A machine vision guided real-time detection method and system for corrugated board defects

Abstract

The application discloses a kind of machine vision guided corrugated board defect real-time detection method and system, it is related to corrugated board detection technical field, the method includes: real-time acquisition target corrugated board production line's gray image;Obtain dynamic statistical feature list, corresponding traversal gray image is statistically extracted, and according to statistical feature extraction result constructs primary feature vector;Based on historical defect detection record, construct dynamic threshold model, and with dynamic threshold model as constraint, in combination with primary feature vector, corrugated board is initially screened;According to the result of initial screening, corresponding extraction gray image is adaptively enhanced, and enhanced gray image is obtained, and based on enhanced gray image, defect detection is carried out.The present application solves the technical problems of poor adaptability and insufficient accuracy in the prior art of corrugated board defect detection.

Description

Technical Field

[0001] This invention relates to the field of equipment condition monitoring technology, specifically to a machine vision-guided real-time defect detection method and system for corrugated cardboard. Background Technology

[0002] In the corrugated cardboard manufacturing industry, product quality control is crucial, and defect detection is a key step in ensuring the quality of products leaving the factory. Currently, machine vision is widely used for surface defect detection in corrugated cardboard, primarily relying on static detection models with preset fixed thresholds. However, due to the diverse specifications of corrugated cardboard products, variations in production line speed, and fluctuations in ambient lighting, fixed thresholds are difficult to adapt to different working conditions, resulting in high false positive and false negative rates. Furthermore, static models lack the ability to perceive and learn from real-time data during production, leading to poor detection accuracy, insufficient defect identification capabilities, and high computational complexity, making it difficult to meet the inspection requirements of corrugated cardboard production. Summary of the Invention

[0003] This application provides a machine vision-guided real-time defect detection method and system for corrugated cardboard, which addresses the technical problems of poor adaptability and insufficient accuracy in the detection of corrugated cardboard defects in the prior art.

[0004] In view of the above problems, this application provides a machine vision-guided real-time defect detection method and system for corrugated cardboard.

[0005] In a first aspect, this application provides a machine vision-guided real-time defect detection method for corrugated cardboard, the method comprising:

[0006] Real-time acquisition of grayscale images of the target corrugated cardboard production line;

[0007] Obtain a dynamic statistical feature list, traverse the grayscale image accordingly to extract statistical features, and construct a primary feature vector based on the statistical feature extraction results;

[0008] A dynamic threshold model is constructed based on historical defect detection records, and the corrugated cardboard is initially screened using the dynamic threshold model as a constraint and combined with the primary feature vector.

[0009] Based on the initial screening results, the corresponding grayscale image is extracted and adaptively enhanced to obtain an enhanced grayscale image, and defect detection is performed based on the enhanced grayscale image.

[0010] Secondly, this application provides a machine vision-guided real-time defect detection system for corrugated cardboard, comprising:

[0011] The image acquisition module is used to acquire grayscale images of the target corrugated cardboard production line in real time.

[0012] The feature extraction module is used to obtain a dynamic statistical feature list, perform statistical feature extraction on the corresponding grayscale image, and construct a primary feature vector based on the statistical feature extraction results.

[0013] The feature screening module is used to construct a dynamic threshold model based on historical defect detection records, and to perform initial screening of corrugated cardboard by combining the dynamic threshold model with the primary feature vector;

[0014] The defect detection module is used to extract the grayscale image according to the initial screening results, perform adaptive grayscale enhancement to obtain the enhanced grayscale image, and perform defect detection based on the enhanced grayscale image.

[0015] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0016] This application proposes a machine vision-guided real-time defect detection method and system for corrugated cardboard. By combining dynamic statistical feature extraction with adaptive threshold adjustment, the accuracy and adaptability of corrugated cardboard defect detection are significantly improved. Compared with traditional methods, the technical solution provided in this application significantly overcomes the limitations of traditional methods in dealing with different product specifications and changes in production environment, achieving the technical effect of efficient and accurate defect detection of corrugated cardboard in complex production environments. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a machine vision-guided real-time defect detection method for corrugated cardboard provided in an embodiment of this application.

[0019] Figure 2 This is a schematic diagram of a machine vision-guided real-time defect detection system for corrugated cardboard provided in an embodiment of this application.

[0020] The components represented by each number in the attached diagram are explained below:

[0021] Image acquisition module 100, feature extraction module 200, feature screening module 300, and defect detection module 400. Detailed Implementation

[0022] This application provides a machine vision-guided real-time defect detection method and system for corrugated cardboard, which addresses the technical problems of poor adaptability and insufficient accuracy in the detection of corrugated cardboard defects in the prior art.

[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0024] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0025] Example 1, as Figure 1 As shown, this application provides a machine vision-guided real-time defect detection method for corrugated cardboard, wherein the method includes:

[0026] S10: Real-time acquisition of grayscale images of the target corrugated cardboard production line.

[0027] In this embodiment, an industrial camera is used to acquire grayscale images of the target corrugated cardboard production line in real time. The target corrugated cardboard production line refers to the production line that produces the corrugated cardboard to be inspected. A grayscale image is an image where each pixel has only one sampled color; such images are typically displayed as grayscale ranging from the darkest black to the brightest white. Unlike black and white images, grayscale images have many levels of color depth between black and white, which can preserve image details while simplifying the matrix and reducing computational load.

[0028] S20: Obtain a dynamic statistical feature list, traverse the grayscale image to extract statistical features, and construct a primary feature vector based on the statistical feature extraction results.

[0029] Traditional methods often employ fixed feature extraction schemes, which cannot adapt to changes in different product specifications and production conditions. This results in features that lack specificity and are computationally redundant, affecting both detection efficiency and the inability to accurately capture real defect features, thus hindering the improvement of detection accuracy.

[0030] Step S20 in the method provided in this application embodiment includes:

[0031] Obtain product specification information for the target corrugated cardboard production line;

[0032] Based on the product specification information, retrieve the historical defect detection image set;

[0033] Based on the preset prior full statistical feature list, the historical defect detection image set is traversed to perform full feature extraction and obtain the historical defect feature set;

[0034] The historical defect feature set is used as verification data and input into the full inspection defect detector to obtain the verification results. The contribution rate of each statistical feature to the detection of the full inspection defect detector is extracted accordingly.

[0035] Based on the detection contribution rate, cumulative calculation is performed, and the top N statistical features that meet the preset contribution rate threshold are extracted and output as the dynamic statistical feature list.

[0036] Using the dynamic statistical feature list as the target, the grayscale image is traversed to extract features and obtain the statistical feature extraction results;

[0037] The statistical feature extraction results are dimensionless and normalized based on the mean of defect-free products to obtain a standard feature set.

[0038] Extract the contribution rate list corresponding to the dynamic statistical feature list, and normalize it accordingly to obtain the feature weight list;

[0039] Based on the feature weight list and the standard feature set, the primary feature vector of the grayscale image is constructed.

[0040] In this embodiment of the application, the product specification information of the corrugated cardboard currently being produced is read from the production management system, including cardboard type, flute type, grammage, etc.

[0041] Based on these product specifications, images of all corrugated cardboard with the same specifications and marked defects are retrieved from the historical database to obtain a historical defect detection image set. The historical defect detection image set contains images of corrugated cardboard with defects and defect markings, such as markings indicating the presence of collapsed corrugations or rough surfaces.

[0042] Based on a pre-defined list of prior full statistical features, the historical defect detection image set is traversed to extract all features and obtain the historical defect feature set. The prior full statistical feature list refers to all existing features that may contribute to defect identification, such as various texture features, grayscale statistical features, geometric shape features, etc.

[0043] A full-inspection defect detector is obtained. This detector refers to an existing detector model with comprehensive detection capabilities for the current factory. Historical defect feature sets are used as validation data and input into the full-inspection defect detector to obtain validation results. The contribution rate of each statistical feature to the detection result is then extracted. The detection contribution rate refers to the degree to which each statistical feature contributes to the final detection result. For example, texture features may contribute significantly to surface roughness defects, while geometric shape features may contribute less. For instance, principal component analysis can be used to extract the detection contribution rate, such as calculating the principal component contribution rate for each feature term. This process quantitatively assesses the relative importance and contribution of each statistical feature in defect detection, providing a basis for subsequent feature selection.

[0044] Based on the detection contribution rate, cumulative calculations are performed to extract the top N statistical features that meet the preset contribution rate threshold, and the output is a dynamic list of statistical features. The contribution rate threshold is set to filter out statistical features with large contribution rates and prevent statistical features with small contribution rates from wasting computing power; for example, it can be set to 0.6.

[0045] The algorithm extracts features from grayscale images by traversing a dynamic statistical feature list, obtaining the statistical feature extraction results. Specifically, a grayscale image feature extractor is constructed based on a convolutional neural network. For example, a four-layer structure can be used: an input layer receives the grayscale image, the convolutional layers use 64 3×3 convolutional kernels activated by ReLU, the pooling layers use 2×2 max pooling, and the output layer outputs the extracted grayscale image features. Annotated historical grayscale images are used for training, with the annotations representing the grayscale image features. These historical grayscale images are then input into the constructed grayscale feature extractor, and the prediction error is calculated using the cross-entropy loss function. This error quantifies the inaccuracy of the current model parameters. The backpropagation algorithm is used to automatically calculate the gradient of the loss function with respect to each parameter in the model. The gradient indicates the direction and magnitude of adjustment for each parameter to reduce the loss. The Adam optimizer is used to iteratively update all parameters in the network based on the calculated gradients. The grayscale image feature extractor training is complete when the loss value decreases to a level that no longer changes significantly and the model accuracy stabilizes, for example, reaching 90%. A grayscale image feature extractor is used to extract features by traversing the grayscale image with a dynamic statistical feature list as the target, and the feature extraction results are obtained.

[0046] The statistical feature extraction results are dimensionless and normalized based on the mean of defect-free products to obtain a standard feature set. For example, the Z-score normalization algorithm is used for dimensionless processing and normalization. The standard feature value = (feature value - mean of all historical values ​​of the feature) ÷ mean of all historical values ​​of the feature.

[0047] Extract the contribution rate list corresponding to the dynamic statistical feature list, normalize it accordingly, and obtain the feature weight list. Feature weight = contribution rate ÷ sum of contribution rates of all features, and the sum of all feature weights is 1.

[0048] Based on the feature weight list and the standard feature set, construct the primary feature vector of the grayscale image.

[0049] By using a dynamic statistical feature list to guide feature extraction and constructing a primary feature vector, intelligent and accurate feature extraction is achieved. It can adaptively select the most effective feature extraction scheme according to the actual production situation, which significantly improves the representativeness of features and computational efficiency, and provides high-quality feature input for subsequent defect identification.

[0050] S30: Construct a dynamic threshold model based on historical defect detection records, and use the dynamic threshold model as a constraint to perform initial screening of corrugated cardboard in combination with the primary feature vector.

[0051] Traditional defect detection methods are difficult to adapt to dynamic changes in the production process, which can easily lead to misjudgments and missed detections, seriously affecting the reliability of detection.

[0052] Step S30 in the method provided in this application embodiment includes:

[0053] Obtain the defect detection records of the first N batches of corrugated cardboard, and obtain the historical defect detection records, where N is greater than or equal to 3;

[0054] Based on the dynamic statistical feature list, the historical defect detection records are traversed to extract features and obtain the historical defect detection feature set.

[0055] Analyze the historical defect detection feature set, determine the feature value of each statistical feature that satisfies the preset confidence value, and merge and output multiple feature values ​​corresponding to the dynamic statistical feature list as the dynamic threshold model;

[0056] The process includes analyzing the historical defect detection feature set, determining the feature value of each statistical feature that satisfies a preset confidence value, and merging and outputting multiple feature values ​​corresponding to the dynamic statistical feature list as the dynamic threshold model, including:

[0057] Based on the dynamic statistical feature list, any one of the statistical features is randomly selected as the first analytical feature;

[0058] Extract the feature value corresponding to the first analysis feature from the historical defect detection feature set to obtain the first analysis feature value set;

[0059] Serialize the first set of analytical features and determine the feature value at the quantile corresponding to the preset confidence value to obtain the first dynamic threshold;

[0060] The dynamic statistical feature list is traversed to obtain multiple dynamic thresholds, which are then merged and output as the dynamic threshold model.

[0061] In this embodiment, defect detection records of the first N batches of corrugated cardboard are obtained. Based on these records, historical defect detection records are then obtained, where N is greater than or equal to 3. Obtaining more defect information from recent batches is more conducive to extracting commonalities and providing a data foundation for subsequent defect analysis.

[0062] Based on the dynamic statistical feature list, historical defect detection records are traversed to extract features and obtain a historical defect detection feature set. For example, the historical defect detection feature set includes one or more features such as the average grayscale value of an image, the standard deviation of the grayscale value of an image, and the image contrast.

[0063] By analyzing historical defect detection feature sets, the feature values ​​that satisfy preset confidence levels for each statistical feature are determined, and multiple feature values ​​corresponding to a dynamic statistical feature list are merged and output as a dynamic threshold model. Using feature values ​​that satisfy preset confidence levels as dynamic thresholds allows the detection to dynamically adjust judgment criteria based on actual production data. This avoids missed detections due to overly lenient fixed thresholds, while preventing false positives due to overly strict thresholds, significantly improving the adaptability and reliability of the defect detection system under different production batches and process parameter variations.

[0064] Specifically, based on a dynamic list of statistical features, any statistical feature is randomly selected as the first analytical feature and the current processing object.

[0065] Extract the feature values ​​corresponding to the first analysis feature from the historical defect detection feature set to obtain the first analysis feature value set.

[0066] The first set of analytical feature values ​​is serialized, and the feature value at the quantile corresponding to the preset confidence level is determined to obtain the first dynamic threshold. For example, if the first analytical feature is the standard deviation of image grayscale values, multiple feature values ​​of the standard deviation of image grayscale values ​​are extracted from the historical defect detection feature set, and these multiple feature values ​​are integrated into a set as the first set of analytical feature values. The multiple feature values ​​in the first set of analytical feature values ​​are sorted from smallest to largest. For example, if there are 10 feature values, the sorting would be "0.1, 0.12, 0.2, 0.3, 0.45, 0.5, 0.52, 0.55, 0.61, 0.8". When the preset confidence level is 0.8, the feature value at the corresponding quantile is 0.55, and the first dynamic threshold is 0.55. This threshold value indicates that 80% of the grayscale value standard deviations in the first set of analytical feature values ​​are less than 0.55.

[0067] The system iterates through the list of dynamic statistical features, obtains multiple dynamic thresholds, and merges them into a dynamic threshold model. The dynamic threshold model sets a dynamic threshold standard for each key feature based on historical defect data statistical experience. This threshold set can then be used as a standard for rapid initial screening of the data.

[0068] Using a dynamic threshold model as a constraint and combining it with primary feature vectors, the corrugated cardboard is initially screened to obtain the initial screening results.

[0069] By constructing a dynamic threshold model based on historical data and combining it with primary feature vectors for initial screening, adaptive optimization of defect judgment criteria was achieved. This enables defect detection to dynamically adjust judgment criteria according to actual production conditions, significantly improving the accuracy and adaptability of initial screening and providing reliable preprocessing results for subsequent fine inspection.

[0070] S40: Based on the initial screening results, extract the corresponding grayscale image, perform adaptive grayscale enhancement, obtain the enhanced grayscale image, and perform defect detection based on the enhanced grayscale image.

[0071] After the initial screening, the final challenge is to accurately identify suspected defects. Traditional methods cannot be optimized for defects with different characteristics, which can lead to the inaccurate detection of weak or special defects. At the same time, the high computational complexity results in low detection efficiency.

[0072] Step S40 in the method provided in this application embodiment includes:

[0073] Based on the initial screening results, the corresponding grayscale image is extracted and adaptively enhanced to obtain an enhanced grayscale image. Defect detection is then performed based on the enhanced grayscale image. Prior to this, the process includes:

[0074] Image segmentation is performed based on the historical defect detection image set to obtain the historical defect image set;

[0075] The historical defect image set is traversed to perform statistical analysis on image grayscale values ​​to obtain the grayscale distribution characteristics of the defect images. Based on the grayscale distribution characteristics of the defect images, a function is fitted, and the function fitting result is output as an interval grayscale enhancement function.

[0076] Combining the aforementioned interval grayscale enhancement function, an adaptive grayscale enhancement function is constructed, wherein the adaptive grayscale enhancement function is a piecewise function;

[0077] Based on the initial screening results, the corresponding grayscale image is extracted and adaptively enhanced to obtain an enhanced grayscale image. Defect detection is then performed based on the enhanced grayscale image, including:

[0078] Using the initial screening results as an index, the grayscale image is extracted, and the set of images to be detected is output.

[0079] The image set to be detected is subjected to grayscale enhancement processing according to the adaptive grayscale enhancement function to generate the enhanced grayscale image;

[0080] Based on the knowledge distillation method, the model of the full inspection defect detector is compressed by combining the historical defect detection feature set and the historical defect detection record to obtain a minor defect detector. The enhanced grayscale image is then input into the minor defect detector for defect detection to obtain the defect detection result.

[0081] The defect detection results include at least the defect location and the defect category.

[0082] In this embodiment, image segmentation is performed based on a historical defect detection image set to obtain a historical defect image set. For example, a grayscale threshold-based segmentation method is used to separate the defective regions in the image from the background, resulting in a historical defect image set containing images of various defects.

[0083] By traversing the historical defect image set and performing statistical analysis on the grayscale values ​​of all defect pixels, such as plotting their grayscale histograms, the grayscale distribution characteristics of the defect images are obtained. Statistical analysis reveals that the grayscale values ​​of defect areas are typically concentrated within a specific interval. Based on these grayscale distribution characteristics, a function fitting method, such as using a piecewise linear function, is employed to construct an interval grayscale enhancement function. The core function of this function is to perform high-contrast stretching on the grayscale intervals within the defect feature set, while compressing non-critical intervals, thereby highlighting the defects.

[0084] By combining the interval grayscale enhancement function, an adaptive grayscale enhancement function is constructed. This adaptive grayscale enhancement function is a piecewise function whose basic logic is: if the grayscale value of the input pixel is within the defect feature interval, the interval grayscale enhancement function is applied for significant enhancement; if it is outside the interval, a smoother mapping function is applied to maintain the overall naturalness of the image. This results in a function that can adaptively and selectively enhance defective regions.

[0085] Using the initial screening results as an index, extract grayscale images and output the set of images to be detected.

[0086] The grayscale enhancement function is used to perform grayscale enhancement processing on the image set to be detected, generating an enhanced grayscale image.

[0087] Based on the knowledge distillation method, a lightweight defect detector is obtained by compressing the full-inspection defect detector model by combining historical defect detection feature sets and historical defect detection records. Specifically, the structurally complex but highly accurate full-inspection defect detector is used as the teacher model to construct a lighter network, such as a 3-layer structure. The input layer receives grayscale images, the convolutional layer contains 32 3×3 convolutional kernels, and the output layer outputs the defect detection results. This lightweight network serves as the student model. Using the historical defect detection feature set and historical defect detection records, the lightweight defect detector not only learns how to predict true defect labels but also learns to mimic the probability distribution output by the full-inspection defect detector. This probability distribution contains rich knowledge from the teacher model, allowing the lightweight defect detector to maintain high accuracy even after compression. Mean squared error is used as the loss function during training, and Adam is used as the optimizer.

[0088] An enhanced grayscale image is input into a light defect detector for defect detection, and the defect detection results are obtained. The defect detection results include at least the defect location and defect category.

[0089] By using adaptive grayscale enhancement and image-based defect detection, accurate identification of suspected defects is achieved. Personalized image enhancement processing can be performed for different defect features, significantly improving the detection capability for various types of defects. At the same time, while ensuring detection accuracy, detection efficiency is improved, ultimately achieving efficient and accurate defect detection.

[0090] Example 2, as Figure 2 As shown, based on the same inventive concept as the machine vision-guided real-time defect detection method for corrugated cardboard provided in Embodiment 1, this embodiment of the invention also provides a machine vision-guided real-time defect detection system for corrugated cardboard, comprising:

[0091] Image acquisition module 100 is used to acquire grayscale images of the target corrugated cardboard production line in real time;

[0092] The feature extraction module 200 is used to obtain a dynamic statistical feature list, perform statistical feature extraction by traversing the grayscale image accordingly, and construct a primary feature vector based on the statistical feature extraction results.

[0093] The feature screening module 300 is used to construct a dynamic threshold model based on historical defect detection records, and to perform initial screening of corrugated cardboard by combining the dynamic threshold model with the primary feature vector;

[0094] The defect detection module 400 is used to extract the grayscale image according to the initial screening result, perform adaptive grayscale enhancement, obtain an enhanced grayscale image, and perform defect detection based on the enhanced grayscale image.

[0095] In one embodiment, the feature extraction module 200 is further configured to:

[0096] Obtain product specification information for the target corrugated cardboard production line;

[0097] Based on the product specification information, retrieve the historical defect detection image set;

[0098] Based on the preset prior full statistical feature list, the historical defect detection image set is traversed to perform full feature extraction and obtain the historical defect feature set;

[0099] The historical defect feature set is used as verification data and input into the full inspection defect detector to obtain the verification results. The contribution rate of each statistical feature to the detection of the full inspection defect detector is extracted accordingly.

[0100] Based on the detection contribution rate, cumulative calculation is performed, and the top N statistical features that meet the preset contribution rate threshold are extracted and output as the dynamic statistical feature list.

[0101] Using the dynamic statistical feature list as the target, the grayscale image is traversed to extract features and obtain the statistical feature extraction results;

[0102] The statistical feature extraction results are dimensionless and normalized based on the mean of defect-free products to obtain a standard feature set.

[0103] Extract the contribution rate list corresponding to the dynamic statistical feature list, and normalize it accordingly to obtain the feature weight list;

[0104] Based on the feature weight list and the standard feature set, the primary feature vector of the grayscale image is constructed.

[0105] In one embodiment, the feature screening module 300 is further configured to:

[0106] Obtain the defect detection records of the first N batches of corrugated cardboard, and obtain the historical defect detection records, where N is greater than or equal to 3;

[0107] Based on the dynamic statistical feature list, the historical defect detection records are traversed to extract features and obtain the historical defect detection feature set.

[0108] Analyze the historical defect detection feature set, determine the feature value of each statistical feature that satisfies the preset confidence value, and merge and output multiple feature values ​​corresponding to the dynamic statistical feature list as the dynamic threshold model;

[0109] Specifically, the process involves analyzing the historical defect detection feature set, determining the feature value of each statistical feature that satisfies a preset confidence level, and merging and outputting multiple feature values ​​corresponding to the dynamic statistical feature list as the dynamic threshold model, including:

[0110] Based on the dynamic statistical feature list, any one of the statistical features is randomly selected as the first analytical feature;

[0111] Extract the feature value corresponding to the first analysis feature from the historical defect detection feature set to obtain the first analysis feature value set;

[0112] Serialize the first set of analytical features and determine the feature value at the quantile corresponding to the preset confidence value to obtain the first dynamic threshold;

[0113] The dynamic statistical feature list is traversed to obtain multiple dynamic thresholds, which are then merged and output as the dynamic threshold model.

[0114] In one embodiment, the defect detection module 400 is further configured to:

[0115] Based on the initial screening results, the corresponding grayscale image is extracted and adaptively enhanced to obtain an enhanced grayscale image. Defect detection is then performed based on the enhanced grayscale image. Prior to this, the process includes:

[0116] Image segmentation is performed based on the historical defect detection image set to obtain the historical defect image set;

[0117] The historical defect image set is traversed to perform statistical analysis on image grayscale values ​​to obtain the grayscale distribution characteristics of the defect images. Based on the grayscale distribution characteristics of the defect images, a function is fitted, and the function fitting result is output as an interval grayscale enhancement function.

[0118] Combining the aforementioned interval grayscale enhancement function, an adaptive grayscale enhancement function is constructed, wherein the adaptive grayscale enhancement function is a piecewise function;

[0119] Based on the initial screening results, the corresponding grayscale image is extracted and adaptively enhanced to obtain an enhanced grayscale image. Defect detection is then performed based on the enhanced grayscale image, including:

[0120] Using the initial screening results as an index, the grayscale image is extracted, and the set of images to be detected is output.

[0121] The image set to be detected is subjected to grayscale enhancement processing according to the adaptive grayscale enhancement function to generate the enhanced grayscale image;

[0122] Based on the knowledge distillation method, the model of the full inspection defect detector is compressed by combining the historical defect detection feature set and the historical defect detection record to obtain a minor defect detector. The enhanced grayscale image is then input into the minor defect detector for defect detection to obtain the defect detection result.

[0123] The defect detection results include at least the defect location and the defect category.

[0124] In summary, the embodiments of this application have at least the following technical effects:

[0125] This application proposes a machine vision-guided real-time defect detection method and system for corrugated cardboard. By combining dynamic statistical feature extraction with adaptive threshold adjustment, the accuracy and adaptability of corrugated cardboard defect detection are significantly improved. Specifically, by constructing a dynamic statistical feature list and a dynamic threshold model based on historical data, adaptive adjustments to production conditions are achieved, effectively reducing false detection and false negative rates. By introducing an adaptive grayscale enhancement mechanism, precise image enhancement processing can be performed on suspected defect areas, improving the ability to identify weak and variant defects. Simultaneously, a lightweight defect detector implemented using knowledge distillation technology significantly improves detection speed while maintaining detection accuracy. Compared to traditional methods, the technical solution provided in this application significantly overcomes the limitations of traditional methods in dealing with different product specifications and changes in production environments, achieving efficient and accurate defect detection of corrugated cardboard in complex production environments.

[0126] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0127] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0128] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A machine vision-guided real-time defect detection method for corrugated cardboard, characterized in that, include: Real-time acquisition of grayscale images of the target corrugated cardboard production line; Obtain a dynamic statistical feature list, traverse the grayscale image accordingly to extract statistical features, and construct a primary feature vector based on the statistical feature extraction results; A dynamic threshold model is constructed based on historical defect detection records, and the corrugated cardboard is initially screened using the dynamic threshold model as a constraint and combined with the primary feature vector. Based on the initial screening results, the corresponding grayscale image is extracted and adaptively enhanced to obtain an enhanced grayscale image, and defect detection is performed based on the enhanced grayscale image. Among them, a dynamic threshold model is constructed based on historical defect detection records, including: Obtain the defect detection records of the first N batches of corrugated cardboard, and obtain the historical defect detection records, where N is greater than or equal to 3; Based on the dynamic statistical feature list, the historical defect detection records are traversed to extract features and obtain the historical defect detection feature set. Analyze the historical defect detection feature set, determine the feature value of each statistical feature that satisfies a preset confidence value, and merge and output multiple feature values ​​corresponding to the dynamic statistical feature list as the dynamic threshold model, including: Based on the dynamic statistical feature list, any one of the statistical features is randomly selected as the first analytical feature; Extract the feature value corresponding to the first analysis feature from the historical defect detection feature set to obtain the first analysis feature value set; Serialize the first set of analytical features and determine the feature value at the quantile corresponding to the preset confidence value to obtain the first dynamic threshold; The dynamic statistical feature list is traversed to obtain multiple dynamic thresholds, which are then merged and output as the dynamic threshold model.

2. The machine vision-guided real-time defect detection method for corrugated cardboard as described in claim 1, characterized in that, Obtain a list of dynamic statistical features, including: Obtain product specification information for the target corrugated cardboard production line; Based on the product specification information, retrieve the historical defect detection image set; Based on the preset prior full statistical feature list, the historical defect detection image set is traversed to perform full feature extraction and obtain the historical defect feature set; The historical defect feature set is used as verification data and input into the full inspection defect detector to obtain the verification results. The contribution rate of each statistical feature to the detection of the full inspection defect detector is extracted accordingly. Based on the detection contribution rate, cumulative calculations are performed to extract the top N statistical features that meet the preset contribution rate threshold, and the output is the dynamic statistical feature list.

3. The machine vision-guided real-time defect detection method for corrugated cardboard as described in claim 2, characterized in that, Statistical features are extracted by traversing the grayscale image, and a primary feature vector is constructed based on the results of the statistical feature extraction, including: Using the dynamic statistical feature list as the target, the grayscale image is traversed to extract features and obtain the statistical feature extraction results; The statistical feature extraction results are dimensionless and normalized based on the mean of defect-free products to obtain a standard feature set. Extract the contribution rate list corresponding to the dynamic statistical feature list, and normalize it accordingly to obtain the feature weight list; Based on the feature weight list and the standard feature set, the primary feature vector of the grayscale image is constructed.

4. The machine vision-guided real-time defect detection method for corrugated cardboard as described in claim 3, characterized in that, Based on the initial screening results, the corresponding grayscale image is extracted and adaptively enhanced to obtain an enhanced grayscale image. Defect detection is then performed based on the enhanced grayscale image. Prior to this, the process includes: Image segmentation is performed based on the historical defect detection image set to obtain the historical defect image set; The historical defect image set is traversed to perform statistical analysis on image grayscale values ​​to obtain the grayscale distribution characteristics of the defect images. Based on the grayscale distribution characteristics of the defect images, a function is fitted, and the output function fitting result is an interval grayscale enhancement function. By combining the aforementioned interval grayscale enhancement function, an adaptive grayscale enhancement function is constructed, wherein the adaptive grayscale enhancement function is a piecewise function.

5. The machine vision-guided real-time defect detection method for corrugated cardboard as described in claim 4, characterized in that, Based on the initial screening results, the corresponding grayscale image is extracted and adaptively enhanced to obtain an enhanced grayscale image. Defect detection is then performed based on the enhanced grayscale image, including: Using the initial screening results as an index, the grayscale image is extracted, and the set of images to be detected is output. The image set to be detected is subjected to grayscale enhancement processing according to the adaptive grayscale enhancement function to generate the enhanced grayscale image; Based on the knowledge distillation method, the model of the full inspection defect detector is compressed by combining the historical defect detection feature set and the historical defect detection record to obtain a minor defect detector. The enhanced grayscale image is then input into the minor defect detector for defect detection to obtain the defect detection result. The defect detection results include at least the defect location and the defect category.

6. A machine vision-guided real-time defect detection system for corrugated cardboard, characterized in that, For implementing a machine vision-guided real-time defect detection method for corrugated cardboard according to any one of claims 1 to 5, the system comprises: The image acquisition module is used to acquire grayscale images of the target corrugated cardboard production line in real time. The feature extraction module is used to obtain a dynamic statistical feature list, perform statistical feature extraction on the corresponding grayscale image, and construct a primary feature vector based on the statistical feature extraction results. The feature screening module is used to construct a dynamic threshold model based on historical defect detection records, and to perform initial screening of corrugated cardboard by combining the dynamic threshold model with the primary feature vector; The defect detection module is used to extract the grayscale image according to the initial screening results, perform adaptive grayscale enhancement to obtain the enhanced grayscale image, and perform defect detection based on the enhanced grayscale image.

Images