Ink-jet printer real-time image calibration system and method based on edge calculation

Abstract

The invention relates to the technical field of image calibration, and discloses an ink-jet printer real-time image calibration system and method based on edge calculation, and the method comprises the steps: constructing a composite parameter matrix, completing the no-load error analysis, and generating a compensation matrix and a dynamic weight distribution function; constructing a multi-scale gradient feature pyramid according to the composite parameter matrix, fusing the features of the pyramid to generate a feature correlation thermodynamic diagram, and establishing a model to predict an ink diffusion trajectory after correction; and finally generating a control signal by combining the compensation matrix and the prediction result, and controlling the motion pose of the nozzle. According to the method, the code spraying process is accurately controlled by integrating multiple factors, the precision and quality of code spraying patterns are remarkably improved, the rejection rate is effectively reduced, and the production efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image calibration, and more specifically, to a real-time image calibration system and method for a coding machine based on edge computing. Background Art

[0002] In the field of industrial production, coding technology, as a key means of product identification and information traceability, is widely used in many industries such as food, medicine, and tobacco. The quality of the coded characters directly affects the readability, traceability, and brand image of the products. In the scenario of cigarette box coding, the quality of the character edges is related to the anti-counterfeiting and traceability effects of the tobacco industry. If the edges of the coded characters are blurred, it will bring great difficulties to work such as two-dimensional code recognition, and thus affect the entire production process and the market circulation of the products.

[0003] In the prior art, the automatic detection method and device for the nozzle of a digital inkjet printer based on machine vision proposed in the Chinese patent application with the publication number CN119459130A mainly detect the inkjet quality of the nozzle and calibrate it by collecting the nozzle needle images and analyzing the inkjet state. This prior art only focuses on the detection of the inkjet performance of the nozzle itself, and lacks in-depth research and targeted treatment of the problem of character edge blurring caused by ink diffusion during the coding process. It cannot predict the diffusion trajectory according to the physical properties of the ink (such as viscosity, surface tension), nor can it effectively distinguish the real edge from the diffusion artifacts, and it is difficult to achieve precise compensation for the blurred edges of the coded characters. When facing the "burr" problem of the edges of the cigarette box coded characters, it cannot provide accurate data support for subsequent image calibration.

[0004] The Chinese patent with the authorization announcement number CN117119115B discloses a calibration method, device, electronic device, and storage medium based on machine vision, which compensates the coding period by obtaining the marker point images to determine the offset data to achieve calibration. However, this prior art does not fully consider the influence of ink diffusion on the edges of the coded characters, cannot predict the diffusion situation based on the ink characteristics, and has no specific measures for sub-pixel calibration, and cannot meet the high requirements for the edge clarity of the coded characters in cigarette box coding. When dealing with the problem of blurred edges of the coded characters, it is difficult to achieve the ideal calibration accuracy.

[0005] In summary, the existing coding machine calibration technologies have deficiencies in dealing with the problem of blurred edges of the coded characters, cannot accurately distinguish the real edge from the diffusion artifacts, are difficult to achieve sub-pixel calibration accuracy, and cannot meet the strict requirements for the edge quality of the coded characters in industries such as the tobacco industry. Summary of the Invention

[0006] To overcome the above-mentioned defects of the prior art, the present invention provides a real-time image calibration system and method for inkjet printers based on edge computing. By constructing a composite parameter matrix, various factors such as ink diffusion, substrate material adsorption, and nozzle movement are comprehensively considered, providing comprehensive data support for image calibration. On this basis, no-load error analysis is carried out to generate a compensation matrix and a dynamic weight distribution function, which can adaptively adjust the compensation strategy. At the same time, a multi-scale gradient feature pyramid is constructed, a heat map is generated by fusing features, and a dynamic prediction model of ink diffusion behavior is established, which can accurately predict the ink diffusion trajectory. Then, a control signal is generated based on the prediction result to perform dynamic feedback control on the movement pose of the nozzle. This series of operations effectively solves the problem of blurred edges of inkjet characters, achieves sub-pixel-level calibration, significantly improves the accuracy and quality of inkjet patterns, greatly improves the recognition rate of inkjet information such as two-dimensional codes, enhances the adaptability of the inkjet printer to different working conditions, meets the high-quality requirements of inkjet in various industries, and brings higher efficiency and lower costs to industrial production.

[0007] The present invention is applicable to various industrial production scenarios with high requirements for inkjet quality, such as batch inkjet for food and beverages, model identification inkjet for electronic products, and cigarette box inkjet for the tobacco industry. In these scenarios, the accuracy, clarity, and stability of inkjet are crucial, and any inkjet defect may lead to product quality problems or difficulties in information traceability.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A real-time image calibration method for an inkjet printer based on edge computing, comprising:

[0010] Constructing a composite parameter matrix, performing no-load error analysis, generating a no-load artifact offset error compensation matrix, establishing a non-linear mapping relationship between the elements of the no-load artifact offset error compensation matrix and the composite parameter matrix, and generating a dynamic weight distribution function for the compensation matrix;

[0011] According to the composite parameter matrix, constructing a multi-scale gradient feature pyramid, the multi-scale gradient feature pyramid including a bottom layer, a middle layer, and a top layer; performing tensor fusion on the features of the bottom layer, middle layer, and top layer of the multi-scale gradient feature pyramid to generate a feature correlation heat map covering different abstraction levels; combining the no-load artifact offset error compensation matrix, correcting the feature correlation heat map, and establishing a dynamic prediction model of ink diffusion behavior to generate a real-time diffusion trajectory prediction result;

[0012] According to the dynamic weight distribution function of the compensation matrix and the real-time diffusion trajectory prediction result, generating an artifact compensation control signal, coupling the artifact compensation control signal to the nozzle drive circuit, and performing dynamic feedback control on the movement pose of the nozzle.

[0013] Further, the constructing of the composite parameter matrix includes:

[0014] Obtain the surface viscosity coefficient and the potential energy gradient value to form a sub-matrix of fluid diffusion field parameters;

[0015] Measure the adsorption attenuation characteristic curve of ink on the surfaces of different materials, extract the adsorption attenuation time constant and the saturation adsorption concentration threshold to form a sub-matrix of substrate material adsorption characteristic data;

[0016] Track the movement process of the nozzle, obtain the nozzle vibration displacement signal, and extract the spatial curvature change characteristics of the nozzle movement trajectory from the nozzle vibration displacement signal to form a sub-matrix of the spatial characteristics of the nozzle movement trajectory;

[0017] Construct a composite parameter matrix according to the sub-matrix of fluid diffusion field parameters, the sub-matrix of substrate material adsorption characteristic data, and the sub-matrix of the spatial characteristics of the nozzle movement trajectory.

[0018] Further, the formation of the sub-matrix of fluid diffusion field parameters includes: collecting the dynamic contact angle change data of ink droplets on the substrate surface, and extracting the surface viscosity coefficient; measuring the spatial difference in the internal pressure distribution of ink droplets, and calculating the potential energy gradient value; quantifying the surface viscosity coefficient and the potential energy gradient value into a sub-matrix of fluid diffusion field parameters.

[0019] Further, the generation of the no-load artifact offset error compensation matrix includes:

[0020] According to the sub-matrix of fluid diffusion field parameters, obtain the artifact width distribution data formed during the ink diffusion process, and generate a diffusion artifact width distribution spectrum;

[0021] Use the diffusion artifact width distribution spectrum as a prior constraint condition at the initial stage of printing in the no-load state, and load it into the lightweight inverse generative adversarial topology network in the pre-constructed edge computing node to obtain a diffusion artifact prior constraint;

[0022] According to the sub-matrix of fluid diffusion field parameters and the sub-matrix of substrate material adsorption characteristic data, obtain the relative contribution rate of the adsorption effect and the relative contribution rate of the ink potential energy driving effect;

[0023] Combined with the diffusion artifact prior constraint, perform error analysis on the printed code image in the no-load state to generate a no-load artifact offset error compensation matrix.

[0024] Further, the obtaining of the relative contribution rate of the adsorption effect and the relative contribution rate of the ink potential energy driving effect includes:

[0025] Based on the surface viscosity coefficient and the potential energy gradient value in the sub-matrix of fluid diffusion field parameters, construct a mathematical and physical model for the diffusion process of ink droplets on the substrate surface, and use the finite element analysis method to numerically simulate the hydrodynamic characteristics inside the droplets to obtain a quantitative description function of the ink potential energy driving effect;

[0026] Based on the adsorption decay time constant and the saturated adsorption concentration threshold in the adsorption characteristic data sub - matrix of the substrate material, an adsorption kinetic model between the ink droplet and the substrate surface is constructed. The molecular dynamics simulation method is used to quantitatively analyze the adsorption process, and a quantitative description function of the adsorption effect is obtained;

[0027] The quantitative description function of the ink potential energy driving effect and the quantitative description function of the adsorption effect are normalized to obtain the relative contribution rate of the adsorption effect and the relative contribution rate of the ink potential energy driving effect.

[0028] Furthermore, in combination with the diffusion artifact prior constraint, error analysis is performed on the ink - jet printing image in the no - load state, and the no - load artifact offset error compensation matrix generated includes:

[0029] Combined with the diffusion artifact prior constraint, the ink - jet printer is used to print a preset reference pattern under no - load conditions, and the actual ink - jet printing image in the no - load state is collected;

[0030] Feature point matching is performed between the collected actual ink - jet printing image and the reference pattern, and the geometric offset of the ink - jet contour is calculated;

[0031] The geometric offset is characterized in matrix form to generate the no - load artifact offset error compensation matrix.

[0032] Furthermore, the calculation of the geometric offset of the ink - jet contour includes:

[0033] Key feature points of the actual ink - jet printing image and the reference pattern are extracted to construct a feature point matching matrix;

[0034] Based on the feature point matching matrix, a geometric offset function of the ink - jet contour is fitted to obtain the local geometric offset and the overall geometric offset;

[0035] The local geometric offset and the overall geometric offset are combined to form the geometric offset of the ink - jet contour.

[0036] Furthermore, the characterization of the geometric offset in matrix form to generate the no - load artifact offset error compensation matrix includes:

[0037] According to the relative contribution rate of the adsorption effect and the relative contribution rate of the ink potential energy driving effect, the local geometric offset and the overall geometric offset are weighted and combined to obtain a comprehensive geometric offset matrix;

[0038] The elements in the comprehensive geometric offset matrix are normalized and discretized to generate the no - load artifact offset error compensation matrix.

[0039] Furthermore, the construction of the multi - scale gradient feature pyramid includes:

[0040] Based on the fluid diffusion field parameter sub - matrix, extract the multi - scale gradient invariant feature clusters formed during the diffusion process of the ink on the substrate surface, and construct the bottom layer of the multi - scale gradient feature pyramid; the multi - scale gradient invariant feature clusters include 64 - dimensional histograms of oriented gradients.

[0041] Based on the substrate material adsorption characteristic data sub - matrix, capture the microscopic porosity distribution mapping on the substrate material surface, and construct the middle layer of the multi - scale gradient feature pyramid.

[0042] Based on the spatial feature sub - matrix of the nozzle movement trajectory, analyze the second - derivative fluctuation characteristics of the nozzle movement trajectory, and construct the top layer of the multi - scale gradient feature pyramid.

[0043] An edge - computing - based real - time image calibration system for an inkjet printer, which is used to implement the above - mentioned edge - computing - based real - time image calibration method for an inkjet printer. The system includes:

[0044] Compensation matrix generation module: used to construct a composite parameter matrix, conduct no - load error analysis, generate a no - load artifact offset error compensation matrix, establish a non - linear mapping relationship between the elements of the no - load artifact offset error compensation matrix and the composite parameter matrix, and generate a compensation matrix dynamic weight distribution function.

[0045] Diffusion trajectory prediction module: According to the composite parameter matrix, construct a multi - scale gradient feature pyramid, which includes a bottom layer, a middle layer, and a top layer; fuse the features of the bottom layer, middle layer, and top layer of the multi - scale gradient feature pyramid tensorially to generate a feature correlation heat map covering different abstraction levels; combine with the no - load artifact offset error compensation matrix to correct the feature correlation heat map, and establish a dynamic prediction model for ink diffusion behavior to generate a real - time diffusion trajectory prediction result.

[0046] Error compensation module: According to the compensation matrix dynamic weight distribution function and the real - time diffusion trajectory prediction result, generate an artifact compensation control signal, couple the artifact compensation control signal to the nozzle drive circuit, and perform dynamic feedback control on the movement pose of the nozzle.

[0047] Compared with the prior art, the beneficial effects of the present invention are:

[0048] The present invention integrates key factors such as ink diffusion, substrate material adsorption, and nozzle movement by constructing a composite parameter matrix, providing comprehensive data support for subsequent analysis and compensation. On this basis, no-load error analysis is carried out to generate a compensation matrix and a dynamic weight distribution function, which can adaptively adjust the compensation strategy according to the actual situation. A multi-scale gradient feature pyramid is constructed and features are fused to generate a heat map. After correction with the no-load artifact offset error compensation matrix, a dynamic prediction model of ink diffusion behavior is established, which can accurately predict the ink diffusion trajectory. Finally, a control signal is generated according to the compensation matrix and the prediction result to feedback control the movement pose of the nozzle, realizing the full-process precise optimization of the inkjet coding process from data acquisition and analysis, error compensation to real-time control, greatly improving the accuracy and quality of the inkjet coding pattern, reducing the rejection rate caused by inkjet coding errors, lowering production costs, enhancing the adaptability of the inkjet printer to different working conditions, and significantly improving production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0050] Figure 1 It is a principle flow chart of the real-time image calibration method for an inkjet printer based on edge computing in the present invention;

[0051] Figure 2 It is a method flow chart of constructing a composite parameter matrix in the real-time image calibration method for an inkjet printer based on edge computing in the present invention;

[0052] Figure 3 It is a method flow chart of forming a fluid diffusion field parameter sub-matrix in the real-time image calibration method for an inkjet printer based on edge computing in the present invention;

[0053] Figure 4 It is a method flow chart of obtaining the relative contribution rate of the adsorption effect and the relative contribution rate of the ink potential energy driving effect in the real-time image calibration method for an inkjet printer based on edge computing in the present invention;

[0054] Figure 5 It is a method flow chart of calculating the geometric offset of the inkjet coding contour in the real-time image calibration method for an inkjet printer based on edge computing in the present invention;

[0055] Figure 6 It is a method flow chart of constructing a multi-scale gradient feature pyramid in the real-time image calibration method for an inkjet printer based on edge computing in the present invention;

[0056] Figure 7This is the functional block diagram of the real-time image calibration system for an inkjet printer based on edge computing in the present invention. Detailed implementation manners

[0057] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0058] Embodiment 1

[0059] Please refer to Figure 1 As shown, this embodiment provides a real-time image calibration method for an inkjet printer based on edge computing, including:

[0060] Step S1000: Construct a composite parameter matrix, perform no-load error analysis, generate a no-load artifact offset error compensation matrix, establish a non-linear mapping relationship between the elements of the no-load artifact offset error compensation matrix and the composite parameter matrix, and generate a compensation matrix dynamic weight distribution function;

[0061] Further, step S1000 includes:

[0062] Step S1100: Construct a composite parameter matrix;

[0063] Further, as Figure 2 shown, step S1100 includes:

[0064] Step S1110: Obtain the surface viscosity coefficient and potential energy gradient value, and form a fluid diffusion field parameter sub-matrix;

[0065] Further, as Figure 3 shown, step S1110 includes:

[0066] Step S1111: Collect the dynamic contact angle change data of the ink droplet on the substrate surface, and extract the surface viscosity coefficient;

[0067] Step S1112: Measure the spatial difference in the internal pressure distribution of the ink droplet, and calculate the potential energy gradient value;

[0068] Step S1113: Quantize the surface viscosity coefficient and potential energy gradient value into a fluid diffusion field parameter sub-matrix.

[0069] Specifically, the dynamic contact angle change data of the ink droplet on the substrate surface is collected by the microfluidic sensing array, and then the surface viscosity coefficient is extracted. The microfluidic sensing array is a micro-sensor array fabricated using microelectromechanical processing technology, which can accurately measure various parameters of fluids at the microscale. During the inkjet coding process, the interaction between the ink droplet and the substrate surface is affected by various factors, and the surface viscosity coefficient is one of them. The surface viscosity coefficient reflects the magnitude of the internal frictional force experienced by the ink droplet when flowing on the substrate surface, and it determines the speed and uniformity of ink diffusion. For example, when the surface viscosity coefficient is large, the ink droplet has difficulty flowing on the substrate surface, and the diffusion speed is slow, which may result in unclear edges of the inkjet coding pattern; conversely, if the surface viscosity coefficient is small, the ink droplet diffuses too quickly, causing problems such as blurring and bleeding in the inkjet coding pattern. By collecting the dynamic contact angle change data to extract the surface viscosity coefficient, the flow characteristics of the ink on the substrate surface can be accurately obtained, providing basic data for subsequent analysis. This helps to evaluate the compatibility of the ink and the substrate material before inkjet coding, adjust the inkjet coding parameters in advance, and ensure the inkjet coding quality. At the same time, accurate surface viscosity coefficient data can provide accurate input parameters for the establishment of a dynamic prediction model of ink diffusion behavior, improve the accuracy of model prediction, and thus achieve precise control of the inkjet coding process.

[0070] The spatial difference in the internal pressure distribution of the ink droplet is measured using a pressure sensing array, and the potential energy gradient value is calculated therefrom. The pressure sensing array is a combined sensor device for measuring pressure distribution, which can obtain the pressure information at different positions inside the ink droplet. The potential energy gradient value represents the degree of change in potential energy per unit distance. During the ink diffusion process, it reflects the magnitude and direction of the driving force for the ink droplet to diffuse. When there is a pressure difference inside the ink droplet, a potential energy gradient is generated, which prompts the ink to diffuse from the high-pressure region to the low-pressure region. For example, if the internal pressure on one side of the ink droplet is greater than that on the other side, the potential energy gradient will push the ink droplet towards the side with lower pressure, affecting the shape and size of the inkjet coding pattern. Measuring the potential energy gradient value can help understand the movement trend of the ink inside the droplet and analyze its impact on the inkjet coding quality. By accurately grasping the potential energy gradient value, the design of the nozzle and the inkjet coding process can be optimized, the injection pressure and angle of the nozzle can be reasonably adjusted, so that the diffusion of the ink on the substrate surface is more uniform, thereby improving the accuracy and quality stability of the inkjet coding pattern. In addition, the data of the potential energy gradient value can also be used to study the kinetic process of ink diffusion, providing a theoretical basis for improving the ink formula to develop ink products more suitable for the inkjet coding process.

[0071] Quantify the surface viscosity coefficient and the potential energy gradient value into a sub-matrix of fluid diffusion field parameters. Quantification is a process of converting continuous physical quantities into discrete digital quantities for easy computer processing and analysis. By quantifying these two parameters into a sub-matrix form, the key physical characteristics during the ink diffusion process can be presented in a structured data form. This structured data is convenient for subsequent integration and analysis with other relevant data (such as substrate material adsorption characteristic data, printhead movement trajectory data, etc.). For example, when constructing a composite parameter matrix, the sub-matrix of fluid diffusion field parameters, as a part of it, is correlated with other sub-matrices to jointly provide comprehensive data support for the calibration of the inkjet printing image. Moreover, the quantified sub-matrix can be directly used as input data for algorithms and models, facilitating efficient numerical calculations and analysis using a computer. This enables the use of more complex and accurate algorithms when studying the ink diffusion behavior and optimizing the inkjet printing process, improving the analysis efficiency and accuracy, and thus achieving refined control of the inkjet printing process, enhancing the inkjet printing quality and production efficiency.

[0072] Step S1120: Measure the adsorption decay characteristic curves of inks on different material surfaces, extract the adsorption decay time constant and the saturated adsorption concentration threshold, and form a sub-matrix of substrate material adsorption characteristic data.

[0073] Specifically, use spectroscopic analysis to measure the adsorption decay characteristic curves of inks on different material surfaces, and then extract the adsorption decay time constant and the saturated adsorption concentration threshold to form a sub-matrix of substrate material adsorption characteristic data. Spectroscopic analysis is a technique based on the absorption, emission, or scattering characteristics of substances for different wavelengths of light to analyze the composition and structure of substances. During the inkjet printing process, the adsorption of the substrate material on the ink affects the diffusion and adhesion effects of the ink. The adsorption decay time constant reflects the speed of ink adsorption on the substrate material surface, and the saturated adsorption concentration threshold represents the maximum amount of ink that the substrate material can adsorb. For example, when inkjet printing on a paper packaging box, if the adsorption decay time constant of the paper is small, it means that the ink adsorbs quickly on the paper surface, which may lead to a fast drying speed of the inkjet printing pattern, but may also make the pattern color uneven; while a lower saturated adsorption concentration threshold may result in insufficient ink adhesion on the paper surface, affecting the clarity and durability of the inkjet printing. By measuring the adsorption decay characteristic curves and extracting relevant parameters to form a sub-matrix, the adsorption characteristics of the substrate material on the ink can be understood in detail. This helps to select a suitable combination of substrate material and ink, optimize the inkjet printing process, such as adjusting the inkjet volume and frequency of the printhead, to ensure the best adsorption and diffusion effects of the ink on the substrate surface, improve the quality and adhesion of the inkjet printing pattern, and at the same time reduce ink waste.

[0074] Step S1130: Track the movement process of the nozzle, obtain the nozzle vibration displacement signal, extract the spatial curvature change characteristics of the nozzle movement trajectory from the nozzle vibration displacement signal, and form a spatial feature sub-matrix of the nozzle movement trajectory.

[0075] Specifically, use a high-speed camera to track and photograph the movement process of the nozzle at a high frame rate to obtain the nozzle vibration displacement signal, and then use Fourier transform to extract the spatial curvature change characteristics of the nozzle movement trajectory from this signal and quantitatively generate a spatial feature sub-matrix of the nozzle movement trajectory. The high-speed camera can capture the movement details of the nozzle at a relatively high frame rate and record the vibration and displacement of the nozzle during the inkjet coding process. Fourier transform is a mathematical method that can convert a time-domain signal into a frequency-domain signal, facilitating the analysis of the frequency components and characteristics of the signal. The spatial curvature change characteristics of the nozzle movement trajectory reflect the stability and accuracy of the nozzle movement. For example, during continuous inkjet coding, if the spatial curvature change of the nozzle movement trajectory is large, it indicates that there are unstable factors in the nozzle movement process, which may lead to problems such as deviation of the inkjet pattern and uneven lines. By extracting this feature and quantifying it into a sub-matrix, the movement state of the nozzle can be monitored in real time. This is of great significance for timely detecting nozzle failures or abnormal conditions, helping to perform equipment maintenance and adjustment in advance to ensure the stability of the inkjet coding quality. At the same time, the spatial feature sub-matrix of the nozzle movement trajectory can also provide data support for optimizing the nozzle movement control algorithm. By adjusting the movement parameters of the nozzle, the nozzle movement can be made more stable and precise, thereby improving the accuracy and quality of the inkjet pattern.

[0076] Step S1140: Construct a composite parameter matrix based on the fluid diffusion field parameter sub-matrix, the substrate material adsorption characteristic data sub-matrix, and the spatial feature sub-matrix of the nozzle movement trajectory.

[0077] Specifically, a composite parameter matrix is constructed based on the fluid diffusion field parameter sub-matrix, the substrate material adsorption characteristic data sub-matrix, and the nozzle movement trajectory spatial characteristic sub-matrix. Integrating these three sub-matrices into a composite parameter matrix realizes the unified description and analysis of multiple key factors in the inkjet coding process. The composite parameter matrix contains information on ink diffusion characteristics, substrate material adsorption characteristics, and nozzle movement characteristics, etc., providing a comprehensive data basis for subsequent no-load error analysis, establishing a compensation matrix, and predicting ink diffusion behavior. For example, when establishing a dynamic prediction model for ink diffusion behavior, the composite parameter matrix, as input data, can comprehensively consider the influence of various factors on ink diffusion, making the model more accurately predict the diffusion trajectory and morphological changes of the ink on the substrate surface. In addition, when performing inkjet coding error compensation, the composite parameter matrix can provide a basis for determining the compensation strategy and calculating the compensation amount. By analyzing the mutual relationship between different sub-matrices, accurate compensation for inkjet coding errors can be achieved, thereby improving the calibration accuracy of the inkjet coding image, ultimately enhancing the quality and production efficiency of inkjet coded products, and reducing the rejection rate and production cost.

[0078] Step S1200: Obtain the data of the artifact width distribution formed during the ink diffusion process according to the fluid diffusion field parameter sub-matrix, and generate a diffusion artifact width distribution spectrum.

[0079] Specifically, step S1200 aims to obtain the data of the artifact width distribution formed during the ink diffusion process according to the fluid diffusion field parameter sub-matrix and generate a diffusion artifact width distribution spectrum. This step plays a key role in subsequent analysis of the error sources in the inkjet coding process and achieving image calibration.

[0080] During the inkjet coding process, the diffusion behavior of the ink on the substrate surface is affected by various factors. The surface viscosity coefficient and potential energy gradient value in the fluid diffusion field parameter sub-matrix are important factors determining the ink diffusion characteristics. The surface viscosity coefficient reflects the magnitude of the internal frictional force experienced by the ink when flowing on the substrate surface, which affects the speed and uniformity of ink diffusion; the potential energy gradient value represents the magnitude and direction of the driving force for the ink to diffuse due to the pressure difference inside the ink. When the ink diffuses, these factors will cause the distribution of the ink on the substrate surface to be uneven, thus generating artifacts. Artifacts refer to the unexpected extra or deformed parts in the inkjet coding pattern, which will affect the quality and accuracy of the inkjet coding.

[0081] By deeply analyzing the surface viscosity characteristics and potential energy distribution reflected by the fluid diffusion field parameter submatrix, the artifact width distribution data formed during the ink diffusion process can be extracted. For example, if the surface viscosity coefficient is small, the ink diffuses faster on the substrate surface, which may cause the artifact width to increase; and a higher potential energy gradient value may intensify the diffusion of ink in a specific direction, causing the artifact to extend in a certain direction. These data are sorted and analyzed to generate a diffusion artifact width distribution spectrum. The diffusion artifact width distribution spectrum is a graph that presents the artifact width data in a specific form. It can intuitively show the changes in the artifact width at different positions, providing an intuitive and quantitative data basis for subsequent error analysis and compensation.

[0082] The beneficial effects of this step are significant. First, the generation of the diffusion artifact width distribution spectrum helps to accurately locate the area and degree of artifacts during the coding process. By analyzing the spectrum, it is possible to clearly understand which areas have more serious artifacts, providing a clear direction for subsequent targeted error compensation. For example, when coding on packaging, if it is found that the width of the artifact in a certain area is large, it may cause the printed production date or batch information to be unclear. After locating the area through the diffusion artifact width distribution spectrum, the area can be adjusted and compensated in the subsequent coding process. Secondly, this step provides key data support for establishing a more accurate coding error model. Based on the diffusion artifact width distribution spectrum, the mechanism of artifact formation can be further studied, and the relationship between it and other factors (such as nozzle movement, substrate material characteristics, etc.) can be analyzed, so as to establish a more complete coding error model and improve the prediction and control capabilities of coding errors. Finally, the generation of the diffusion artifact width distribution spectrum is conducive to optimizing the coding process parameters. By comparing and analyzing the diffusion artifact width distribution spectra generated under different process parameters, the optimal parameter ranges such as surface viscosity coefficient and potential energy gradient value can be determined, and then the process parameters such as ink formula and nozzle injection pressure can be adjusted to reduce the generation of artifacts and improve the quality of inkjet printing.

[0083] Step S1300, the diffusion artifact width distribution spectrum is used as a priori constraint condition for the initial stage of printing in an unloaded state, and loaded into a lightweight inverse generative adversarial topology network in a pre-built edge computing node to obtain a priori constraint on the diffusion artifact;

[0084] Specifically, step S1300 is to use the diffusion artifact width distribution spectrum as a priori constraint condition for the initial stage of printing in the no-load state, and load it into the lightweight reverse generation adversarial topology network in the pre-built edge computing node, so as to obtain the diffusion artifact priori constraint. This step, with the help of advanced network models and edge computing technology, is conducive to the subsequent accurate coding error compensation.

[0085] A prior constraint refers to the restrictive conditions set based on existing knowledge or experience before conducting certain analyses or calculations. It can guide the model to train and predict in a direction more consistent with the actual situation. In this step, the diffusion artifact width distribution spectrum serves as a prior constraint condition, providing initial information about the artifacts in the inkjet coding process for the lightweight inverse generative adversarial topology network.

[0086] The lightweight inverse generative adversarial topology network is a network model specifically designed for edge devices. Edge computing is a distributed computing paradigm that locates computing and data storage close to the data source or user, capable of reducing data transmission latency and improving system response speed. Through carefully designed compact network structures and optimized computational operations, the scale of this network model is controlled below 9.8MB, enabling it to perform real-time inference on resource-constrained edge devices. Real-time inference means that the network can quickly process input data during the inkjet coding process and output prediction results in a timely manner, meeting the requirements for real-time calibration of the inkjet printer.

[0087] Loading the diffusion artifact width distribution spectrum into this network, the network learns the features and patterns in these data to generate diffusion artifact prior constraints. For example, the network can learn information such as the changing trends of artifact widths in different regions and the probability distribution of artifact occurrences from the diffusion artifact width distribution spectrum, thereby generating corresponding constraint conditions. These constraint conditions can be understood as a kind of "expectation" of the possible artifact situations in the inkjet coding process. Subsequently, when conducting inkjet coding error analysis and compensation, these "expectations" can be used to more accurately judge the errors in the actual inkjet coding images and make corresponding adjustments.

[0088] The beneficial effects of this step are reflected in multiple aspects. From the perspective of the accuracy of error compensation, the prior constraint of diffusion artifacts provides a more reliable reference basis for subsequent error analysis. In the actual inkjet coding process, due to the interference of various factors, various errors often occur in the inkjet coding image. With the prior constraint, it is possible to more accurately identify whether these errors are caused by artifacts or other factors, thereby improving the pertinence and accuracy of error compensation. For example, when inkjet coding cigarette packages, the inkjet coding image may be affected by various factors such as uneven surface material of the outer shell and minute vibrations of the nozzle. Through the prior constraint of diffusion artifacts, it is possible to more accurately determine which errors are caused by ink diffusion artifacts, and then take more effective compensation measures. From the perspective of real-time performance, the real-time inference ability of the lightweight inverse generative adversarial topology network on edge devices ensures that the inkjet printer can obtain prior constraint information in a timely manner during the inkjet coding process and adjust parameters such as nozzle movement in real time according to this information, greatly improving the working efficiency of the inkjet printer. In addition, the lightweight design of this network model reduces the demand for hardware resources of edge devices, enabling the inkjet printer to achieve efficient image calibration functions even with low-cost hardware configurations, with good economy and practicality.

[0089] Step S1400: Obtain the relative contribution rate of the adsorption effect and the relative contribution rate of the ink potential energy driving effect according to the fluid diffusion field parameter sub-matrix and the substrate material adsorption characteristic data sub-matrix.

[0090] Further, as Figure 4 shown, step S1400 includes:

[0091] Step S1410: Based on the surface viscosity coefficient and potential energy gradient value in the fluid diffusion field parameter sub-matrix, construct a mathematical and physical model for the diffusion process of ink droplets on the substrate surface, and use the finite element analysis method to numerically simulate the hydrodynamic characteristics inside the droplets to obtain a quantitative description function of the ink potential energy driving effect.

[0092] Step S1420: Based on the adsorption decay time constant and saturated adsorption concentration threshold in the substrate material adsorption characteristic data sub-matrix, construct an adsorption kinetic model between the ink droplets and the substrate surface, and use the molecular dynamics simulation method to quantitatively analyze the adsorption process to obtain a quantitative description function of the adsorption effect.

[0093] Step S1430: Normalize the quantitative description function of the ink potential energy driving effect and the quantitative description function of the adsorption effect to obtain the relative contribution rate of the adsorption effect and the relative contribution rate of the ink potential energy driving effect.

[0094] Specifically, the purpose of step S1400 is to obtain the relative contribution rate of the adsorption effect and the relative contribution rate of the ink potential energy driving effect based on the fluid diffusion field parameter sub-matrix and the substrate material adsorption characteristic data sub-matrix, which is crucial for deeply understanding the diffusion mechanism of the ink on the substrate surface and achieving precise inkjet coding process control. In step S1410, the mathematical physics model is a tool for mathematical abstraction and simplification of actual physical phenomena. By establishing such a model, the diffusion process of the ink droplet on the substrate surface can be described in mathematical language. The surface viscosity coefficient and the potential energy gradient value, as the key parameters of the model, directly affect the diffusion behavior of the ink droplet. The finite element analysis method is a numerical calculation method that discretizes the continuous solution domain into a finite number of elements. It can transform complex physical problems into algebraic equations for solution. In this step, by using the finite element analysis method to simulate the hydrodynamic characteristics inside the ink droplet, the distribution of the velocity field, pressure field, etc. inside the droplet can be analyzed in detail, and then a quantitative description function of the ink potential energy driving effect can be obtained. For example, during the simulation process, it can be observed how the fluid inside the ink droplet flows under different combinations of the surface viscosity coefficient and the potential energy gradient value, and what kind of influence this flow has on the diffusion range and speed of the droplet on the substrate surface. Through such simulation and analysis, the obtained quantitative description function can accurately express the relationship between the ink potential energy driving effect and the relevant parameters, providing a quantitative basis for accurately evaluating the influence of the ink potential energy driving effect on the inkjet coding process in the future. This helps to adjust the relevant parameters according to different requirements during the inkjet coding process design stage, optimize the diffusion effect of the ink, and improve the inkjet coding quality.

[0095] In step S1420, the adsorption kinetics model is used to describe the change of the adsorption process of the ink droplets on the substrate surface over time. The adsorption decay time constant reflects the speed of the ink adsorption on the substrate surface, and the saturated adsorption concentration threshold indicates the maximum amount of ink that the substrate material can adsorb. The molecular dynamics simulation method is a computer simulation technology that studies the behavior of matter from the atomic scale. It studies the microscopic mechanism of the adsorption process by simulating the movement and interaction of molecules. In this step, the molecular dynamics simulation method can be used to observe in detail the interaction process between the ink molecules and the molecules on the substrate surface, and how this interaction affects the adsorption amount and adsorption speed of the ink on the substrate surface. For example, through simulation, it can be found that under different adsorption decay time constants and saturated adsorption concentration thresholds, there are differences in the distribution of adsorption sites of ink molecules on the substrate surface, the formation process of the adsorption layer, etc. Through such simulation and analysis, the quantitative description function of the adsorption effect can accurately reflect the relationship between the adsorption process and related parameters, and provides a microscopic basis for in-depth understanding of the influence of the adsorption effect on the inkjet printing process. This is of great significance for selecting the appropriate substrate material and ink combination, as well as optimizing parameters such as drying time in the coding process, which can improve the adhesion of the ink on the substrate surface and the clarity of the coding pattern.

[0096] In step S1430, normalization is a method of converting data of different dimensions or value ranges to a unified scale, which can eliminate the dimensional differences between the data and facilitate comparison and analysis. By normalizing the two quantitative description functions, the ink potential energy driving effect and the adsorption effect can be measured on the same scale, so as to obtain their relative importance in the entire ink diffusion process, that is, the relative contribution rate. For example, after normalization, if it is found that the relative contribution rate of the adsorption effect is high, it means that under the current inkjet conditions, the adsorption effect has a greater impact on the ink diffusion. Then, when optimizing the inkjet process, it is necessary to pay more attention to the adsorption characteristics of the substrate material, select a substrate material with more suitable adsorption performance, or adjust the drying conditions during the inkjet process, etc., to optimize the ink diffusion effect; on the contrary, if the relative contribution rate of the ink potential energy driving effect is high, the parameters related to the ink potential energy, such as the injection pressure of the nozzle, etc., should be adjusted. By determining these two relative contribution rates, the inkjet process parameters can be optimized more targetedly, and the ink diffusion process can be accurately controlled, thereby improving the stability and reliability of the inkjet quality.

[0097] Step S1500, combining the diffusion artifact prior constraint, performing error analysis on the inkjet coding image in the no-load state, and generating a no-load artifact offset error compensation matrix;

[0098] Further, step S1500 includes:

[0099] Step S1510, in combination with the prior constraint of diffusion artifacts, use an inkjet printer to print a preset reference pattern under no-load conditions, and collect the actual inkjet image in the no-load state;

[0100] Specifically, the prior constraint of diffusion artifacts is obtained by loading the diffusion artifact width distribution spectrum into a lightweight inverse generative adversarial topology network. It reflects an expected information about the possible diffusion artifact situations in the no-load state of the current inkjet system, including the estimation of the shape, position, and size of the artifacts. During actual operation, the inkjet printer runs in the no-load state, that is, it does not perform the actual inkjet work on products, but only prints the reference pattern on a specific substrate material according to the preset inkjet parameters. For an inkjet printer applied to the inkjet of cigarette wrapping paper, the substrate material selected at this time can be a paper similar to the material of the cigarette wrapping paper to simulate the actual inkjet environment to the greatest extent. The preset reference pattern is usually carefully designed, including various combinations of shapes and lines, such as a pattern composed of rectangles, circles, and lines of different sizes. These patterns can cover various geometric features that may appear during the inkjet process, facilitating a comprehensive analysis of the inkjet image in the subsequent stage. After the inkjet is completed, use a high-precision image acquisition device, such as an industrial camera, to take pictures of the pattern printed on the substrate material. When collecting the image, it is necessary to ensure the stability of the acquisition environment, avoid factors such as light changes and equipment vibrations from interfering with the image quality, so as to obtain a clear and accurate actual inkjet image. For example, by adjusting parameters such as the focal length, aperture, and exposure time of the camera, ensure that the details of the inkjet pattern in the image can be clearly presented.

[0101] This step has beneficial effects in multiple aspects. From the perspective of error analysis, the actual inkjet printing image collected provides an intuitive data source for determining the errors in the inkjet printing process. By comparing the actual inkjet printing image with the reference pattern, the deviations in the shape, position, size, etc. of the inkjet printing pattern can be clearly observed. These deviations are the errors existing in the inkjet printing process and provide the basic data for subsequent precise error compensation. In the inkjet printing of cigarette wrapping paper, if the position of a certain character in the actual inkjet printing image is offset, the offset amount can be determined through subsequent analysis, and corresponding adjustments can be made. From the perspective of improving the inkjet printing quality, collecting images by combining the diffusion artifact prior constraint can more specifically focus on the error situations related to artifacts. Because the prior constraint provides information about artifacts, when collecting images, the areas that may be affected by artifacts can be specifically observed and recorded, which helps to deeply analyze the influence mechanism of artifacts on the inkjet printing quality, thereby providing a basis for optimizing the inkjet printing process. For example, if the prior constraint shows that artifacts may occur due to ink diffusion in a certain area, this area can be focused on when collecting images. If it is found that the actual inkjet printing image has problems such as blurring or deformation in this area, further analysis can be carried out on what factors cause this situation, and then measures can be taken for improvement. In addition, collecting images in the no-load state can eliminate the interference of the actual product content on the inkjet printing effect, more purely reflect the performance and error situation of the inkjet printer itself, make the subsequent generated no-load artifact offset error compensation matrix more targeted and accurate, ultimately improve the inkjet printing quality, and reduce product defects caused by inkjet printing errors.

[0102] Step S1520: Perform feature point matching on the collected actual inkjet printing image and the reference pattern, and calculate the geometric offset of the inkjet printing contour;

[0103] Furthermore, as Figure 5 shown, step S1520 includes:

[0104] Step S1521: Extract the key feature points of the actual inkjet printing image and the reference pattern, and construct a feature point matching matrix;

[0105] Step S1522: Based on the feature point matching matrix, fit the geometric offset function of the inkjet printing contour to obtain the local geometric offset and the overall geometric offset;

[0106] Step S1523: Combine the local geometric offset and the overall geometric offset to form the geometric offset of the inkjet printing contour.

[0107] Specifically, the core task of step S1520 is to perform feature point matching on the collected actual inkjet printing image and the reference pattern, calculate the geometric offset of the inkjet printing contour, determine the errors in the inkjet printing process, and provide key data support for subsequent error compensation.

[0108] In step S1521, the Scale-Invariant Feature Transform (SIFT) algorithm is used to extract the key feature points of the actual inkjet printing image and the reference pattern, and a feature point matching matrix is constructed. The Scale-Invariant Feature Transform (SIFT) algorithm is a feature extraction technology widely used in the field of computer vision. Its principle is based on feature detection and description in different scale spaces of the image. Gaussian filtering is performed on the image at different scales to construct a Gaussian pyramid, and then Difference of Gaussian (DoG) operations are used to detect the key points in the image. For each key point, the gradient direction histogram of its surrounding neighborhood is calculated to form a feature descriptor. These feature descriptors have scale invariance, rotation invariance, and a certain degree of illumination invariance, and can stably represent the features of the image under different conditions. In inkjet printing image analysis, due to factors such as nozzle jitter, ink diffusion, and differences in the surface characteristics of the substrate material during the inkjet printing process, there are differences in scale, rotation, and illumination between the actual inkjet printing image and the reference pattern. The SIFT algorithm can effectively handle these changes and accurately extract the key feature points in the image. For example, in the inkjet printing on cigarette packaging paper, the characters in the actual inkjet printing image may rotate and scale due to the slight jitter of the nozzle, and at the same time, due to the different textures on the surface of the packaging paper, the light reflection situation will also be different. The SIFT algorithm can extract key feature points such as the corner points and endpoint of the character edge from these complex changes. By extracting the key feature points of the actual inkjet printing image and the reference pattern and constructing a feature point matching matrix, the elements in the matrix represent the pairs of mutually matching feature points in the two images. The beneficial effect of this step is that the application of the SIFT algorithm greatly improves the accuracy and stability of feature point extraction. Accurate feature point extraction is the basis for subsequent accurate calculation of geometric offsets, and can reduce mis-matching caused by image changes. By constructing a feature point matching matrix, a reliable data basis is provided for subsequent calculation of the geometric offset of the inkjet printing contour, making subsequent error analysis and compensation more accurate, helping to improve the inkjet printing quality, and ensuring the accuracy and consistency of the inkjet printing pattern.

[0109] In step S1522, the fitting geometric offset function is obtained by mathematically processing the data in the feature point matching matrix to find a function that describes the geometric transformation relationship between the actual inkjet printing image and the reference pattern. Since the errors in the inkjet printing process may vary at different positions of the inkjet printing contour, it is necessary to calculate the local geometric offset and the overall geometric offset separately. The local geometric offset reflects the offset of each local area of the inkjet printing contour, and it can accurately locate the deviation of specific positions in the inkjet printing pattern. For example, in the brand logo printed on the cigarette packaging paper, the local stroke of a certain letter may be offset due to the momentary jitter of the nozzle. By calculating the local geometric offset, the offset direction and distance of this stroke can be accurately determined. The overall geometric offset describes the overall offset trend of the entire inkjet printing contour relative to the reference pattern, and it reflects the systematic deviation in the inkjet printing process. For example, if the entire brand logo is offset in a certain direction during printing, the overall geometric offset can reflect this overall deviation. By fitting the geometric offset function to obtain the local and overall geometric offsets, the offset situation of the inkjet printing contour can be comprehensively understood. This helps to deeply analyze the source and nature of the inkjet printing errors. The local geometric offset can help to detect abnormal situations in local areas during the inkjet printing process, such as local deviations caused by the failure of individual nozzles, providing a basis for targeted adjustment of the nozzles. The overall geometric offset helps to evaluate the overall performance and stability of the inkjet printer, providing a reference for adjusting the parameters of the inkjet printer, thereby improving the accuracy and stability of the inkjet printing, reducing the deformation and offset of the inkjet printing pattern.

[0110] Step S1523 is the integration of the results of the previous two sub-steps, combining the local and overall offset information to form a complete geometric offset data of the inkjet printing contour. The geometric offset of the inkjet printing contour comprehensively describes the differences in shape and position between the actual inkjet printing image and the reference pattern. It is a comprehensive indicator that contains the offset information of each part of the inkjet printing pattern. For example, in the inkjet printing on the cigarette packaging paper, the geometric offset of the inkjet printing contour not only reflects the offset of individual characters but also the overall offset of the entire inkjet printing pattern (such as the pattern composed of brand logo, production date, etc.) relative to the reference pattern. This complete geometric offset data is crucial for generating the no-load artifact offset error compensation matrix later, as it provides accurate error information for the compensation matrix. With the accurate geometric offset, the compensation amount can be calculated more precisely, making the subsequent inkjet printing error compensation more accurate and effective. When generating the compensation matrix, corresponding compensation strategies can be formulated according to the specific values of the geometric offset for different positions and degrees of offset, thereby effectively reducing the offset and deformation of the inkjet printing pattern, improving the inkjet printing quality, ensuring that the inkjet printing pattern meets the expected design requirements, and enhancing the appearance quality and readability of the product.

[0111] Step S1530: Matrix-represent the geometric offset to generate an offset error compensation matrix for no-load artifacts.

[0112] Further, step S1530 includes:

[0113] Step S1531: Weightedly combine the local geometric offset and the overall geometric offset according to the relative contribution rate of the adsorption effect and the relative contribution rate of the ink potential energy driving effect to obtain a comprehensive geometric offset matrix;

[0114] Step S1532: Normalize and discretize the elements in the comprehensive geometric offset matrix to generate an offset error compensation matrix for no-load artifacts.

[0115] Specifically, the main purpose of step S1530 is to matrix-represent the geometric offset of the inkjet profile to generate an offset error compensation matrix for no-load artifacts, providing key data support for subsequent precise error compensation of the nozzle movement pose.

[0116] In step S1531, the relative contribution rate of the adsorption effect and the relative contribution rate of the ink potential energy driving effect are calculated through the previous steps, which respectively reflect the relative importance of the adsorption effect and the ink potential energy driving effect in the ink diffusion process. During the inkjet coding process, both the adsorption effect and the ink potential energy driving effect will affect the diffusion of the ink, thereby causing the inkjet coding pattern to shift. For example, in the inkjet coding of cigarette wrapping paper, the adsorption ability of wrapping paper with different materials to the ink is different. When the relative contribution rate of the adsorption effect is high, it indicates that the adsorption effect of the wrapping paper on the ink has a greater impact on the shift of the inkjet coding pattern; while the ink potential energy driving effect is related to factors such as the internal pressure and surface tension of the ink. When the relative contribution rate of the ink potential energy driving effect is high, it shows that the potential energy driving effect of the ink itself has a more significant impact on the shift of the inkjet coding pattern. The local geometric shift amount and the overall geometric shift amount respectively describe the shift situation of the inkjet coding contour from different angles. The weighted combination assigns different weights to the local and overall geometric shift amounts according to the relative importance of the adsorption effect and the ink potential energy driving effect, and then performs a combined calculation. For example, if the relative contribution rate of the adsorption effect is 0.6 and the relative contribution rate of the ink potential energy driving effect is 0.4, when calculating the comprehensive geometric shift amount, the weight corresponding to the local geometric shift amount may be 0.6, and the weight corresponding to the overall geometric shift amount is 0.4. Through this weighted calculation, a comprehensive geometric shift amount matrix is obtained. The beneficial effect of this step is that it takes into account the contribution differences of different factors to the shift of the inkjet coding pattern, integrates the local and overall geometric shift amounts through the weighted combination method, so that the comprehensive geometric shift amount matrix can more accurately reflect the comprehensive impact of various factors on the shift of the inkjet coding pattern. Compared with only considering a single factor or a simple combination without weighting, this method can more comprehensively consider the complex situation in the inkjet coding process, provides a more reliable data basis for generating a more accurate no-load artifact shift error compensation matrix in the subsequent stage, helps to improve the accuracy of inkjet coding error compensation, thereby improving the inkjet coding quality, reducing the errors and defects of the inkjet coding pattern, and ensuring that the presentation effect of the inkjet coding pattern on the cigarette wrapping paper is more accurate and clear.

[0117] In step S1532, the normalization process maps the elements in the comprehensive geometric offset matrix to a specific interval, usually the interval [0, 1]. The purpose is to eliminate the dimensional differences between different elements, making each element comparable. During the inkjet coding process, the geometric offsets may have different numerical ranges due to different measurement units or calculation methods. Through the normalization process, these numerical values in different ranges can be unified into a standard interval, facilitating subsequent calculations and analyses. The discretization process converts continuous numerical values into discrete numerical values to facilitate computer processing and storage. When generating the no-load artifact offset error compensation matrix, the number of rows and columns is consistent with the number of rows and columns of the inkjet printer nozzle array because the compensation matrix needs to compensate for the coding errors of each nozzle. Corresponding to the nozzle array can ensure the accuracy and pertinence of the compensation. For example, assuming that the elements in the comprehensive geometric offset matrix represent the offset distances of the inkjet pattern at different positions, the unit of these distances may be pixels. Through the normalization process, these offset distances are converted into relative values, enabling the offset situations at different positions to be compared on the same scale. After further discretization, these relative values are converted into discrete numerical values, such as integers, which are convenient for subsequent calculations in the compensation algorithm. The generated no-load artifact offset error compensation matrix can be directly used to adjust the movement of the print head or the ink jetting parameters to compensate for the errors during the inkjet coding process. For example, according to the element values in the compensation matrix, the jetting time, jetting volume of each nozzle, or the movement trajectory of the print head can be adjusted to correct the offset of the inkjet pattern. The beneficial effect of this step is that the normalization and discretization processes make the comprehensive geometric offset matrix more suitable for subsequent calculations and applications. The no-load artifact offset error compensation matrix generated through these processes can be more effectively used for inkjet coding error compensation, improving the accuracy and efficiency of the compensation. The design consistent with the number of rows and columns of the inkjet printer nozzle array ensures that the compensation matrix can accurately correct the coding errors of each nozzle, thereby significantly improving the inkjet coding quality, reducing the errors and defects of the inkjet pattern, and ensuring that the pattern on the cigarette wrapping paper is clear and accurate, meeting the quality requirements of the product.

[0118] In step S1600, a non-linear mapping relationship is established between the elements of the no-load artifact offset error compensation matrix and the composite parameter matrix to generate a compensation matrix dynamic weight distribution function.

[0119] Furthermore, step S1600 includes:

[0120] In step S1610, the relative contribution rate of the adsorption effect, the relative contribution rate of the ink potential energy driving effect, the fluid diffusion field parameter sub-matrix, the substrate material adsorption characteristic sub-matrix, and the print head movement trajectory spatial characteristic sub-matrix are used as input parameters to construct a composite input vector;

[0121] Step S1620: Learn the non - linear mapping relationship between the composite input vector and each element of the no - load artifact offset error compensation matrix through a neural network method;

[0122] Step S1630: Generate a compensation matrix dynamic weight distribution function according to the learned non - linear mapping relationship.

[0123] Specifically, step S1600 aims to establish the non - linear mapping relationship between each element of the no - load artifact offset error compensation matrix and the composite parameter matrix, and generate a compensation matrix dynamic weight distribution function, so as to realize the adaptive adjustment of the inkjet coding error compensation and improve the inkjet coding accuracy.

[0124] In step S1610, the relative contribution rate of the adsorption effect and the relative contribution rate of the ink potential energy driving effect reflect the relative importance of the adsorption and the ink potential energy driving effects in the ink diffusion process; the fluid diffusion field parameter sub - matrix includes the surface viscosity coefficient and the potential gradient value, etc., which reflects the hydrodynamic characteristics of the ink diffusion; the substrate material adsorption characteristic sub - matrix describes the adsorption characteristics of the substrate material to the ink through the adsorption decay time constant and the saturated adsorption concentration threshold; the nozzle movement trajectory space characteristic sub - matrix characterizes the stability and accuracy of the nozzle movement. In the actual application scenario of cigarette packaging paper inkjet coding, different packaging paper materials have different adsorption properties for the ink. For example, for paper packaging and plastic film packaging, there will be differences in their adsorption decay time constant and saturated adsorption concentration threshold, which affect the relative contribution rate of the adsorption effect. And the vibration and displacement of the nozzle during the inkjet coding process are reflected by the nozzle movement trajectory space characteristic sub - matrix. Combining these parameters to construct a composite input vector can comprehensively reflect the key factors related to ink diffusion, substrate material characteristics, and nozzle movement in the inkjet coding process. By constructing the composite input vector, multiple factors affecting the inkjet coding error compensation are integrated into one vector, providing comprehensive and structured data for the subsequent learning of the neural network, enabling the neural network to simultaneously consider the comprehensive influence of multiple factors on the inkjet coding error compensation, helping to improve the accuracy of the neural network learning mapping relationship, and further laying a foundation for generating a more accurate compensation matrix dynamic weight distribution function and improving the accuracy of the inkjet coding error compensation.

[0125] In step S1620, the neural network is a method based on representation learning of data. It consists of a large number of nodes (neurons) and the edges connecting these nodes, and learns the patterns and relationships in the data by adjusting the connection weights between the nodes. In this step, the input of the neural network is the composite input vector, and the output is each element of the no-load artifact offset error compensation matrix. During the training process, the neural network continuously adjusts its own weights to minimize the error between the predicted output and the elements of the actual no-load artifact offset error compensation matrix. For example, in the inkjet coding of cigarette wrapping paper, when the movement trajectory of the print head changes, the spatial feature sub-matrix of the print head movement trajectory in the composite input vector will change accordingly. The neural network will learn how this change affects the inkjet coding error compensation, and thus adjust the elements of the output compensation matrix. Due to the complex relationships among various factors during the inkjet coding process, showing non-linear characteristics, traditional linear methods are difficult to accurately describe these relationships. However, the powerful non-linear fitting ability of the neural network can effectively capture the complex non-linear mapping relationship between the composite input vector and each element of the no-load artifact offset error compensation matrix. By learning this relationship, the neural network can accurately predict the corresponding compensation matrix elements according to the changes in the input parameters, providing an accurate basis for generating the dynamic weight allocation function in the subsequent stage, so as to achieve precise control of the inkjet coding error compensation, improve the inkjet coding quality, and reduce the errors and defects of the inkjet coding pattern.

[0126] In step S1630, the compensation matrix dynamic weight allocation function is a function that can adaptively adjust the no-load artifact offset error compensation matrix according to the real-time changes of the hydrodynamic parameters and material property parameters during the inkjet printing process. During the inkjet coding process of cigarette wrapping paper, as the wrapping paper material changes (such as the adsorption characteristics change due to the water absorption difference of papers in different batches), or the movement accuracy of the print head changes after long-term operation, the hydrodynamic parameters and material property parameters will change in real time. At this time, the compensation matrix dynamic weight allocation function will adjust the weights of each element in the no-load artifact offset error compensation matrix according to these changes. For example, if the adsorption effect in a certain area is enhanced, this function will correspondingly increase the weights of the compensation matrix elements related to the adsorption effect, making the compensation more focused on correcting the inkjet coding errors caused by adsorption. By generating the compensation matrix dynamic weight allocation function, it is possible to dynamically adjust the no-load artifact offset error compensation matrix under different inkjet printing conditions, ensuring the accuracy of the inkjet coding error compensation. This helps to improve the adaptability of the inkjet printer in various complex situations. No matter what kind of cigarette wrapping paper material is encountered, or what kind of minor changes occur to the print head, accurate inkjet coding error compensation can be achieved, thus improving the stability of the inkjet coding quality, reducing the rejection rate, and lowering the production cost.

[0127] Step S2000: Construct a multi-scale gradient feature pyramid according to the composite parameter matrix. The multi-scale gradient feature pyramid includes a bottom layer, a middle layer, and a top layer. Tensor fusion is performed on the features of the bottom, middle, and top layers of the multi-scale gradient feature pyramid to generate a feature correlation heat map covering different abstraction levels. In combination with the no-load artifact offset error compensation matrix, the feature correlation heat map is corrected, and a dynamic prediction model for ink diffusion behavior is established to generate a real-time diffusion trajectory prediction result.

[0128] Further, step S2000 includes:

[0129] Step S2100: Construct a multi-scale gradient feature pyramid according to the composite parameter matrix. The multi-scale gradient feature pyramid includes a bottom layer, a middle layer, and a top layer.

[0130] Further, as Figure 6 shown, step S2100 includes:

[0131] Step S2110: Based on the fluid diffusion field parameter sub-matrix, extract the multi-scale gradient invariant feature clusters formed during the ink diffusion process on the substrate surface, and construct the bottom layer of the multi-scale gradient feature pyramid. The multi-scale gradient invariant feature clusters include 64-dimensional histograms of oriented gradients.

[0132] Step S2120: Based on the substrate material adsorption characteristic data sub-matrix, capture the microscopic porosity distribution mapping on the substrate material surface, and construct the middle layer of the multi-scale gradient feature pyramid.

[0133] Step S2130: Based on the nozzle movement trajectory spatial feature sub-matrix, analyze the second-order derivative fluctuation characteristics of the nozzle movement trajectory, and construct the top layer of the multi-scale gradient feature pyramid.

[0134] Specifically, the purpose of step S2100 is to construct a multi-scale gradient feature pyramid according to the composite parameter matrix, extract features related to ink diffusion, substrate material properties, and nozzle movement from different levels, and provide a basis for subsequent analysis and prediction of ink diffusion behavior.

[0135] In step S2110, the 64-dimensional histogram of oriented gradients in the multi-scale gradient invariant feature cluster is used to characterize the scale-space structural features of the ink diffusion profile. The histogram of oriented gradients forms a histogram by calculating the gradient direction and amplitude of each pixel point in the image and statistically analyzing the gradient direction distribution within a certain area. By calculating at different scales, multi-scale histograms of oriented gradients are obtained, which can reflect the structural features of the ink diffusion profile at different sizes and resolutions. Taking the inkjet printing on cigarette wrapping paper as an example, in the initial stage of ink diffusion, the histogram of oriented gradients at a small scale can capture the subtle changes at the edge of the ink droplets; as the diffusion progresses, the histogram of oriented gradients at a large scale can reflect the overall diffusion trend. By extracting these multi-scale gradient invariant feature clusters to construct the bottom layer of the pyramid, basic feature information of ink diffusion is provided for subsequent analysis. These features can reflect the structural changes during the early and continuous processes of ink diffusion, helping to understand the dynamic process of ink diffusion more deeply. At the same time, it provides underlying data support for subsequent feature fusion and analysis, enabling a comprehensive analysis of ink diffusion behavior by gradually combining other factors starting from the most basic diffusion profile structure, thus improving the accuracy and comprehensiveness of the analysis of ink diffusion behavior.

[0136] In step S2120, the microscopic porosity distribution on the surface of the substrate material affects the adsorption behavior of the ink on its surface. By introducing a fractal dimension descriptor to establish a correlation model between the microscopic structural features of the substrate material and the ink adsorption behavior, the discrimination ability of the middle-layer features can be improved. The fractal dimension descriptor is used to quantify the complexity of the microscopic structure of the material, and its value reflects the irregularity and self-similarity of the microscopic porosity distribution. In the application of cigarette wrapping paper, the microscopic porosities of wrapping papers made of different materials are different, and so are their fractal dimensions. For example, the paper wrapping paper has a relatively large and complex microscopic porosity distribution and a relatively high fractal dimension; while the plastic film wrapping paper has a relatively small microscopic porosity and a lower fractal dimension. By capturing the mapping of the microscopic porosity distribution to construct the middle layer of the pyramid, the microscopic characteristics of the substrate material can be related to the ink adsorption behavior. This helps to analyze the influence mechanism of different substrate materials on ink diffusion. For example, it can study how the microscopic porosity affects the adsorption rate and adsorption amount of the ink, and thus affects the ink diffusion process. At the same time, when the middle-layer features are fused with the bottom-layer and top-layer features, the interaction of multiple factors during the ink diffusion process can be more comprehensively reflected, improving the accuracy of predicting the ink diffusion behavior.

[0137] In step S2130, the second-order derivative fluctuation feature of the nozzle movement trajectory reflects the acceleration change of the nozzle movement and can characterize the instability of the nozzle movement. The short-time Fourier transform is used to extract the time-frequency domain features of the nozzle vibration displacement signal, and then the chaotic theory is used to analyze the nonlinear dynamic behavior of the nozzle movement trajectory to construct the top-level feature mapping. The short-time Fourier transform can convert the time-domain signal into a time-frequency domain signal and display the energy distribution of the signal at different times and frequencies. The chaotic theory is used to study the seemingly random but actually regular behavior in nonlinear systems. When jet coding the cigarette wrapping paper, the vibration and displacement of the nozzle will cause changes in the coding position and pattern quality. If the nozzle movement is unstable, the second-order derivative fluctuation will be large. By analyzing these fluctuation features to construct the top-level feature mapping, the nozzle movement state can be monitored in real time. For example, when abnormal vibration occurs during the jet coding process of the nozzle, the top-level features will change accordingly, which helps to detect nozzle failures or abnormal conditions in a timely manner. After the top-level features are fused with the bottom-level and middle-level features, various factors such as ink diffusion, substrate material adsorption, and nozzle movement can be comprehensively considered to more accurately predict the ink diffusion behavior and provide an important basis for real-time adjustment of jet coding parameters and optimization of jet coding quality.

[0138] Step S2200: Tensor-fuse the bottom-level, middle-level, and top-level features of the multi-scale gradient feature pyramid to generate a feature correlation heat map covering different abstraction levels;

[0139] Furthermore, step S2200 includes:

[0140] Step S2210: Fuse the multi-scale gradient invariant feature cluster at the bottom level with the microscopic porosity distribution mapping at the middle level to obtain the local ink diffusion behavior features;

[0141] Step S2220: Fuse the second-order derivative fluctuation feature of the nozzle movement trajectory at the top level with the local ink diffusion behavior features to obtain the ink distribution and nozzle movement correlation matrix;

[0142] Step S2230: Thermodynamically encode the ink distribution and nozzle movement correlation matrix to generate a feature correlation heat map reflecting the correlation strength between the local features at the bottom level and the global features at the top level.

[0143] Specifically, the purpose of step S2200 is to tensor-fuse the bottom-level, middle-level, and top-level features of the multi-scale gradient feature pyramid to generate a feature correlation heat map covering different abstraction levels, so as to reveal the internal relationship between ink diffusion, substrate material properties, and nozzle movement, and provide a basis for subsequent accurate analysis and prediction of ink diffusion behavior.

[0144] In step S2210, the recursive tensor product algorithm is adopted to fuse the multi-scale gradient invariant feature clusters at the bottom layer with the mapping of the micro-porosity distribution in the middle layer, so as to obtain the local ink diffusion behavior characteristics. The recursive tensor product algorithm is a method for processing tensor operations. By gradually calculating the products between tensors, it can effectively fuse information in different dimensions. The multi-scale gradient invariant feature clusters contain 64-dimensional histograms of oriented gradients, which depict the scale-space structural characteristics of the ink diffusion contour and reflect the contour changes at different scales during the ink diffusion process. The mapping of the micro-porosity distribution in the middle layer, on the other hand, reflects the influence of the surface microstructure of the substrate material on ink adsorption. Taking the inkjet printing on cigarette wrapping paper as an example, during the inkjet printing process, the ink diffuses on the surface of the wrapping paper. The bottom layer features can capture the detailed changes in the ink diffusion contour, such as the performance of the serrated structure at the edge of the ink droplet at different scales. The middle layer features reflect how the micro-porosity of the wrapping paper affects ink adsorption and thus the initiation and continuation of diffusion. By fusing these two layers of features through the recursive tensor product algorithm, the diffusion characteristics of the ink itself and the influence of the substrate material can be comprehensively considered to obtain the local ink diffusion behavior characteristics. Such fused features can more comprehensively describe the ink diffusion behavior in the local area, providing richer information for subsequent analysis. It helps to accurately analyze the differences in ink diffusion on different substrate material regions. For example, in different parts of the cigarette wrapping paper, due to different micro-porosities, the ink diffusion speed and morphology will be different. The local ink diffusion behavior characteristics can accurately capture these differences, thus providing a basis for precisely controlling ink diffusion and improving the quality and stability of the inkjet printing pattern.

[0145] In step S2220, the high-order singular value decomposition algorithm is used to fuse the second-order derivative fluctuation characteristics of the top-layer nozzle movement trajectory and the local ink diffusion behavior characteristics, obtaining the correlation matrix between the ink distribution and the nozzle movement. The high-order singular value decomposition algorithm is an extension of the traditional singular value decomposition, capable of processing multi-dimensional tensor data and extracting key information from the data. The second-order derivative fluctuation characteristics of the top-layer nozzle movement trajectory reflect the acceleration change of the nozzle movement, characterizing the instability of the nozzle movement; the local ink diffusion behavior characteristics describe the diffusion characteristics of the ink in the local area. During the inkjet coding process of cigarette wrapping paper, the instability of the nozzle movement will cause changes in the inkjet position and amount, thus affecting the distribution of the ink on the wrapping paper. For example, when the acceleration of the nozzle movement suddenly changes, the inkjet direction and speed will change, resulting in a deviation in the distribution of the ink on the wrapping paper. By fusing these two characteristics through the high-order singular value decomposition algorithm, a correlation mapping between the global ink distribution pattern and the nozzle movement pattern can be established, obtaining the correlation matrix between the ink distribution and the nozzle movement. This matrix can clearly show how the nozzle movement state affects the ink distribution, providing a powerful tool for analyzing the error sources in the inkjet coding process. Through the analysis of the correlation matrix, the impact of abnormal nozzle movement on the ink distribution can be detected in a timely manner, and then the nozzle movement parameters can be adjusted to optimize the inkjet coding process, reduce the inkjet coding quality problems caused by the instability of the nozzle movement, and improve the accuracy and consistency of the inkjet coding.

[0146] In step S2230, thermodynamic coding is performed on the correlation matrix between the ink distribution and the nozzle movement, generating a feature correlation degree heat map reflecting the correlation strength between the bottom-layer local feature distribution and the top-layer global features. Among them, the feature correlation degree heat map adopts an adaptive color mapping scheme, enhancing the visual prominence effect of the key areas of ink diffusion by dynamically adjusting the color saturation and brightness gradient. Thermodynamic coding is a method of converting data into an intuitively representable form. By coding the correlation matrix between the ink distribution and the nozzle movement, the data information in the matrix can be transformed into a visual heat map. In the scenario of inkjet coding of cigarette wrapping paper, the heat map can intuitively display the correlation strength between the ink distribution and the nozzle movement in different regions. For example, on the heat map, the darker or more vivid the color of a region, the greater the influence of the nozzle movement on the ink distribution in that region. The adaptive color mapping scheme dynamically adjusts the color saturation and brightness gradient according to the characteristics of the data, making the key areas of ink diffusion (such as the ink concentration area, diffusion edge, etc.) more prominent in the heat map.

[0147] This visual feature correlation heat map has beneficial effects in many aspects. On the one hand, it provides an intuitive analysis tool for operators. By observing the heat map, they can quickly understand the relationship between ink distribution and nozzle movement during the inkjet coding process and discover potential problem areas, such as the correlation between abnormal ink distribution and unstable nozzle movement in certain areas. On the other hand, for the subsequent analysis and prediction of ink diffusion behavior, the heat map provides a clear data visualization basis, which helps researchers analyze the interaction between different factors more deeply and provides strong support for optimizing the inkjet coding process and improving the inkjet coding quality. At the same time, the heat map is also convenient for comparison and combined analysis with other data, further improving the understanding and control ability of the inkjet coding process.

[0148] Step S2300: Combine the no-load artifact offset error compensation matrix to correct the feature correlation heat map, establish a dynamic prediction model for ink diffusion behavior, and generate a real-time diffusion trajectory prediction result.

[0149] The purpose of step S2300 is to combine the no-load artifact offset error compensation matrix to correct the feature correlation heat map, establish a dynamic prediction model for ink diffusion behavior, and generate a real-time diffusion trajectory prediction result for refined control of the inkjet coding profile, so as to achieve precise control and quality improvement of the inkjet coding process.

[0150] Furthermore, step S2300 includes:

[0151] Step S2310: Construct a kinetic evolution model for ink diffusion behavior during the inkjet coding process, introduce the no-load artifact offset error compensation matrix into the kinetic evolution model for ink diffusion behavior, establish an ink diffusion offset correction equation, and perform offset correction on the feature correlation heat map to obtain a sequence of corrected feature correlation heat maps;

[0152] Specifically, the kinetic evolution model is based on the fluid diffusion field parameter sub-matrix, the substrate material adsorption characteristic data sub-matrix, and the nozzle movement trajectory spatial characteristic sub-matrix, and uses the finite element analysis method to numerically simulate the ink diffusion process. The finite element analysis method is a numerical calculation method that discretizes the continuous solution domain into a finite number of elements and can effectively handle complex physical problems. In cigarette wrapper inkjet coding, this model comprehensively considers the hydrodynamic characteristics of the ink, the adsorption effect of the substrate material on the ink, and the influence of nozzle movement on ink jetting. The no-load artifact offset error compensation matrix is a data matrix obtained from the previous steps for compensating the coding error. By introducing it into the kinetic evolution model, an ink diffusion offset correction equation is established, and the mapping relationship between the no-load artifact offset error compensation matrix and the feature correlation degree heat map is fitted using the least squares method, so as to adaptively adjust the position and shape of the ink diffusion region in the heat map. For example, if during the cigarette wrapper inkjet coding process, the ink diffusion region shows an offset due to nozzle movement or ink diffusion characteristics, the ink diffusion offset correction equation can accurately adjust the ink diffusion region in the heat map according to the information in the compensation matrix. This correction can improve the accuracy of the feature correlation degree heat map, provide a more reliable data basis for subsequent analysis and prediction. The corrected heat map sequence can more realistically reflect the actual situation of ink diffusion, help accurately analyze the ink diffusion behavior, timely detect potential inkjet coding quality problems, provide an accurate basis for optimizing the inkjet coding process, and further improve the stability and consistency of the inkjet coding quality.

[0153] Step S2320: Real-time collect the viscosity response curve of the ink droplets on the substrate surface, and combine the relative contribution rate of the adsorption effect and the relative contribution rate of the ink potential energy driving effect to adaptively correct the kinetic evolution model of the ink diffusion behavior, and obtain the corrected kinetic evolution model;

[0154] Specifically, by deploying a microfluidic sensor array, the viscosity response curve of the ink droplet on the substrate surface is collected in real time, and the microfluidic sensor array can monitor the viscosity change of the ink droplet on the substrate surface in real time. The relative contribution rate of the adsorption effect and the relative contribution rate of the ink potential energy driving effect reflect the relative importance of the adsorption and the ink potential energy driving effect in the ink diffusion process. During the inkjet coding process of cigarette wrapping paper, as the coding progresses, the viscosity of the ink may change due to environmental factors or interactions with the substrate material. For example, when the humidity of the wrapping paper changes, the viscosity of the ink will change, thereby affecting the diffusion of the ink. The contact angle boundary condition of the solid-liquid interface in the kinetic evolution model is dynamically updated according to the relative contribution rate of the adsorption effect, the volume force load distribution is corrected according to the relative contribution rate of the ink potential energy driving effect, and the adsorption-diffusion coupling field theory is introduced to establish a cross-scale multi-field collaborative analysis framework. This can make the kinetic evolution model more accurately describe the ink diffusion process, taking into account the real-time changes of various factors. Through the adaptive correction of the model, the diffusion state of the ink at different times can be predicted more accurately, providing a more accurate model basis for the subsequent prediction of the diffusion trajectory, helping to improve the control accuracy of the ink diffusion during the inkjet coding process, reduce the error of the inkjet pattern, and improve the inkjet quality.

[0155] Step S2330, based on the corrected kinetic evolution model, construct a dynamic prediction model for the ink diffusion behavior, and input the corrected sequence of the feature correlation degree heat maps into the dynamic prediction model for the ink diffusion behavior to predict the probability distribution cloud map of the ink diffusion profile within a predetermined future time;

[0156] Specifically, the dynamic prediction model of ink diffusion behavior uses a long short-term memory neural network to perform spatio-temporal modeling on the evolution process of the heat map. By designing a reasonable network structure and loss function, it extracts the spatio-temporal correlation patterns contained in the ink diffusion process to achieve the prediction of the future ink diffusion trend. The long short-term memory neural network can effectively process time series data and remember long-term dependencies. In the inkjet coding of cigarette wrapping paper, by inputting the corrected heat map sequence of feature correlation degrees, the model can learn the variation rules of ink diffusion in time and space. For example, the model can learn the speed, direction of ink diffusion at different times, and the variation patterns affected by the movement of the nozzle and the substrate material. Using the previously obtained heat map sequence of feature correlation degrees for training can enable the model to better fit the actual ink diffusion situation. The output of the model is a probability distribution cloud map of the ink diffusion contour within a predetermined future time, such as 3 ms. This cloud map shows the possible diffusion range and probability distribution of the ink in the short future. By accurately predicting the probability distribution cloud map of the ink diffusion contour, the trend of ink diffusion can be understood in advance, providing a basis for real-time adjustment during the inkjet coding process, such as adjusting the movement trajectory of the nozzle or the ink injection volume in advance to ensure the accuracy of the inkjet pattern, reduce the inkjet error, and improve the inkjet quality and production efficiency.

[0157] Step S2340: Perform morphological analysis on the probability distribution cloud map of the ink diffusion contour, extract the key deformation feature points on the ink diffusion edge, and construct an ink diffusion edge deformation prediction grid containing n1 nodes;

[0158] Specifically, a multi-scale curvature analysis algorithm is used to perform morphological analysis on the probability distribution cloud map of the ink diffusion profile, extract the key deformation feature points of the ink diffusion edge, and adaptively determine the number and distribution of the key feature points. At the same time, the B-spline curve fitting method is introduced to generate a prediction grid with the key feature points as control vertices and optimize the topological structure of the grid, and adaptive grid encryption is implemented in the rapid ink diffusion area. The multi-scale curvature analysis algorithm can accurately capture the local and global shape changes of the curve by calculating the curvature of the curve at different scales, so as to extract the key deformation feature points of the ink diffusion edge. In the inkjet coding of cigarette wrapping paper, the shape change of the ink diffusion edge reflects the dynamic process of diffusion. By extracting the key deformation feature points, the boundary change of the ink diffusion can be better described. The B-spline curve fitting method can generate a smooth curve with these key feature points as control vertices to construct a prediction grid. Adaptive grid encryption is to increase the density of the grid in the rapid ink diffusion area according to the speed and trend of the ink diffusion, so as to more accurately describe the diffusion situation in this area. For example, on the cigarette wrapping paper, if the ink diffusion speed in a certain area is fast, grid encryption can more accurately track the front of the ink diffusion and improve the prediction accuracy of the diffusion profile. This method helps to more accurately predict the boundary and shape change of the ink diffusion, provides accurate boundary information for the subsequent generation of high-precision real-time diffusion trajectory prediction results, improves the ability of fine control of the inkjet coding profile, ensures that the edge of the inkjet coding pattern is clear and accurate, and meets the fine control requirements of the inkjet coding quality.

[0159] Step S2350: Integrate the probability distribution cloud map of the ink diffusion profile and the deformation prediction grid of the ink diffusion edge to generate a real-time diffusion trajectory prediction result for fine control of the inkjet coding profile.

[0160] Specifically, the graph cut algorithm is used to perform adaptive threshold segmentation on the probability distribution cloud map of the ink diffusion contour, and the main ink diffusion contour is extracted. On this basis, the ink diffusion edge deformation prediction grid is mapped to the main ink diffusion contour, and the grid deformation and contour deformation are synchronized through affine transformation to obtain a high-precision real-time diffusion trajectory prediction result. The spatial resolution of the prediction result is optimized to 0.01 pixels, meeting the refined control requirements for the inkjet printing quality. The graph cut algorithm is an image segmentation technology that separates the target area from the background by finding the optimal segmentation boundary in the image, realizing the extraction of the main ink diffusion contour. In cigarette wrapping paper inkjet printing, accurately extracting the main ink diffusion contour is crucial for controlling the shape of the inkjet printing pattern. Mapping the previously generated prediction grid to the main contour and synchronizing the grid deformation and contour deformation through affine transformation can accurately simulate the actual trajectory of ink diffusion. Affine transformation is a linear transformation that can maintain the "linearity" and "parallelism" of the graph. By adjusting the transformation parameters, the grid can accurately deform as the ink diffusion contour changes. For example, when the ink diffusion contour changes, the grid can be adjusted synchronously to accurately reflect the real-time situation of ink diffusion. The high-precision real-time diffusion trajectory prediction result obtained through these operations, with the spatial resolution optimized to 0.01 pixels, can provide accurate control information for the inkjet printer, enabling the inkjet printer to adjust the movement of the print head and the ink injection in real time according to the prediction result, ensuring the accuracy and quality of the inkjet printing pattern, meeting the strict requirements for the refined control of the pattern in cigarette wrapping paper inkjet printing, reducing the deformation and error of the inkjet printing pattern, and improving the appearance quality and readability of the product.

[0161] Step S3000: Generate an artifact compensation control signal according to the compensation matrix dynamic weight distribution function and the real-time diffusion trajectory prediction result, and couple the artifact compensation control signal to the print head drive circuit to perform dynamic feedback control on the movement pose of the print head.

[0162] Furthermore, step S3000 includes:

[0163] Step S3100: Obtain a pre-estimated value of the inkjet printing error according to the real-time diffusion trajectory prediction result and the compensation matrix dynamic weight distribution function.

[0164] Furthermore, step S3100 includes:

[0165] Step S3110: Extract the curvature data of the ink diffusion edge curve according to the real-time diffusion trajectory prediction result, and construct a matrix for describing the shape of the ink diffusion contour.

[0166] Step S3120: Perform feature matching between the matrix for describing the shape of the ink diffusion contour and the preset inkjet printing pattern, calculate the Euclidean distance between the ink diffusion shape and the geometric contour of the inkjet printing pattern, and obtain a shape matching degree index.

[0167] Step S3130: Input the morphological matching degree index into the compensation matrix dynamic weight distribution function, superimpose the weight coefficients corresponding to the fluid diffusion field parameter sub-matrix, the substrate material adsorption characteristic data sub-matrix, and the nozzle movement trajectory spatial feature sub-matrix, calculate the spatial geometric offset compensation amount, and generate a predicted value of the inkjet coding error.

[0168] Specifically, in step S3110, the real-time diffusion trajectory prediction result contains the real-time information of the ink diffusion on the cigarette wrapper. This information is obtained through previous steps such as constructing a multi-scale gradient feature pyramid, performing feature fusion, and establishing a dynamic prediction model. The curvature data of the ink diffusion edge curve can reflect the bending degree and change trend of the ink diffusion edge, which is crucial for describing the morphology of the ink diffusion profile. For example, when inkjet coding on a cigarette wrapper, the ink diffusion edge may show different degrees of bending, and the curvature data can quantify these bending situations. By extracting this curvature data, a morphological description matrix of the ink diffusion profile is constructed. The elements in the matrix can be curvature values or curvature-related parameters at different positions. Such a matrix can present the morphological information of the ink diffusion profile in a structured data form, facilitating subsequent comparison and analysis with the preset printing pattern. The beneficial effect of this step is that by accurately extracting the curvature data to construct the description matrix, the morphological characteristics of the ink diffusion profile can be captured more accurately. Compared with simple profile description methods, the matrix description based on curvature data can more precisely reflect the changes in the ink diffusion edge. When performing feature matching with the preset printing pattern subsequently, it can provide more accurate information, improve the accuracy of the matching, and thus more accurately evaluate the inkjet coding error, providing more reliable data support for subsequent compensation operations, helping to improve the quality of the inkjet coding pattern and reduce inkjet coding defects caused by the inconsistent ink diffusion morphology with the expectation.

[0169] In step S3120, feature matching is a commonly used technique in the field of computer vision for comparing the similarity between two objects or images. In this step, by performing feature matching between the ink diffusion contour shape description matrix and the preset printing pattern, the degree of difference between the actual shape of the current ink diffusion and the ideal printing pattern can be determined. The Euclidean distance is a commonly used method for measuring the distance between two vectors. When calculating the Euclidean distance between the ink diffusion shape and the geometric contour of the printing pattern, the ink diffusion contour shape description matrix and the geometric contour information of the printing pattern are converted into vector forms, and the difference between the two is quantified by calculating the Euclidean distance between the vectors. For example, assume that the preset printing pattern is a regular rectangle, while the contour formed by the actual ink diffusion presents an irregular shape due to various factors (such as uneven ink diffusion, nozzle movement deviation, etc.). By calculating the Euclidean distance between the two, a numerical value can be obtained. The smaller this value, the more similar the ink diffusion shape is to the geometric contour of the printing pattern, that is, the higher the shape matching degree; conversely, the larger the value, the greater the difference. The obtained shape matching degree index is a quantified numerical value used to intuitively evaluate the degree of conformity between the ink diffusion shape and the expected pattern during the inkjet coding process. The beneficial effect of this step is that the shape matching degree index provides a quantified standard for evaluating the inkjet coding error. Through this index, it is possible to clearly understand whether the ink diffusion during the current inkjet coding process meets the expectations, thereby judging the quality of the inkjet coding. In practical applications, when the shape matching degree index exceeds a certain threshold, problems existing in the inkjet coding process, such as nozzle blockage, abnormal ink supply, etc., can be detected in a timely manner, which helps to quickly locate the root cause of the problem, take corresponding measures for adjustment and improvement, ensure the stability of the inkjet coding quality, reduce the scrap rate, and improve production efficiency.

[0170] In step S3130, the compensation matrix dynamic weight distribution function is established based on the previous steps. It can adaptively adjust the compensation matrix according to the real-time changes of hydrodynamic parameters and material property parameters during the inkjet printing process. The fluid diffusion field parameter sub-matrix, the substrate material adsorption characteristic data sub-matrix, and the nozzle movement trajectory spatial feature sub-matrix respectively contain the key information affecting ink diffusion, the effect of the substrate material on the ink, and the nozzle movement state. The morphology matching degree index is input into the compensation matrix dynamic weight distribution function, and calculations are performed in combination with the weight coefficients corresponding to the above three sub-matrices. For example, when the morphology matching degree index is low, indicating a large difference between the ink diffusion morphology and the printed pattern, according to the calculation rules of the compensation matrix dynamic weight distribution function, the adjustment amplitude of the spatial geometric offset compensation amount will be increased. The spatial geometric offset compensation amount calculated in this way takes into account multiple factors affecting the coding error and can more accurately reflect the error amount that needs to be compensated during the actual coding process, thereby generating a coding error prediction value. The beneficial effect of this step is that the coding error prediction value calculated by comprehensively considering multiple factors is more accurate and reliable. By analyzing and integrating multiple key factors, the reasons for generating errors during the coding process can be captured more comprehensively, so as to more accurately estimate the error amount. The accurate error prediction value provides an accurate basis for generating the artifact compensation control signal in the subsequent stage, helps to achieve more precise coding error compensation, effectively improves the coding quality, ensures that the presentation effect of the coded pattern on the cigarette wrapping paper meets the design requirements, reduces product appearance defects caused by coding errors, and improves the overall quality and market competitiveness of the product.

[0171] Step S3200: Based on the coding error prediction value, in combination with the no-load artifact offset error compensation matrix, generate an artifact compensation control signal;

[0172] Specifically, step S3200 aims to generate an artifact compensation control signal based on the coding error prediction value, in combination with the no-load artifact offset error compensation matrix, so as to provide a control instruction for adjusting the nozzle movement pose to compensate for the coding error in the subsequent stage. First, the coding error prediction value is fused with the no-load artifact offset error compensation matrix to construct a composite error compensation tensor. The coding error prediction value reflects the deviation estimation from the preset pattern due to factors such as ink diffusion during the current coding process, while the no-load artifact offset error compensation matrix is a data set for compensating the possible artifact offset error during coding in the no-load state. The fusion of the two to construct a composite error compensation tensor can comprehensively consider error factors in different situations and more comprehensively describe the coding error situation. For example, when coding on a cigarette wrapping paper, the coding error prediction value may show that the pattern is locally blurred due to ink diffusion, while the no-load artifact offset error compensation matrix records the offset information that may be generated due to the slight jitter of the nozzle in the no-load state. Combining the two can more accurately grasp the overall error.

[0173] Next, according to the numerical magnitudes of different elements in the composite error compensation tensor, determine the weight allocation scheme for each control node in the nozzle drive circuit, and generate a weight allocation matrix. The elements in the composite error compensation tensor contain error information of different positions and types. By analyzing these elements, determine the contribution degree of each control node to the compensated error, so as to allocate corresponding weights to each control node of the nozzle drive circuit. For example, if at a certain position, the coding error is mainly caused by the horizontal offset of the nozzle, then in the weight allocation matrix, the weights of the nodes related to the horizontal movement control of the nozzle will be relatively large. Such a weight allocation scheme can adjust the control signal of the nozzle drive circuit specifically according to the specific error situation.

[0174] Finally, use the weight allocation matrix to modulate the standardized electronic control pulse signal to synthesize a composite pulse sequence injected with high-frequency micro-perturbations, forming an artifact compensation control signal. The standardized electronic control pulse signal is the basic control signal of the nozzle drive circuit. By modulating it with the weight allocation matrix, parameters such as the amplitude and frequency of the pulse signal can be adjusted according to the requirements of error compensation. The composite pulse sequence injected with high-frequency micro-perturbations can control the movement of the nozzle more precisely. For example, when compensating for the small offset of the nozzle, the high-frequency micro-perturbations can make the nozzle adjust its position more accurately, avoiding new errors caused by over-compensation or under-compensation. The beneficial effect of this step is that the generated artifact compensation control signal can comprehensively consider various error factors and achieve precise control of the nozzle movement by precisely adjusting the control signal of the nozzle drive circuit. This precise control helps to more effectively compensate the coding error, improve the accuracy and clarity of the coding pattern. In cigarette packaging paper coding, it can ensure that information such as brand logos and production dates printed is clear and accurate, reduce product quality problems caused by coding errors, improve the appearance quality and readability of products, and enhance the competitiveness of products in the market. At the same time, due to considering various error factors for compensation, the adaptability of the coder to different working conditions is improved. Even when facing cigarette packaging papers of different materials or slight performance changes in the nozzle, the stability of the coding quality can be ensured.

[0175] Step S3300: Couple the artifact compensation control signal to the nozzle drive circuit to perform dynamic feedback control on the movement pose of the nozzle.

[0176] Specifically, the purpose of step S3300 is to couple the artifact compensation control signal to the nozzle drive circuit to perform dynamic feedback control on the movement pose of the nozzle, so as to achieve real-time correction of the coding error and improve the coding quality. The artifact compensation control signal is generated through previous steps by comprehensively considering the predicted coding error value, the no-load artifact offset error compensation matrix, and multiple factors affecting the coding process. This signal contains the instruction information for precisely adjusting the movement of the nozzle.

[0177] The nozzle driving circuit is a key component for controlling the movement of the nozzle. It receives the artifact compensation control signal and converts it into actual driving actions, thereby adjusting the movement pose of the nozzle. During the inkjet coding process of cigarette wrapping paper, the movement pose of the nozzle directly affects the inkjet position of the ink and the formation of the pattern. For example, when there is a slight deviation of the nozzle in the horizontal or vertical direction, it will cause the position of the inkjet pattern to be inaccurate or the shape to be deformed. By coupling the artifact compensation control signal to the nozzle driving circuit, the movement of the nozzle can be dynamically adjusted according to the real-time error situation.

[0178] Dynamic feedback control is a closed-loop control process. It continuously adjusts the movement of the nozzle based on the current inkjet coding error situation to achieve the purpose of reducing errors. During the inkjet coding process, continuously monitor the actual situation of the inkjet pattern (such as obtaining the real-time inkjet coding image through an image acquisition device), compare it with the preset ideal pattern, and calculate the error. Generate the artifact compensation control signal according to the error, and then feedback this signal to the nozzle driving circuit to adjust the movement pose of the nozzle. Then monitor the inkjet pattern again, repeat the above process, continuously optimize the movement of the nozzle, and gradually reduce the inkjet coding error. For example, if it is found that a certain part of the inkjet pattern is offset, the dynamic feedback control will timely adjust the movement direction and speed of the nozzle according to the artifact compensation control signal, so that the ink ejected subsequently can accurately fall on the expected position, thereby correcting the offset of the pattern.

[0179] By dynamically feedback controlling the movement pose of the nozzle in step S3300, the errors occurring in the inkjet coding process can be corrected in real time, greatly improving the accuracy and quality of the inkjet coding pattern. In the inkjet coding of cigarette wrapping paper, it ensures that the information such as the printed text and pattern is accurate, improves the appearance quality of the product, and enhances the brand image of the product. At the same time, the dynamic feedback control improves the adaptability and stability of the inkjet printer. Even when encountering various interference factors during the inkjet coding process (such as slight vibrations of the nozzle, slight changes in the ink properties, etc.), it can timely adjust the movement of the nozzle to ensure that the inkjet coding quality is not affected. In addition, this precise control method can also reduce the scrap rate caused by inkjet coding errors, reduce production costs, improve production efficiency, and bring better economic benefits to the enterprise.

[0180] Embodiment 2

[0181] Based on Embodiment 1, this embodiment provides a real-time image calibration system for an inkjet printer based on edge computing, as Figure 7 shown, including:

[0182] Compensation matrix generation module: used to construct a composite parameter matrix, perform no-load error analysis, generate a no-load artifact offset error compensation matrix, establish a non-linear mapping relationship between the elements of the no-load artifact offset error compensation matrix and the composite parameter matrix, and generate a compensation matrix dynamic weight distribution function;

[0183] Diffusion trajectory prediction module: Based on the composite parameter matrix, construct a multi-scale gradient feature pyramid, which includes a bottom layer, a middle layer, and a top layer; fuse the features of the bottom layer, middle layer, and top layer of the multi-scale gradient feature pyramid through tensors to generate a feature correlation heat map covering different abstraction levels; combine with the no-load artifact offset error compensation matrix to correct the feature correlation heat map and establish a dynamic prediction model for ink diffusion behavior to generate a real-time diffusion trajectory prediction result;

[0184] Error compensation module: Generate an artifact compensation control signal according to the compensation matrix dynamic weight distribution function and the real-time diffusion trajectory prediction result, couple the artifact compensation control signal to the nozzle drive circuit, and perform dynamic feedback control on the motion pose of the nozzle.

[0185] In the compensation matrix generation module, the construction of the composite parameter matrix includes:

[0186] Step S1110: Obtain the surface viscosity coefficient and potential gradient value to form a fluid diffusion field parameter sub-matrix;

[0187] Step S1120: Measure the adsorption decay characteristic curve of ink on different material surfaces, extract the adsorption decay time constant and the saturated adsorption concentration threshold to form a substrate material adsorption characteristic data sub-matrix;

[0188] Step S1130: Track the movement process of the nozzle, obtain the nozzle vibration displacement signal, and extract the spatial curvature change characteristics of the nozzle movement trajectory from the nozzle vibration displacement signal to form a nozzle movement trajectory spatial characteristic sub-matrix;

[0189] Step S1140: Construct a composite parameter matrix according to the fluid diffusion field parameter sub-matrix, the substrate material adsorption characteristic data sub-matrix, and the nozzle movement trajectory spatial characteristic sub-matrix.

[0190] In the compensation matrix generation module, the generation of the compensation matrix dynamic weight distribution function includes:

[0191] Step S1610: Use the relative contribution rate of adsorption effect, the relative contribution rate of ink potential energy driving effect, the fluid diffusion field parameter sub-matrix, the substrate material adsorption characteristic sub-matrix, and the nozzle movement trajectory spatial characteristic sub-matrix as input parameters to construct a composite input vector;

[0192] Step S1620: Learn the non-linear mapping relationship between the composite input vector and each element of the no-load artifact offset error compensation matrix through a neural network method;

[0193] Step S1630: Generate a compensation matrix dynamic weight distribution function according to the learned non-linear mapping relationship.

[0194] In the diffusion trajectory prediction module, the construction of the multi-scale gradient feature pyramid includes:

[0195] Step S2110: Based on the fluid diffusion field parameter sub-matrix, extract the multi-scale gradient invariant feature clusters formed during the diffusion of the ink on the substrate surface, and construct the bottom layer of the multi-scale gradient feature pyramid; the multi-scale gradient invariant feature clusters include 64-dimensional histograms of oriented gradients.

[0196] Step S2120: Based on the substrate material adsorption characteristic data sub-matrix, capture the microscopic porosity distribution mapping on the substrate material surface, and construct the middle layer of the multi-scale gradient feature pyramid.

[0197] Step S2130: Based on the spatial feature sub-matrix of the nozzle movement trajectory, analyze the second-order derivative fluctuation characteristics of the nozzle movement trajectory, and construct the top layer of the multi-scale gradient feature pyramid.

[0198] In the diffusion trajectory prediction module, the generation of the feature correlation degree heat map covering different abstraction levels includes:

[0199] Step S2210: Fuse the multi-scale gradient invariant feature clusters in the bottom layer with the microscopic porosity distribution mapping in the middle layer to obtain the local ink diffusion behavior characteristics.

[0200] Step S2220: Fuse the second-order derivative fluctuation characteristics of the nozzle movement trajectory in the top layer with the local ink diffusion behavior characteristics to obtain the ink distribution and nozzle movement correlation matrix.

[0201] Step S2230: Perform thermodynamic coding on the ink distribution and nozzle movement correlation matrix to generate a feature correlation degree heat map reflecting the correlation strength between the local feature distribution in the bottom layer and the global features in the top layer.

[0202] In the error compensation module, the generation of the artifact compensation control signal includes:

[0203] Obtain the estimated value of the coding error according to the real-time diffusion trajectory prediction result and the compensation matrix dynamic weight distribution function; based on the estimated value of the coding error, combine it with the no-load artifact offset error compensation matrix to generate the artifact compensation control signal.

[0204] The obtaining of the estimated value of the coding error includes:

[0205] Step S3110: According to the real-time diffusion trajectory prediction result, extract the curvature data of the ink diffusion edge curve, and construct the ink diffusion contour shape description matrix.

[0206] Step S3120: Perform feature matching between the ink diffusion contour shape description matrix and the preset printing pattern, calculate the Euclidean distance between the ink diffusion shape and the geometric contour of the printing pattern, and obtain the shape matching degree index.

[0207] Step S3130: Input the morphological matching degree index into the compensation matrix dynamic weight distribution function, superimpose the weight coefficients corresponding to the fluid diffusion field parameter sub-matrix, the substrate material adsorption characteristic data sub-matrix, and the spatial feature sub-matrix of the nozzle movement trajectory, calculate the spatial geometric offset compensation amount, and generate a predicted value of the coding error.

[0208] The methods and systems of the present application can be implemented in many ways. For example, the methods and systems of the present application can be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is only for illustration, and the steps of the method of the present application are not limited to the specific order described above, unless otherwise specifically stated.

[0209] In addition, parts of the above technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive elaboration.

[0210] As described in the specific embodiments above, the purpose, technical solutions, and beneficial effects of the present invention have been further described in detail. It should be understood that the above description is only the specific embodiments of the present invention and is not used to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. A real-time image calibration method for an inkjet printer based on edge computing, characterized in that, The method includes: Constructing a composite parameter matrix, performing no-load error analysis, generating a no-load artifact offset error compensation matrix, establishing a non-linear mapping relationship between the elements of the no-load artifact offset error compensation matrix and the composite parameter matrix, and generating a compensation matrix dynamic weight distribution function; According to the composite parameter matrix, constructing a multi-scale gradient feature pyramid, where the multi-scale gradient feature pyramid includes a bottom layer, a middle layer, and a top layer; performing tensor fusion on the features of the bottom layer, middle layer, and top layer of the multi-scale gradient feature pyramid to generate a feature correlation heat map covering different abstraction levels; combining with the no-load artifact offset error compensation matrix, correcting the feature correlation heat map, and establishing a dynamic prediction model for ink diffusion behavior to generate a real-time diffusion trajectory prediction result; According to the compensation matrix dynamic weight distribution function and the real-time diffusion trajectory prediction result, generating an artifact compensation control signal, coupling the artifact compensation control signal to the nozzle drive circuit, and performing dynamic feedback control on the motion pose of the nozzle.

2. The real-time image calibration method of the inkjet printer based on edge computing according to claim 1, wherein, The constructing of the composite parameter matrix includes: Obtaining the surface viscosity coefficient and the potential gradient value to form a fluid diffusion field parameter sub-matrix; Measuring the adsorption attenuation characteristic curve of ink on different material surfaces, extracting the adsorption attenuation time constant and the saturated adsorption concentration threshold to form a substrate material adsorption characteristic data sub-matrix; Tracking the nozzle movement process, obtaining the nozzle vibration displacement signal, and extracting the spatial curvature change characteristics of the nozzle movement trajectory from the nozzle vibration displacement signal to form a nozzle movement trajectory spatial feature sub-matrix; Constructing a composite parameter matrix according to the fluid diffusion field parameter sub-matrix, the substrate material adsorption characteristic data sub-matrix, and the nozzle movement trajectory spatial feature sub-matrix.

3. The real-time image calibration method of the inkjet printer based on edge computing according to claim 2, wherein, The forming of the fluid diffusion field parameter sub-matrix includes: collecting the dynamic contact angle change data of ink droplets on the substrate surface, extracting the surface viscosity coefficient; measuring the spatial difference in the internal pressure distribution of ink droplets, calculating the potential gradient value; quantifying the surface viscosity coefficient and the potential gradient value into a fluid diffusion field parameter sub-matrix.

4. The real-time image calibration method of the inkjet printer based on edge computing according to claim 3, characterized in that The generating of the no-load artifact offset error compensation matrix includes: According to the fluid diffusion field parameter sub-matrix, obtaining the artifact width distribution data formed during the ink diffusion process, and generating a diffusion artifact width distribution spectrum; Using the diffusion artifact width distribution spectrum as a prior constraint condition in the starting stage of printing in the no-load state, loading it into the lightweight inverse generative adversarial topology network in the pre-constructed edge computing node to obtain a diffusion artifact prior constraint; According to the fluid diffusion field parameter sub-matrix and the substrate material adsorption characteristic data sub-matrix, obtaining the relative contribution rate of the adsorption effect and the relative contribution rate of the ink potential driving effect; Combining with the diffusion artifact prior constraint, performing error analysis on the coding image in the no-load state to generate a no-load artifact offset error compensation matrix.

5. The real-time image calibration method of the inkjet printer based on edge computing according to claim 4, wherein, The combining with the diffusion artifact prior constraint, performing error analysis on the coding image in the no-load state, and generating a no-load artifact offset error compensation matrix includes: Combining with the diffusion artifact prior constraint, using the coder to print a preset reference pattern under no-load conditions, and collecting the actual coding image in the no-load state; Performing feature point matching on the collected actual coding image and the reference pattern, and calculating the geometric offset of the coding contour; Matrixize the geometric offset to generate an offset error compensation matrix for no-load artifacts.

6. The real-time image calibration method of the inkjet printer based on edge computing according to claim 5, characterized in that The calculation of the geometric offset of the inkjet profile includes: Extract the key feature points of the actual inkjet image and the reference pattern, and construct a feature point matching matrix; Based on the feature point matching matrix, fit the geometric offset function of the inkjet profile to obtain the local geometric offset and the overall geometric offset; Combine the local geometric offset and the overall geometric offset to form the geometric offset of the inkjet profile.

7. The real-time image calibration method of the inkjet printer based on edge computing according to claim 6, characterized in that, The matrixization of the geometric offset to generate an offset error compensation matrix for no-load artifacts includes: According to the relative contribution rate of the adsorption effect and the relative contribution rate of the ink potential energy driving effect, perform weighted combination on the local geometric offset and the overall geometric offset to obtain a comprehensive geometric offset matrix; Perform normalization and discretization processing on the elements in the comprehensive geometric offset matrix to generate an offset error compensation matrix for no-load artifacts.

8. The real-time image calibration method of the inkjet printer based on edge computing according to claim 2, wherein, The construction of the multi-scale gradient feature pyramid includes: Based on the fluid diffusion field parameter sub-matrix, extract the multi-scale gradient invariant feature clusters formed during the diffusion process of the ink on the substrate surface, and construct the bottom layer of the multi-scale gradient feature pyramid; the multi-scale gradient invariant feature clusters include 64-dimensional direction gradient histograms; Based on the substrate material adsorption characteristic data sub-matrix, capture the microscopic porosity distribution mapping on the substrate material surface, and construct the middle layer of the multi-scale gradient feature pyramid; Based on the spatial feature sub-matrix of the nozzle movement trajectory, analyze the second-order derivative fluctuation characteristics of the nozzle movement trajectory, and construct the top layer of the multi-scale gradient feature pyramid.

9. The real-time image calibration method of the inkjet printer based on edge computing according to claim 8, characterized in that, The generation of the feature correlation heat map covering different abstraction levels includes: Fuse the multi-scale gradient invariant feature clusters in the bottom layer with the microscopic porosity distribution mapping in the middle layer to obtain the local ink diffusion behavior characteristics; Fuse the second-order derivative fluctuation characteristics of the nozzle movement trajectory in the top layer with the local ink diffusion behavior characteristics to obtain an ink distribution and nozzle movement correlation matrix; Perform thermodynamic coding on the ink distribution and nozzle movement correlation matrix to generate a feature correlation heat map reflecting the correlation strength between the local feature distribution in the bottom layer and the global feature in the top layer.

10. An edge-computing-based real-time image calibration system for an inkjet printer, which is used to implement the edge-computing-based real-time image calibration method described in any one of claims 1-9, characterized in that, The system includes: Compensation matrix generation module: used to construct a composite parameter matrix, perform no-load error analysis, generate an offset error compensation matrix for no-load artifacts, establish a non-linear mapping relationship between the elements of the offset error compensation matrix for no-load artifacts and the composite parameter matrix, and generate a compensation matrix dynamic weight distribution function; Diffusion trajectory prediction module: According to the composite parameter matrix, construct a multi-scale gradient feature pyramid, which includes a bottom layer, a middle layer, and a top layer; perform tensor fusion on the features of the bottom layer, middle layer, and top layer of the multi-scale gradient feature pyramid to generate a feature correlation heat map covering different abstraction levels; combine the offset error compensation matrix for no-load artifacts to correct the feature correlation heat map, and establish a dynamic prediction model for ink diffusion behavior to generate a real-time diffusion trajectory prediction result; Error compensation module: According to the compensation matrix dynamic weight distribution function and the real-time diffusion trajectory prediction result, generate an artifact compensation control signal, couple the artifact compensation control signal to the nozzle drive circuit, and perform dynamic feedback control on the movement pose of the nozzle.

Images