A product whole-process tracing system and method for identifying a two-dimensional code

Abstract

The application discloses a product whole-process tracing system and method for identifying a two-dimensional code, belongs to the technical field of digital image processing and machine vision detection, and comprises a data acquisition module, an irregular template module, a background color processing module, a steganographic coding module and an anti-fake generation module.

Description

Technical Field

[0001] This invention relates to the fields of digital image processing and machine vision inspection technology, and in particular to a product traceability system and method with QR codes. Background Technology

[0002] Product anti-counterfeiting and end-to-end traceability are crucial for ensuring supply chain security. In the fields of data recognition and optical recording technology, machine-readable two-dimensional data matrices are widely used for product identification. In actual industrial production lines, traceability labels need to be stably assigned to the surfaces of various products and then optically read and decoded using vision devices.

[0003] Among related technologies, Chinese patent CN223770636U discloses a graded traceability cardboard face sheet and a graded traceability cardboard face sheet inkjet printing system, including a substrate pattern layer, a visible code non-contact printing layer, and a hidden code non-contact printing layer, wherein: the visible code non-contact printing layer is printed with a first micro dot code and a QR code; the hidden code non-contact printing layer is printed with a second micro dot code; and the substrate pattern layer is printed with a legal text area.

[0004] Regarding the aforementioned technologies, QR codes and hidden micro-dot codes are physically dispersed and printed in different printing layers or spatial areas, failing to form a deeply nested and integrated single optical recording carrier structure. This results in a low overall anti-counterfeiting integration of the label, making it easy to be segmented and counterfeited or partially damaged. At the same time, this encoding and reading method highly depends on the color matching of the paper substrate. When facing complex material carriers with highly reflective properties, optical reading devices cannot reliably extract the underlying encoding features, easily creating blind spots and causing a significant drop in data recognition rate. This makes it difficult to meet the high-precision and high-reliability end-to-end traceability requirements of industrial production lines. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a product end-to-end traceability system and method using QR codes. For complex material carriers such as bottle caps, a dynamic coding mechanism that integrates multimodal physical features with micro-matrix steganography is employed to achieve concealed information embedding even under strong light interference, thereby enhancing the product's anti-counterfeiting level and traceability reliability.

[0006] The above objectives can be achieved through the following approach: A product traceability system with QR codes includes a data acquisition module for acquiring original images and multispectral scanning data of a target bottle cap surface, extracting the closed boundary pixel set of the original image, and calculating the reflectivity value set of the target bottle cap surface based on the multispectral scanning data; an irregular QR code template module for extracting a polygonal outline from the closed boundary pixel set, mapping the polygonal outline to a standard QR code grid, and eliminating data nodes overlapping with the polygonal outline within the standard QR code grid to generate an irregularly shaped, white-spaced QR code template; and a dynamic background color processing module for extracting grayscale values ​​from the original image, performing point-by-point difference calculations with the reflectivity value set to filter out pixels, and outputting a dynamic background color. The system comprises: a color removal identifier image; a micro-matrix steganography module, used to acquire the product traceability base code through external input, convert the product traceability base code into a binary data stream, adjust the spatial distribution of the dynamic background color removal identifier image according to the binary data stream, and generate a micro-matrix frequency-modulated steganography image; a composite anti-counterfeiting generation module, used to scale the micro-matrix frequency-modulated steganography image to the boundary size of the polygonal outline image, embed the irregularly shaped blank QR code template, perform data redundancy rearrangement calculation, and generate a composite anti-counterfeiting inkjet printing data package; and an inkjet printing traceability execution module, used to transmit the composite anti-counterfeiting inkjet printing data package to the inkjet printing equipment to perform ink spraying operations, acquire real-shot images of the finished product surface through a camera at the end of the production line, extract the product traceability base code, and write it into the traceability database system.

[0007] Optionally, the data acquisition module includes: an image contour extraction unit, used to perform grayscale processing and edge detection on the original image, extract the edge pixel coordinate sequence, and perform connected component topology analysis on the edge pixel coordinate sequence to generate a closed boundary pixel set; and a light intensity feature analysis unit, used to analyze the multispectral scanning data to extract the reflectivity matrix, perform spatial grid division on the surface of the target bottle cap, calculate the average light wave distribution value of the reflectivity matrix in each grid, and summarize to construct a set of reflectivity values.

[0008] Optionally, the irregular QR code template module includes: a contour vector fitting unit, used to perform polygon approximation on the closed boundary pixel set, generate a vector polygon contour map, extract the geometric center point of the standard QR code grid, translate and align the vector polygon contour map to the geometric center point, and establish a grid mapping coordinate system; and a node collision elimination unit, used to perform ray cross-testing based on the grid mapping coordinate system, determine the positional relationship between the standard QR code grid and the vector polygon contour map, delete overlapping data nodes, and output an irregular blank QR code template.

[0009] Optionally, the dynamic background color processing module includes: a grayscale feature construction unit, used to perform color space conversion operations on the original image and extract grayscale values ​​of pixel coordinates, and construct a two-dimensional grayscale feature matrix based on the grayscale values; and a differential zeroing and filtering unit, used to perform point-by-point differential calculations on the same coordinates between the two-dimensional grayscale feature matrix and the set of reflectance intensity values, delete pixels in the negative value range, and output a dynamic background color removal marker image.

[0010] Optionally, the micro-dot matrix steganography module includes: a basic encoding conversion unit, used to perform character set feature mapping on the product traceability basic encoding to extract character encoding values, and perform bit serialization and error correction bit appending operations on the character encoding values ​​to generate a binary data stream; and a grid coloring modulation unit, used to segment the dynamic background color removal identifier image into a spatial pixel grid, traverse the bit state of the binary data stream, perform pixel filling and erasing operations on the spatial pixel grid according to the bit state to reconstruct the grid coloring rate, and output a micro-dot matrix frequency-modulated steganography image.

[0011] Optionally, the system further includes: extracting the geometric topological distribution features of the closed boundary pixel set, and performing an inner product hash operation on the geometric topological distribution features and the binary data stream to output a cross-modal security feature vector.

[0012] Optionally, the composite anti-counterfeiting generation module includes: a polar coordinate constraint scaling unit, used to extract the maximum polar coordinate radius value of the polygonal outline image, perform proportional scaling on the micro-matrix frequency-modulated stegogram with the maximum polar coordinate radius value as a constraint condition, and align and embed it into the irregularly shaped blank QR code template to generate an initial embedded data packet; and a cross-modal perturbation rearrangement unit, used to extract the cross-modal security feature vector as an address mapping index, perform a randomized rearrangement calculation of the redundant verification data block on the initial embedded data packet using the address mapping index, update the spatial arrangement coordinates of the redundant verification data block, and output the composite anti-counterfeiting inkjet data packet.

[0013] Optionally, the inkjet printing traceability execution module includes: a hardware driver conversion unit, used to parse the composite anti-counterfeiting inkjet data packet to extract the array layout coordinates, map the array layout coordinates to the ink droplet driving pulse sequence of the piezoelectric printhead, and instruct the inkjet printing device to perform ink spraying operation according to the ink droplet driving pulse sequence; and a morphological reverse analysis unit, used to perform morphological filtering and positioning cropping on the real-shot image of the finished product surface to extract the central imprint area, quantify the pixel distribution frequency of the central imprint area, and perform demodulation decoding operation on the pixel distribution frequency to recover the product traceability basic code and write it into the traceability database system.

[0014] Based on the same inventive concept, this invention also provides a product end-to-end traceability method using QR codes. The method includes: acquiring an original image and multispectral scanning data of a target bottle cap surface; extracting a set of closed boundary pixels from the original image; and calculating a set of reflective intensity values ​​for the target bottle cap surface based on the multispectral scanning data; extracting a polygonal contour map from the closed boundary pixel set; mapping the polygonal contour map to a standard QR code grid; removing data nodes overlapping with the polygonal contour map within the standard QR code grid to generate an irregularly shaped, blank-spaced QR code template; and extracting grayscale values ​​from the original image and performing point-by-point difference calculations with the reflective intensity value set. The process involves filtering out pixels to output a dynamic background color removal marker image; acquiring the product traceability base code through external input, converting the product traceability base code into a binary data stream, adjusting the spatial distribution of the dynamic background color removal marker image based on the binary data stream, and generating a micro-dot matrix frequency-modulated steganalysis image; scaling the micro-dot matrix frequency-modulated steganalysis image to the boundary size of the polygonal outline image and embedding it into the irregularly shaped blank QR code template, performing data redundancy rearrangement calculation, and generating a composite anti-counterfeiting inkjet printing data package; transmitting the composite anti-counterfeiting inkjet printing data package to the inkjet printing equipment to perform ink spraying operations, acquiring a real-shot image of the finished product surface through a camera at the end of the production line, extracting the product traceability base code, and writing it into the traceability database system.

[0015] Compared with the prior art, the present invention has the following advantages: 1. It avoids localized exposure caused by the reflection of metallic base, and does not require additional manual intervention or white base spraying process. Without affecting the original production rhythm, it improves the success rate of scanning QR codes on industrial production lines. 2. While maintaining a high fault tolerance rate and reading speed, it retains the data capacity of the QR code, giving the traceability code intuitive visual recognition and brand promotion value; 3. It not only provides anti-cloning barriers, but also ensures that even under extreme working conditions, the underlying traceability data can still be read through the micro-matrix identified by the reverse analysis center.

[0016] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

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

[0018] Figure 1 This is a framework diagram of a product traceability system with QR code identification according to an embodiment of the present invention.

[0019] Figure 2 This is a schematic diagram of the structure of a product full-process traceability system with QR code identification according to an embodiment of the present invention.

[0020] Figure 3 This is a dynamic background color difference screening two-dimensional feature density distribution map according to an embodiment of the present invention.

[0021] Figure 4 This is a comparison diagram of the mesh shading rate distribution before and after micro-matrix frequency modulation steganography in an embodiment of the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Reference Figure 1 One embodiment of the present invention proposes a product full-process traceability system with QR code identification. For complex material carriers such as bottle caps, a dynamic coding mechanism that integrates multimodal physical features and micro-dot matrix steganography is adopted to achieve the embedding of hidden information in strong light interference environment, thereby improving the anti-counterfeiting level and traceability reliability of the product.

[0024] like Figure 2 As shown, the system in this embodiment specifically includes: The data acquisition module is used to acquire original images and multispectral scanning data of the target bottle cap surface, extract the closed boundary pixel set of the original image, and calculate the set of reflective intensity values ​​of the target bottle cap surface based on the multispectral scanning data. Optionally, the data acquisition module includes: The image contour extraction unit is used to perform grayscale processing and edge detection on the original image, extract the edge pixel coordinate sequence, and perform connected component topology analysis on the edge pixel coordinate sequence to generate a closed boundary pixel set. The system receives the raw image captured by an industrial color camera and reads the red, green, and blue color channel values ​​at each pixel coordinate in the raw image. It then performs a linear weighted sum of the three color channel values ​​using specific coefficients, and performs grayscale processing. The physical meaning of this process is to remove redundant information in the color dimension, retaining only the brightness and photometric gradient features of the bottle cap surface. The grayscale processing calculation relationship satisfies the following formula: , Among them, letters Represents the grayscale value after converting the corresponding pixel coordinates; letters Represents the red color channel value; letters Represents the green color channel value; letters The values ​​represent the blue color channel values; the numbers 0.299, 0.587, and 0.114 are standard photometric weighting constants that match the visual sensitivity of the human eye. After grayscale processing, edge detection based on the Canny operator is performed on the generated grayscale image. This action locates the physical location of abrupt changes in image brightness by calculating the first spatial derivative of the grayscale values ​​of adjacent pixels, and outputs the coordinates of points greater than the judgment threshold to form an edge pixel coordinate sequence. The judgment threshold is not set by human experience, but is determined by statistical fitting based on 500 optical sampling experiments on the surface roughness of the same batch of bottle cap materials. The optimal signal-to-noise ratio critical value for filtering out metal brush texture noise in the experimental statistics is taken as the low threshold, and three times the low threshold is taken as the high threshold, thereby ensuring the complete detection of real physical edges. Subsequently, eight-connected topological analysis is performed on the extracted edge pixel coordinate sequence to track the interconnected edge points in two-dimensional space, remove broken or forked line segments, retain continuous closed loops with overlapping first and last coordinates, and generate a closed boundary pixel set.

[0025] For example, the image contour extraction unit acquires an original image with a resolution of 1920 pixels by 1080 pixels. It locates the pixel at coordinates 100th row and 150th column, and reads the red channel value of 200, green channel value of 150, and blue channel value of 50. The image contour extraction unit substitutes these values ​​into the grayscale processing formula, calculating an approximate grayscale value of 153 for this pixel. After completing grayscale conversion for all pixels, the image contour extraction unit performs edge detection, extracting a set high threshold of 180 and a low threshold of 60 based on experimental data. After gradient calculation and threshold filtering, the image contour extraction unit extracts an edge pixel coordinate sequence containing 8500 spatial coordinates. Further, the image contour extraction unit performs connected component topology analysis on this sequence, removing discontinuous scattered points and line segments, ultimately extracting a complete ring-shaped region composed of 1200 connected pixel coordinates, generating a closed boundary pixel set.

[0026] The light intensity feature analysis unit is used to analyze the multispectral scanning data to extract the reflectivity matrix, perform spatial grid division on the surface of the target bottle cap, calculate the average light wave distribution value of the reflectivity matrix in each grid, and summarize and construct a set of reflectivity values.

[0027] The intensity feature analysis unit analyzes the raw binary data stream from the multispectral sensor, extracts the light reflectance ratio of the bottle cap surface under different spatial coordinates and wavelengths, and constructs a three-dimensional tensor containing spatial plane coordinates and wavelength channels, i.e., the reflectance matrix. To reduce the computational load of high-dimensional matrix operations and match the working physical granularity of industrial printheads, the intensity feature analysis unit performs spatial meshing on the two-dimensional spatial plane corresponding to the target bottle cap surface. The method for determining the mesh side length is based on the equivalent mapping ratio between the minimum physical droplet diameter of the back-end inkjet printer and the spatial resolution of the multispectral sensor, ensuring that one mesh cell corresponds exactly to an independent and controllable working block of the inkjet printer in terms of physical size. For each independent mesh after division, the intensity feature analysis unit calculates the mathematical expectation of the reflectance values ​​of all sampling points within the mesh under all wavelength channels, thereby quantifying the macroscopic reflective characteristics of the local area and generating an average light wave distribution value. The calculation relationship of the average light wave distribution value satisfies the following formula: , Among them, letters Representing the Line 1 The average light wave distribution value of the grid corresponding to the column; letter Represents the total number of spatial sampling points contained within a single grid; letters Represents the total number of wavelength channels included in the multispectral scan data; letters This represents the spatial coordinates within the corresponding grid. and And the wavelength channel is The reflectance values ​​are obtained from specific physical sampling points. The light intensity feature analysis unit traverses all the divided grids, combines the calculated average light wave distribution values ​​into a one-dimensional array according to the spatial row and column index order, and summarizes them to construct a set of reflectance intensity values. For example, the light intensity feature analysis unit analyzes the multispectral scanning data and extracts a reflectance matrix with a spatial resolution of 640 by 640 sampling points and containing 8 wavelength channels. Based on the physical droplet size mapping ratio of the inkjet printer, the light intensity feature analysis unit determines that the grid side length is 32 sampling points, performs spatial grid division on the entire scanning surface, and divides it into 20 rows by 20 columns, totaling 400 independent grids. For the grid with coordinates of row 1 and column 1, the light intensity feature analysis unit counts that the grid contains 1024 spatial sampling points. The light intensity feature analysis unit extracts a total of 8192 reflectance values ​​for these 1024 sampling points on the 8 wavelength channels. The light intensity feature analysis unit sums these 8192 values ​​and divides the sum by the total number of samples, 8192, to calculate the average light wave distribution value of the grid, which is 0.82. The light intensity feature analysis unit repeats the above process, and after calculating the values ​​of all 400 grids, it summarizes them in row and column order to generate a numerical sequence containing 400 elements, thus constructing a set of reflected light intensity values.

[0028] The irregular QR code template module is used to extract a polygonal outline image through the closed boundary pixel point set, map the polygonal outline image to a standard QR code grid, and remove data nodes that overlap with the polygonal outline image within the standard QR code grid to generate an irregular blank QR code template. Optionally, the irregularly shaped QR code template module includes: The contour vector fitting unit is used to perform polygon approximation on the closed boundary pixel set, generate a vector polygon contour map, extract the geometric center point of the standard QR code grid, translate and align the vector polygon contour map to the geometric center point, and establish a grid mapping coordinate system. The contour vector fitting unit receives the set of closed boundary pixels extracted by the front-end module and performs polygon approximation on this set. This approximation process calculates the vertical spatial distance from each pixel coordinate on the contour curve to the line segment connecting the beginning and end. If this vertical spatial distance is greater than the approximation distance threshold, the pixel coordinate is retained as a new vertex of the polygon, and the original curve is divided into two segments using this pixel coordinate as the boundary. This vertical spatial distance calculation and filtering process is recursively executed. The value of this approximation distance threshold is not subjectively set, but is based on fitting analysis of 500 sets of bottle cap contour sample data under different lighting and vibration conditions on the production line, selecting the distance pixel value that maintains a contour area change rate of less than 2% and minimizes the number of vertices. After approximation processing, the contour vector fitting unit generates a vector polygon contour map composed of a series of discrete vertex coordinates connected end-to-end. Subsequently, the contour vector fitting unit reads the externally input standard QR code grid, extracts the two-dimensional matrix scale of the standard QR code grid, and calculates half the number of rows and columns of the two-dimensional matrix as the coordinates of the geometric center point. Simultaneously, the contour vector fitting unit calculates the arithmetic mean of the coordinates of all vertices in the vector polygon contour map, using this as the geometric center point of the vector polygon contour map. To align the contour from the physical image space to the QR code logical space, the contour vector fitting unit performs a translation alignment operation, the calculation of which satisfies the following formula: , Among them, letters Represents the vertex coordinates of the vector polygon outline after translation and alignment; letters Represents the original vertex coordinates of the vector polygon outline before translation and alignment; letters Represents the geometric center coordinates of a standard QR code grid; letters This represents the coordinates of the geometric center point of the vector polygon outline. By traversing all vertices and performing formula calculations, the outline vector fitting unit translates and aligns the entire vector polygon outline to the geometric center point of the standard QR code grid, thereby eliminating the spatial offset between the physical coordinate system and the logical coordinate system and establishing a grid-mapped coordinate system.

[0029] For example, the contour vector fitting unit obtains a closed boundary pixel set containing 1200 pixel coordinates. The approximation distance threshold determined based on production line sample data is 4 pixels. The contour vector fitting unit performs polygon approximation operations, removing redundant points on smooth curves and extracting 10 key turning vertices to generate a vector polygon contour map. The contour vector fitting unit calculates the arithmetic mean of the coordinates of these 10 key turning vertices, obtaining the geometric center coordinates of the vector polygon contour map as horizontal 500 and vertical 500. Simultaneously, the contour vector fitting unit extracts a standard QR code grid of version 7 used by the target, with a size of 41 rows by 41 columns. The contour vector fitting unit calculates the geometric center coordinates of this standard QR code grid as horizontal 21 and vertical 21. Substituting the above coordinates into the translation alignment formula, taking the original vertex coordinates of horizontal 510 and vertical 520 as an example, subtracting the original center point coordinates of horizontal 500 and vertical 500, and adding the grid center point coordinates of horizontal 21 and vertical 21, the calculated vertex coordinates after translation alignment are horizontal 31 and vertical 41. The contour vector fitting unit performs the same translation calculation on the remaining 9 vertices, successfully establishing a grid mapping coordinate system in which the physical contour and the data grid completely coincide.

[0030] The node collision elimination unit is used to perform ray cross-test based on the grid mapping coordinate system, determine the positional relationship between the standard QR code grid and the vector polygon outline, delete overlapping data nodes, and output an irregular blank QR code template.

[0031] The node collision elimination unit, based on the established grid-mapped coordinate system, performs a ray crossing test on each data node within the standard QR code grid. The physical significance of this test lies in using the ray casting method in computational geometry to determine whether a discrete data node falls within the physical contour of a closed polygon. The node collision elimination unit constructs an infinitely long unidirectional ray horizontally to the right, using the coordinates of the currently tested data node as the geometric starting point, and calculates the total number of physical intersections between this unidirectional ray and all boundary segments of the vector polygon contour. To accurately determine positional relationships, the node collision elimination unit performs a parity modulo operation on this total number of intersections. The calculation relationship of this judgment logic satisfies the following formula: , Among them, letters Represents the numerical value of the parity check result; letters Represents the total number of physical intersections between a horizontal unidirectional ray and the boundary line segment of a vector polygon contour; symbol This represents performing a modulo operation on a constant 2. When the parity check result is equal to 1, the node collision elimination unit determines that the ray has intersected the boundary line segment an odd number of times. Based on the principle of closed curves in topology, this determines that the data node is completely located within the interior space of the vector polygon outline and marks it as an overlapping data node. When the parity check result is equal to 0, the node collision elimination unit determines that the data node is located in the exterior space of the vector polygon outline. After completing the ray cross-test for all data nodes, the node collision elimination unit completely removes all data points marked as overlapping data nodes from the complete data sequence of the standard QR code grid. The blank areas formed by these deletions are reserved for subsequent graphic printing, while the remaining non-overlapping data node combinations are ultimately output as irregularly shaped blank QR code templates.

[0032] For example, the node collision elimination unit reads a standard QR code grid with dimensions of 41 rows by 41 columns in the grid-mapped coordinate system. This standard QR code grid contains a total of 1681 discrete data nodes. The node collision elimination unit selects the data node with coordinates of horizontal 21 and vertical 21 for ray crossing test. A horizontal unidirectional ray is emitted to the right from horizontal 21 and vertical 21. After geometric comparison, the unidirectional ray passes through the right boundary segment of the vector polygon contour map once, and the total number of physical crossings is recorded as 1. Substituting the total number of physical crossings of 1 into the parity check formula, the modulo operation of the constant 2 is performed, and the parity check result value is calculated to be 1. Based on this parity check result value of 1, the node collision elimination unit determines that the data node with coordinates of horizontal 21 and vertical 21 is an overlapping data node. Subsequently, the node collision elimination unit selects the data node with coordinates of horizontal 1 and vertical 1 and emits a unidirectional ray. It does not intersect with any boundary segment, the total number of physical crossings is 0, the modulo result is 0, and it is determined that it is located outside. After traversing 1681 data nodes, the node collision elimination unit identified 120 overlapping data nodes. These 120 overlapping data nodes were then completely removed from the standard QR code grid data sequence. The remaining 1561 data nodes were then matrix-recombined to successfully output an irregularly shaped, blank QR code template for data stream writing.

[0033] The dynamic background color processing module is used to extract grayscale values ​​from the original image and perform point-by-point difference calculation with the set of reflectivity values ​​to filter out pixels and output a dynamic background color removal mark image. Optionally, the dynamic background color processing module includes: The grayscale feature construction unit is used to perform color space conversion operations on the original image and extract the grayscale values ​​of pixel coordinates, and construct a two-dimensional grayscale feature matrix based on the grayscale values; The grayscale feature construction unit receives the raw image from the data acquisition module. It performs color space transformation on the raw image, traversing its two-dimensional physical space to extract the red, green, and blue color channel values ​​corresponding to each pixel coordinate. To reduce the dimensionality of the three-dimensional color channel data to single-channel data that directly reflects the light intensity on the bottle cap surface, the grayscale feature construction unit performs a linear transformation calculation according to the photometric standards defined by the International Commission on Illumination (ICI). The calculation relationship satisfies the following formula: , Among them, letters The horizontal coordinates of the pixels are and vertical The grayscale value obtained after position transformation; letters The red color channel value representing the same pixel coordinate; letters The green color channel value representing the same pixel coordinate; letters The blue color channel value represents the same pixel coordinate; the numbers 0.299, 0.587, and 0.114 represent standard constant weights that match the visual sensitivity of the human eye. After completing the color space transformation operation for all pixel coordinates, the grayscale feature construction unit combines and concatenates all the extracted grayscale values ​​according to the spatial physical arrangement order of the original image to construct a two-dimensional matrix, namely the two-dimensional grayscale feature matrix.

[0034] For example, the grayscale feature construction unit acquires the original image with a spatial resolution of 1920 pixels multiplied by 1080 pixels. The grayscale feature construction unit locates the pixel at coordinates 500 horizontally and 600 vertically, extracting the red color channel value of 180, the green color channel value of 160, and the blue color channel value of 100 for that pixel. The grayscale feature construction unit substitutes these color channel values ​​into the color space conversion formula, performing multiplication and addition calculations to obtain an approximate grayscale value of 159 for that coordinate. The grayscale feature construction unit performs the same color space conversion operation on each of the 2,073,600 pixels in the original image, ultimately constructing a two-dimensional grayscale feature matrix containing 2,073,600 elements.

[0035] The differential zeroing and filtering unit is used to perform point-by-point differential calculations on the same coordinates between the two-dimensional grayscale feature matrix and the set of reflective intensity values, delete pixels in the negative value range, and output a dynamic background color removal mark image.

[0036] The differential zeroing and filtering unit extracts the grayscale feature matrix output by the grayscale feature construction unit, and simultaneously extracts the reflective intensity value set output by the data acquisition module. Since the reflective intensity value set is generated based on spatial grid partitioning, its physical data density is lower than that of the two-dimensional grayscale feature matrix. Therefore, the differential zeroing and filtering unit first performs bilinear spatial interpolation resampling calculation on the reflective intensity value set. Bilinear spatial interpolation resampling calculation refers to performing distance-weighted linear interpolation once in the horizontal and once in the vertical directions of the two-dimensional plane, smoothly expanding the low-resolution discrete grid values ​​to a continuous coordinate system with the same spatial size as the two-dimensional grayscale feature matrix, generating a resampled reflective matrix. Subsequently, the differential zeroing and filtering unit performs point-by-point difference calculation on the two-dimensional grayscale feature matrix and the resampled reflective matrix using the same coordinates. The physical meaning of this point-by-point difference calculation is to evaluate whether the actual visual grayscale of each physical pixel is swallowed up by the high-intensity background reflection in that local area. The calculation relationship satisfies the following formula: , Among them, letters The horizontal coordinates of the pixels are and vertical The difference value obtained by calculating the position; letters Represents the grayscale value of the two-dimensional grayscale feature matrix at this coordinate; letter This represents the average light wave distribution value of the resampled reflection matrix at that coordinate; the letter represents the average light wave distribution value of the resampled reflection matrix at that coordinate. This represents the dimensional alignment constant. Since the average light wave distribution value extracted from multispectral analysis is a reflectance ratio physical quantity between zero and one, the dimensional alignment constant is strictly set to the maximum theoretical physical boundary value of grayscale depth, 255, to ensure strict consistency of the data scale on both sides of the equation. The differential zeroing and filtering unit performs numerical condition judgments on each of the calculated differential values. When the differential value is less than zero, the differential zeroing and filtering unit determines that the pixel at that coordinate belongs to the negative value range. Pixels in the negative value range mean that the actual pixel features at that location are completely masked by the metallic background reflection, belonging to the optical interference area. The differential zeroing and filtering unit forcibly deletes the pixel state of all pixels in the negative value range and assigns a value of zero, while assigning a value of one to pixels with a differential value greater than or equal to zero. Through binarization assignment operations, the differential zeroing and filtering unit constructs a binarized matrix to characterize the interference-free coding physical region, and finally outputs it as a dynamic background color removal marker map. Figure 3 As shown, the spatial mapping relationship of millions of pixels in the original grayscale value and the reflective intensity dimension after alignment is presented. By setting the filtering threshold, the high-density noise reduction points located below the threshold are accurately identified as negative value pixels and completely filtered out, effectively avoiding the absorption of the underlying coding data by the metallic background reflection.

[0037] For example, the differential zeroing filter unit extracts a two-dimensional grayscale feature matrix of size 1920 pixels by 1080 pixels, and a set of reflectance intensity values ​​consisting of 400 grid values. The differential zeroing filter unit performs bilinear spatial interpolation resampling calculation, smoothly mapping and expanding the 400 grid values ​​into a resampled reflectance matrix of 1920 pixels by 1080 pixels. The differential zeroing filter unit selects positions at coordinates 500 horizontally and 600 vertically for numerical comparison. The differential zeroing filter unit reads the grayscale value of the two-dimensional grayscale feature matrix at this coordinate as 159, and reads the average light wave distribution value of the resampled reflectance matrix at the same coordinate as 0.8. The differential zeroing filter unit multiplies the average light wave distribution value 0.8 by the dimensional alignment constant 255, calculating the aligned reflectance intensity threshold as 204. The differential zeroing filter unit subtracts the aligned reflectance intensity threshold 204 from the grayscale value 159, obtaining a difference value of -45 for this coordinate. Because -45 is less than zero, the differential zeroing filtering unit determines that the coordinate point belongs to the negative value range and completely deletes its value at that coordinate, changing it to zero. For another position with coordinates of horizontal 10 and vertical 10, its grayscale value is 200 and the aligned reflectivity threshold is 100, with a differential value of 100, which is greater than zero. Therefore, the differential zeroing filtering unit changes this coordinate to one. After traversing all matrix coordinates, the differential zeroing filtering unit outputs a full-size dynamic background color removal marker image composed of zero and one values.

[0038] The micro-dot matrix steganography module is used to obtain the product traceability basic code through external input, convert the product traceability basic code into a binary data stream, adjust the spatial distribution of the dynamic background color removal mark image according to the binary data stream, and generate a micro-dot matrix frequency modulation steganography image. Optionally, the micro-matrix steganography module includes: The basic encoding conversion unit is used to perform character set feature mapping on the product traceability basic encoding to extract character encoding values, and to perform bit serialization and error correction bit appending operations on the character encoding values ​​to generate a binary data stream; The basic encoding conversion unit acquires the product traceability basic code, composed of English letters and Arabic numerals, through an external communication bus. The unit performs character set feature mapping on this code, using the universally accepted ASCII information exchange standard to map each visible character in the code to a decimal character code value. After mapping, the unit performs bit serialization on the extracted character code values, converting the decimal values ​​into eight-bit machine code, and then concatenates all the machine code along the physical memory address. To prevent data loss or incompleteness in harsh industrial environments, the unit adds error correction bits to the end of the combined sequence. Finally, the unit merges the original information bits with the generated check bits, outputting a complete binary data stream.

[0039] The grid shading modulation unit is used to divide the dynamic background color removal identifier map into a spatial pixel grid, traverse the bit state of the binary data stream, perform pixel filling and erasing operations of the spatial pixel grid according to the bit state to reconstruct the grid shading rate, and output a micro-dot matrix frequency-modulated stegogram.

[0040] The grid coloring modulation unit receives the dynamic background color removal marker image output by the dynamic background color processing module. The grid coloring modulation unit uniformly divides this dynamic background color removal marker image into a discrete spatial pixel grid composed of multiple pixels along the horizontal and vertical physical coordinate axes. The physical side length of this spatial pixel grid is strictly equal to the number of pixels corresponding to the minimum physical diameter of the control spot of a single piezoelectric nozzle in the back-end coding device, ensuring that the adjustment of the grid coloring rate can be executed by the hardware with absolutely no error. The grid coloring modulation unit sequentially extracts the logical state of each independent bit in the binary data stream and establishes a one-to-one spatial mapping relationship between these logical states and the divided spatial pixel grid. Subsequently, the grid coloring modulation unit performs pixel filling and erasing operations within the spatial pixel grid based on the bit state. This process of reconstructing the grid coloring rate follows a clear binary feedback rule, and its calculation relationship satisfies the following formula: , Among them, letters Representing the The target mesh shading rate to be achieved after reconstruction of a spatial pixel grid; letters Represents the first in the binary data stream The logical state of each bit; letters Represents the threshold for high shading rate constant; letters This represents the low shading rate constant threshold. The high and low shading rate constant thresholds are derived from the optical contrast detection limit parameters of the camera on the back-end production line. When the bit state is one, the grid shading modulation unit calculates the actual number of pixels in the current grid, randomly locates and illuminates blank points that originally belonged to the background removal area using an algorithm, until the proportion of colored pixels in the current grid reaches 90%. When the bit state is zero, the grid shading modulation unit randomly locates and erases existing colored pixels in the current grid until their proportion drops to 10%. After traversing all bits of the binary data stream and completing the shading rate reconstruction of the corresponding grid, the grid shading modulation unit outputs a micro-matrix frequency-modulated stegmap carrying the steganographic data. For example... Figure 4 As shown, the modulation effect of binary data stream on physical pixel grid density is demonstrated. After pixel padding and erasure operations, the originally randomly distributed grid color rate is forced to polarize into two clusters of corresponding logic bits "1" and "0". Without destroying the macro background color, the covert writing of high-frequency digital signals on physical entities is successfully achieved.

[0041] For example, the grid coloring modulation unit divides the dynamic background color removal marker image into a spatial pixel grid of size five pixels by five pixels, with each grid containing twenty-five physical pixels. The grid coloring modulation unit extracts the first bit at index 1 in the binary data stream and reads its logical state as 1. According to the feedback rule, the grid coloring modulation unit determines that the target grid coloring rate of the first spatial pixel grid needs to reach 90%, which means 22.5 pixels need to be colored, rounded to twenty-three pixels. After scanning, the first spatial pixel grid initially has only fifteen colored pixels. The grid coloring modulation unit randomly selects eight positions from the remaining ten blank points in the grid and performs a pixel filling operation, assigning them a dark color state. The grid coloring modulation unit extracts the second bit at index 2 in the binary data stream and reads its logical state as 0. It determines that the target grid coloring rate of the second spatial pixel grid needs to be reduced to 10%, which means only 2.5 pixels are allowed to be colored, rounded to three pixels. After scanning, this grid initially has twelve colored pixels. The grid shading modulation unit randomly selects nine colored dots and performs a pixel erasure operation, restoring them to a blank background. This operation changes the density and arrangement of ink dots at the microscopic level, and completes data steganography at the macroscopic level.

[0042] Optionally, the system further includes: Extract the geometric topological distribution features of the closed boundary pixel set, and perform inner product hash operation on the geometric topological distribution features and the binary data stream to output a cross-modal security feature vector.

[0043] After completing polygon approximation and bitstream construction, to achieve a deep cryptographic binding between the physical image appearance and the underlying encoded data, the relative spatial positions and Euclidean distance arrays of all coordinate points in the closed boundary pixel set are extracted to construct a geometric topological distribution feature tensor describing the shape of the image's periphery. Simultaneously, the binary data stream generated by the basic encoding conversion unit is extracted. An inner product hash operation is performed on the feature tensor in the image modality and the bitstream in the digital modality. The physical meaning of this operation is to cross-encrypt the immutability of visual geometry and the uniqueness of data information. Its computational relationship satisfies the following formula: , Among them, letters Represents the cross-modal security feature vector of the output; letters Represents the one-dimensional geometric topological distribution feature tensor extracted from the set of closed boundary pixels; letters Represents the converted digital binary data stream; symbol Represents the dot product and inner product operation of two tensor vectors; letters This represents an irreversible cryptographic hash function based on the SHA standard. First, the inner product of the geometric and data features is calculated, generating a scalar product that characterizes the degree of correlation between the two. This scalar product is then input into the hash function, outputting a fixed-length, unique, and irreversibly derivable cross-modal secure feature vector. This feature vector serves as the core comparison benchmark for anti-counterfeiting verification.

[0044] The composite anti-counterfeiting generation module is used to scale the micro-dot matrix frequency-modulated stegogram to the boundary size of the polygonal outline map, embed the irregularly shaped blank QR code template, perform data redundancy rearrangement calculation, and generate a composite anti-counterfeiting inkjet data package. Optionally, the composite anti-counterfeiting generation module includes: The polar coordinate constraint scaling unit is used to extract the maximum polar coordinate radius value of the polygonal outline map, perform proportional scaling on the micro-matrix frequency-modulated stegmap with the maximum polar coordinate radius value as a constraint condition, and align and embed it into the irregular blank QR code template to generate an initial embedded data packet. The polar coordinate constraint scaling unit receives the vector polygon contour map generated by the contour vector fitting unit and extracts the coordinates of all boundary vertices in the map. To define the physical circumscribed safe boundary of the irregular polygon, the polar coordinate constraint scaling unit transforms all boundary vertex coordinates from the Cartesian coordinate system to a polar coordinate system with the geometric center of the shape as the pole. In the polar coordinate system, the polar coordinate constraint scaling unit calculates each polar radius value, which physically represents the straight-line physical distance from the shape center to the boundary vertex. Subsequently, the polar coordinate constraint scaling unit extracts the mathematical maximum value among all polar radius values ​​and defines it as the maximum polar radius value. The calculation relationship of the maximum polar radius value satisfies the following formula: , Among them, letters Represents the extracted maximum radius value in polar coordinates; letters and The first vector polygon outline diagram Two-dimensional horizontal and vertical coordinates of each boundary vertex; letters and The two-dimensional horizontal and vertical coordinates representing the geometric center of the vector polygon outline; symbol This represents the operation of extracting the maximum value from the set. After obtaining the maximum radius value in polar coordinates, the polar coordinate constraint scaling unit extracts the micro-dot matrix frequency-modulated steganography output by the grid coloring modulation unit and obtains the original physical length of the diagonal of the image. The polar coordinate constraint scaling unit multiplies the maximum radius value in polar coordinates by a constant two to obtain the target constraint outer diameter, and uses the ratio of the target constraint outer diameter to the physical length of the diagonal as the scaling factor to perform a proportional scaling matrix multiplication operation on the micro-dot matrix frequency-modulated steganography. After scaling, the polar coordinate constraint scaling unit performs a spatial coordinate translation and overlap operation between the image center point of the micro-dot matrix frequency-modulated steganography and the grid center point of the irregular blank QR code template to achieve alignment and embedding, thereby generating an initial embedded data package containing the surrounding QR code data and the central steganography image features.

[0045] For example, the polar coordinate constraint scaling unit extracts a vector polygon outline containing 10 vertices. The unit calculates its geometric center coordinates as horizontal 500 and vertical 500. For a vertex with coordinates of horizontal 500 and vertical 600, the unit substitutes these coordinates into the distance formula, calculating the polar radius from this point to the center as 100 pixels. After calculating the polar radius values ​​for all 10 vertices, the unit compares the results and finds the maximum value to be 120 pixels, which is the maximum polar radius. The unit multiplies the maximum polar radius value of 120 pixels by 2, obtaining a target constraint outer diameter of 240 pixels. Simultaneously, the unit reads the original diagonal physical length of the micro-matrix FM steganography as 480 pixels. The unit calculates a scaling factor of 0.5. The unit then performs a scaling matrix operation of multiplying all pixel coordinates of the micro-matrix FM steganography by 0.5, shrinking it to within the constraint boundaries. Finally, the polar coordinate constraint scaling unit moves the center coordinates of the scaled-down image to the geometric center coordinates of the irregular blank QR code template, completing seamless alignment and embedding, and generating the initial embedded data package.

[0046] The cross-modal perturbation rearrangement unit is used to extract the cross-modal security feature vector as an address mapping index, use the address mapping index to perform a redundancy check data block reordering calculation on the initial embedded data packet, update the spatial arrangement coordinates of the redundancy check data block, and output the composite anti-counterfeiting inkjet data packet.

[0047] Because conventional QR codes have a fixed data layout paradigm, they are easily reverse-engineered. The cross-modal perturbation and rearrangement unit introduces a cryptographic perturbation mechanism to break this conventional paradigm. The unit extracts the cross-modal security feature vector output by the system's preceding modules. This vector is a sequence of irreversible hash values. Each byte in this hash value sequence is converted into a decimal integer, serving as an address mapping index sequence. Subsequently, the unit extracts redundant check data blocks from the QR code grid surrounding the initial embedded data packet. Based on the generated address mapping index sequence, it performs a reordering calculation on the physical array positions of the redundant check data blocks in memory space. This reordering calculation mechanism forces the originally continuous sequential coordinates of the data blocks to be mapped to discrete coordinates driven by hash features. The calculation relationship satisfies the following formula: , Among them, letters Representing the The new spatial arrangement coordinates of the redundant check data blocks after being rearranged in disorder; letters This represents the default original spatial arrangement coordinate sequence in the standard QR code grid before scrambling; letters Represents the first feature extracted from the cross-modal security feature vector. A decimal mapping index value; letters Represents the total physical number of redundancy check data blocks in the current initial chimeric data packet; symbol This represents performing a modulo loop to prevent array out-of-bounds errors; function This represents the mapping operation that extracts the corresponding position coordinates. Through formula traversal calculation, the cross-modal perturbation rearrangement unit completely disrupts the spatial arrangement coordinates of redundant verification data blocks, realizing the private encryption of QR code decoding rules, and finally assembling all the rearranged data to output a composite anti-counterfeiting inkjet data packet.

[0048] For example, the cross-modal perturbation rearrangement unit extracts a series of cross-modal security feature vectors, extracts the first three bytes and converts them to decimal integers, resulting in the address mapping index sequence 105, 42, and 88. The cross-modal perturbation rearrangement unit extracts the initial embedded data packet and scans it to confirm that it contains 20 redundant check data blocks that need to be rearranged. For the first target position after rearrangement, the cross-modal perturbation rearrangement unit extracts the first mapping index value 105. The cross-modal perturbation rearrangement unit substitutes 105 into the modulo calculation formula, performs a modulo operation with 20, and obtains a remainder of 5. Based on this remainder, the cross-modal perturbation rearrangement unit forcibly moves the redundant check data block located at the 5th position in the original spatial arrangement coordinate sequence to the 1st physical coordinate position of the new matrix. For the second target position, the index value 42 is extracted, and the remainder when modulo 20 is 2. The cross-modal perturbation rearrangement unit moves the data block originally located at the 2nd position to the 2nd physical coordinate position of the new matrix. For the third target position, the index value 88 is extracted. Taking the remainder 8 modulo 20, the cross-modal perturbation rearrangement unit moves the data block originally positioned at the eighth position to the third physical coordinate position in the new matrix. The cross-modal perturbation rearrangement unit iterates through all 20 data blocks, performs the forced physical displacement, updates the underlying layout coordinates, and outputs an encrypted, out-of-order composite anti-counterfeiting inkjet data packet.

[0049] The inkjet printing traceability execution module is used to transmit the composite anti-counterfeiting inkjet printing data packet to the inkjet printing equipment to perform ink spraying operations, acquire real-shot images of the finished product surface through the end-of-line camera, extract the product traceability basic code, and write it into the traceability database system.

[0050] Optionally, the inkjet printing traceability execution module includes: The hardware driver conversion unit is used to parse the composite anti-counterfeiting inkjet data packet to extract the array layout coordinates, map the array layout coordinates into the ink droplet driving pulse sequence of the piezoelectric printhead, and instruct the inkjet printing device to perform ink spraying operation according to the ink droplet driving pulse sequence. The hardware driver conversion unit receives the composite anti-counterfeiting inkjet data packet output by the composite anti-counterfeiting generation module. The hardware driver conversion unit performs protocol-level stripping and parsing operations on this composite anti-counterfeiting inkjet data packet, extracting a two-dimensional dot matrix composed of zero and one values. The hardware driver conversion unit traverses all row and column indices of this two-dimensional dot matrix, converting the position information of all pixels with a value of one into array arrangement coordinates for the horizontal and vertical stepping of the physical printhead. To convert the purely digital coordinates into analog electrical signals controlling the backend mechanical hardware, the hardware driver conversion unit maps the extracted array arrangement coordinates into a droplet driving pulse sequence for the piezoelectric printhead. The physical meaning of this mapping is to specify when the piezoelectric ceramic inside the printhead deforms to squeeze the ink and form flying droplets; the calculation relationship of its electrical pulse period satisfies the following formula: , Among them, letters Represents the number of ink droplet drive pulses required for each movement of the piezoelectric printhead; the letter Represents the physical distance between two adjacent ink dots calculated from the array layout coordinates; letters Represents the actual physical transport speed of the bottle cap on the industrial production line; letters This represents the pulse duty cycle constant, which is a fixed value of 50%, determined by the original manufacturer's hardware electrical characteristics. Through the above mapping calculation, the hardware driver conversion unit converts the static array arrangement coordinates into a dynamic timing level sequence accompanying the time axis distribution, i.e., the ink droplet driving pulse sequence. The hardware driver conversion unit sends execution commands to the marking equipment via the industrial control bus based on the ink droplet driving pulse sequence, driving the piezoelectric printhead of the marking equipment to perform precise ink spraying on the high-speed moving target bottle cap surface, completing the cross-physical domain transfer of digital information to a physical entity pattern.

[0051] For example, the hardware-driven conversion unit parses the composite anti-counterfeiting inkjet printing data packet and extracts a two-dimensional dot matrix with a size of 1,000 columns by 1,000 rows. The hardware-driven conversion unit locks the coordinates of a black ink dot to be printed in the 10th row and 20th column, extracting it as a set of array arrangement coordinates. According to the inkjet printing equipment installation specifications, the physical spacing between adjacent ink dots in the array arrangement coordinates is 0.1 mm. The production line encoder provides real-time feedback that the physical transmission speed of the bottle cap on the conveyor belt is 200 mm per second. The hardware-driven conversion unit divides the physical spacing of 0.1 mm by the physical transmission speed of 200 mm per second, calculating the time interval between adjacent actions to be 0.0005 seconds, or 500 microseconds. The hardware-driven conversion unit further multiplies 500 microseconds by 50% of the pulse duty cycle constant, calculating that the high-level ink droplet driving pulse period of the piezoelectric printhead driver chip is 250 microseconds. The hardware-driven conversion unit generates a level pulse command with a duration of 250 microseconds and sends it to the inkjet printer via the communication bus. The piezoelectric ceramic inside the inkjet printer physically contracts the instant it receives the pulse sequence, precisely spraying a drop of black ink onto the designated array coordinates on the surface of the target bottle cap.

[0052] The morphological reverse analysis unit is used to perform morphological filtering and positioning cropping on the real-shot image of the finished product surface to extract the central imprint region, quantize the pixel distribution frequency of the central imprint region, and perform demodulation decoding operation on the pixel distribution frequency to recover the product traceability basic code and write it into the traceability database system.

[0053] The morphological reverse engineering unit acquires real-world images of the finished product surface after printing using an industrial vision camera installed at the end of the production line. Due to the prevalent uneven lighting and metallic brushed texture interference in industrial environments, the morphological reverse engineering unit first performs a top-hat morphological filtering operation on the image. The physical meaning of this top-hat morphological filtering operation is to separate and highlight the fine ink structures that are darker than the surrounding background, thereby filtering out large areas of slow lighting changes. After morphological filtering, the morphological reverse engineering unit performs a positioning and cropping operation on the entire image based on the finder pattern features of the QR code's outer perimeter, accurately removing redundant background and extracting the central imprint region containing steganographic information. Subsequently, the morphological reverse engineering unit performs pixel distribution frequency quantization calculations on the spatial grid of the central imprint region. This quantization calculation refers to calculating the spatial density of black pixels within each physical grid relative to the total number of pixels in the grid. After acquiring the pixel distribution frequency, the morphology reverse analysis unit performs demodulation and decoding operations to determine whether the density value of each grid belongs to a high or low tinting rate range. Based on the tinting rate range, it maps the grid back to a bit state stream at a logic high or low level. After collecting all bit state streams, the morphology reverse analysis unit performs redundancy check comparison and error repair of physical ink-stained areas. Then, according to the character set reverse mapping table, it converts the corrected bit stream into a human-readable string, successfully restoring the product traceability base code. Finally, the morphology reverse analysis unit packages the product traceability base code, along with the corresponding printing time and production batch parameters, into a structured query language message and writes it to the remote traceability database system via network protocol.

[0054] Based on the same inventive concept, the present invention also provides a method for full-process product traceability using QR codes, the method comprising: The original image and multispectral scanning data of the target bottle cap surface are acquired, the closed boundary pixel set of the original image is extracted, and the reflectance intensity value set of the target bottle cap surface is calculated based on the multispectral scanning data. A polygonal outline is extracted from the closed boundary pixel set, the polygonal outline is mapped to a standard QR code grid, and data nodes that overlap with the polygonal outline within the standard QR code grid are removed to generate an irregularly shaped blank QR code template. The grayscale values ​​are extracted from the original image and then differentially calculated point by point with the set of reflectance intensity values ​​to filter out pixels, outputting a dynamic background color removal marker image. The product traceability base code is obtained through external input, the product traceability base code is converted into a binary data stream, and the spatial distribution of the dynamic background color removal mark map is adjusted according to the binary data stream to generate a micro-matrix frequency modulation stegmap. The micro-dot matrix frequency-modulated stegogram is scaled to the boundary size of the polygonal outline map and embedded in the irregularly shaped blank QR code template. Data redundancy rearrangement calculation is performed to generate a composite anti-counterfeiting inkjet data package. The composite anti-counterfeiting inkjet data packet is transmitted to the inkjet printing equipment to perform ink spraying operations. The actual image of the finished product surface is obtained through the camera at the end of the production line, the basic traceability code of the product is extracted and written into the traceability database system.

[0055] It should be noted that the formulas described above, through the principle of dimensional consistency and mathematical standardization methods (such as normalization, dimensionless parameter conversion, or unit system unification), can translate physical quantities with different properties into unitless standard values ​​or parameters that can be superimposed in the same dimension. This eliminates the interference of different dimensions on the computational logic, allowing the formulas to retain the original data distribution characteristics while possessing mathematical rationality and adaptability to objective laws. These are conventional technical methods and will not be elaborated further. The electrical connections between the various units described above do not necessarily represent direct or indirect connections; any indirect connection method is applicable to the embodiments of this invention as long as it achieves the purpose of this invention. The above descriptions are merely exemplary embodiments of this invention and should not be construed as limiting the scope of this invention.

[0056] All equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or conventional techniques in the art not described herein.

Claims

1. A product end-to-end traceability system using QR codes, characterized in that, The system includes: The data acquisition module is used to acquire original images and multispectral scanning data of the target bottle cap surface, extract the closed boundary pixel set of the original image, and calculate the set of reflective intensity values ​​of the target bottle cap surface based on the multispectral scanning data. The irregular QR code template module is used to extract a polygonal outline image through the closed boundary pixel point set, map the polygonal outline image to a standard QR code grid, and remove data nodes that overlap with the polygonal outline image within the standard QR code grid to generate an irregular blank QR code template. The dynamic background color processing module is used to extract grayscale values ​​from the original image and perform point-by-point difference calculation with the set of reflectivity values ​​to filter out pixels and output a dynamic background color removal mark image. The micro-dot matrix steganography module is used to obtain the product traceability basic code through external input, convert the product traceability basic code into a binary data stream, adjust the spatial distribution of the dynamic background color removal mark image according to the binary data stream, and generate a micro-dot matrix frequency modulation steganography image. The composite anti-counterfeiting generation module is used to scale the micro-dot matrix frequency-modulated stegogram to the boundary size of the polygonal outline map, embed the irregularly shaped blank QR code template, perform data redundancy rearrangement calculation, and generate a composite anti-counterfeiting inkjet data package. The inkjet printing traceability execution module is used to transmit the composite anti-counterfeiting inkjet printing data packet to the inkjet printing equipment to perform ink spraying operations, acquire real-shot images of the finished product surface through the end-of-line camera, extract the product traceability basic code, and write it into the traceability database system.

2. The product end-to-end traceability system with QR code identification as described in claim 1, characterized in that, The data acquisition module includes: The image contour extraction unit is used to perform grayscale processing and edge detection on the original image, extract the edge pixel coordinate sequence, and perform connected component topology analysis on the edge pixel coordinate sequence to generate a closed boundary pixel set. The light intensity feature analysis unit is used to analyze the multispectral scanning data to extract the reflectivity matrix, perform spatial grid division on the surface of the target bottle cap, calculate the average light wave distribution value of the reflectivity matrix in each grid, and summarize and construct a set of reflectivity values.

3. The product end-to-end traceability system with QR code identification as described in claim 1, characterized in that, The irregular-shaped QR code template module includes: The contour vector fitting unit is used to perform polygon approximation on the closed boundary pixel set, generate a vector polygon contour map, extract the geometric center point of the standard QR code grid, translate and align the vector polygon contour map to the geometric center point, and establish a grid mapping coordinate system. The node collision elimination unit is used to perform ray cross-test based on the grid mapping coordinate system, determine the positional relationship between the standard QR code grid and the vector polygon outline, delete overlapping data nodes, and output an irregular blank QR code template.

4. The product end-to-end traceability system with QR code identification as described in claim 1, characterized in that, The dynamic background color processing module includes: The grayscale feature construction unit is used to perform color space conversion operations on the original image and extract the grayscale values ​​of pixel coordinates, and construct a two-dimensional grayscale feature matrix based on the grayscale values; The differential zeroing and filtering unit is used to perform point-by-point differential calculations on the same coordinates between the two-dimensional grayscale feature matrix and the set of reflective intensity values, delete pixels in the negative value range, and output a dynamic background color removal mark image.

5. A product end-to-end traceability system with QR code identification as described in claim 1, characterized in that, The micro-matrix steganography module includes: The basic encoding conversion unit is used to perform character set feature mapping on the product traceability basic encoding to extract character encoding values, and to perform bit serialization and error correction bit appending operations on the character encoding values ​​to generate a binary data stream; The grid shading modulation unit is used to divide the dynamic background color removal identifier map into a spatial pixel grid, traverse the bit state of the binary data stream, perform pixel filling and erasing operations of the spatial pixel grid according to the bit state to reconstruct the grid shading rate, and output a micro-dot matrix frequency-modulated stegogram.

6. The product end-to-end traceability system with QR code identification as described in claim 1, characterized in that, The system also includes: Extract the geometric topological distribution features of the closed boundary pixel set, and perform inner product hash operation on the geometric topological distribution features and the binary data stream to output a cross-modal security feature vector.

7. A product end-to-end traceability system with QR code identification as described in claim 6, characterized in that, The composite anti-counterfeiting generation module includes: The polar coordinate constraint scaling unit is used to extract the maximum polar coordinate radius value of the polygonal outline map, perform proportional scaling on the micro-matrix frequency-modulated stegmap with the maximum polar coordinate radius value as a constraint condition, and align and embed it into the irregular blank QR code template to generate an initial embedded data packet. The cross-modal perturbation rearrangement unit is used to extract the cross-modal security feature vector as an address mapping index, use the address mapping index to perform a redundancy check data block reordering calculation on the initial embedded data packet, update the spatial arrangement coordinates of the redundancy check data block, and output the composite anti-counterfeiting inkjet data packet.

8. A product end-to-end traceability system with QR code identification as described in claim 1, characterized in that, The inkjet printing traceability execution module includes: The hardware driver conversion unit is used to parse the composite anti-counterfeiting inkjet data packet to extract the array layout coordinates, map the array layout coordinates into the ink droplet driving pulse sequence of the piezoelectric printhead, and instruct the inkjet printing device to perform ink spraying operation according to the ink droplet driving pulse sequence. The morphological reverse analysis unit is used to perform morphological filtering and positioning cropping on the real-shot image of the finished product surface to extract the central imprint region, quantize the pixel distribution frequency of the central imprint region, and perform demodulation decoding operation on the pixel distribution frequency to recover the product traceability basic code and write it into the traceability database system.

9. A product end-to-end traceability method using QR codes, applied to a product end-to-end traceability system using QR codes as described in any one of claims 1-8, characterized in that, The method includes: The original image and multispectral scanning data of the target bottle cap surface are acquired, the closed boundary pixel set of the original image is extracted, and the reflectance intensity value set of the target bottle cap surface is calculated based on the multispectral scanning data. A polygonal outline is extracted from the closed boundary pixel set, the polygonal outline is mapped to a standard QR code grid, and data nodes that overlap with the polygonal outline within the standard QR code grid are removed to generate an irregularly shaped blank QR code template. The grayscale values ​​are extracted from the original image and then differentially calculated point by point with the set of reflectance intensity values ​​to filter out pixels, and a dynamic background color removal mark image is output. The product traceability base code is obtained through external input, the product traceability base code is converted into a binary data stream, and the spatial distribution of the dynamic background color removal mark map is adjusted according to the binary data stream to generate a micro-matrix frequency modulation stegmap. The micro-dot matrix frequency-modulated stegogram is scaled to the boundary size of the polygonal outline map and embedded in the irregularly shaped blank QR code template. Data redundancy rearrangement calculation is performed to generate a composite anti-counterfeiting inkjet data package. The composite anti-counterfeiting inkjet data packet is transmitted to the inkjet printing equipment to perform ink spraying operations. The actual image of the finished product surface is obtained through the camera at the end of the production line, the basic traceability code of the product is extracted and written into the traceability database system.

Images