Self-adhesive label production quality monitoring method and system based on image analysis

Abstract

The invention discloses a self-adhesive label production quality monitoring method and system based on image analysis, and relates to the technical field of label quality monitoring, and the method comprises the steps: obtaining a self-adhesive label; performing image analysis according to the plane material layer monitoring image of the plane material layer, and determining a label plane material quality coefficient; comprehensively detecting the adhesion performance of the adhesive layer to obtain a label adhesion quality coefficient; performing bilateral characteristic detection integration on the backing paper layer according to a backing paper bilateral characteristic detection weight to obtain a label backing paper quality coefficient; setting a label quality expectation condition based on the self-adhesive label; and carrying out expectation coincidence judgment, and outputting a self-adhesive label quality detection result. The technical problems that in the prior art, non-setting adhesive production quality monitoring is lack of accuracy and comprehensiveness, and the quality control efficiency is low are solved, and the technical effects that the non-setting adhesive label production quality monitoring accuracy and comprehensiveness are improved, and the quality control efficiency is improved are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of label quality monitoring, and particularly relates to a method and system for monitoring production quality of adhesive labels based on image analysis. BACKGROUND

[0002] In the field of adhesive label production, quality monitoring has always been a crucial link. With the growing demand for adhesive labels in the market and the increasing diversification of application scenarios, the requirements for the quality of adhesive labels are becoming higher and higher. However, in the prior art, there are many problems in the production quality monitoring of adhesive labels. On the one hand, traditional quality monitoring methods often focus on a single indicator or aspect, lacking comprehensive detection of each layer of adhesive labels (such as the face material layer, the adhesive layer, and the backing paper layer). This one-sided detection method leads to a lack of accuracy and comprehensiveness in the quality monitoring results, making it difficult to truly reflect the overall quality of adhesive labels. On the other hand, the existing quality monitoring methods also have deficiencies in quality control efficiency. Due to the lack of advanced and intelligent detection means, the detection process is tedious and time-consuming, not only increasing production costs, but also failing to timely discover and solve quality problems in the production process. This makes the produced adhesive labels may have unstable quality.

[0003] The prior art has the technical problems of lack of accuracy and comprehensiveness in the production quality monitoring of adhesive labels, and low efficiency in quality control. SUMMARY

[0004] The present application provides a method and system for monitoring production quality of adhesive labels based on image analysis, which is used to solve the technical problems of lack of accuracy and comprehensiveness in the production quality monitoring of adhesive labels, and low efficiency in quality control in the prior art.

[0005] In view of the above problems, the present application provides a method and system for monitoring production quality of adhesive labels based on image analysis.

[0006] In a first aspect of the present application, a method for monitoring production quality of adhesive labels based on image analysis is provided, the method comprising: The label face material quality coefficient is determined by image analysis according to a face material layer monitoring image of the face material layer; the label adhesive quality coefficient is obtained by comprehensive adhesion performance detection of the adhesive layer according to multi-dimensional adhesion performance detection factors, wherein the multi-dimensional adhesion performance detection factors include initial adhesion, holding adhesion and adhesion migration; the label backing paper quality coefficient is obtained by integrating double-side characteristic detection of the backing paper layer according to a backing paper double-side characteristic detection weight; based on the label, a label quality expectation condition is set, wherein the label quality expectation condition includes an expected face material quality coefficient, an expected label adhesive quality coefficient and an expected label backing paper quality coefficient; based on the label quality expectation condition, expected compliance judgment is performed on the label face material quality coefficient, the label adhesive quality coefficient and the label backing paper quality coefficient, and a label quality detection result is output.

[0007] In a second aspect of the present application, an image analysis-based label production quality monitoring system is provided, which comprises: A label acquisition module is configured to acquire a label, wherein the label comprises a face material layer, an adhesive layer and a backing paper layer; a label face material quality coefficient determination module is configured to determine a label face material quality coefficient by image analysis according to a face material layer monitoring image of the face material layer; a label adhesive quality coefficient acquisition module is configured to obtain a label adhesive quality coefficient by comprehensive adhesion performance detection of the adhesive layer according to multi-dimensional adhesion performance detection factors, wherein the multi-dimensional adhesion performance detection factors include initial adhesion, holding adhesion and adhesion migration; a label backing paper quality coefficient acquisition module is configured to obtain a label backing paper quality coefficient by integrating double-side characteristic detection of the backing paper layer according to a backing paper double-side characteristic detection weight; a label quality expectation condition setting module is configured to set a label quality expectation condition based on the label, wherein the label quality expectation condition includes an expected face material quality coefficient, an expected label adhesive quality coefficient and an expected label backing paper quality coefficient; and a label quality detection result output module is configured to perform expected compliance judgment on the label face material quality coefficient, the label adhesive quality coefficient and the label backing paper quality coefficient based on the label quality expectation condition, and output a label quality detection result.

[0008] One or more technical solutions provided in the present application have at least the following technical effects or advantages: The label face material quality coefficient is determined according to image analysis of a face material layer monitoring image of the face material layer; the adhesive quality coefficient of the label is obtained according to comprehensive detection of the adhesive layer based on a plurality of adhesion performance detection factors; the label base paper quality coefficient is obtained by integrating bilateral property detection of the base paper layer; the label quality expectation condition is set based on the adhesive label; the label face material quality coefficient, the label adhesive quality coefficient and the label base paper quality coefficient are judged to meet the expectation based on the label quality expectation condition, and the adhesive label quality detection result is output. The technical effect of improving the accuracy and comprehensiveness of adhesive label production quality monitoring and improving the quality control efficiency is achieved. BRIEF DESCRIPTION OF DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0010] Figure 1 The flow chart of the adhesive label production quality monitoring method based on image analysis provided by the embodiments of the present application is shown. Figure 2 The structure diagram of the adhesive label production quality monitoring system based on image analysis provided by the embodiments of the present application is shown.

[0011] The label face material quality coefficient is determined according to image analysis of a face material layer monitoring image of the face material layer; the adhesive quality coefficient of the label is obtained according to comprehensive detection of the adhesive layer based on a plurality of adhesion performance detection factors; the label base paper quality coefficient is obtained by integrating bilateral property detection of the base paper layer; the label quality expectation condition is set based on the adhesive label; the label face material quality coefficient, the label adhesive quality coefficient and the label base paper quality coefficient are judged to meet the expectation based on the label quality expectation condition, and the adhesive label quality detection result is output. The technical effect of improving the accuracy and comprehensiveness of adhesive label production quality monitoring and improving the quality control efficiency is achieved. DETAILED DESCRIPTION

[0012] The present application provides an adhesive label production quality monitoring method and system based on image analysis, which is used to solve the technical problems of lack of accuracy and comprehensiveness and low quality control efficiency in the prior art adhesive production quality monitoring.

[0013] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings of the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0014] Embodiment one, as Figure 1As shown, the present application provides a method for monitoring the production quality of self-adhesive labels based on image analysis, the method comprising: Step S100: obtaining a self-adhesive label, wherein the self-adhesive label comprises a surface material layer, an adhesive layer and a backing paper layer.

[0015] Specifically, self-adhesive labels are a common material used for marking and pasting. They consist of three main components: a facestock layer, an adhesive layer, and a backing paper layer. The facestock layer is the outermost layer of the self-adhesive label and comes into direct contact with the outside world. It typically has a certain degree of printability and wear resistance to enable clear printing of various identification information and maintain surface integrity during use. The facestock layer can be made of a variety of materials, such as paper and plastic film, and different materials are suitable for different application scenarios. The adhesive layer, located between the facestock layer and the backing paper layer, is the key component that enables the self-adhesive label to be adhered to the surface of an object. The performance of the adhesive layer directly affects the label's adhesion and durability. Its characteristics include initial tack, persistent tack, and adhesion migration, which require detailed evaluation in subsequent testing. The backing paper layer is the bottom layer of the self-adhesive label, which supports and protects the adhesive layer. The backing paper layer needs to have a certain strength to prevent damage during label production and use. At the same time, it needs to have appropriate anti-sticking properties to facilitate the label's peeling from the backing paper during use. The process of obtaining self-adhesive labels involves taking samples directly from the production line, or randomly selecting a certain number of samples from a batch of labels that have already been produced. These samples will be used as the objects of subsequent quality monitoring to ensure that the quality of the entire self-adhesive label production process is controllable.

[0016] Step S200: performing image analysis based on the surface material layer monitoring image of the surface material layer to determine the label surface material quality coefficient.

[0017] Specifically, a monitoring image of the face material layer is obtained, which is captured by professional image acquisition equipment that can accurately capture the surface features of the face material layer, including color, texture, flatness, and the presence of defects, etc. Then, image analysis is performed on the obtained face material layer monitoring image, which involves multiple technical and algorithmic aspects. For example, color analysis algorithms are used to evaluate the color consistency of the face material layer. By comparing the color values of different areas in the image, it is determined whether there is a color deviation that would affect the appearance quality and printing effect of the label. Meanwhile, texture analysis is also an important part of the process. By analyzing the uniformity and clarity of the texture, the material characteristics and surface treatment of the face material layer can be understood. If the texture is not uniform or clear, it means that there are problems with the quality of the face material layer, such as uneven material mixing or imperfect surface processing. Flatness analysis determines whether the face material layer is flat by detecting the ups and downs of the surface in the image. An uneven face material layer would make printing difficult, affecting the clarity and accuracy of the label. In addition, it is necessary to detect whether there are defects such as scratches, holes, and spots in the image. These defects would directly affect the appearance quality and performance of the label. A comprehensive evaluation of the face material layer monitoring image ultimately determines a label face material quality coefficient that quantifies the quality status of the face material layer. This coefficient serves as an important basis for determining whether the quality of the face material layer meets the requirements and provides key data support for subsequent adhesive label quality monitoring.

[0018] Step S300: Perform comprehensive adhesion performance detection on the adhesive layer according to multi-dimensional adhesion performance detection factors, and obtain a label adhesive quality coefficient, wherein the multi-dimensional adhesion performance detection factors include initial adhesion, holding adhesion, and adhesion migration.

[0019] Specifically, to obtain the label adhesive quality coefficient, comprehensive adhesion performance detection of the adhesive layer is required based on multi-dimensional adhesion performance detection factors including initial adhesion, holding adhesion, and adhesion migration. For initial adhesion, a test instrument is activated based on relevant factor generation detection instructions to obtain data and process the results, and then a model is built using deep learning to output an evaluation coefficient to be added to the detection results. For holding adhesion, the test instrument is activated according to instructions to obtain results, and a neural network is trained using the result record set to generate an evaluation channel to obtain a coefficient to be added to the detection results when the loss coefficient condition is met. For adhesion migration, multiple detection scenarios are built by activating scene factors to test the adhesive layer to obtain the migration adhesive amount, and an evaluation coefficient is calculated to be added to the detection results. Finally, the multi-dimensional adhesion performance detection factors are considered for weight setting and calculation, and the comprehensive detection results are weighted to obtain the label adhesive quality coefficient.

[0020] Step S400: Perform double-sided characteristic detection integration of the base paper layer according to double-sided characteristic detection weights, and obtain a label base paper quality coefficient.

[0021] Specifically, the work is carried out according to the weight of the double-side characteristic detection of the base paper. For the base paper layer, the anti-adhesion characteristic detection and the strength characteristic detection are carried out respectively. In the anti-adhesion characteristic detection, a plurality of base paper samples (which have the same production parameters as the base paper layer) are obtained, and then a peel force tester is used to test the peel force of the samples respectively to obtain a plurality of test peel forces. The peel force test result is obtained by central value calculation, and the pre-constructed anti-adhesion characteristic detection model is activated. The peel force test result is input into the model, and the anti-adhesion characteristic coefficient of the base paper is output. In the strength characteristic detection, the strength characteristic coefficient of the base paper is obtained by corresponding detection means. Then, the obtained anti-adhesion characteristic coefficient and strength characteristic coefficient of the base paper are normalized to obtain the double-side characteristic detection result of the base paper. Finally, the detection result is weighted based on the weight of the double-side characteristic detection of the base paper to generate the label base paper quality coefficient, which can reflect the quality status of the base paper layer and provide an important basis for the overall quality evaluation of the adhesive label.

[0022] Step S500: setting label quality expectation conditions based on the adhesive label, wherein the label quality expectation conditions include expected facestock quality coefficient, expected label adhesive quality coefficient and expected label base paper quality coefficient.

[0023] Specifically, the label quality expectation conditions are set based on the obtained adhesive label, and the set label quality expectation conditions include expected facestock quality coefficient, expected label adhesive quality coefficient and expected label base paper quality coefficient. The expected facestock quality coefficient is a quantitative index of the ideal quality that the facestock layer should achieve according to the application scenario and quality standard of the adhesive label. It comprehensively considers the requirements of the facestock layer in color, texture, flatness and flawlessness, etc. The expected label adhesive quality coefficient is a quantitative index set for the adhesive layer considering the ideal standards that should be met in the multi-dimensional adhesion performance such as initial adhesion, adhesion retention and adhesion migration. The expected label base paper quality coefficient is a quantitative index set based on the ideal state that the base paper layer should achieve in terms of anti-adhesion characteristic and strength characteristic. By clearly defining these expectation conditions, a clear reference standard is provided for the judgment of the adhesive label quality detection result.

[0024] Step S600: making expected compliance judgments on the label facestock quality coefficient, the label adhesive quality coefficient and the label base paper quality coefficient based on the label quality expectation conditions, and outputting an adhesive label quality detection result.

[0025] Specifically, based on the set label quality expectation conditions, the label face material quality coefficient is expected to meet the judgment, and the actual detected label face material quality coefficient is compared with the expected face material quality coefficient. If the two values are equal or the difference is within a reasonable error allowed range, it can be determined that the label face material quality meets the expected conditions, which means that the face material layer has reached the expected quality standards in terms of color, texture, flatness and no defects. Then, the label adhesive quality coefficient is judged similarly, and the actual label adhesive quality coefficient is compared with the expected label adhesive quality coefficient. If the actual coefficient meets the requirements set by the expected label adhesive quality coefficient, including the standards of initial adhesion, adhesion and adhesion migration, etc. Multidimensional adhesion performance, then it is determined that the adhesive layer quality meets the expectations. Then, the label paper quality coefficient is judged to meet the expectations, and the actual label paper quality coefficient is compared with the expected label paper quality coefficient. When the actual coefficient meets the expected coefficient, considering that the bottom paper layer reaches the expected state in terms of anti-sticking and strength characteristics, etc. It is determined that the bottom paper layer quality meets the expectations. Finally, the judgment results of the above three aspects are combined to output the adhesive label quality detection result. If the label face material quality coefficient, label adhesive quality coefficient and label paper quality coefficient all meet the respective expectations, the overall adhesive label quality detection result is qualified. Otherwise, as long as one coefficient does not meet the expected conditions, the adhesive label quality detection result is determined to be unqualified, thereby providing accurate evaluation conclusion for the production quality of adhesive labels.

[0026] In one possible implementation, step S300 further includes: Step S310: detecting the adhesion performance of the adhesive layer according to the multi-dimensional adhesion performance detection factor to obtain an adhesion performance detection result.

[0027] Step S320: initializing the weight setting according to the multi-dimensional adhesion performance detection factor to determine an adhesion performance detection first weight condition.

[0028] Step S330: calculating the proportion according to the adhesion performance detection result to obtain an adhesion performance detection second weight condition.

[0029] Step S340: calculating the central value of each dimension weight based on the adhesion performance detection first weight condition and the adhesion performance detection second weight condition to obtain an adhesion performance detection third weight condition.

[0030] Step S350: calculating the adhesion performance detection result according to the adhesion performance detection third weight condition to output the label adhesive quality coefficient.

[0031] Specifically, the adhesion performance detection of the adhesive layer is a process of comprehensively considering multi-dimensional adhesion performance detection factors to obtain the adhesion performance detection result containing the initial adhesion detection evaluation coefficient, the holding adhesion detection evaluation coefficient and the adhesion migration detection evaluation coefficient. First, the initial adhesion is detected, and the initial adhesion detection instruction is generated based on the initial adhesion requirement in the multi-dimensional adhesion performance detection factors, which activates the initial adhesion tester to detect the initial adhesion of the adhesive layer multiple times to obtain multiple initial adhesion detection data. The central value calculation of these data obtains the initial adhesion detection result, and the initial adhesion detection result record set and the initial adhesion detection evaluation record set are used for deep learning to build the initial adhesion detection evaluation model. Finally, the initial adhesion detection evaluation coefficient is output through the model based on the initial adhesion detection result, and is added to the adhesion performance detection result. Then, the holding adhesion is detected, and the holding adhesion tester is activated according to the holding adhesion detection instruction to test the holding adhesion confidence of the adhesive layer to obtain the holding adhesion test result. The residual neural network is trained with the holding adhesion test result record set as the input information and the holding adhesion detection evaluation record set as the output information. The holding adhesion detection evaluation loss coefficient is obtained every predetermined number of times of training, and the holding adhesion detection evaluation channel is generated when the loss coefficient is less than the holding adhesion detection evaluation loss threshold. The holding adhesion detection evaluation coefficient is obtained through the channel based on the holding adhesion test result, and is added to the adhesion performance detection result. Finally, the adhesion migration is detected, and the adhesion migration detection instruction is generated based on the adhesion migration requirement in the multi-dimensional adhesion performance detection factors to activate the adhesion migration detection scene factors, including the adhesion target, the adhesion duration, the test environment and the applied pressure, etc. Multiple adhesion migration detection scenes are built according to these scene factors, and the adhesion migration of the adhesive layer is tested according to these scenes to obtain multiple migration adhesive amounts. The adhesion migration evaluation is performed based on these scenes and the migration adhesive amounts to obtain multiple adhesion migration evaluation coefficients, and the adhesion migration detection evaluation coefficient is generated by the central value calculation of these coefficients and is added to the adhesion performance detection result.

[0032] According to the multi-dimensional adhesion performance detection factor to initialize the weight setting, so as to determine the adhesion performance detection first weight condition. This process needs to consider the various aspects contained in the multi-dimensional adhesion performance detection factor. For the factors of initial adhesion, adhesion and adhesion migration, the weight is allocated according to their relative importance in the overall performance of the adhesive and the degree of influence on the final quality of the adhesive label. For example, if the initial adhesion has a greater impact on the initial effect of label pasting in the actual application scene, a relatively high initial weight will be given to the initial adhesion. Similarly, if the adhesion is crucial to the stability of the label during long-term use, a suitable weight will be set for it. For adhesion migration, if it has a significant impact on the quality of label pasting in different environments, a corresponding weight will also be determined. In this way, the weight condition determined by considering various factors is the adhesion performance detection first weight condition, which will be the basis for further calculation and adjustment of the weight.

[0033] The analysis is performed on each part of the adhesion performance detection result, which involves the initial adhesion detection evaluation coefficient, the adhesion detection evaluation coefficient and the adhesion migration detection evaluation coefficient. The proportion of each coefficient in the overall adhesion performance detection result is calculated, for example, the proportion of initial adhesion is obtained by dividing the initial adhesion detection evaluation coefficient by the sum of all coefficients. In this way, the proportions of initial adhesion, adhesion and adhesion migration in the adhesion performance detection result are calculated respectively. These proportion values are the adhesion performance detection second weight condition, which reflects the relative importance of each adhesion performance factor in the actual detection result, and provides a basis for further accurate adjustment of the weight.

[0034] The first weight condition for adhesion performance detection is obtained according to the initialization weight setting of the multi-dimensional adhesion performance detection factor, which reflects the weight distribution of each adhesion performance factor (such as initial adhesion, holding adhesion, and adhesion migration) in theory. The second weight condition for adhesion performance detection is obtained according to the proportion calculation of the adhesion performance detection result, which reflects the weight of each factor in actual detection. Then, the two weight conditions are considered comprehensively. For each adhesion performance factor, the weight value in the first weight condition and the weight value corresponding to the proportion of the initial adhesion in the second weight condition are calculated by, for example, a weighted average method, to obtain a weight value that comprehensively considers the theory and the actual situation. Finally, by calculating each adhesion performance factor in this way, a new set of weight values is obtained, which is the third weight condition for adhesion performance detection, and comprehensively considers the theory and the actual situation, and can more accurately reflect the weight of each adhesion performance factor in the final adhesion performance evaluation, providing a more reasonable basis for subsequent weighted calculation of the adhesion performance detection result according to the weight condition.

[0035] The third weight condition for adhesion performance detection is obtained by comprehensively considering the initialization weight in theory (the first weight condition) and the proportion weight of the actual detection result (the second weight condition), and accurately reflects the weight distribution of each adhesion performance factor (such as initial adhesion, holding adhesion, and adhesion migration) in the final adhesion performance evaluation. Then, the adhesion performance detection result is weighted. The adhesion performance detection result includes the initial adhesion detection evaluation coefficient, the holding adhesion detection evaluation coefficient, and the adhesion migration detection evaluation coefficient. For the initial adhesion detection evaluation coefficient, multiply it by the weight value corresponding to the initial adhesion in the third weight condition for adhesion performance detection; for the holding adhesion detection evaluation coefficient, multiply it by the weight value corresponding to the holding adhesion; and for the adhesion migration detection evaluation coefficient, multiply it by the weight value corresponding to the adhesion migration. Finally, add the results of the above weighted calculation, and the sum is the label adhesive quality coefficient. This coefficient comprehensively considers each adhesion performance factor of the adhesive layer and its weight, and can accurately reflect the quality status of the adhesive layer, providing an important basis for the overall quality evaluation of the adhesive label.

[0036] In one possible implementation, step S310 further includes: Step S311: generating an initial adhesion detection instruction based on the multi-dimensional adhesion performance detection factor, and activating an initial adhesion tester based on the initial adhesion detection instruction.

[0037] Step S312: Based on the adhesive layer, multiple initial adhesion detections are performed according to the initial adhesion tester, and multiple initial adhesion detection data are obtained.

[0038] Step S313: Central value calculation is performed according to the multiple initial adhesion detection data, and an initial adhesion detection result is obtained.

[0039] Step S314: Deep learning is performed according to the initial adhesion detection result record set and the initial adhesion detection evaluation record set, and an initial adhesion detection evaluation model is built.

[0040] Step S315: Based on the initial adhesion detection result, the initial adhesion detection evaluation model is used to output an initial adhesion detection evaluation coefficient, and the initial adhesion detection evaluation coefficient is added to the adhesion performance detection result.

[0041] Specifically, the initial step of initial adhesion detection of the adhesive layer based on multi-dimensional adhesion performance detection factors is to generate initial adhesion detection instructions. The multi-dimensional adhesion performance detection factors include related requirements and parameter settings for initial adhesion detection. According to these settings, instructions specifically for initial adhesion detection are generated. This instruction is the key signal to start the initial adhesion detection process, which will activate the initial adhesion tester and prepare for subsequent detection operations.

[0042] In order to more accurately obtain the initial adhesion information of the adhesive layer, multiple initial adhesion detections are performed based on the adhesive layer and using the initial adhesion tester. In this process, multiple samples with the same production parameters as the adhesive layer are particularly used. This is because the initial adhesion of the adhesive layer may be affected by multiple factors. Using samples with the same production parameters can minimize the interference of other factors and ensure that the detection results truly reflect the initial adhesion characteristics of the adhesive layer. Each time the initial adhesion tester is used to detect a sample, an initial adhesion detection data is obtained. Since multiple detections are performed, multiple initial adhesion detection data are ultimately obtained, which will serve as an important basis for subsequent analysis and evaluation of the initial adhesion of the adhesive layer.

[0043] Central value calculation of multiple initial adhesion detection data is to extract a result that can represent the overall initial adhesion characteristics from numerous data. The average value is calculated by adding multiple initial adhesion detection data and then dividing by the total number of data. The average value obtained can reflect the overall level of these data. This has the advantage of considering all detection data and reducing the influence of individual outliers on the result. Through such central value calculation, the initial adhesion detection result obtained can serve as an important basis for subsequent evaluation and analysis, providing strong support for accurately judging the initial adhesion performance of the adhesive layer.

[0044] The initial adhesion detection result record set is taken as input data, and the initial adhesion detection evaluation record set is taken as expected output data. The data is preprocessed, such as normalization, so that the data is in a suitable range for better training. Then, a suitable BPNN structure is designed, the number of input layer nodes is determined as the number of features in the initial adhesion detection result record set, and the number of output layer nodes is determined as the number of categories of initial adhesion evaluation or specific evaluation values. The number of hidden layers is determined according to experience and experiment, and generally can be tried from a small number of nodes and gradually increased to find the best structure. Next, the BPNN is trained using the training data. During the training process, the data enters the network from the input layer, is calculated and transmitted through each layer, and the prediction result is generated at the output layer. The prediction result is compared with the expected output, the error is calculated, then the error is propagated from the output layer to the input layer through the back propagation algorithm, and the weights and biases of each layer in the network are adjusted according to the error. This process is repeated until the error reaches an acceptable range or a preset number of training times is reached. After training, the obtained BPNN model can quickly and accurately output the initial adhesion detection evaluation coefficient according to the new initial adhesion detection result. This model can automatically learn the complex relationship between the initial adhesion detection result and the evaluation, and provides an effective method for the evaluation of initial adhesion.

[0045] Based on the obtained initial adhesion detection result, it is input into the initial adhesion detection evaluation model. This model is built by deep learning on the initial adhesion detection result record set and the initial adhesion detection evaluation record set, and can accurately evaluate the input initial adhesion detection result. After the model receives the initial adhesion detection result, it outputs the initial adhesion detection evaluation coefficient through internal calculation and analysis. This coefficient is a quantitative evaluation of the initial adhesion detection result, reflecting the performance of the adhesive layer in terms of initial adhesion. For example, a higher evaluation coefficient indicates good initial adhesion, while a lower coefficient means that there is a problem with the initial adhesion. Finally, this initial adhesion detection evaluation coefficient is added to the adhesion performance detection result, which is a comprehensive consideration of the overall adhesion performance of the adhesive layer, including initial adhesion, holding adhesion and adhesion migration. After adding the initial adhesion detection evaluation coefficient, the adhesion performance detection result is more comprehensive and accurate, providing more information for subsequent comprehensive analysis and judgment of the adhesion performance of the adhesive layer.

[0046] In one possible implementation, step S310 further includes: Step S316: Based on the multi-dimensional adhesion performance detection factor, generate the holding adhesion detection instruction, and activate the holding adhesion tester based on the holding adhesion detection instruction; based on the adhesive layer, perform holding adhesion confidence test according to the holding adhesion tester to obtain holding adhesion test results; take the holding adhesion test result record set as input information, take the holding adhesion detection evaluation record set as output information, train the residual neural network, and obtain the holding adhesion detection evaluation loss coefficient every predetermined number of times of training; when the holding adhesion detection evaluation loss coefficient is less than the holding adhesion detection evaluation loss threshold, generate a holding adhesion detection evaluation channel; based on the holding adhesion test results, obtain a holding adhesion detection evaluation coefficient according to the holding adhesion detection evaluation channel, and add the holding adhesion detection evaluation coefficient to the adhesion performance detection results.

[0047] Specifically, the adhesive layer of the self-adhesive label is subjected to holding adhesion detection. Based on the holding adhesion related requirements in the multi-dimensional adhesion performance detection factor, a holding adhesion detection instruction is generated. The multi-dimensional adhesion performance detection factor comprehensively considers initial adhesion, holding adhesion and adhesion migration, etc. The holding adhesion is crucial for evaluating the adhesion performance of the adhesive layer during long-term use. The generated holding adhesion detection instruction serves as a specific signal for activating the holding adhesion tester, which is a device specially designed for measuring the holding adhesion of the adhesive layer. When receiving the holding adhesion detection instruction, the holding adhesion tester starts running and prepares to test the holding adhesion of the adhesive layer of the self-adhesive label.

[0048] The same multiple samples as the adhesive layer production parameters are used for detection, because in actual production, the performance of the adhesive layer will be affected by many factors, and the use of samples with the same production parameters can minimize the interference of other factors, ensuring that the detection results can truly reflect the holding adhesion characteristics of the adhesive layer. Multiple detections are performed to improve the accuracy and reliability of the detection results. Since holding adhesion has certain fluctuations, a single detection result cannot fully reflect the holding adhesion performance of the adhesive layer. Through multiple detections, more data can be obtained to more accurately evaluate the holding adhesion of the adhesive layer. After multiple detections, the central value is taken as the holding adhesion test result. The central value is the average value, and the purpose of taking the central value is to reduce the influence of individual abnormal values on the results, making the test results more stable and reliable. Through such a detection process, a test result that can more accurately reflect the holding adhesion of the adhesive layer is obtained, providing an important basis for subsequent evaluation and analysis.

[0049] The holding adhesion test result record set is taken as input information, which contains various data obtained from previous holding adhesion tests, such as holding adhesion test results under different conditions, results obtained by testing different samples, etc., which reflect the actual performance of the adhesive layer in terms of holding adhesion. At the same time, the holding adhesion detection evaluation record set is taken as output information, which contains information after evaluating the holding adhesion test results. Then, the residual neural network is trained, which has strong learning ability and generalization ability and can effectively handle complex nonlinear relationships. During the training process, the network continuously adjusts its weights and parameters to make the output results as close as possible to the expected output in the holding adhesion detection evaluation record set. The holding adhesion detection evaluation loss coefficient is obtained every predetermined number of times of training. This loss coefficient is used to measure the gap between the current network output and the expected output. The smaller the loss coefficient, the closer the network's prediction results to the actual evaluation results. By continuously monitoring the loss coefficient, the training progress and performance of the network can be understood. If the loss coefficient gradually decreases and reaches a stable low value, it means that the network has learned the relationship between the holding adhesion test results and the evaluation well. Through such a training process, a model that can accurately evaluate the holding adhesion of the adhesive layer can be obtained, providing strong support for the quality monitoring of the adhesive label.

[0050] When the holding adhesion detection evaluation loss coefficient is less than the holding adhesion detection evaluation loss threshold, it means that after training the residual neural network, the difference between the network's evaluation results of holding adhesion and the actual expected evaluation results is small enough to be acceptable. At this time, the holding adhesion detection evaluation channel is generated, which can be understood as a path or mechanism specially used for accurate evaluation of holding adhesion. Through this channel, new holding adhesion test results can be quickly and accurately converted into holding adhesion detection evaluation coefficients. Once this evaluation channel is generated, subsequent holding adhesion evaluation of the adhesive label can directly use this channel without the need for complex neural network training process again. Only by inputting new holding adhesion test results into this channel, the corresponding evaluation coefficients can be quickly obtained and added to the adhesion performance detection results, providing important basis for comprehensive evaluation of the quality of the adhesive label.

[0051] Based on the obtained tackiness test results, a tackiness detection evaluation channel is utilized to determine a tackiness detection evaluation coefficient. The tackiness test results are data obtained through tackiness confidence tests on the adhesive layer, which reflect the adhesive's ability to maintain adhesion under certain conditions. The tackiness detection evaluation channel is generated when the tackiness detection evaluation loss coefficient is less than a threshold value, and it is a trained and optimized evaluation mechanism that can accurately give an evaluation coefficient based on the tackiness test results. By analyzing and processing the tackiness test results through the tackiness detection evaluation channel, the tackiness detection evaluation coefficient is obtained. This coefficient is a quantitative evaluation of the tackiness of the adhesive layer, reflecting the degree of tackiness of the adhesive layer within a certain standard or desired range. Finally, the tackiness detection evaluation coefficient is added to the adhesion performance detection results, which are a comprehensive consideration of the overall adhesion performance of the adhesive layer, including initial adhesion, tackiness, and adhesion migration. The addition of the tackiness detection evaluation coefficient makes the adhesion performance detection results more comprehensive and accurate, providing more information for subsequent quality evaluation of the adhesive label.

[0052] In one possible implementation, step S310 further includes: Step S317: Based on the multi-dimensional adhesion performance detection factor, generate adhesion migration detection instructions, and activate adhesion migration detection scene factors according to the adhesion migration detection instructions, wherein the adhesion migration detection scene factors include adhesion targets, adhesion duration, test environment, and applied pressure; build K adhesion migration detection scenes according to the adhesion migration detection scene factors, wherein K is a positive integer greater than 1; based on the adhesive layer, perform adhesion migration tests according to the K adhesion migration detection scenes to obtain K migration adhesive amounts; perform adhesion migration evaluation based on the K adhesion migration detection scenes and the K migration adhesive amounts to obtain K adhesion migration evaluation coefficients; perform central value calculation according to the K adhesion migration evaluation coefficients to generate an adhesion migration detection evaluation coefficient, and add the adhesion migration detection evaluation coefficient to the adhesion performance detection results.

[0053] Specifically, based on the multi-dimensional adhesion performance detection factors, adhesion migration detection instructions are generated, which include initial adhesion, adhesion retention, and adhesion migration. When the adhesion migration of the adhesive layer needs to be detected, specific detection instructions are generated according to these factors. The role of this detection instruction is to start the adhesion migration detection process. Once the detection instruction is generated, it will trigger the activation of the adhesion migration detection scene factors. Adhesion migration detection scene factors include adhesion target, adhesion duration, test environment, and applied pressure. These factors are crucial for accurately evaluating the adhesion migration performance of the adhesive layer. The adhesion target refers to the object to which the adhesive layer is attached, which can be objects of different materials and surface properties. Different adhesion targets will have different effects on adhesion migration. Adhesion duration determines the observation of adhesion migration within a certain time range. Longer adhesion duration will result in more adhesion migration phenomena. Test environment includes factors such as temperature, humidity, and light, which will affect the performance of the adhesive and in turn affect the degree of adhesion migration. The applied pressure simulates the pressure conditions that the adhesive layer is subjected to in actual use, and different applied pressures will also have different effects on adhesion migration. By activating these scene factors, specific detection conditions and parameters can be provided for subsequent adhesion migration tests, ensuring the accuracy and reliability of the test results.

[0054] Adhesion migration detection scene factors include adhesion target, adhesion duration, test environment, and applied pressure. Different combinations of these parameters can create various detection conditions to comprehensively evaluate the adhesion migration performance of the adhesive layer. Setting K as a positive integer greater than 1 means that multiple different detection scenes are to be set up. This is done to test the adhesion migration of the adhesive layer from multiple angles and under various conditions, avoiding the limitations that may arise from a single detection scene.

[0055] Before conducting the adhesion migration test, the initial weights of each sample for testing and the corresponding adhesion target are accurately measured and recorded as W1 (initial weight of the sample) and W2 (initial weight of the adhesion target), respectively. Then, the sample is subjected to adhesion testing according to specific adhesion migration detection scenarios (such as applying a certain pressure to a specific material adhesion target under a specific temperature and humidity environment and maintaining a specific adhesion duration). After the test, the weights of the sample and the adhesion target are measured again, recorded as W3 (weight of the sample after testing) and W4 (weight of the adhesion target after testing), respectively. At this time, the migrated adhesive amount is calculated by the following formula: migrated adhesive amount = (W4 - W2) - (W1 - W3). This method determines the amount of adhesive that migrates from the adhesive layer to the adhesion target by measuring the change in weight, which is relatively intuitive and accurate.

[0056] In the adhesion migration evaluation to obtain K adhesion migration evaluation coefficients, K adhesion migration detection scenarios and K migration adhesive amounts corresponding thereto are comprehensively considered. For each detection scenario, analysis is performed from multiple aspects. First, the migration adhesive amount is compared with a preset standard value. If the migration adhesive amount is low, it indicates that the adhesion migration degree of the adhesive in this scenario is small, and the adhesion stability is good, and accordingly a higher evaluation coefficient is given. Conversely, if the migration adhesive amount is high, it indicates that the adhesion migration is more significant, and the adhesion stability of the adhesive has a problem, and the evaluation coefficient is lowered. At the same time, the change of the migration adhesive amount is focused on. If the migration adhesive amount changes more smoothly at different times or stages in the same detection scenario, it indicates that the adhesion migration performance of the adhesive has good consistency and reliability, which will promote the evaluation coefficient to increase. If the migration adhesive amount fluctuates greatly, it means that the performance of the adhesive in this scenario is not stable and is easily affected by external factors, and the evaluation coefficient will be lowered accordingly. In addition, other factors in the detection scenario cannot be ignored, such as the material and surface characteristics of the adhesion target, the adhesion time, the test environment (including temperature, humidity, pressure, etc.). Different adhesion targets have different effects on the adhesion migration of the adhesive, and the length of the adhesion time also affects the migration adhesive amount, and the change of the test environment can also speed up or slow down the migration process of the adhesive. By comprehensively analyzing these factors, an accurate adhesion migration evaluation coefficient is determined for each adhesion migration detection scenario, and finally K evaluation coefficients are obtained, which will provide an important basis for subsequent product quality evaluation and process improvement, etc.

[0057] According to the K adhesion migration evaluation coefficients, a central value is calculated to generate an adhesion migration detection evaluation coefficient. The central value calculation adopts the average value method, and the purpose is to extract an index from multiple evaluation coefficients that can represent the overall adhesion migration performance. Then, the adhesion migration detection evaluation coefficient is added to the adhesion performance detection result, which is a comprehensive consideration of the performance of the adhesive layer, including initial adhesion, adhesion retention and adhesion migration. After adding the adhesion migration detection evaluation coefficient, the adhesion performance detection result is more comprehensive and accurate, and provides more information for subsequent quality evaluation of the adhesive label.

[0058] In one possible implementation, step S400 further includes: Step S410: performing a release property detection according to the base paper layer to obtain a base paper release property coefficient.

[0059] Step S420: performing a strength property detection according to the base paper layer to obtain a base paper strength property coefficient.

[0060] Step S430: performing a normalization processing according to the base paper release property coefficient and the base paper strength property coefficient to obtain a base paper double-side property detection result.

[0061] Step S440: Based on the double-side characteristic detection weight of the base paper, the double-side characteristic detection result of the base paper is weighted and calculated to generate the label base paper quality coefficient, wherein the double-side characteristic detection weight of the base paper includes a predetermined weight of the anti-sticking characteristic and a predetermined weight of the strength characteristic.

[0062] Specifically, the anti-sticking characteristic detection of the base paper layer, the anti-sticking characteristic is directly related to the ease of use and reliability of the adhesive label in actual use. In order to accurately obtain the base paper anti-sticking characteristic coefficient, a series of detection methods and means are adopted. First, by simulating the actual label peeling process, the adhesive label is attached to the base paper layer, and then the professional peeling equipment is used to peel the label from the base paper at a certain speed and angle. In this process, the size of the force required to peel the label is measured. If the peeling force is small, it means that the anti-sticking effect between the base paper layer and the adhesive is good, and the base paper anti-sticking characteristic coefficient will be correspondingly higher; on the contrary, if the peeling force is large, it means that the anti-sticking performance of the base paper is poor, and the base paper anti-sticking characteristic coefficient will be lower. In addition, the surface condition of the base paper after the label is peeled off is observed. If the surface of the base paper is clean and there is no adhesive residue, it also indicates that the anti-sticking characteristic of the base paper is good; if there is a lot of adhesive residue on the surface of the base paper, it means that the anti-sticking performance of the base paper has problems. At the same time, multiple repeated tests are carried out to obtain more accurate and reliable data. Through statistical analysis of multiple test results, the average value and standard deviation are calculated to more comprehensively evaluate the anti-sticking characteristic of the base paper layer. Finally, according to the results of these tests and analysis, the base paper anti-sticking characteristic coefficient is determined, which can quantitatively reflect the anti-sticking performance of the base paper layer, and provide an important reference for subsequent base paper quality evaluation and product application.

[0063] The strength properties of the base paper have an important influence on the stability and reliability of the adhesive label during production, storage, transportation and use. In order to obtain the base paper strength property coefficient, the strength of the base paper is detected from multiple aspects. Among them, the tensile strength is an important indicator. A professional material tensile testing machine is used to fix the base paper sample on the testing machine, and then stretch the base paper in a certain direction at a certain stretching speed until the base paper breaks. During this process, the testing machine records the maximum tension that the base paper can withstand during stretching. According to this maximum tension and the cross-sectional area of the base paper and other parameters, the tensile strength value of the base paper is obtained. The higher the tensile strength value, the stronger the base paper's resistance to tensile deformation, and the higher the base paper strength property coefficient. In addition to tensile strength, the tear strength of the base paper cannot be ignored. The tear strength reflects the base paper's resistance to tearing when subjected to local external forces. Through the Elmendorf tear method, the base paper is tested. In the test, the base paper sample is cut open in a small opening according to the specified method, and then the base paper is torn at a certain speed to measure the force required to tear the base paper. The greater the tearing force, the higher the tear strength of the base paper, and the base paper strength property coefficient will also be correspondingly improved. In addition, the base paper is subjected to bending strength test. By placing the base paper sample on the bending test device and applying a certain bending moment, the deformation and damage of the base paper during bending are observed. The base paper with high bending strength is not easy to break or deform when subjected to bending force, which also reflects its good strength properties. When testing the strength properties, multiple repeated tests are also required to reduce test errors and improve the accuracy and reliability of the test results. The test results of tensile strength, tear strength, bending strength and other aspects are considered comprehensively, and through certain data analysis and processing methods, the base paper strength property coefficient is finally determined. This coefficient can comprehensively and accurately reflect the strength properties of the base paper layer, and provide an important basis for evaluating the overall quality of the adhesive label.

[0064] According to the base paper anti-adhesion property coefficient and the base paper strength property coefficient, normalization processing is performed. Through normalization processing, the two coefficients are in the same numerical range, between 0 and 1. After normalization processing, the base paper double-sided property detection result comprehensively reflects the performance of the base paper in terms of anti-adhesion properties and strength properties.

[0065] The double-side property detection weight of the base paper includes a predetermined weight of the anti-adhesion property and a predetermined weight of the strength property, which are preset according to actual needs and importance levels, and are used to adjust the contribution proportion of the anti-adhesion property and the strength property in the comprehensive evaluation. For example, if the anti-adhesion property of the base paper is more important in a specific application scenario, a higher weight is given to the anti-adhesion property; on the contrary, if the strength property is more critical, the weight of the strength property is increased. By weighting the base paper double-side property detection result with the corresponding weight, the label base paper quality coefficient is obtained, which comprehensively considers the anti-adhesion property and the strength property of the base paper, can more accurately reflect the quality level of the base paper, and provides an important basis for the overall quality evaluation of the adhesive label.

[0066] In one possible implementation, step S410 further includes: Step S411: obtaining a plurality of base paper samples, wherein the plurality of base paper samples have the same production parameters as the base paper layer.

[0067] Step S412: performing a peel force test on each of the plurality of base paper samples according to a peel force tester to obtain a plurality of test peel forces.

[0068] Step S413: performing a central value calculation according to the plurality of test peel forces to obtain a peel force test result.

[0069] Step S414: activating a pre-constructed anti-adhesion property detection model.

[0070] Step S415: inputting the peel force test result into the anti-adhesion property detection model to output the base paper anti-adhesion property coefficient.

[0071] Specifically, a plurality of base paper samples are obtained, which have the same production parameters as the base paper layer actually used, so as to ensure that the test result can accurately reflect the real performance of the base paper layer, because the production parameters are the same, the interference of other factors is minimized.

[0072] The plurality of base paper samples are respectively tested for peel force using a peel force tester. The peel force tester can accurately measure the force required to peel the label from the base paper. During the test, the standard test method and operation process are strictly followed to ensure that the test conditions of each base paper sample are consistent. Through such a test, a plurality of test peel forces are obtained.

[0073] The central value calculation is performed according to the plurality of test peel forces. The central value calculation adopts the average value method, and by calculating the central value, the influence of the deviation of individual test data on the result is reduced, so as to obtain a peel force test result which can better represent the overall situation.

[0074] The pre-constructed anti-sticking property detection model is activated. The model is trained and optimized by a large amount of data, and can accurately evaluate the anti-sticking property of the base paper according to the input peeling force data.

[0075] The peeling force test result obtained in the foregoing is input into the anti-sticking property detection model. The model outputs a base paper anti-sticking property coefficient through internal calculation and analysis. The coefficient quantitatively reflects the anti-sticking property of the base paper, so that the performance of the base paper is better understood, and strong support is provided for overall quality control of the adhesive label.

[0076] In the second embodiment, based on the same inventive concept as the image analysis-based adhesive label production quality monitoring method in the foregoing embodiments, as shown in Figure 2 The present application provides an image analysis-based adhesive label production quality monitoring system. The system and method embodiments in the present application are based on the same inventive concept. The system comprises: An adhesive label acquisition module 10 is configured to acquire an adhesive label, wherein the adhesive label comprises a facestock layer, an adhesive layer, and a base paper layer.

[0077] A label facestock quality coefficient determination module 20 is configured to perform image analysis based on the facestock layer monitoring image of the facestock layer, and determine a label facestock quality coefficient.

[0078] A label adhesive quality coefficient acquisition module 30 is configured to perform comprehensive adhesion performance detection on the adhesive layer based on a plurality of adhesion performance detection factors, and obtain a label adhesive quality coefficient, wherein the plurality of adhesion performance detection factors include initial adhesion, holding adhesion, and adhesion migration.

[0079] A label base paper quality coefficient acquisition module 40 is configured to perform double-side property detection integration on the base paper layer based on a double-side property detection weight, and obtain a label base paper quality coefficient.

[0080] A label quality expectation condition setting module 50 is configured to set label quality expectation conditions based on the adhesive label, wherein the label quality expectation conditions include an expected facestock quality coefficient, an expected label adhesive quality coefficient, and an expected label base paper quality coefficient.

[0081] An adhesive label quality detection result output module 60 is configured to perform expected compliance judgment on the label facestock quality coefficient, the label adhesive quality coefficient, and the label base paper quality coefficient based on the label quality expectation conditions, and output an adhesive label quality detection result.

[0082] Further, the label adhesive quality coefficient acquisition module 30 further comprises: An adhesion performance detection result acquisition unit, configured to perform adhesion performance detection on the adhesive layer according to the multi-dimensional adhesion performance detection factor, and obtain an adhesion performance detection result.

[0083] A first weight condition determination unit, configured to determine an adhesion performance detection first weight condition according to initialization weight setting of the multi-dimensional adhesion performance detection factor.

[0084] A second weight condition determination unit, configured to obtain an adhesion performance detection second weight condition according to proportion calculation of the adhesion performance detection result.

[0085] A third weight condition determination unit, configured to obtain an adhesion performance detection third weight condition based on each-dimensional weight central value calculation of the adhesion performance detection first weight condition and the adhesion performance detection second weight condition.

[0086] A weighted calculation unit, configured to perform weighted calculation on the adhesion performance detection result according to the adhesion performance detection third weight condition, and output the label adhesive quality coefficient.

[0087] Further, the adhesion performance detection result acquisition unit further comprises: A primer adhesion tester activation unit, configured to generate a primer adhesion detection instruction based on the multi-dimensional adhesion performance detection factor, and activate a primer adhesion tester based on the primer adhesion detection instruction.

[0088] A primer adhesion detection data acquisition unit, configured to perform multiple primer adhesion detections according to the primer adhesion tester based on the adhesive layer, and obtain multiple primer adhesion detection data.

[0089] A primer adhesion detection result acquisition unit, configured to obtain a primer adhesion detection result according to central value calculation of the multiple primer adhesion detection data.

[0090] A primer adhesion detection evaluation model building unit, configured to build a primer adhesion detection evaluation model according to deep learning of a primer adhesion detection result record set and a primer adhesion detection evaluation record set.

[0091] An initial tack detection evaluation coefficient output unit is configured to output an initial tack detection evaluation coefficient based on the initial tack detection result and the initial tack detection evaluation model, and add the initial tack detection evaluation coefficient to the adhesion performance detection result.

[0092] Furthermore, the adhesion performance test result acquisition unit further includes: The adhesion tester activation unit generates an adhesion test instruction based on the multi-dimensional adhesion performance detection factor, and activates the adhesion tester based on the adhesion test instruction.

[0093] The tack test result obtaining unit performs a tack confidence test on the adhesive layer using the tack tester to obtain a tack test result.

[0094] A stickiness detection and evaluation loss coefficient acquisition unit is used to train the residual neural network with a stickiness test result record set as input information and a stickiness detection and evaluation record set as output information, and obtain the stickiness detection and evaluation loss coefficient after each predetermined number of trainings.

[0095] The stickiness detection and evaluation channel generation unit is configured to generate a stickiness detection and evaluation channel when the stickiness detection and evaluation loss coefficient is less than a stickiness detection and evaluation loss threshold.

[0096] A stickiness detection evaluation coefficient acquisition unit is configured to obtain a stickiness detection evaluation coefficient based on the stickiness test result and the stickiness detection evaluation channel, and add the stickiness detection evaluation coefficient to the adhesion performance test result.

[0097] Furthermore, the adhesion performance test result acquisition unit further includes: An adhesion mobility detection scenario factor activation unit generates an adhesion mobility detection instruction based on the multidimensional adhesion performance detection factor, and activates the adhesion mobility detection scenario factor according to the adhesion mobility detection instruction, wherein the adhesion mobility detection scenario factor includes an adhesion target, adhesion duration, test environment and applied pressure.

[0098] An adhesion mobility detection scenario building unit is used to build K adhesion mobility detection scenarios according to the adhesion mobility detection scenario factors, wherein K is a positive integer greater than 1.

[0099] A migration adhesive quantity acquisition unit is configured to perform an adhesion migration test based on the adhesive layer and according to the K adhesion migration detection scenarios to obtain K migration adhesive quantities.

[0100] An adhesion migration evaluation coefficient acquisition unit is configured to perform adhesion migration evaluation based on the K adhesion migration detection scenarios and the K migration adhesive amounts to obtain K adhesion migration evaluation coefficients.

[0101] An adhesion migration detection evaluation coefficient generation unit is used to perform centralized value calculation based on the K adhesion migration evaluation coefficients to generate an adhesion migration detection evaluation coefficient, and add the adhesion migration detection evaluation coefficient to the adhesion performance test result.

[0102] Furthermore, the label backing paper quality coefficient acquisition module 40 further includes: The base paper anti-sticking characteristic coefficient obtaining unit is used to perform anti-sticking characteristic detection on the base paper layer to obtain the base paper anti-sticking characteristic coefficient.

[0103] A base paper strength characteristic coefficient acquisition unit is used to perform strength characteristic detection on the base paper layer to obtain a base paper strength characteristic coefficient.

[0104] The base paper double-sided characteristic detection result acquisition unit is used to perform normalization processing according to the base paper anti-sticking characteristic coefficient and the base paper strength characteristic coefficient to obtain the base paper double-sided characteristic detection result.

[0105] A label base paper quality coefficient generating unit is configured to generate the label base paper quality coefficient by performing weighted calculation on the base paper double-sided characteristic detection result based on the base paper double-sided characteristic detection weight, wherein the base paper double-sided characteristic detection weight includes a predetermined weight for the anti-adhesive characteristic and a predetermined weight for the strength characteristic.

[0106] Furthermore, the base paper anti-sticking characteristic coefficient obtaining unit further includes: A base paper sample obtaining unit is used to obtain a plurality of base paper samples, wherein the production parameters of the plurality of base paper samples are the same as those of the base paper layer.

[0107] The test peeling force obtaining unit is used to perform peeling force tests on the plurality of base paper samples respectively using a peeling force testing machine to obtain a plurality of test peeling forces.

[0108] A peeling force test result acquisition unit is configured to perform a central value calculation according to the plurality of test peeling forces, and obtain a peeling force test result.

[0109] A detection model activation unit is configured to activate a pre-constructed anti-sticking property detection model.

[0110] A base paper anti-sticking property coefficient output unit is configured to input the peeling force test result into the anti-sticking property detection model, and output the base paper anti-sticking property coefficient.

[0111] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. Moreover, the above-mentioned embodiments of the present application are described. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous.

[0112] The above-mentioned is only the preferred embodiment of the present application, and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0113] The specification and drawings are merely exemplary of the present application, and any and all modifications, variations, combinations or equivalents that fall within the scope of the present application should be considered. Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is intended to include these modifications and variations.

Claims

1. A method for monitoring the quality of production of adhesive labels based on image analysis, characterized by, The method comprises: acquiring an adhesive label, wherein the adhesive label comprises a face material layer, an adhesive layer and a backing paper layer; performing image analysis according to a face material layer monitoring image of the face material layer to determine a label face material quality coefficient; performing comprehensive adhesion performance detection on the adhesive layer according to multi-dimensional adhesion performance detection factors to obtain a label adhesive quality coefficient, wherein the multi-dimensional adhesion performance detection factors comprise initial adhesion, holding adhesion and adhesion migration; performing bilateral characteristic detection integration on the backing paper layer according to a backing paper bilateral characteristic detection weight to obtain a label backing paper quality coefficient; setting label quality expectation conditions based on the adhesive label, wherein the label quality expectation conditions comprise an expected face material quality coefficient, an expected label adhesive quality coefficient and an expected label backing paper quality coefficient; performing expected compliance judgment on the label face material quality coefficient, the label adhesive quality coefficient and the label backing paper quality coefficient based on the label quality expectation conditions, and outputting an adhesive label quality detection result.

2. The method of claim 1, wherein, performing comprehensive adhesion performance detection on the adhesive layer according to multi-dimensional adhesion performance detection factors to obtain a label adhesive quality coefficient, comprising: performing adhesion performance detection on the adhesive layer according to the multi-dimensional adhesion performance detection factors to obtain an adhesion performance detection result; performing initialization weight setting according to the multi-dimensional adhesion performance detection factors to determine an adhesion performance detection first weight condition; performing proportion calculation according to the adhesion performance detection result to obtain an adhesion performance detection second weight condition; performing each-dimensional weight central value calculation based on the adhesion performance detection first weight condition and the adhesion performance detection second weight condition to obtain an adhesion performance detection third weight condition; performing weighted calculation on the adhesion performance detection result according to the adhesion performance detection third weight condition to output the label adhesive quality coefficient.

3. The method of claim 2, wherein, performing adhesion performance detection on the adhesive layer according to the multi-dimensional adhesion performance detection factors to obtain an adhesion performance detection result, comprising: generating an initial adhesion detection instruction based on the multi-dimensional adhesion performance detection factors, and activating an initial adhesion tester based on the initial adhesion detection instruction; performing multiple initial adhesion detections according to the initial adhesion tester based on the adhesive layer to obtain multiple initial adhesion detection data; performing central value calculation according to the multiple initial adhesion detection data to obtain an initial adhesion detection result; performing deep learning according to an initial adhesion detection result record set and an initial adhesion detection evaluation record set to build an initial adhesion detection evaluation model; outputting an initial adhesion detection evaluation coefficient based on the initial adhesion detection result according to the initial adhesion detection evaluation model, and adding the initial adhesion detection evaluation coefficient to the adhesion performance detection result.

4. The method of claim 2, wherein, performing adhesion performance detection on the adhesive layer according to the multi-dimensional adhesion performance detection factors to obtain an adhesion performance detection result, comprising: generating a holding adhesion detection instruction based on the multi-dimensional adhesion performance detection factors, and activating a holding adhesion tester based on the holding adhesion detection instruction; performing holding adhesion confidence testing according to the holding adhesion tester based on the adhesive layer to obtain a holding adhesion test result; The residual neural network is trained by taking the set of adhesion test results as input information and taking the set of adhesion detection evaluation records as output information. When the adhesion detection evaluation loss coefficient is less than the adhesion detection evaluation loss threshold, an adhesion detection evaluation channel is generated. Based on the adhesion test results, the adhesion detection evaluation coefficient is obtained according to the adhesion detection evaluation channel, and the adhesion detection evaluation coefficient is added to the adhesion performance detection result. According to the multi-dimensional adhesion performance detection factor, the adhesion performance of the adhesive layer is detected to obtain an adhesion performance detection result, including:

5. The method of claim 2, wherein, Based on the multi-dimensional adhesion performance detection factor, an adhesion migration detection instruction is generated, and an adhesion migration detection scene factor is activated according to the adhesion migration detection instruction, wherein the adhesion migration detection scene factor includes an adhesion target, an adhesion duration, a test environment, and an applied pressure. According to the adhesion migration detection scene factor, K adhesion migration detection scenes are built, wherein K is a positive integer greater than 1. Based on the adhesive layer, adhesion migration tests are performed according to the K adhesion migration detection scenes to obtain K migration adhesive amounts. Based on the K adhesion migration detection scenes and the K migration adhesive amounts, adhesion migration evaluation is performed to obtain K adhesion migration evaluation coefficients. According to the K adhesion migration evaluation coefficients, a central value is calculated to generate an adhesion migration detection evaluation coefficient, and the adhesion migration detection evaluation coefficient is added to the adhesion performance detection result. According to the double-side characteristic detection weight of the base paper, the double-side characteristic detection of the base paper layer is integrated to obtain a label base paper quality coefficient, including:

6. The method of claim 1, wherein, According to the base paper layer, an anti-adhesion characteristic detection is performed to obtain a base paper anti-adhesion characteristic coefficient. According to the base paper layer, a strength characteristic detection is performed to obtain a base paper strength characteristic coefficient. According to the base paper anti-adhesion characteristic coefficient and the base paper strength characteristic coefficient, a normalization processing is performed to obtain a base paper double-side characteristic detection result. Based on the base paper double-side characteristic detection weight, a weighted calculation is performed on the base paper double-side characteristic detection result to generate the label base paper quality coefficient, wherein the base paper double-side characteristic detection weight includes an anti-adhesion characteristic predetermined weight and a strength characteristic predetermined weight. According to the base paper layer, an anti-adhesion characteristic detection is performed to obtain a base paper anti-adhesion characteristic coefficient, including:

7. The method of claim 6, wherein, A plurality of base paper samples are obtained, wherein the plurality of base paper samples have the same production parameters as the base paper layer. According to a peel force testing machine, peel force tests are respectively performed on the plurality of base paper samples to obtain a plurality of test peel forces. According to the plurality of test peel forces, a central value is calculated to obtain a peel force test result. A pre-constructed anti-adhesion characteristic detection model is activated. The peel force test result is input into the anti-adhesion characteristic detection model, and the base paper anti-adhesion characteristic coefficient is output. The system includes:

8. An image analysis-based monitoring system for the quality of production of adhesive labels, characterized by the fact that, A self-adhesive label acquisition module is configured to acquire a self-adhesive label, wherein the self-adhesive label includes a face material layer, an adhesive layer, and a base paper layer. ​ The label face material quality coefficient determination module is configured to perform image analysis according to the face material layer monitoring image of the face material layer, and determine a label face material quality coefficient. The label adhesive quality coefficient acquisition module is configured to perform comprehensive adhesion performance detection on the adhesive layer according to multi-dimensional adhesion performance detection factors, and obtain a label adhesive quality coefficient, wherein the multi-dimensional adhesion performance detection factors include initial adhesion, holding adhesion, and adhesion migration. The label base paper quality coefficient acquisition module is configured to perform double-edge characteristic detection integration on the base paper layer according to a base paper double-edge characteristic detection weight, and obtain a label base paper quality coefficient. The label quality expectation condition setting module is configured to set a label quality expectation condition based on the adhesive label, wherein the label quality expectation condition includes an expected face material quality coefficient, an expected label adhesive quality coefficient, and an expected label base paper quality coefficient. The adhesive label quality detection result output module is configured to perform expected compliance judgment on the label face material quality coefficient, the label adhesive quality coefficient, and the label base paper quality coefficient based on the label quality expectation condition, and output an adhesive label quality detection result.

Images