features include texture features (gray-level co-occurrence matrix, Gabor filter, etc.) and shape features (Hu moment invariant moment, Fourier descriptor, etc.). After training, the classification model of impurities and defects is obtained, and the corrected image data is predicted to identify the impurities and defect areas in the image. The impurities and defect areas are morphologically processed to extract their contours and position information, and the area, length and other parameters are calculated. The detection results are compared with the preset threshold (area>5mm 2 , length>10mm) for comparison. If it exceeds the threshold, it is judged as a defect and a copper foil quality inspection report is output, including information such as defect quantity, location, area, etc., to provide data support for copper foil quality assessment.

[0056] In this embodiment, the construction module includes: a region extraction unit, a defect recognition unit and a recognition model construction unit;

[0057] The region extraction unit is used to extract the background region, normal texture region and defect suspected region on the surface of the copper foil based on the preprocessed multi-angle multi-spectral image by using a semantic segmentation network based on deep learning;

[0058] The defect recognition unit is used to determine whether the suspected defect area is a real defect based on the area shape and texture features, and classify and identify the defect type to obtain a defect recognition result;

[0059] The recognition model building unit is used to build a copper foil surface defect recognition model based on the defect recognition result.

[0060] Among them, the semantic segmentation network based on deep learning adopts an improved SegNet network structure, extracts the multi-scale features of the image through a convolutional neural network, and then upsamples the feature map through a deconvolution network, and finally obtains the pixel-level segmentation result of the copper foil surface image.

[0061] Among them, the regional shape features are: area, perimeter, and circularity;

[0062] The regional texture features are: gray-level co-occurrence matrix and local binary pattern.

[0063] The defect recognition unit includes: a defect determination subunit, a defect information output subunit and a defect determination subunit;

[0064] The defect discrimination subunit is used to input the extracted regional shape features and texture features into the defect judgment model based on the support vector machine, and judge whether the region is a real defect with a confidence level of 90% by comparing the training sample data;

[0065] The defect information output subunit is used to output and save the center coordinates, defect type, and confidence information of the area if it is judged to be a real defect area;

[0066] The defect determination subunit is used to traverse all areas and if any real defect area is detected, it is determined that the copper foil image has defects.

[0067] Specifically, the defect detection results are compared with the preset qualified standard threshold, such as the defect area does not exceed 2mm 2 , the number of defects does not exceed 5, etc. If it exceeds the threshold, it is judged as a defective product, triggering alarm prompts and subsequent processing processes such as automatic rejection, thereby realizing fully automatic detection and quality grading of copper foil defects.

[0068] In this embodiment, the detection module includes: an image acquisition unit, an image convolution unit, an image decomposition unit and a defect detection unit;

[0069] The image acquisition unit is used to acquire the copper foil surface texture image;

[0070] The image convolution unit is used to convolve the copper foil surface texture image using a directional Gabor filter according to the directional characteristics of the copper foil surface texture to obtain a texture image with enhanced directionality;

[0071] The image decomposition unit is used to perform multi-scale decomposition on the directionally enhanced texture image using Gabor filter groups of different scales and directions according to the periodic characteristics of the copper foil surface texture, so as to obtain multi-scale texture features;

[0072] The defect detection unit is used to classify and identify the multi-scale texture features according to the copper foil surface defect recognition model to determine whether there are defects on the copper foil surface; if there are defects, the defect information is output and the defective copper foil is removed.

[0073] Specifically, first, according to the directional characteristics of the copper foil surface texture, four directions (0°, 45°, 90°, 135°) of Gabor filters are selected to convolve the image, where the parameters of the Gabor filter are set as follows: wavelength λ = 8, directions θ are 0°, 45°, 90°, 135°, phase offset ψ = 0, and Gaussian function standard deviation σ = 4. The directional information of the texture can be effectively enhanced by Gabor filtering. Then, three different scales (1:1, 1:2, 1:4) of Gabor filter groups are used to perform multi-scale decomposition on the image. Each scale of the Gabor filter group contains eight filters in different directions (0°~180°, with an interval of 25°). Multi-scale Gabor filtering can extract the multi-scale features of the copper foil surface texture. The Gabor filter responses of different scales and directions are weighted fused, and the weights are learned and optimized through training sample data to obtain the enhanced copper foil surface texture image. Compared with the original image, the contrast and clarity of the texture are significantly improved. In order to further improve the clarity of the texture edge, the enhanced texture image is binarized by an adaptive threshold segmentation algorithm. The threshold is automatically determined according to the histogram distribution of the image, and the texture image after binarization is clearer and more regular. The binary image is analyzed by the connected domain analysis algorithm, and the geometric features such as the area, perimeter, and circularity of each connected domain are calculated. The integrity and uniformity of the texture are judged according to the preset threshold. The connected domain with an area less than 100 pixels or a circularity less than 5 is considered to be a texture defect. Finally, the extracted texture features are input into the SVM classifier for defect recognition. The classifier is trained by a large number of defects and normal samples. If scratches, stains and other defects are detected on the copper foil surface, the location coordinates and type information of the defect are output, and the ID of the defective copper foil is recorded in the database, notifying the equipment to remove it. If the surface quality of the copper foil meets the standard, it is released and the ID is recorded to enter the next coating process. Through the above method, the comprehensive automatic detection and sorting of the surface quality of the copper foil can be realized, ensuring product quality while improving production efficiency.

[0074] The present invention introduces an adaptive parameter adjustment mechanism in view of the differences in surface characteristics of copper foils of different production batches and different models. Through real-time analysis of the surface texture characteristics of the copper foil, the detection threshold, feature extraction method and other parameters of the algorithm are dynamically adjusted, so that the detection system can adaptively adapt to the detection needs of different copper foil products. At the same time, an incremental learning strategy is adopted to update and optimize the detection model online by continuously accumulating defect sample data in the production process, so as to continuously improve the stability and reliability of the detection system.

[0075] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. An intelligent detection system for copper foil surface quality, characterized in that: include: Acquisition modules, construction modules and detection modules; The acquisition module is used to acquire multi-angle and multi-spectral images of the copper foil surface and perform preprocessing; The construction module is used to automatically identify and locate the copper foil surface based on the preprocessed multi-angle multi-spectral image and to construct a copper foil surface defect recognition model by using a deep learning-based defect detection algorithm; The detection module is used to use the copper foil surface defect recognition model to achieve comprehensive detection of the surface defects of the copper foil to be detected.

2. The intelligent detection system for copper foil surface quality according to claim 1 is characterized in that: The acquisition module includes: a collection unit, a preprocessing unit and a feature extraction unit; The acquisition unit is used to acquire the copper foil surface image; The preprocessing unit is used to preprocess the collected copper foil surface image using adaptive image enhancement technology; The feature extraction unit is used to design a special light source layout scheme in combination with the surface characteristics of the copper foil, and to obtain the pre-processed image features using multi-angle and multi-spectral imaging technology.

3. The intelligent detection system for copper foil surface quality according to claim 2 is characterized in that: The feature extraction unit includes: a model building subunit, an optimization subunit, a data acquisition subunit and a feature acquisition subunit; The model building subunit is used to obtain the reflectivity distribution of the copper foil surface at different incident angles and wavelengths, and to establish a reflection model of the copper foil surface; The optimization subunit is used to optimize the light source layout and multi-spectral imaging scheme on the copper foil surface based on the reflection model; The data acquisition subunit is used to use a multispectral camera to acquire image data of different bands on the copper foil surface under the optimized light source layout and multispectral imaging scheme; The feature acquisition subunit is used to determine the area and position of impurities and defects in the image data through a support vector machine algorithm, and to acquire the surface texture features and shape features of the copper foil.

4. The intelligent detection system for copper foil surface quality according to claim 3 is characterized in that: The construction module includes: a region extraction unit, a defect recognition unit and a recognition model construction unit; The region extraction unit is used to extract the background region, normal texture region and defect suspected region on the surface of the copper foil based on the preprocessed multi-angle multi-spectral image by using a semantic segmentation network based on deep learning; The defect recognition unit is used to determine whether the suspected defect area is a real defect by combining the area shape and texture features, and classify and identify the defect type to obtain a defect recognition result; The recognition model building unit is used to build a copper foil surface defect recognition model based on the defect recognition result.

5. The intelligent detection system for copper foil surface quality according to claim 4, characterized in that: The deep learning-based semantic segmentation network adopts an improved SegNet network structure, extracts multi-scale features of the image through a convolutional neural network, and then upsamples the feature map through a deconvolution network, and finally obtains a pixel-level segmentation result of the copper foil surface image.

6. The intelligent detection system for copper foil surface quality according to claim 4, characterized in that: The regional shape characteristics are: area, perimeter, and circularity; The regional texture features are: gray-level co-occurrence matrix and local binary pattern.

7. The intelligent detection system for copper foil surface quality according to claim 4, characterized in that: The defect recognition unit includes: a defect determination subunit, a defect information output subunit and a defect determination subunit; The defect determination subunit is used to input the extracted regional shape features and texture features into the defect determination model based on the support vector machine, and by comparing the training sample data, determine whether the region is a real defect with a confidence level of 90%; The defect information output subunit is used to output and save the center coordinates, defect type, and confidence information of the area if it is determined to be a real defect area; The defect determination subunit is used to traverse all areas and, if any real defect area is detected, determine that the copper foil image has defects.

8. The intelligent detection system for copper foil surface quality according to claim 1, characterized in that: The detection module includes: an image acquisition unit, an image convolution unit, an image decomposition unit and a defect detection unit; The image acquisition unit is used to acquire the copper foil surface texture image; The image convolution unit is used to convolve the copper foil surface texture image using a directional Gabor filter according to the directional characteristics of the copper foil surface texture, so as to obtain a texture image with enhanced directionality; The image decomposition unit is used to perform multi-scale decomposition on the directionally enhanced texture image using Gabor filter groups of different scales and directions according to the periodic characteristics of the copper foil surface texture, so as to obtain multi-scale texture features; The defect detection unit is used to classify and identify the multi-scale texture features according to the copper foil surface defect recognition model to determine whether there are defects on the copper foil surface; if there are defects, the defect information is output and the defective copper foil is removed.

Images