An anti-counterfeiting identification and monitoring system and method based on machine vision shrink film morphology

Abstract

The present invention relates to the field of machine vision and image processing technology, and in particular to an anti-counterfeiting identification and monitoring system and method based on machine vision shrink film morphology. It includes the following parts: a reference feature acquisition module: obtaining three morphological features of the shrink film after the product is packaged, namely, a watermark pattern, a closing seam, and a hole; taking photos and recording based on the three morphological features, and using the three morphological features as a standard reference for the reference image; a transportation damage simulation module: obtaining the product transportation route from the production line to the hands of consumers, and recording the key nodes in the product transportation route. The present invention improves the accuracy and reliability of anti-counterfeiting verification through advanced information technology and image recognition technology; dynamically monitors supply chain environmental parameters to reduce the misjudgment rate; image recognition captures subtle changes in packaging to improve accuracy; and full-process data monitoring enhances transparency and traceability, improves consumer trust, and ensures product quality and safety.

Description

Technical Field

[0001] The present invention relates to the technical field of machine vision and image processing, and in particular to an anti-counterfeiting identification and monitoring system and method based on machine vision shrink film morphology. Background Art

[0002] With the development of global trade and increasing consumer awareness of product safety, ensuring transparency and authenticity throughout the entire product process, from production to delivery, has become crucial. In the fast-moving consumer goods (FMCG) sector, traditional anti-counterfeiting methods are struggling due to logistical complexity and the environmental susceptibility of packaging materials. Physical damage during transportation and regional variations in light intensity can cause packaging to vary in appearance, making traditional anti-counterfeiting methods based on static templates prone to misjudgment or missed detection. When shrink film is applied to products, the placement of watermarks, the shape of the seams, and the location and size of ventilation holes all exhibit unique randomness.

[0003] To address these challenges, a new monitoring mechanism that can dynamically adapt to changes in actual application scenarios is urgently needed. Leveraging advanced information technology, integrated sensors, and data analysis algorithms, real-time data collection and analysis of commodity flow and surrounding environmental parameters within the supply chain can be used to achieve more accurate and reliable anti-counterfeiting verification. Furthermore, advances in image recognition technology are providing new solutions for capturing subtle changes in structural features, facilitating the development of effective anti-counterfeiting tools. Summary of the Invention

[0004] In order to overcome the shortcomings of backward anti-counterfeiting measures and easy misjudgment, the present invention provides an anti-counterfeiting identification and monitoring system and method based on machine vision shrink film morphology.

[0005] The technical solution is: an anti-counterfeiting identification and monitoring system based on machine vision shrink film morphology, including:

[0006] The benchmark feature acquisition module captures the three morphological features of the shrink film after product packaging: the watermark pattern, the closing seam, and the holes; takes photos and records these three morphological features, and uses them as the standard reference for the benchmark image;

[0007] The transportation damage simulation module obtains the product's transportation route from the production line to the consumer's hands and records key nodes along the product's transportation route, including the transportation route status and the length of time the product has been exposed. Based on this information, the module uses the watermark pattern change formula to calculate a first watermark pattern change image. This first change image represents the change in the watermark pattern compared to its factory-issued state due to damage caused by handling, bumps, or exposure during transportation.

[0008] Environmental damage modeling module: Based on the first change image and the final geographic location of the product, local solar radiation intensity parameters are obtained and a ghosting superposition formula is used to calculate a second change image of the watermark pattern; the second change image is used for product anti-counterfeiting identification and verification;

[0009] Anti-counterfeiting verification decision module: based on the reference image, the first changed image, the second changed image, combined with the change in the shape of the seam and the change in the shape of the hole, obtains the anti-counterfeiting identification result.

[0010] Preferably, the reference feature acquisition module includes:

[0011] Obtain grayscale image matrix data of the original watermark pattern before printing and grayscale image matrix data of the watermark pattern after printing; calculate the reference ghosting coefficient of the watermark pattern using the reference ghosting coefficient formula.

[0012] Preferably, the use of a reference ghosting coefficient formula to calculate the reference ghosting coefficient of the watermark pattern includes: the reference ghosting coefficient formula is as follows, in, is the reference ghosting coefficient, is the grayscale image matrix of the original watermark pattern; is the ghost convolution kernel; is the convolution operation; is the total number of image pixels; For pixels.

[0013] Preferably, the transport damage simulation module includes:

[0014] Based on the reference ghosting coefficient, the transport route condition is obtained, wherein the transport route condition specifically includes road smoothness, driver's driving skill, and transport route accident type;

[0015] At the same time, obtain the duration of time the product is exposed to the outside, specifically the duration of time the product watermark pattern is exposed to natural light;

[0016] Based on the reference ghosting coefficient, the transportation route conditions, and the length of time the product is exposed outdoors, a watermark pattern change formula is used to calculate a first change image of the watermark pattern.

[0017] Preferably, the calculation of the first changed image of the watermark pattern using a watermark pattern change formula based on the reference ghosting coefficient, the transportation route conditions, and the duration of the product being exposed outdoors includes: the watermark pattern change formula is as follows,

[0018] in, is the grayscale value of the watermark after transportation; is the original watermark grayscale value; is the material photosensitivity coefficient; The duration of nudity; is the solar radiation intensity; is the baseline ghosting coefficient; is the transport shock sensitivity coefficient; is the road smoothness; Score driving proficiency; is the accident impact equivalent; is the random noise term .

[0019] Preferably, the environmental damage modeling module includes:

[0020] Based on the sunlight exposure conditions in the consumer's area, ultraviolet radiation data and day-night temperature difference data are extracted; based on the ultraviolet radiation data and day-night temperature difference data, a ghost superposition formula is used to calculate a second change image of the watermark pattern.

[0021] Preferably, the method of calculating the second changed image of the watermark pattern using a ghost superposition formula includes: the ghost superposition formula is as follows: in, For the Total environmental damage value of a product; , The first UV dose and temperature difference data of sampled products, =1,2,……, ,in, ≪N1, is the number of sampled products in the batch, and N1 is the total amount of the batch; For the Position weight of each product; For the The local environmental correction amount for each product.

[0022] Preferably, the anti-counterfeiting verification decision module includes:

[0023] Using the watermark pattern features in the reference image, the first changed image, and the second changed image as a first comparison feature sequence;

[0024] Using the seam features in the reference image and the seam morphology change image as a second comparison feature sequence;

[0025] using the hole features in the reference image and the hole morphology change image as a third comparison feature sequence;

[0026] The images with similarity lower than the preset threshold in the three comparison feature sequences are extracted as the final comparison images for anti-counterfeiting verification on the consumer side.

[0027] Preferably, the step of extracting an image with a similarity lower than a preset threshold value from the three comparison feature sequences as a final comparison image for anti-counterfeiting verification at the consumer end includes:

[0028] The similarity threshold refers to a quantitative index of image difference calculated by a feature extraction algorithm. When the similarity is lower than a preset threshold, the image is determined to be a valid comparison image.

[0029] A method for anti-counterfeiting identification and monitoring of shrink film morphology based on machine vision, comprising:

[0030] S1: Use a high-resolution industrial camera to capture the original form of the shrink film watermark pattern, seams, and holes from multiple angles, and align the feature areas using an image registration algorithm. Extract the watermark grayscale matrix, construct a Gaussian difference convolution kernel, and calculate the baseline ghosting coefficient after performing the convolution operation to quantify the printing process error.

[0031] S2: Integrate GPS and IoT sensors to collect transportation parameters; establish a multi-factor damage model and dynamically generate a watermark degradation image after transportation based on photosensitivity and impact sensitivity;

[0032] S3: Access the meteorological database to obtain the target area's UV intensity and daily temperature difference; use the spatial interpolation algorithm to calculate the stacking position weight, execute the ghost overlay formula, and generate the environmental aging characteristic map;

[0033] S4: Deploy a multi-scale feature comparison engine to perform joint analysis of watermark patterns, seams, and holes; set dynamic thresholds to trigger anti-counterfeiting alarms, and support mobile image hash value comparison and verification.

[0034] Beneficial effects: The present invention significantly improves the accuracy and reliability of anti-counterfeiting verification by introducing advanced information technology and image recognition technology. First, the ability to dynamically monitor the flow of goods and their environmental parameters in the supply chain ensures that the anti-counterfeiting mechanism can adapt to changes in actual application scenarios and reduces the misjudgment rate caused by logistics complexity and environmental factors. Secondly, the advancement of image recognition technology makes it possible to capture subtle changes in the structural characteristics of product packaging, further improving the accuracy of anti-counterfeiting verification. By monitoring and analyzing data from all links in the supply chain throughout the process, not only is the misjudgment caused by physical damage or environmental factors reduced, but the transparency and traceability of the entire process from production to delivery of the product are also enhanced. Ultimately, this will enhance consumer trust, ensure product quality and safety, and provide brands with more reliable anti-counterfeiting protection. In addition, the real-time data feedback mechanism can help companies quickly respond to potential problems, optimize supply chain management, and improve operational efficiency. Overall, the present invention not only solves the limitations of traditional anti-counterfeiting methods, but also brings a safer, more transparent and efficient supply chain experience to companies and consumers. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a schematic structural diagram of the anti-counterfeiting identification and monitoring system based on machine vision shrink film morphology of the present invention;

[0036] Figure 2 The figure is a flow chart of the anti-counterfeiting identification and monitoring method based on machine vision shrink film morphology of the present invention. DETAILED DESCRIPTION

[0037] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0038] Example 1: An anti-counterfeiting identification and monitoring system based on machine vision shrink film morphology, such as Figure 1 Shown, including:

[0039] The benchmark feature acquisition module captures the three morphological features of the shrink film after product packaging: the watermark pattern, the closing seam, and the holes; takes photos and records these three morphological features, and uses them as the standard reference for the benchmark image;

[0040] The transportation damage simulation module obtains the product's transportation route from the production line to the consumer's hands and records key nodes along the product's transportation route, including the transportation route status and the length of time the product has been exposed. Based on this information, the module uses the watermark pattern change formula to calculate a first watermark pattern change image. This first change image represents the change in the watermark pattern compared to its factory-issued state due to damage caused by handling, bumps, or exposure during transportation.

[0041] Environmental damage modeling module: Based on the first change image and the final geographic location of the product, local solar radiation intensity parameters are obtained and a ghosting superposition formula is used to calculate a second change image of the watermark pattern; the second change image is used for product anti-counterfeiting identification and verification;

[0042] Anti-counterfeiting verification decision module: based on the reference image, the first changed image, the second changed image, combined with the change in the shape of the seam and the change in the shape of the hole, obtains the anti-counterfeiting identification result.

[0043] Benchmark feature acquisition module, including:

[0044] Obtain grayscale image matrix data of the original watermark pattern before printing and grayscale image matrix data of the watermark pattern after printing; calculate the reference ghosting coefficient of the watermark pattern using the reference ghosting coefficient formula.

[0045] Further explanation: A high-resolution industrial camera captures grayscale images of the pre-printing design and the post-printing shrink film under standardized lighting. These images are aligned using an image registration algorithm to ensure pixel-level spatial consistency. A baseline ghosting coefficient, B, normalizes process errors to a [0,1] scalar using the mean pixel absolute difference, providing an initial reference for subsequent transportation / environmental damage modeling. This eliminates inherent equipment error interference, allowing damage simulation to focus solely on external factors. When capturing the watermark pattern, seam, and hole features after shrink film packaging, attention is paid to their random nature. During shrinkage, the seam shape, watermark attachment position, and stretching and perforation location and size are all random. Using a high-resolution camera to capture these random variations, the baseline ghosting coefficient is calculated, incorporating the random variations of the watermark after printing to fully reflect the randomness of the shrink film's initial state.

[0046] The base ghosting coefficient of the watermark pattern is calculated using the base ghosting coefficient formula, including: the base ghosting coefficient formula is as follows, in, is the reference ghosting coefficient, is the grayscale image matrix of the original watermark pattern; is the ghost convolution kernel; is the convolution operation; is the total number of image pixels; For pixels.

[0047] Further explanation is that the formula for normalizing the grayscale image matrix of the original watermark pattern to [0,1] is, , the normalization formula is well known to those skilled in the art, so the normalization formula will not be described separately below; The normalized grayscale matrix of the original watermark pattern (range 0, 1), representing the pixel values ​​under ideal printing conditions. is the Gaussian difference convolution kernel (DoG kernel), defined as ,in is the Gaussian kernel, standard deviation < ; Simulate ink diffusion during printing (wide Gaussian kernel ) and edge blurring caused by mechanical vibration of the device (narrow Gaussian kernel ), which comprehensively characterizes material properties and process errors. Original watermark image Perform convolution operation to generate a theoretical "after process damage" watermark image, reflecting the inherent defects of the printing process. Used to normalize the sum of absolute differences to ensure the baseline ghosting coefficient is a dimensionless scalar (range 0,1). , quantifying the combined effects of material properties (such as ink penetration) and equipment errors (such as overprint offset) during the printing process, providing a benchmark for subsequent transportation / environmental damage modeling. Compressing the image difference into a single scalar makes it easier to link with dynamic damage models (such as photochemical fading and mechanical impact) to avoid multi-factor coupling interference. Parameters ( , ) can be adapted to different printing equipment through calibration experiments.

[0048] Transport damage simulation module, including:

[0049] Based on the reference ghosting coefficient, the transport route condition is obtained, wherein the transport route condition specifically includes road smoothness, driver's driving skill, and transport route accident type;

[0050] At the same time, obtain the duration of time the product is exposed to the outside, specifically the duration of time the product watermark pattern is exposed to natural light;

[0051] Based on the reference ghosting coefficient, the transportation route conditions, and the length of time the product is exposed outdoors, a watermark pattern change formula is used to calculate a first change image of the watermark pattern.

[0052] Further explanation is that the road roughness (R, vibration spectrum energy normalization), driving level (D, rapid acceleration and deceleration frequency score), accident type (A, emergency stop / collision event classification) are collected in real time through the on-board IoT sensor (three-axis accelerometer + GPS), and the data are standardized to the [0,1] range. Based on the spatiotemporal matching of the photosensor and the transportation trajectory, the light exposure period of the watermark area is located, and the effective radiation time T (hours) is accumulated. After the benchmark ghosting coefficient B separates the process error, R, D, and A quantify the mechanical impact, and T×S (solar radiation intensity) quantifies the photochemical damage, realizing multi-factor independent modeling; through the logic function ( Dynamically adjust the contribution of the ghosting coefficient B to match the nonlinear superposition characteristics of actual damage. The randomness of shrink film wrapping further complicates transportation damage. Different products have different initial forms, and under the same transportation conditions, watermarks exhibit varying mechanical impact and photochemical damage. When collecting parameters and calculating the changing image, the ghosting coefficient is adjusted using a logical function, taking into account the randomness of shrink film, ensuring more realistic results.

[0053] Based on the reference ghosting coefficient, the transportation route conditions, and the length of time the product is exposed to the outside, a watermark pattern change formula is used to calculate a first change image of the watermark pattern, including: the watermark pattern change formula is as follows, in, is the grayscale value of the watermark after transportation; is the original watermark grayscale value; is the material photosensitivity coefficient; The duration of nudity; is the solar radiation intensity; is the baseline ghosting coefficient; is the transport shock sensitivity coefficient; is the road smoothness; Score driving proficiency; is the accident impact equivalent; is the random noise term .

[0054] Further explanation is, : The original watermark pixel grayscale value is normalized to represent the ideal state. : Material photosensitivity (m² / kW·h), an experimental measure of light sensitivity. T and S are exposure time (hours) and solar radiation intensity (kW / m²), quantifying light exposure. B: Baseline ghosting coefficient, which is normalized to isolate inherent errors in the printing process. : Calibrate mechanical damage weight. 、 、 : Normalize the transportation parameters to correspond to road smoothness, driving level score, and accident impact equivalent. : Random noise, simulating environmental disturbances. Exponential term Describes light fading, logical function Correlate impact with process errors to simulate edge blur. 、 Calibration is adapted to different materials and transportation scenarios. Eliminate interference from printing defects and improve damage assessment accuracy. Consider a packaging film (λ=0.05, k=2.0) after transportation: parameters: T=8h, S=1.2kW / m², R=0.6, D=0.3, A=0.1, B=0.2. Calculation: Light damage retention rate 62%, mechanical damage contribution 0.18 (logistic function term). Result: The original grayscale of 0.8 decreased to 0.67, with a deviation >5% triggering an alarm.

[0055] Environmental damage modeling module, including:

[0056] Based on the sunlight exposure conditions in the consumer's area, ultraviolet radiation data and day-night temperature difference data are extracted; based on the ultraviolet radiation data and day-night temperature difference data, a ghost superposition formula is used to calculate a second change image of the watermark pattern.

[0057] Further explanation is that the annual average ultraviolet intensity (UVI) and daily average temperature difference (ΔT) of the target area are extracted through the meteorological database API, and the product stacking position is located by combining the IoT temperature and humidity sensor to obtain the local microenvironment correction value ( , The synergistic effect of ultraviolet rays (photooxidation) and temperature difference (thermal expansion and contraction) leads to nonlinear aging of watermark materials, and its cumulative damage needs to be quantified; the ghost superposition formula ( ) Fusion batch sampling data ( , ) and geographic location interpolation ( ), addressing regional climate differences and uneven local exposure. Predicting watermark fading / cracking patterns supports anti-counterfeiting verification across climate zones. The random morphology of shrink film causes different parts of the product to be affected differently by UV rays and temperature differences. Local variations caused by wrapping randomness are considered when extracting environmental data and calculating the change image. The random distribution of shrink film features is also taken into account when determining position weights, resulting in more accurate environmental damage modeling.

[0058] The second variation image of the watermark pattern is calculated using a ghost superposition formula, including: the ghost superposition formula is as follows, in, For the Total environmental damage value of a product; , The first UV dose and temperature difference data of sampled products, =1,2,……, ,in, ≪N1, is the number of sampled products in the batch, and N1 is the total amount of the batch; For the Position weight of each product; For the The local environmental correction amount for each product.

[0059] Further explanation is, , :The first in the batch The ultraviolet dose (UVI·h) and daily average temperature difference (℃) of each sampled product are analyzed, and the data collection cost is reduced by sparse sampling (M≪N). : The position weight of the nth product, and the position weight is normalized, determined by the transport stacking position (such as the outer layer exposure weight is 0.9, the inner layer is 0.3). , :Based on the local UV intensity and temperature difference correction calculated based on the final geographic location of the product, it can solve the regional climate differences. Batch sampling item (∑ ) Use the statistical laws of batch data to reduce the cost of full monitoring; local correction items ( ) Refine individual exposure differences through spatial interpolation to improve model accuracy. Environmental damage quantification: formula output As the environmental damage value, input into the watermark pattern change formula ( ), driving the photochemical aging term ( ) and the co-evolution of the mechanical damage term. Benchmark ghosting coefficient After separating the printing defects, Focus on quantifying the effects of the external environment and avoid cross-interference. For example, randomly sample M=50 products from a batch (N=1000) and measure the average = 12. The nth product is located in the outer layer ( =0.9), interpolation value =15, =10°C. Calculation: =12+0.9×15×10=147. Application: Input into the watermark change formula to drive the grayscale value of the product to reduce by an additional 10%, which is consistent with the baseline ghosting coefficient. =0.1, the total damage reaches 20%, exceeding the threshold and triggering an alarm.

[0060] Anti-counterfeiting verification decision module, including:

[0061] Using the watermark pattern features in the reference image, the first changed image, and the second changed image as a first comparison feature sequence;

[0062] Using the seam features in the reference image and the seam morphology change image as a second comparison feature sequence;

[0063] using the hole features in the reference image and the hole morphology change image as a third comparison feature sequence;

[0064] The images with similarity lower than the preset threshold in the three comparison feature sequences are extracted as the final comparison images for anti-counterfeiting verification on the consumer side.

[0065] Further explanation: A multi-dimensional anti-counterfeiting verification system is constructed by independently comparing three sets of features (multimodal): watermark pattern, seam, and hole. The variation patterns of each feature type are driven by different physical mechanisms (photochemical aging, mechanical deformation, and material cracking), making it difficult for counterfeiters to simultaneously reproduce the true degradation trajectory of all features. Joint analysis of multiple features significantly reduces the cost of counterfeiting, while single-feature imitations cannot pass verification. Dynamic threshold determination: Preset similarity thresholds (e.g., SSIM < 0.85) adapt to different environmental damage intensities, reducing false positives. Only images with significant differences are displayed to consumers, simplifying verification logic and supporting fast mobile comparison. For example, consider the anti-counterfeiting verification of a certain alcoholic beverage packaging: The watermark pattern sequence: The baseline watermark is clear, but faded due to light after transportation (similarity 0.72), while the counterfeit exhibits no fading (similarity 0.95); the seam sequence: The seam on the authentic product deforms due to vibration (similarity 0.68), while the counterfeit exhibits perfect morphology (similarity 0.98); the hole sequence: The authentic product exhibits cracked hole edges (similarity 0.65), while the counterfeit exhibits no cracks (similarity 0.99). Verification: Only the authentic product has three similarity groups below the threshold (0.85), triggering the "authentic" confirmation. When building an anti-counterfeiting verification system, the randomness of shrink film wrapping is crucial for distinguishing authenticity. Counterfeiters find it difficult to replicate random features and variations. When setting thresholds and extracting comparison images, emphasize random morphology and variations. Images that fail to reflect these variations are more likely to be counterfeit.

[0066] Extract the images with similarity below the preset threshold from the three comparison feature sequences as the final comparison images for consumer-side anti-counterfeiting verification, including:

[0067] The similarity threshold refers to a quantitative index of image difference calculated by a feature extraction algorithm. When the similarity is lower than a preset threshold, the image is determined to be a valid comparison image.

[0068] To further explain, the similarity threshold is based on feature extraction algorithms (such as SIFT and CNN) that calculate structural and texture differences in key image regions. This difference is often quantified using the SSIM (Structural Similarity Index) or the cosine distance of eigenvectors. The preset threshold is calibrated experimentally: authentic damaged and forged samples are collected, and the distribution of the two data sets is statistically analyzed to select the optimal dividing point (such as the optimal point on the ROC curve). When the similarity is below the threshold (e.g., SSIM < 0.8), the damage conforms to the expected physical degradation pattern and is considered valid. Above the threshold, the damage is considered abnormal (e.g., no damage or artificial forgery).

[0069] Example 2: Based on Example 1, Figure 2 As shown, a method for anti-counterfeiting identification and monitoring based on machine vision shrink film morphology includes:

[0070] S1: Use a high-resolution industrial camera to capture the original form of the shrink film watermark pattern, seams, and holes from multiple angles, and align the feature areas using an image registration algorithm. Extract the watermark grayscale matrix, construct a Gaussian difference convolution kernel, and calculate the baseline ghosting coefficient after performing the convolution operation to quantify the printing process error.

[0071] S2: Integrate GPS and IoT sensors to collect transportation parameters; establish a multi-factor damage model and dynamically generate a watermark degradation image after transportation based on photosensitivity and impact sensitivity;

[0072] S3: Access the meteorological database to obtain the target area's UV intensity and daily temperature difference; use the spatial interpolation algorithm to calculate the stacking position weight, execute the ghost overlay formula, and generate the environmental aging characteristic map;

[0073] S4: Deploy a multi-scale feature comparison engine to perform joint analysis of watermark patterns, seams, and holes; set dynamic thresholds to trigger anti-counterfeiting alarms, and support mobile image hash value comparison and verification.

[0074] The above is a detailed introduction to the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method and core idea of ​​the present application. At the same time, for those skilled in the art, based on the idea of ​​the present application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. An anti-counterfeiting identification and monitoring system based on machine vision shrink film morphology, characterized by: include: Reference feature acquisition module: This module acquires the three morphological features of the shrink film after product packaging: the watermark pattern, the seam, and the holes. Taking photos and recording based on the three morphological features, and using the three morphological features as standard references for a baseline image; The base ghosting coefficient of the watermark pattern is calculated using the base ghosting coefficient formula, including: the base ghosting coefficient formula is as follows, in, is the reference ghosting coefficient, is the grayscale image matrix of the original watermark pattern; is the ghost convolution kernel; is the convolution operation; is the total number of image pixels; is a pixel; Transportation damage simulation module: obtain the product transportation route from the production line to the hands of consumers, record the key nodes in the product transportation route, the key nodes include the transportation route status and the length of time the product is exposed to the outside; based on the transportation route status and the length of time the product is exposed to the outside, use the watermark pattern change formula to calculate the first change image of the watermark pattern; the first change image refers to the change in the watermark pattern compared with the factory state due to the loss caused by handling or exposure of the product during transportation; based on the reference ghosting coefficient, transportation route status, and the length of time the product is exposed to the outside, use the watermark pattern change formula to calculate the first change image of the watermark pattern, including: the watermark pattern change formula is as follows, in, is the grayscale value of the watermark after transportation; is the original watermark grayscale value; is the material photosensitivity coefficient; The duration of nudity; is the solar radiation intensity; is the baseline ghosting coefficient; is the transport shock sensitivity coefficient; is the road smoothness; Score driving proficiency; is the accident impact equivalent; is the random noise term ; Environmental damage modeling module: Based on the first change image, local solar radiation intensity parameters are obtained based on the final geographical location of the product, and a second change image of the watermark pattern is calculated using a ghost superposition formula; the second change image is used for product anti-counterfeiting identification and verification; the second change image of the watermark pattern is calculated using a ghost superposition formula, including: the ghost superposition formula is as follows, in, For the Total environmental damage value of a product; , The first UV dose and temperature difference data of sampled products, =1,2,……, ,in, , is the number of sampled products in the batch, N1 is the total batch size; For the Position weight of each product; For the Local environmental correction for each product; Anti-counterfeiting verification decision module: based on the reference image, the first changed image, the second changed image, combined with the change in the shape of the seam and the change in the shape of the hole, obtains the anti-counterfeiting identification result.

2. The anti-counterfeiting identification and monitoring system based on machine vision shrink film morphology according to claim 1 is characterized in that: The reference feature acquisition module includes: Obtain grayscale image matrix data of the original watermark pattern before printing and grayscale image matrix data of the watermark pattern after printing.

3. The anti-counterfeiting identification and monitoring system based on machine vision shrink film morphology according to claim 1 is characterized in that: The transport damage simulation module includes: Based on the reference ghosting coefficient, the transport route condition is obtained, wherein the transport route condition specifically includes road smoothness, driver's driving skill, and transport route accident type; At the same time, the length of time the product is exposed to the outside is obtained, and the exposure length is specifically the length of time the product watermark pattern is exposed to natural light.

4. The anti-counterfeiting identification and monitoring system based on machine vision shrink film morphology according to claim 1 is characterized in that: The environmental damage modeling module includes: Based on the sunlight exposure conditions in the consumer's area, ultraviolet radiation data and day and night temperature difference data are extracted; based on the ultraviolet radiation data and day and night temperature difference data.

5. The anti-counterfeiting identification and monitoring system based on machine vision shrink film morphology according to claim 1 is characterized in that: The anti-counterfeiting verification decision module includes: Using the watermark pattern features in the reference image, the first changed image, and the second changed image as a first comparison feature sequence; Using the seam features in the reference image and the seam morphology change image as a second comparison feature sequence; using the hole features in the reference image and the hole morphology change image as a third comparison feature sequence; The images with similarity lower than the preset threshold in the three comparison feature sequences are extracted as the final comparison images for anti-counterfeiting verification on the consumer side.

6. The anti-counterfeiting identification and monitoring system based on machine vision shrink film morphology according to claim 5 is characterized in that: The step of extracting an image whose similarity is lower than a preset threshold value from the three comparison feature sequences as the final comparison image for the consumer-side anti-counterfeiting verification includes: The similarity threshold refers to a quantitative index of image difference calculated by a feature extraction algorithm. When the similarity is lower than a preset threshold, the image is determined to be a valid comparison image.

7. A method for anti-counterfeiting identification and monitoring based on machine vision of shrink film morphology, used in the anti-counterfeiting identification and monitoring system based on machine vision of shrink film morphology according to any one of claims 1 to 6, characterized in that: include: S1: Use a high-resolution industrial camera to capture the original form of the shrink film watermark pattern, seam, and holes from multiple angles, and align the feature areas using an image registration algorithm; Extract the watermark grayscale matrix, construct the Gaussian difference convolution kernel, perform the convolution operation and calculate the reference ghosting coefficient to achieve the quantification of printing process error; S2: Integrate GPS and IoT sensors to collect transportation parameters; establish a multi-factor damage model and dynamically generate a watermark degradation image after transportation based on photosensitivity and impact sensitivity; S3: Access the meteorological database to obtain the target area's UV intensity and daily temperature difference; use the spatial interpolation algorithm to calculate the stacking position weight, execute the ghost overlay formula, and generate the environmental aging characteristic map; S4: Deploy a multi-scale feature comparison engine to perform joint analysis of watermark patterns, seams, and holes; set dynamic thresholds to trigger anti-counterfeiting alarms, and support mobile image hash value comparison and verification.

Images