Method and system for matching decorative paper printing pigments

Abstract

The application discloses a decorative paper printing pigment color matching method and system, relates to the printing technical field, and discloses the decorative paper printing pigment color matching method and system, which obtains multispectral data and decomposes total spectral residual to identify dominant factors of equipment mechanical deviation and pigment batch deviation, generates a robust formula under multi-scene constraints in combination with a pre-trained formula generation model, effectively distinguishes color deviation sources and ensures color consistency under multiple light environments, can intelligently identify dominant factors of color deviation and generate a multi-scene color consistent pigment formula, thereby reducing manual intervention, avoiding misjudgment repair measures, improving printing efficiency and product quality.

Description

Technical Field

[0001] This application relates to the field of printing technology, and in particular to a method and system for color matching of pigments for decorative paper printing. Background Technology

[0002] Decorative paper, as a core material for surface veneer of engineered wood panels, directly impacts the visual quality and market competitiveness of furniture and decoration materials through accurate color reproduction and long-term stability. In the actual production process of gravure printing on decorative paper, printing equipment is inevitably affected by gradual changes in mechanical conditions such as doctor blade wear, increased cumulative printing volume on the printing rollers, and fluctuations in printing pressure. Simultaneously, different batches of pigments exhibit inherent differences in particle size distribution and spectral absorption characteristics. These factors combined cause the printed colors to gradually deviate from the standard sample, resulting in a systematic color deviation. Currently, the industry's commonly used repair methods rely primarily on manual intervention: operators must frequently stop the machine to extract printed samples, obtain color difference data through offline colorimetry equipment, and then manually adjust the ink volume of each color group or re-sample based on personal experience. While some systems have introduced online RGB imaging monitoring, the insufficient dimensions of the RGB color space prevent effective differentiation between the independent contributions of mechanical factors and batch-specific pigment factors to color deviation, often leading to a lack of targeted repair measures, such as misjudging pigment batch defects by using ink volume compensation. More importantly, existing color matching methods only focus on color matching under a single standard light source. However, decorative paper products need to adapt to various lighting environments such as shopping mall lighting and warm home lighting in actual use. Metamerism means that printed materials that are calibrated under a standard light source may still show significant visual color differences under other lighting conditions, forcing companies to repeatedly stop production for adjustments, which restricts production continuity and product delivery quality.

[0003] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The main purpose of this application is to provide a method and system for color matching of printing pigments for decorative paper, which aims to improve printing efficiency and product quality.

[0005] To achieve the above objectives, this application proposes a method for color matching of printing pigments on decorative paper, the method comprising:

[0006] Acquire current multispectral data of online printed materials from the printing press, as well as pre-registered standard multispectral data;

[0007] The total spectral residual is determined based on the current multispectral data and the standard multispectral data, and the total spectral residual is decomposed into equipment mechanical offset component and pigment batch offset component according to the preset offset separation rule.

[0008] Analyze the statistical characteristics of the equipment mechanical offset component and the pigment batch offset component to determine the dominant factors causing color shift; the dominant factors include equipment mechanical offset and pigment batch offset.

[0009] When the batch shift of pigment is determined to be the dominant factor, the standard multispectral data is used as the target spectrum and input into the pre-trained formula generation model. The pre-trained formula generation model generates a new pigment volume concentration formula. The pre-trained formula generation model is constrained by the metamerism index of multiple scenarios during training, so that the generated formula meets the preset color consistency requirements under multiple application scenario light sources.

[0010] The new pigment volume concentration formulation is output to the printing press to replace the currently used pigment formulation.

[0011] In one embodiment, the steps of determining the total spectral residual based on the current multispectral data and the standard multispectral data, and decomposing the total spectral residual into a device mechanical offset component and a pigment batch offset component according to a preset offset separation rule, include:

[0012] Acquire pre-established device-sensitive directional features and material-sensitive directional features; wherein, the device-sensitive directional features characterize the migration pattern of the printed spectrum as the mechanical state of the equipment changes and is independent of the pigment batch, and the material-sensitive directional features characterize the absorption change pattern of the printed spectrum as the pigment batch changes and is independent of the mechanical state of the equipment;

[0013] The total spectral residual is projected onto the sensitive direction feature of the device to obtain the mechanical offset component of the device;

[0014] The total spectral residual is projected onto the material-sensitive directional features to obtain the pigment batch-specific offset component.

[0015] In one embodiment, the establishment of the device-sensitive direction features and the material-sensitive direction features includes:

[0016] Standard samples were printed using the same batch of pigments under multiple different equipment status parameters. Multispectral data was collected, and spectral transfer characteristics were extracted from the collected multispectral data.

[0017] Principal component analysis was performed on the spectral transfer characteristics obtained under multiple different equipment state parameters. The first principal component with the largest contribution was extracted as the first feature component that only changes with the equipment state parameters and is not sensitive to the pigment batch. This first feature component was determined as the equipment sensitive direction feature.

[0018] Standard samples were printed under conditions of multiple different batches of pigments and constant equipment parameters. Multispectral data was collected, and spectral transfer characteristics were extracted from the collected multispectral data.

[0019] Principal component analysis was performed on the spectral transfer characteristics obtained from multiple different batches of pigments. The second principal component with the largest contribution was extracted as the second feature component that only varies with the pigment batch and is not sensitive to the equipment state parameters. This second feature component was determined as the material-sensitive directional feature.

[0020] Store the device's sensitive direction features and the material's sensitive direction features for subsequent decomposition and retrieval.

[0021] In one embodiment, the step of analyzing the statistical characteristics of the device mechanical offset component and the pigment batch offset component to determine the dominant factor causing the color shift includes:

[0022] Calculate the energy percentage of the mechanical offset component of the equipment in the total spectral residual, and use it as the equipment offset index;

[0023] The energy percentage of the pigment batch-specific offset component in the total spectral residual is calculated and used as the material offset index;

[0024] When the equipment offset index is greater than the preset equipment offset threshold and the equipment offset index is greater than the material offset index, the mechanical offset of the equipment is determined to be the dominant factor; otherwise, the batch offset of the pigment is determined to be the dominant factor.

[0025] In one embodiment, when mechanical misalignment of the equipment is determined to be the dominant factor, the method further includes:

[0026] The spectral compensation amount is calculated based on the mechanical offset component of the equipment, and the spectral compensation amount makes the compensated predicted spectrum consistent with the standard multispectral data in the sensitive direction of the equipment.

[0027] The spectral compensation amount is mapped to the ink bond opening adjustment value of each color group according to a preset conversion relationship;

[0028] The ink key opening adjustment value is used as a compensation signal and output to the ink key actuator of the printing press through PID control.

[0029] In one embodiment, the method further includes:

[0030] Construct a neural network that includes a recipe generation pathway and a scenario robustness evaluation pathway;

[0031] Obtain a training sample set containing target spectral samples, primary color spectral dictionary, and spectral power distributions of light sources from multiple application scenarios;

[0032] The formulation generation pathway outputs a predicted pigment volume concentration formulation based on the input target spectral sample, and calculates the spectral reconstruction error between the reconstructed spectrum corresponding to the predicted pigment volume concentration formulation and the target spectral sample.

[0033] The reconstructed spectrum is obtained by the scenario robustness assessment pathway based on the predicted pigment volume concentration formulation and the primary color spectral dictionary, and the color difference between the reconstructed spectrum and the target spectral sample is calculated under multiple application scenario light sources. The multi-scenario metamerism index is obtained by combining the color differences of each scenario.

[0034] With the simultaneous reduction of the spectral reconstruction error and the multi-scene metamerism index as the joint optimization objective, the parameters of the recipe generation pathway and the scene robustness evaluation pathway are alternately and adversarially adjusted until the convergence condition is met, so that the trained recipe generation pathway can be used as the recipe generation model.

[0035] In one embodiment, the alternating antagonistic adjustment includes:

[0036] The parameters of the scenario robustness evaluation pathway are fixed, and the parameters of the formulation generation pathway are updated with the goal of reducing the spectral reconstruction error, so that the formulation generation pathway learns to map the target spectrum into a pigment volume concentration formulation.

[0037] The parameters of the formulation generation pathway are fixed, and the multi-scenario metamerism index is calculated based on the predicted pigment volume concentration formulation output by the current formulation generation pathway. The parameters of the scenario robustness assessment pathway are updated using the multi-scenario metamerism index as a feedback signal, so that the scenario robustness assessment pathway can learn to identify the multi-scenario color difference risk of the formulation.

[0038] In the next iteration, the metamerism index of the multi-scene calculations of the scenario robustness evaluation pathway is added as a penalty term to the loss function of the recipe generation pathway, forcing the recipe generation pathway to prioritize color schemes with lower metamerism indices when updating parameters.

[0039] Repeat the above steps until both the spectral reconstruction error and the multi-scene metamerism index are reduced to the preset convergence range.

[0040] In one embodiment, when batch-to-batch pigment shift is determined to be the dominant factor, the method further includes:

[0041] Before inputting the standard multispectral data into the pre-trained formulation generation model, the material absorption core features of the current pigment are extracted from the pigment batch offset component. These material absorption core features reflect the difference in absorption spectrum between the current batch of pigment and the standard batch.

[0042] The material absorption core features are used as auxiliary inputs and input together with the standard multispectral data into the pre-trained formula generation model;

[0043] The pre-trained formulation generation model, combined with the material absorption core characteristics, generates a new pigment volume concentration formulation that is modified to suit the characteristics of the current batch of pigments.

[0044] In one embodiment, the method further includes:

[0045] The total spectral residual, the decomposition results of the equipment mechanical offset component and the pigment batch offset component generated during this closed-loop repair process, the old formula that was replaced, the new pigment volume concentration formula, and the color difference data measured in multiple application scenarios after repair are integrated into a repair case and stored in the case library.

[0046] After obtaining the new total spectral residual and completing the decomposition, the similarity between the decomposition result of the currently obtained total spectral residual and the decomposition result of the existing cases in the case library is calculated.

[0047] When the similarity exceeds a preset matching threshold, a new formula is extracted from the existing matched cases as an initial corrected formula, and the initial corrected formula is directly applied to the printing press.

[0048] Furthermore, to achieve the above objectives, this application also proposes a color matching system for decorative paper printing pigments, the color matching system for decorative paper printing pigments comprising: a memory, a processor, and a color matching program for decorative paper printing pigments stored in the memory and executable on the processor, the color matching program for decorative paper printing pigments being configured to implement the steps of the color matching method for decorative paper printing pigments.

[0049] The decorative paper printing pigment color matching method and system proposed in this application acquires multispectral data and decomposes the total spectral residual to identify the dominant factors of mechanical deviation of equipment and batch deviation of pigment. Combined with a pre-trained formula generation model, it generates robust formulas under multi-scenario constraints, effectively distinguishes the root causes of color deviation and ensures color consistency under multi-lighting environments. It can intelligently identify the dominant factors of color deviation and generate pigment formulas with consistent color in multiple scenarios, thereby reducing manual intervention, avoiding misjudgment and repair measures, and improving printing efficiency and product quality. Attached Figure Description

[0050] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0051] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 A schematic flowchart of an embodiment of the method for color matching of printing pigments for decorative paper according to this application;

[0053] Figure 2 For this application Figure 1 A detailed flowchart of step S200;

[0054] Figure 3 For this application Figure 1 Detailed flowchart of step S300;

[0055] Figure 4 This is a schematic diagram of a structural embodiment of the decorative paper printing pigment color matching system of this application.

[0056] Explanation of icon numbers:

[0057] 10. Memory; 20. Processor.

[0058] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0059] The technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of this application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0060] It should be understood that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0061] In existing technologies, after prolonged operation, gravure printing on decorative paper can cause the printed colors to deviate from the standard due to gradual changes in the mechanical condition of the equipment and differences in pigments between different batches. Existing repair methods rely on manual adjustment based on experience or low-dimensional RGB monitoring, which cannot accurately distinguish between mechanical equipment deviations and batch-specific pigment color deviations, resulting in a lack of targeted repair. Furthermore, existing color matching methods only consider a single standard light source, failing to address metamerism issues in multiple scenarios, leading to repeated downtime for color correction, which restricts production efficiency and delivery quality.

[0062] Based on this, embodiments of this application provide a method for color matching of printing pigments for decorative paper, referring to... Figure 1 The method for color matching of printing pigments on decorative paper includes steps S100 to S500, wherein:

[0063] Step S100: Obtain the current multispectral data of the online printed matter of the printing press, as well as the pre-registered standard multispectral data;

[0064] Step S200: Determine the total spectral residual based on the current multispectral data and the standard multispectral data, and decompose the total spectral residual into equipment mechanical offset component and pigment batch offset component according to the preset offset separation rule;

[0065] Step S300: Analyze the statistical characteristics of the equipment mechanical offset component and the pigment batch offset component to determine the dominant factors causing the color shift; the dominant factors include equipment mechanical offset and pigment batch offset.

[0066] Step S400: When it is determined that the batch shift of pigment is the dominant factor, the standard multispectral data is used as the target spectrum and input into the pre-trained formula generation model. The pre-trained formula generation model generates a new pigment volume concentration formula. The pre-trained formula generation model is constrained by the metamerism index of multiple scenarios during training, so that the generated formula meets the preset color consistency requirements under multiple application scenario light sources.

[0067] Step S500: Output the new pigment volume concentration formula to the printing press to replace the currently used pigment formula.

[0068] In this embodiment, multispectral data refers to spectral information collected across multiple discrete or continuous wavelength ranges. It contains more comprehensive information on color and material optical properties than traditional tristimulus values, enabling a more accurate characterization of the printed material's color state. The total spectral residual refers to the spectral difference between the current printed material's multispectral data and standard multispectral data, quantifying the degree to which the printed material's color deviates from the standard. The equipment mechanical offset component refers to the spectral shift in the printed material caused by changes in the mechanical state of the printing equipment (e.g., doctor blade wear, increased printing roller length, shift in printing force, etc.), and this component is independent of pigment batches. The pigment batch offset component refers to the spectral shift in the printed material caused by differences in pigment particle size, absorption characteristics, etc., between different batches of pigment, and this component is independent of the equipment mechanical state.

[0069] In this embodiment, the dominant factor refers to one or a group of factors that contribute significantly to color shift in printed materials among various factors. In this embodiment, it mainly refers to mechanical deviation of the equipment or batch-specific deviation of the pigment. The pre-trained formula generation model refers to a machine learning model trained on a large amount of data, capable of generating a corresponding pigment volume concentration formula based on the input target spectral data. This model has completed the training process before practical application. The multi-scene metamerism index is a comprehensive quantitative indicator of the color difference between the color printed from the same pigment formula and the target color under multiple different application scenario light sources. It is used to evaluate the color consistency performance of the formula under different lighting conditions. The pigment volume concentration formula refers to the volume concentration ratio combination of various primary color pigments that constitute the printing pigment, which directly determines the final color of the printed material.

[0070] In this embodiment, the method for color matching of pigments in decorative paper printing first acquires the current multispectral data of the printed matter on the printing press, as well as pre-registered standard multispectral data. The current multispectral data can be obtained by periodically scanning and measuring with a spectrometer on the printing production line, or by measuring printed matter samples offline. The pre-registered standard multispectral data can be obtained by accurately measuring standard samples before production and storing them in a database as a benchmark for subsequent comparisons. For example, printed matter can be randomly sampled from the production line periodically, measured using a handheld spectrophotometer, and the measurement results used as the current multispectral data. Simultaneously, pre-made standard color patches are measured, and their spectral data is used as standard multispectral data.

[0071] In this embodiment, the total spectral residual is determined based on the current multispectral data and the standard multispectral data. Then, according to a preset offset separation rule, the total spectral residual is decomposed into a device mechanical offset component and a pigment batch offset component. The total spectral residual can be obtained by subtracting the current multispectral data from the standard multispectral data wavelength by wavelength. The offset separation rule can be set based on experience. For example, through historical data analysis, when a large deviation occurs in a specific band of the total spectral residual, it is allocated to the device mechanical offset component and the pigment batch offset component in a fixed proportion. Alternatively, a simple threshold can be set; when the total spectral residual exceeds this threshold, it is roughly attributed to the device or the pigment based on its changing trend.

[0072] In this embodiment, the statistical characteristics of the equipment mechanical offset component and the pigment batch offset component are analyzed to determine the dominant factor causing the color shift. The dominant factor includes both equipment mechanical offset and pigment batch offset. The statistical characteristics can be obtained by simply summing or averaging each component to obtain its overall magnitude. For example, the sum of the spectral energies of the equipment mechanical offset component and the pigment batch offset component can be calculated, and then the magnitudes of these two sums can be compared; the component with the larger energy sum is determined to be the dominant factor.

[0073] In this embodiment, when batch-specific pigment shift is determined to be the dominant factor, the standard multispectral data is used as the target spectrum and input into a pre-trained formulation generation model. This model then generates a new pigment volume concentration formulation. During training, the pre-trained formulation generation model is constrained by the metamerism index for multiple scenarios, ensuring that the generated formulation meets preset color consistency requirements under various application lighting conditions. For example, the model can be a multilayer perceptron network, with standard multispectral data as input and volume concentrations of each primary color pigment as output. During model training, the multi-scenario metamerism index constraint can be achieved by adding a penalty term to the loss function. This penalty term is obtained by simulating and summing the color differences under several typical light sources (such as sunlight, fluorescent lamps, and incandescent lamps). Finally, the new pigment volume concentration formulation is output to the printing press, replacing the currently used pigment formulation. This output process can involve displaying the generated formulation parameters on the printing press's operating interface, allowing the operator to manually input them into the printing press's ink volume control system. Alternatively, the formulation can be directly sent to the printing press's ink volume adjustment unit via a data interface, which will automatically perform the formulation replacement operation.

[0074] The color matching method for decorative paper printing pigments provided in this embodiment can automatically acquire multispectral data of printed materials and compare it with standard data to accurately quantify color shifts. By decomposing the total spectral residual into mechanical offset components of the equipment and batch-specific offset components of the pigments, this method can accurately identify the root causes of color shifts in decorative paper printing, avoiding the lack of targeted remedial measures due to factor confusion in traditional methods. Furthermore, when generating new pigment formulations, this method inherently considers the constraints of metamerism indices in multiple scenarios, ensuring that the generated formulations meet color consistency requirements under various practical application light sources. This effectively solves the problem in existing technologies where the pigments are acceptable under a single light source but still exhibit color differences in the final application scenario, thereby reducing the frequency of repeated downtime for color correction and improving the production efficiency and product quality of decorative paper printing.

[0075] In one feasible implementation, refer to Figure 2 Step S200 includes steps S210 to S230, wherein:

[0076] Step S210: Obtain pre-established device-sensitive direction features and material-sensitive direction features; wherein, the device-sensitive direction features characterize the migration pattern of the printed spectrum as the mechanical state of the equipment changes and is independent of the pigment batch, and the material-sensitive direction features characterize the absorption change pattern of the printed spectrum as the pigment batch changes and is independent of the mechanical state of the equipment.

[0077] Step S220: Project the total spectral residual onto the sensitive direction feature of the device to obtain the mechanical offset component of the device;

[0078] Step S230: Project the total spectral residual onto the material-sensitive direction feature to obtain the pigment batch offset component.

[0079] In this embodiment, pre-established device-sensitive direction features and material-sensitive direction features are acquired. The device-sensitive direction features characterize the migration pattern of the printed material's spectrum as a function of the equipment's mechanical state, independent of the pigment batch. The material-sensitive direction features characterize the absorption change pattern of the printed material's spectrum as a function of the pigment batch, independent of the equipment's mechanical state. These features are obtained in advance through analysis and modeling of a large amount of historical data. They can be considered as orthogonal or quasi-orthogonal basis vectors in the color shift space, representing the color change directions independently caused by the equipment state and pigment batch, respectively. For example, under controlled experimental conditions, a large amount of printed material spectral data can be collected by changing the equipment state (keeping the pigment batch constant) and the pigment batch (keeping the equipment state constant), and these physically meaningful feature directions can be extracted using statistical analysis methods (such as principal component analysis). Once established, these features can be stored for subsequent real-time decomposition.

[0080] In this embodiment, the total spectral residual is projected onto the device-sensitive direction feature to obtain the device mechanical offset component. This projection operation is achieved by calculating the inner product between the total spectral residual vector and the device-sensitive direction feature vector, thereby quantifying the portion of the total spectral residual related to device mechanical offset. In this way, color deviations caused by changes in device mechanical state can be extracted from the complex total spectral residual. The total spectral residual is then projected onto the material-sensitive direction feature to obtain the pigment batch offset component. Similar to the above, this projection operation is achieved by calculating the inner product between the total spectral residual vector and the material-sensitive direction feature vector, thereby accurately extracting the portion of the total spectral residual related to pigment batch differences.

[0081] In this embodiment, by utilizing pre-established equipment-sensitive direction features and material-sensitive direction features with clear physical meaning, the total spectral residual is precisely decomposed into equipment mechanical offset components and pigment batch-specific offset components. This decomposition method based on feature direction projection effectively avoids confusion between different offset sources, ensuring the accuracy and independence of the decomposition results. Therefore, when subsequently analyzing the statistical characteristics of the equipment mechanical offset components and pigment batch-specific offset components, the dominant factors can be determined based on more reliable data, thereby providing printing presses with more targeted correction strategies. Whether adjusting equipment mechanical parameters or updating pigment formulations, this can improve the accuracy and efficiency of pigment matching methods for decorative paper printing, reducing resource waste and production delays caused by misjudgments.

[0082] In one feasible implementation, the establishment of the device-sensitive directional features and material-sensitive directional features includes: printing standard samples using the same batch of pigments under multiple different device state parameters, collecting multispectral data, and extracting spectral transfer features from the collected multispectral data; performing principal component analysis on the spectral transfer features obtained under multiple different device state parameters, extracting the first principal component with the largest contribution as the first feature component that only changes with the device state parameters and is insensitive to the pigment batch, and determining this first feature component as the device-sensitive directional feature; printing standard samples under multiple different batches of pigments while keeping the device state parameters unchanged, collecting multispectral data, and extracting spectral transfer features from the collected multispectral data; performing principal component analysis on the spectral transfer features obtained under multiple different batches of pigments, extracting the second principal component with the largest contribution as the second feature component that only changes with the pigment batch and is insensitive to the device state parameters, and determining this second feature component as the material-sensitive directional feature; and storing the device-sensitive directional features and the material-sensitive directional features for subsequent decomposition and retrieval.

[0083] In this embodiment, standard samples are printed using the same batch of pigments under multiple different equipment state parameters, and multispectral data is collected. Spectral transfer characteristics are then extracted from the collected multispectral data. This step aims to isolate the influence of mechanical deviations in the equipment by using a controlled variable method. By printing standard samples using the same batch of stable pigments under different equipment state parameters (e.g., printing speed, ink volume, pressure, temperature, humidity, etc.), it can be ensured that the collected multispectral data primarily reflects the influence of equipment state changes on the spectral characteristics of the printed matter, while excluding interference from batch-to-batch pigment differences. The collected multispectral data is the reflectance or transmission spectrum information of the printed matter at different wavelengths. Extracting spectral transfer characteristics from this multispectral data refers to converting the raw spectral data into a more representative or easier-to-analyze feature vector using a color management system or spectral analysis algorithm. For example, this can be done by calculating the K / S value (the ratio of the absorption coefficient K to the scattering coefficient S in Kubelka-Munk theory) or by directly using specific wavelength points of the spectral curve as features.

[0084] In this embodiment, principal component analysis (PCA) is performed on the spectral transfer features obtained under multiple different equipment state parameters. The first principal component with the largest contribution is extracted as the first feature component that only changes with the equipment state parameters and is insensitive to pigment batches. This first feature component is then identified as the equipment-sensitive direction feature. Principal component analysis (PCA) is a commonly used data dimensionality reduction technique that can extract the most important feature components from high-dimensional data. In this step, performing PCA on the spectral transfer features obtained under different equipment state parameters can identify the direction of change with the largest variance and the most representative characteristics in the dataset. The first principal component with the largest contribution represents the dominant mode of influence of changes in equipment state parameters on spectral transfer features. Since pigment batch variations were excluded in the experimental design, this first principal component can accurately characterize the migration mode of the printed spectrum with changes in the mechanical state of the equipment and is insensitive to pigment batches, thus being identified as the equipment-sensitive direction feature.

[0085] In this embodiment, standard samples are printed using multiple batches of pigments while keeping the equipment parameters constant. Multispectral data is collected, and spectral transfer characteristics are extracted from the collected multispectral data. Similar to the previous steps, this step also employs the controlled variable method, but its purpose is to isolate the influence of batch-specific pigment shifts. By printing standard samples using multiple batches of pigments (e.g., from different suppliers, with different production dates, or different batch numbers) while keeping the equipment parameters constant (e.g., set to standard or stable printing conditions), it can be ensured that the collected multispectral data primarily reflects the influence of pigment batch differences on the spectral characteristics of the printed matter, while excluding interference from changes in equipment conditions. Similarly, spectral transfer characteristics are extracted from these multispectral data for subsequent analysis.

[0086] In this embodiment, principal component analysis (PCA) was performed on the spectral transfer characteristics obtained from multiple different batches of pigment. The second principal component with the largest contribution was extracted as the second feature component that varies only with the pigment batch and is insensitive to equipment state parameters. This second feature component was then identified as the material-sensitive directional feature. In this step, principal component analysis was performed on the spectral transfer characteristics obtained from different pigment batches. The second principal component with the largest contribution represents the dominant mode of influence of pigment batch variation on spectral transfer characteristics. Since equipment state variation was excluded in the experimental design, this second principal component can accurately characterize the absorption mode of the printed spectrum as the pigment batch varies, and it is insensitive to equipment state parameters, thus being identified as the material-sensitive directional feature.

[0087] In this embodiment, the device-sensitive direction features and the material-sensitive direction features are stored for subsequent decomposition. Once the device-sensitive direction features and material-sensitive direction features are successfully established and determined, they will be stored in the system's database or memory. These features are the key basis for subsequent total spectral residual decomposition. By storing them, it can be ensured that during the actual printing process, the system can quickly and accurately call these preset features to analyze and decompose the real-time acquired spectral data, thereby achieving accurate diagnosis of the causes of color shift.

[0088] In this embodiment, by printing standard samples and acquiring multispectral data under controlled experimental conditions for changes in equipment status and pigment batches, and combining this with principal component analysis (PCA) techniques, the system can effectively decouple a first characteristic component related only to mechanical equipment offset and a second characteristic component related only to batch-specific pigment offset from complex spectral data. This method ensures that the established sensitive direction features are highly specific and accurate, avoiding feature confusion. Based on this, when performing subsequent total spectral residual decomposition, projecting the total spectral residual onto these accurately established sensitive direction features improves the accuracy and reliability of separating the mechanical equipment offset component and the batch-specific pigment offset component. This allows the system to more accurately determine the dominant factors causing color shifts, providing a solid data foundation for subsequent targeted corrective measures (whether adjusting equipment parameters or updating pigment formulations), and significantly improving the accuracy and efficiency of the pigment color matching method for decorative paper printing.

[0089] In one feasible implementation, refer to Figure 3 Step S300 includes steps S310 to S330, wherein:

[0090] Step S310: Calculate the energy proportion of the mechanical offset component of the device in the total spectral residual, and use it as the device offset index;

[0091] Step S320: Calculate the energy percentage of the pigment batch offset component in the total spectral residual, and use it as the material offset index;

[0092] Step S330: When the equipment offset index is greater than the preset equipment offset threshold and the equipment offset index is greater than the material offset index, the mechanical offset of the equipment is determined to be the dominant factor; otherwise, the batch offset of the pigment is determined to be the dominant factor.

[0093] In this embodiment, the energy percentage can be obtained by calculating the sum of squares of the device mechanical offset component in the spectral dimension and comparing it with the sum of squares of the total spectral residual in the spectral dimension. This aims to quantify the contribution of device mechanical offset to the overall color deviation. The energy percentage of the pigment batch offset component in the total spectral residual is calculated and used as a material offset index. Similar to the calculation method of the device offset index, this material offset index reflects the contribution of pigment batch offset to the total spectral residual. In this way, two potential offset sources can be quantitatively compared.

[0094] In this embodiment, the dominant factor is determined by comparing these two indices with a preset threshold. When the equipment offset index is greater than the preset equipment offset threshold, and the equipment offset index is also greater than the material offset index, the system determines that the mechanical offset of the equipment is the dominant factor. This means that the change in equipment status is the main cause of the current color deviation, and its influence has reached a level that requires attention. In other cases, i.e., when the equipment offset index does not meet the above conditions, the system determines that the batch offset of pigment is the dominant factor. This provides a clear and quantitative decision-making logic, avoiding errors that may be caused by subjective judgment.

[0095] In this embodiment, by calculating the equipment offset index and the material offset index, and combining them with preset thresholds for logical judgment, misdiagnosis caused by subjective experience or vague judgment can be avoided. This precise diagnostic mechanism ensures the targeted nature of subsequent corrective measures. For example, when mechanical equipment offset is the dominant factor, the system will guide equipment adjustment; while when batch-specific pigment offset is the dominant factor, it will trigger the regeneration of the pigment formula, improving the efficiency and accuracy of color correction, reducing unnecessary trial and error costs, and thus ensuring the consistency and stability of printed colors.

[0096] In one feasible implementation, when the mechanical offset of the equipment is determined to be the dominant factor, the method further includes: calculating a spectral compensation amount based on the mechanical offset component of the equipment, the spectral compensation amount making the compensated predicted spectrum consistent with the standard multispectral data in the sensitive direction of the equipment; mapping the spectral compensation amount to the ink key opening adjustment value of each color group according to a preset conversion relationship; and outputting the ink key opening adjustment value as a compensation signal to the ink key actuator of the printing press through PID control.

[0097] In this embodiment, the mechanical offset component of the equipment is quantitative data characterizing the spectral deviation caused by changes in the mechanical state of the printing equipment. The purpose of calculating the spectral compensation is to counteract this deviation, allowing the actual spectral performance of the printed material to return to a standard state. This calculation process can be based on a pre-established equipment characteristic model, for example, determined through inverse spectral prediction or optimization algorithms. For instance, a mapping model from the equipment's mechanical state to its spectral response can be constructed, and then a spectral adjustment amount can be found through inverse solving, such that after adjustment, the component of the spectrum currently affected by the mechanical offset in the equipment-sensitive direction is consistent with the component of the standard multispectral data in that direction. This ensures the targeted and effective nature of the compensation, i.e., correction is only performed on the specific spectral dimension affected by the mechanical offset.

[0098] In this embodiment, the spectral compensation amount, as an abstract spectral domain adjustment value, needs to be converted into a physical parameter executable by the printing press. The preset conversion relationship refers to a known, calibrated, or modeled mapping function that can translate the spectral domain adjustment requirements into a physical control quantity—the ink key opening. For example, the printing press's color profile (ICC Profile) or a physical model-based ink volume-spectral response model can be used. Through this conversion relationship, the system can determine how much opening adjustment is needed for each color group (e.g., CMYK) of ink keys to achieve the required spectral compensation. This mapping is a crucial aspect of printing press control, tightly integrating color science with mechanical control.

[0099] In this embodiment, the key opening adjustment value is a specific physical adjustment command. To ensure that these commands can be executed accurately and stably, this application employs a PID (Proportional-Integral-Derivative) control method. A PID controller is a classic feedback control algorithm that calculates the output signal based on the current error (i.e., the difference between the target adjustment value and the actual key opening), the accumulated error, and the rate of change of the error. Using the key opening adjustment value as the setpoint or target value of the PID controller, the controller generates a compensation signal that drives the printing press key actuator (e.g., a servo motor or pneumatic device) to perform corresponding physical adjustments. The introduction of PID control effectively overcomes the inertia, hysteresis, and nonlinearity problems of the printing press key actuator, achieving rapid response and precise control of the key opening, thereby ensuring the accurate implementation of spectral compensation.

[0100] In this embodiment, when the system determines that the color shift is mainly caused by mechanical deviation of the equipment, it no longer merely stops at the identification level, but can calculate a precise spectral compensation amount based on the quantified mechanical deviation component. Subsequently, this spectral compensation amount is effectively converted into the ink key opening adjustment value of each color group of the printing press, achieving a seamless connection from the spectral domain to the physical control domain. Furthermore, by employing PID control, these adjustment values ​​are output to the ink key actuator of the printing press, ensuring the accuracy, stability, and response speed of the ink key opening adjustment. This enables the printing press to compensate for color deviations caused by changes in mechanical state in real time and dynamically, improving color consistency and stability in the decorative paper printing process, effectively reducing scrap rates and production interruptions caused by mechanical problems of the equipment, thereby improving overall production efficiency and product quality.

[0101] In one feasible implementation, the method further includes: constructing a neural network comprising a recipe generation pathway and a scene robustness evaluation pathway; acquiring a training sample set comprising a target spectral sample, a primary color spectral dictionary, and spectral power distributions of multiple application scenario light sources; the recipe generation pathway outputting a predicted pigment volume concentration recipe based on the input target spectral sample, and calculating the spectral reconstruction error between the reconstructed spectrum corresponding to the predicted pigment volume concentration recipe and the target spectral sample; the scene robustness evaluation pathway obtaining a reconstructed spectrum based on the predicted pigment volume concentration recipe and the primary color spectral dictionary, and calculating the color difference between the reconstructed spectrum and the target spectral sample under multiple application scenario light sources, and obtaining a multi-scene metamerism index by combining the color differences of each scenario; using the simultaneous reduction of the spectral reconstruction error and the multi-scene metamerism index as a joint optimization objective, alternately and adversarially adjusting the parameters of the recipe generation pathway and the scene robustness evaluation pathway until the convergence condition is met, so that the trained recipe generation pathway is used as the recipe generation model.

[0102] In this embodiment, the formulation generation pathway is responsible for converting the target spectrum into a pigment formulation, while the scene robustness assessment pathway is responsible for evaluating the color consistency of the formulation under different light sources. This dual-pathway architecture allows the model to optimize formulation accuracy while also considering its robustness under complex lighting conditions. The neural network can employ a deep learning model, such as one built based on convolutional neural networks, recurrent neural networks, or fully connected networks. The formulation generation pathway can be an encoder-decoder structure that maps high-dimensional spectral data to low-dimensional pigment volume concentration formulations. The scene robustness assessment pathway can be a classifier or regressor that receives the formulation and a primary color spectral dictionary, and outputs multi-scene color difference assessment results.

[0103] In this embodiment, to train the neural network, a training sample set is needed, including target spectral samples, a primary color spectral dictionary, and the spectral power distributions of light sources from multiple application scenarios. Target spectral samples can be obtained through spectral measurements of actual printed materials or standard color blocks. The primary color spectral dictionary can be constructed by measuring the spectral reflectance curves of each primary color pigment at different concentrations. The spectral power distributions of light sources from multiple application scenarios can be obtained from standard light source databases (such as CIE standard light sources D65, A, F2, etc.) or by actually measuring the spectral characteristics of different lighting environments. These data collectively provide the model with a basis for learning color matching rules and evaluating metamerism.

[0104] In this embodiment, during training, the formulation generation pathway outputs a predicted pigment volume concentration formulation based on the input target spectral sample, and calculates the spectral reconstruction error between the reconstructed spectrum corresponding to the predicted pigment volume concentration formulation and the target spectral sample. The formulation generation pathway receives the target spectral sample as input, processes it through its internal neural network structure, and outputs a vector containing the volume concentrations of multiple primary color pigments, i.e., the predicted pigment volume concentration formulation. Then, using a primary color spectral dictionary and a color mixing model (such as the Kubelka-Munk theory or its variants), the predicted pigment volume concentration formulation is converted into a reconstructed spectrum. The spectral reconstruction error can be quantified by calculating indicators such as the Euclidean distance, mean square error, or CIE ΔE between the reconstructed spectrum and the target spectral sample.

[0105] In this embodiment, the scene robustness assessment pathway obtains the reconstructed spectrum based on the predicted pigment volume concentration formulation and the primary color spectral dictionary, and calculates the color difference between the reconstructed spectrum and the target spectral sample under multiple application scenario light sources. The multi-scene metamerism index is obtained by combining the color differences across scenarios. The scene robustness assessment pathway first receives the predicted pigment volume concentration formulation and the primary color spectral dictionary output by the formulation generation pathway, and generates the reconstructed spectrum. Then, for each application scenario light source, using the spectral power distribution of that light source, the reconstructed spectrum, and the target spectral sample, the color difference between the reconstructed spectrum and the target spectral sample under that light source is calculated according to CIE colorimetric principles (such as CIE L*a*b* space). The multi-scene metamerism index can be obtained by weighted averaging of the color differences under all application scenario light sources, taking the maximum value, or using other statistical methods to reflect the overall color consistency performance of the formulation across multiple scenarios.

[0106] In this embodiment, the joint optimization objective is to simultaneously reduce spectral reconstruction error and multi-scene metamerism index. The parameters of the recipe generation pathway and the scene robustness assessment pathway are alternately and adversarially adjusted until a convergence condition is met. The trained recipe generation pathway is then used as the recipe generation model. The joint optimization objective can be represented as a loss function, which is the weighted sum of spectral reconstruction error and multi-scene metamerism index. Alternating adversarial adjustments allow the two pathways to mutually promote each other, ultimately reaching a balance and generating both accurate and robust recipes. The convergence condition can be set as the change in the loss function being less than a certain threshold, or reaching a preset number of training epochs.

[0107] Through the above technical solution, this application can train a highly robust formula generation model. This model can not only accurately convert the target spectrum into a pigment volume concentration formula, but more importantly, it fully considers the color performance under different application scenarios and light sources during the training process. This ensures that the generated formula can maintain excellent color consistency under various lighting conditions, effectively solving the common metamerism problem in decorative paper printing, improving the color quality and stability of printed materials, and meeting the high standard color requirements of decorative paper products in complex usage environments.

[0108] In one feasible implementation, the alternating adversarial adjustment includes: fixing the parameters of the scene robustness assessment pathway, updating the parameters of the formulation generation pathway with the goal of reducing the spectral reconstruction error, so that the formulation generation pathway learns to map the target spectrum to a pigment volume concentration formulation; fixing the parameters of the formulation generation pathway, calculating the multi-scene metamerism index based on the predicted pigment volume concentration formulation output by the current formulation generation pathway, and using the multi-scene metamerism index as a feedback signal to update the parameters of the scene robustness assessment pathway, so that the scene robustness assessment pathway learns to identify the multi-scene color difference risk of the formulation; in the next iteration, adding the multi-scene metamerism index calculated by the scene robustness assessment pathway as a penalty term to the loss function of the formulation generation pathway, forcing the formulation generation pathway to prioritize color schemes with lower multi-scene metamerism indices when updating parameters; repeating the above steps until both the spectral reconstruction error and the multi-scene metamerism index are reduced to a preset convergence interval.

[0109] In this embodiment, during training, the parameters of the scene robustness evaluation pathway are initially kept constant, serving as the fixed evaluator for the current stage. At this point, the optimization focus is on the formulation generation pathway, aiming to minimize the spectral reconstruction error between the reconstructed spectrum corresponding to the predicted pigment volume concentration formulation generated by it and the target spectral sample. In this way, the formulation generation pathway can focus on learning how to accurately convert the input target spectrum into the corresponding pigment volume concentration formulation, ensuring the spectral accuracy of the generated formulation. This process can employ optimization algorithms such as gradient descent, iteratively adjusting the weights of the formulation generation pathway based on the spectral reconstruction error.

[0110] In this embodiment, after the parameters of the formulation generation pathway are fixed, the training focus shifts to the scene robustness assessment pathway. This pathway receives the predicted pigment volume concentration formulation output by the current formulation generation pathway and calculates the multi-scene metamerism index under multiple application scenario light sources based on this. This multi-scene metamerism index is used as a feedback signal to guide the update of the scene robustness assessment pathway parameters. Through this step, the scene robustness assessment pathway can learn and improve its ability to identify and quantify the color difference risk of different color schemes under multiple scenarios, thus becoming an effective metamerism risk assessor.

[0111] In this embodiment, in subsequent training iterations, the multi-scene metamerism index calculated by the scene robustness evaluation pathway is cleverly integrated into the loss function of the recipe generation pathway, specifically as a penalty term. This means that when the recipe generation pathway updates its parameters again, its optimization objective not only includes reducing spectral reconstruction errors but also must simultaneously consider minimizing the multi-scene metamerism index determined by the scene robustness evaluation pathway. This mechanism effectively forces the recipe generation pathway to actively avoid color schemes that are prone to significant color differences under different light sources when generating pigment volume concentration recipes, thereby prioritizing recipes with better multi-scene color consistency.

[0112] In this embodiment, the parameter update processes in the two stages described above will be performed alternately, forming an iterative loop. In each iteration, the recipe generation pathway and the scenario robustness evaluation pathway will be adjusted and learned according to their respective optimization objectives. This iterative process will continue until the two key indicators, spectral reconstruction error and multi-scenario metamerism index, both reach the preset convergence conditions. For example, when their values ​​are below a specific threshold or change only slightly within a certain number of iterations, it indicates that the model has reached a stable and high-performance state.

[0113] In this embodiment, through the above-described technical solution, this application employs an alternating adversarial adjustment training strategy, effectively solving the balance problem that may arise when jointly optimizing spectral reconstruction error and metamerism index in multiple scenarios. This strategy enables the formulation generation pathway to fully utilize the metamerism risk feedback provided by the scenario robustness assessment pathway while learning spectral accuracy. This allows it to proactively avoid schemes prone to color differences under different light sources when generating pigment volume concentration formulations. This iterative and mutually reinforcing training method ensures that the final trained formulation generation model not only provides formulations with high spectral accuracy, but more importantly, these formulations meet the preset color consistency requirements under multiple application scenario light sources, improving the robustness and reliability of decorative paper printing pigment formulations and reducing the risk of color deviation in actual production.

[0114] In one feasible implementation, when the batch-specific shift of pigment is determined to be the dominant factor, the method further includes: extracting the material absorption core feature of the current pigment from the pigment batch-specific shift component before inputting the standard multispectral data into the pre-trained formulation generation model; the material absorption core feature reflecting the difference in absorption spectrum between the current batch of pigment and the standard batch; inputting the material absorption core feature as an auxiliary input, together with the standard multispectral data, into the pre-trained formulation generation model; and generating a new pigment volume concentration formulation that is modified for the characteristics of the current batch of pigment by combining the pre-trained formulation generation model with the material absorption core feature.

[0115] In this embodiment, the material absorption core feature refers to a set of parameters that can quantify and characterize the light absorption capacity of a pigment at different wavelengths. Essentially, it reflects the inherent optical properties of the pigment and its molecular structure's efficiency in absorbing light energy. When extracting this feature, the batch-specific shift component of the pigment can be inverted based on a physical model (e.g., Kubelka-Munk theory) to obtain the deviation of the absorption coefficient or scattering coefficient of the current batch of pigment relative to the standard pigment. For example, by analyzing the spectral shape and intensity of the batch-specific shift component, spectral decomposition techniques (such as nonnegative matrix factorization, independent component analysis) or machine learning methods can be used to separate the core components directly related to the pigment's absorption characteristics from the complex shift signal. The extraction of this feature aims to accurately capture the differences in spectral absorption between the current batch of pigment and the standard batch, providing a quantitative basis for subsequent formulation adjustments.

[0116] In this embodiment, the material absorption kernel features serve as auxiliary input, which can be concatenated or fused with standard multispectral data in vector, matrix, or specific encoding form, or fed as an independent input channel into the pre-trained recipe generation model. For example, in a neural network model, the material absorption kernel features can be used as an additional input layer or fused with the feature representation of standard multispectral data in an intermediate layer. This auxiliary input mechanism allows the model to consider not only the target color itself but also the specific optical characteristics of the current pigment batch when generating recipes.

[0117] In this embodiment, the pre-trained formula generation model, such as a deep learning-based neural network, after receiving standard multispectral data and material absorption kernel features, integrates the two types of information using the complex mapping relationships learned internally. The model dynamically adjusts the proportions of each primary color pigment in the pigment volume concentration formula based on the absorption differences between the current pigment batch and the standard pigment batch, as reflected in the material absorption kernel features, to compensate for these differences. This ensures that even with fluctuations in pigment batches, the final printed color accurately matches the target standard.

[0118] In this embodiment, when batch-specific shift of pigment is determined to be the dominant factor, the material absorption core feature of the current pigment can be extracted from the batch-specific shift component. This feature accurately quantifies the difference in absorption spectrum between the current pigment batch and the standard batch. This material absorption core feature is used as auxiliary input and fed along with standard multispectral data into a pre-trained formulation generation model. This allows the model to intelligently correct itself by fully incorporating the actual optical characteristics of the current pigment batch when generating new pigment volume concentration formulations. This avoids formulation deviations that may result from relying solely on standard multispectral data, improving the accuracy of pigment formulations and their adaptability to actual pigment batch variations. Ultimately, this solution ensures high-precision color matching of printed materials across different pigment batches, effectively reducing color inconsistencies caused by pigment batch fluctuations, thereby improving the color quality and production stability of decorative paper printing.

[0119] In one feasible implementation, the method further includes: integrating the total spectral residual, the decomposition results of the equipment mechanical offset component and the pigment batch offset component generated during the closed-loop repair process, the replaced old formula, the new pigment volume concentration formula, and the color difference data measured in multiple application scenarios after repair into a repair case, and storing the repair case in a case library; after obtaining a new total spectral residual and completing the decomposition, calculating the similarity between the decomposition result of the currently obtained total spectral residual and the decomposition result of the existing cases in the case library; when the similarity exceeds a preset matching threshold, extracting the corresponding new formula from the matched existing cases as the initial correction formula, and directly applying the initial correction formula to the printing press.

[0120] In this embodiment, after a color shift occurs in the printing press and the aforementioned closed-loop repair process is completed, the system automatically or manually collects key data related to the repair. This data includes the total spectral residual reflecting the overall color shift, the decomposition results of the equipment mechanical shift component and pigment batch shift component revealing the root cause and extent of the shift, the old formula used before repair, the new pigment volume concentration formula generated and applied to the printing press after precise calculation and verification, and color difference data measured under different application scenarios and light sources after repair to ensure color consistency. This data is structured and integrated into a repair case and stored in a searchable and reusable case library, thereby transforming the experience gained from each problem-solving into usable knowledge assets.

[0121] In this embodiment, during subsequent printing production, after obtaining a new total spectral residual and completing its decomposition, the system calculates the similarity between the decomposition result of the currently obtained total spectral residual and the decomposition results of existing cases in the case library. For example, various measurement methods such as Euclidean distance, cosine similarity, or weighted similarity can be used to quantify the similarity between the current color shift problem and previously solved problems. In this way, the system can intelligently identify whether the current problem is highly similar to a previously successfully solved problem.

[0122] In this embodiment, when the calculated similarity exceeds a preset matching threshold, the system extracts the corresponding new formula from the matched existing cases as the initial correction formula and directly applies the initial correction formula to the printing press. This means that for color shift problems that are highly similar to historical cases, there is no need to restart the complex formula generation model to calculate from scratch; instead, the already verified and effective solution can be directly reused. For example, if the current total spectral residual decomposition result shows extremely high similarity to the decomposition result of a historical case in the case library in terms of equipment mechanical offset component and pigment batch offset component, and the repair effect of the historical case is good, the system will directly extract the new pigment volume concentration formula stored in the historical case and send it to the printing press as the initial correction formula to quickly adjust the pigment formula.

[0123] In this embodiment, a repair case library is established to structure and store the experience and results of each closed-loop repair. When a new color shift problem is encountered, the system can intelligently compare the current problem with historical cases and quickly identify similar, already resolved problems. Once the similarity reaches a preset standard, the validated pigment formula from the historical cases can be directly extracted and applied as the initial correction formula to the printing press. This case-based reasoning mechanism reduces the need for complex analysis and formula generation calculations from scratch every time a color shift is encountered, thereby significantly shortening the time for fault diagnosis and correction, improving the efficiency and response speed of printing production, and reducing computational resource consumption. Its advantages are particularly evident for frequently occurring similar problems.

[0124] In the embodiments of this application, the color matching method for decorative paper printing pigments acquires multispectral data and decomposes the total spectral residual to identify the dominant factors of mechanical deviation of equipment and batch deviation of pigments. Combined with a pre-trained formula generation model, it generates robust formulas under multi-scenario constraints, effectively distinguishes the root causes of color deviation and ensures color consistency under multi-lighting environments. It can intelligently identify the dominant factors of color deviation and generate pigment formulas with consistent colors in multiple scenarios, thereby reducing manual intervention, avoiding misjudgment and repair measures, and improving printing efficiency and product quality.

[0125] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the color matching method of decorative paper printing pigments in this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0126] This application also provides a color matching system for decorative paper printing pigments, see reference. Figure 4 The decorative paper printing pigment color matching system includes: a memory 10, a processor 20, and a decorative paper printing pigment color matching program stored on the memory 10 and executable on the processor 20. The decorative paper printing pigment color matching program is configured to implement the steps of the decorative paper printing pigment color matching method.

[0127] The decorative paper printing pigment color matching system provided in this application, employing the decorative paper printing pigment color matching method in the above embodiments, can improve printing efficiency and product quality. Compared with the prior art, the beneficial effects of the decorative paper printing pigment color matching system provided in this application are the same as those of the decorative paper printing pigment color matching method provided in the above embodiments, and other technical features of the decorative paper printing pigment color matching system are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0128] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0129] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. All equivalent structural transformations made under the technical concept of this application using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included within the scope of patent protection of this application.

Claims

1. A method for color matching of printing pigments for decorative paper, characterized in that, The method includes: Acquire current multispectral data of online printed materials from the printing press, as well as pre-registered standard multispectral data; The total spectral residual is determined based on the current multispectral data and the standard multispectral data, and the total spectral residual is decomposed into equipment mechanical offset component and pigment batch offset component according to the preset offset separation rule. Analyze the statistical characteristics of the equipment mechanical offset component and the pigment batch offset component to determine the dominant factors causing color shift; the dominant factors include equipment mechanical offset and pigment batch offset. When the batch shift of pigment is determined to be the dominant factor, the standard multispectral data is used as the target spectrum and input into the pre-trained formula generation model. The pre-trained formula generation model generates a new pigment volume concentration formula. The pre-trained formula generation model is constrained by the metamerism index of multiple scenarios during training, so that the generated formula meets the preset color consistency requirements under multiple application scenario light sources. The new pigment volume concentration formulation is output to the printing press to replace the currently used pigment formulation.

2. The method for color matching of printing pigments on decorative paper as described in claim 1, characterized in that, The steps of determining the total spectral residual based on the current multispectral data and the standard multispectral data, and decomposing the total spectral residual into equipment mechanical offset components and pigment batch offset components according to a preset offset separation rule, include: Acquire pre-established device-sensitive directional features and material-sensitive directional features; wherein, the device-sensitive directional features characterize the migration pattern of the printed spectrum as the mechanical state of the equipment changes and is independent of the pigment batch, and the material-sensitive directional features characterize the absorption change pattern of the printed spectrum as the pigment batch changes and is independent of the mechanical state of the equipment; The total spectral residual is projected onto the sensitive direction feature of the device to obtain the mechanical offset component of the device; The total spectral residual is projected onto the material-sensitive directional features to obtain the pigment batch-specific offset component.

3. The method for color matching of printing pigments on decorative paper as described in claim 2, characterized in that, The establishment of the equipment-sensitive direction characteristics and material-sensitive direction characteristics includes: Standard samples were printed using the same batch of pigments under multiple different equipment status parameters. Multispectral data was collected, and spectral transfer characteristics were extracted from the collected multispectral data. Principal component analysis was performed on the spectral transfer characteristics obtained under multiple different equipment state parameters. The first principal component with the largest contribution was extracted as the first feature component that only changes with the equipment state parameters and is not sensitive to the pigment batch. This first feature component was determined as the equipment sensitive direction feature. Standard samples were printed under conditions of multiple different batches of pigments and constant equipment parameters. Multispectral data was collected, and spectral transfer characteristics were extracted from the collected multispectral data. Principal component analysis was performed on the spectral transfer characteristics obtained from multiple different batches of pigments. The second principal component with the largest contribution was extracted as the second feature component that only varies with the pigment batch and is not sensitive to the equipment state parameters. This second feature component was determined as the material-sensitive directional feature. Store the device's sensitive direction features and the material's sensitive direction features for subsequent decomposition and retrieval.

4. The method for color matching of printing pigments on decorative paper as described in claim 1, characterized in that, The steps for analyzing the statistical characteristics of the mechanical offset component of the equipment and the batch-specific offset component of the pigment to determine the dominant factor causing the color shift include: Calculate the energy percentage of the mechanical offset component of the equipment in the total spectral residual, and use it as the equipment offset index; The energy percentage of the pigment batch-specific offset component in the total spectral residual is calculated and used as the material offset index; When the equipment offset index is greater than the preset equipment offset threshold and the equipment offset index is greater than the material offset index, the mechanical offset of the equipment is determined to be the dominant factor; otherwise, the batch offset of the pigment is determined to be the dominant factor.

5. The method for color matching of printing pigments on decorative paper as described in claim 1, characterized in that, When mechanical misalignment of the equipment is determined to be the dominant factor, the method further includes: The spectral compensation amount is calculated based on the mechanical offset component of the equipment, and the spectral compensation amount makes the compensated predicted spectrum consistent with the standard multispectral data in the sensitive direction of the equipment. The spectral compensation amount is mapped to the ink bond opening adjustment value of each color group according to a preset conversion relationship; The ink key opening adjustment value is used as a compensation signal and output to the ink key actuator of the printing press through PID control.

6. The method for color matching of printing pigments on decorative paper as described in claim 1, characterized in that, The method further includes: Construct a neural network that includes a recipe generation pathway and a scenario robustness evaluation pathway; Obtain a training sample set containing target spectral samples, primary color spectral dictionary, and spectral power distributions of light sources from multiple application scenarios; The formulation generation pathway outputs a predicted pigment volume concentration formulation based on the input target spectral sample, and calculates the spectral reconstruction error between the reconstructed spectrum corresponding to the predicted pigment volume concentration formulation and the target spectral sample. The reconstructed spectrum is obtained by the scenario robustness assessment pathway based on the predicted pigment volume concentration formulation and the primary color spectral dictionary, and the color difference between the reconstructed spectrum and the target spectral sample is calculated under multiple application scenario light sources. The multi-scenario metamerism index is obtained by combining the color differences of each scenario. With the simultaneous reduction of the spectral reconstruction error and the multi-scene metamerism index as the joint optimization objective, the parameters of the recipe generation pathway and the scene robustness evaluation pathway are alternately and adversarially adjusted until the convergence condition is met, so that the trained recipe generation pathway can be used as the recipe generation model.

7. The method for color matching of printing pigments on decorative paper as described in claim 6, characterized in that, The alternating antagonistic adjustment includes: The parameters of the scenario robustness evaluation pathway are fixed, and the parameters of the formulation generation pathway are updated with the goal of reducing the spectral reconstruction error, so that the formulation generation pathway learns to map the target spectrum into a pigment volume concentration formulation. The parameters of the formulation generation pathway are fixed, and the multi-scenario metamerism index is calculated based on the predicted pigment volume concentration formulation output by the current formulation generation pathway. The parameters of the scenario robustness assessment pathway are updated using the multi-scenario metamerism index as a feedback signal, so that the scenario robustness assessment pathway can learn to identify the multi-scenario color difference risk of the formulation. In the next iteration, the metamerism index of the multi-scene calculations of the scenario robustness evaluation pathway is added as a penalty term to the loss function of the recipe generation pathway, forcing the recipe generation pathway to prioritize color schemes with lower metamerism indices when updating parameters. Repeat the above steps until both the spectral reconstruction error and the multi-scene metamerism index are reduced to the preset convergence range.

8. The method for color matching of printing pigments on decorative paper as described in claim 1, characterized in that, When batch shift of pigments is determined to be the dominant factor, the method further includes: Before inputting the standard multispectral data into the pre-trained formulation generation model, the material absorption core features of the current pigment are extracted from the pigment batch offset component. These material absorption core features reflect the difference in absorption spectrum between the current batch of pigment and the standard batch. The material absorption core features are used as auxiliary inputs and input together with the standard multispectral data into the pre-trained formula generation model; The pre-trained formulation generation model, combined with the material absorption core characteristics, generates a new pigment volume concentration formulation that is modified to suit the characteristics of the current batch of pigments.

9. The method for color matching of printing pigments on decorative paper as described in claim 1, characterized in that, The method further includes: The total spectral residual, the decomposition results of the equipment mechanical offset component and the pigment batch offset component generated during this closed-loop repair process, the old formula that was replaced, the new pigment volume concentration formula, and the color difference data measured in multiple application scenarios after repair are integrated into a repair case and stored in the case library. After obtaining the new total spectral residual and completing the decomposition, the similarity between the decomposition result of the currently obtained total spectral residual and the decomposition result of the existing cases in the case library is calculated. When the similarity exceeds a preset matching threshold, the corresponding new formula is extracted from the existing matched cases as the initial corrected formula, and the initial corrected formula is directly applied to the printing press.

10. A color matching system for printing pigments on decorative paper, characterized in that, The decorative paper printing pigment color matching system includes: a memory, a processor, and a decorative paper printing pigment color matching program stored in the memory and executable on the processor, the decorative paper printing pigment color matching program being configured to implement the steps of the decorative paper printing pigment color matching method as described in any one of claims 1 to 9.

Images