Efficient and accurate frequency modulation screen printing color management color matching auxiliary system

Abstract

The invention provides an efficient and accurate frequency modulation screen printing color management color matching auxiliary system which comprises a collector, the collector comprises an image collection module, the image collection module can capture a multi-channel filter array of a printing manuscript through a multispectral imaging device, and an analyzer comprises a spectral analysis module and a color matching prediction module. The spectral analysis module is in communication connection with the image acquisition module, after low-dimensional feature vectors are generated, candidate schemes are listed, and the color matching prediction module predicts an ink matching scheme based on the low-dimensional feature vectors and a deep convolutional neural network; relates to the technical field of frequency modulation screen printing toning. According to the method, the limitation of traditional single-factor color matching is broken through by introducing a multi-modal data fusion (spectrum + base material + ambient light) and Bayesian optimization feedback mechanism, the nonlinear mapping problem of high-dimensional spectral data is solved through cooperation of manifold learning dimension reduction and deep neural network prediction, and efficient and accurate frequency modulation network color matching adjustment is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of FM screening printing color matching, and specifically to an efficient and accurate color management color matching auxiliary system for FM screening printing. Background Art

[0002] Due to advantages such as uniform dot distribution and smooth color transition, FM screening printing technology is widely used in high-end packaging, art reproduction and other fields. Currently, the mainstream color management solutions in the industry are mostly based on spectral reflectance measurement or CIELAB colorimetric model, and color matching is achieved by matching the chromaticity parameters of the original manuscript and the printed matter. Some systems use empirical formulas or linear regression models to predict ink ratios and rely on manual calibration to correct color differences. In addition, there have been attempts in the prior art to store printing equipment parameters and substrate characteristics through a static database, such as adjusting the dot generation algorithm by presetting ICC profile files.

[0003] However, the above technologies have significant defects:

[0004] 1. Traditional solutions only rely on spectral or chromaticity data, ignoring the influence of substrate ink absorption characteristics and environmental light changes on color presentation, resulting in poor color reproduction consistency between the laboratory environment and the production site, leading to single data dimension and poor color matching effect;

[0005] 2. Static color matching models cannot respond to dynamic interference factors such as ink viscosity fluctuations and equipment wear during the printing process, and it is necessary to frequently stop the machine for manual parameter adjustment, which seriously reduces production efficiency, resulting in insufficient dynamic adaptability and slow adjustment efficiency. Summary of the Invention

[0006] The purpose of the present invention is to provide an efficient and accurate color management color matching auxiliary system for FM screening printing to solve the problems raised in the background art above and overcome the existing technical defects.

[0007] To solve the above technical problems, the technical solution adopted by the present invention is: an efficient and accurate color management color matching auxiliary system for FM screening printing, including a collector, the collector includes an image acquisition module, and the image acquisition module can capture a multi-channel filter array of a printed original manuscript through a multi-spectral imaging device;

[0008] An analyzer, the analyzer includes a spectral analysis module and a color matching prediction module, the spectral analysis module is communicatively connected to the image acquisition module, generates a low-dimensional feature vector and then lists candidate solutions, and the color matching prediction module predicts an ink ratio solution based on the low-dimensional feature vector and a deep convolutional neural network and generates a probability distribution of multiple candidate solutions; and

[0009] An adjuster, the adjuster includes a printing control module and a feedback calibration module. The printing control module is used for the FM screening printing equipment to dynamically adjust the parameters of the ink ratio scheme. The feedback calibration module iteratively corrects the parameters through the Bayesian algorithm and synchronously updates the correction results to the color matching prediction module.

[0010] A usage method of an efficient and accurate FM screening printing color management color matching auxiliary system, including

[0011] Step S1: Obtain the high-resolution spectral cube data of the printed original through a multispectral imaging device, and the sampling interval does not exceed 10nm;

[0012] Step S2: Denoise and normalize the original spectral data, and extract the light scattering characteristics caused by the micro-structure of the substrate surface;

[0013] Step S3: Input the preprocessed data into a multi-task deep learning model to simultaneously predict the ink ratio, dot distribution, and drying time parameters;

[0014] Step S31: Construct a heterogeneous data set, including spectral data under laboratory standard conditions and noise data collected from industrial sites;

[0015] Step S32: Adopt an attention mechanism to fuse spectral features and device state features to enhance the adaptability of the model to dynamic working conditions;

[0016] Step S33: Train the model in stages through a curriculum learning strategy, first optimize the color difference index, and then gradually introduce cost and efficiency constraints;

[0017] Step S4: Dynamically adjust the dot generation algorithm according to the real-time sensor data of the printing equipment (including ink viscosity, environmental humidity);

[0018] Step S41: Establish a physical model of ink diffusion - substrate penetration, and solve the partial differential equation to predict the dot edge blur;

[0019] Step S42: Adopt a meta-learning framework to quickly adapt the dot distribution parameters based on the test data of a small number of new substrates;

[0020] Step S43: In the high-speed printing mode, enable a lightweight convolutional neural network to optimize the dot spatial arrangement in real time;

[0021] Step S5: During the continuous printing process, perform time series analysis on the color difference data based on the sliding window mechanism to trigger local or global color matching parameter calibration.

[0022] As a further solution of the present invention: an efficient and accurate color management color matching auxiliary system for FM screening printing, the spectral analysis module includes a spectral dimensionality reduction unit and a gamut matching unit. The spectral dimensionality reduction unit maps high-dimensional spectral data into a three-dimensional feature space through a locally linear embedding algorithm and preserves the local non-linear structure of the spectral reflection curve. The gamut matching unit uses a non-dominated sorting genetic algorithm to solve a multi-objective optimization problem within the target gamut, and optimizes printing metrics including color difference, ink consumption, and printing efficiency.

[0023] As a further solution of the present invention: an efficient and accurate color management color matching auxiliary system for FM screening printing, the color matching prediction module includes a neural network prediction model, a generative adversarial network, and a solution optimization unit. The neural network prediction model has an input layer that fuses spectral feature vectors, substrate surface roughness parameters, and ambient light color temperature data, and the output layer generates an ink ratio probability matrix through a Softmax function. The generative adversarial network generates synthetic spectral data under limited sample conditions and expands the training data set. The solution optimization unit screens Pareto optimal solutions from candidate solutions according to a preset gamut coverage threshold and cost constraint conditions.

[0024] As a further solution of the present invention: an efficient and accurate color management color matching auxiliary system for FM screening printing, the feedback calibration module includes a color difference detection unit and a dynamic compensation unit. The color difference detection unit is configured to collect multi-point reflection spectra on the surface of the printed matter through an optical fiber spectrometer and calculate the root mean square error between it and the target spectrum. The dynamic compensation unit establishes a response surface model of color difference-ink ratio, updates the color matching weight coefficient through a gradient descent algorithm, and simultaneously introduces a coupling constraint condition of printing speed-drying time.

[0025] As a further solution of the present invention: an efficient and accurate color management color matching auxiliary system for FM screening printing further includes a cross-media database. The cross-media database includes a substrate optical property library, a device property library, and an environmental light library. The substrate optical property library stores absorption / reflection characteristic curves of different substrates in the ultraviolet-visible-near-infrared band. The device property library records the maximum dot frequency of multiple FM screening printers, the ink viscosity-temperature response curve, and historical calibration data. The environmental light library associates spectral power distribution data of standard light sources D50 / D65 and user-defined light sources.

[0026] As a further solution of the present invention: an efficient and accurate color management color matching auxiliary system for FM screening printing, the cross-media database establishes a substrate-ink-device ternary relationship network through knowledge graph technology, supports multi-condition retrieval based on semantic similarity, and the cross-media database adopts a federated learning framework to synchronously update the weight parameters of the neural network model across devices while protecting local data privacy.

[0027] As a further solution of the present invention: An efficient and accurate frequency modulation network printing color management color matching auxiliary system further includes a user interaction interface, and the user interaction interface includes a three-dimensional color gamut visualization unit, a dynamic simulation unit, and a decision support unit. The three-dimensional color gamut visualization unit displays the color space coverage ranges of the target color gamut and the candidate color matching schemes through volume rendering technology. The dynamic simulation unit generates a color rendering preview of the printed matter under different lighting conditions based on the ray tracing algorithm. The decision support unit provides a cost-quality Pareto frontier analysis of the color matching scheme and an interpretable feature importance ranking.

[0028] Compared with the prior art, the beneficial effects of the present invention include:

[0029] This application breaks through the limitations of traditional single-factor color matching by introducing multi-modal data fusion (spectrum + substrate + ambient light) and the Bayesian optimization feedback mechanism, and solves the non-linear mapping problem of high-dimensional spectral data through the cooperation of manifold learning dimensionality reduction and deep neural network prediction, realizing efficient and accurate frequency modulation network color matching adjustment. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The disclosure of the present invention will be described with reference to the accompanying drawings. It should be understood that the drawings are only for illustrative purposes and are not intended to limit the scope of protection of the present invention. In the drawings, the same reference numerals are used to refer to the same components. Among them:

[0031] Figure 1 Schematically shows a schematic structural diagram of a process method proposed according to an embodiment of the present invention;

[0032] Figure 2 Schematically shows a schematic flowchart of step S3 proposed according to an embodiment of the present invention;

[0033] Figure 3 Schematically shows a schematic flowchart of step S4 proposed according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0034] It is easy to understand that according to the technical solution of the present invention, without changing the essence of the present invention, those of ordinary skill in the art can propose various structural ways and implementation ways that can be mutually replaced. Therefore, the following detailed description and the accompanying drawings are only exemplary descriptions of the technical solution of the present invention and should not be regarded as all of the present invention or as a limitation or restriction on the technical solution of the present invention.

[0035] An efficient and accurate frequency modulation network printing color management color matching auxiliary system according to an embodiment of the present invention is shown in combination with the accompanying drawings, including a collector, and the collector includes an image acquisition module, and the image acquisition module can capture a multi-channel filter array of a printed original through a multi-spectral imaging device;

[0036] An analyzer, the analyzer includes a spectral analysis module and a color matching prediction module, the spectral analysis module is communicatively connected to the image acquisition module, generates a low-dimensional feature vector and then lists candidate solutions, and the color matching prediction module predicts an ink ratio solution based on the low-dimensional feature vector and a deep convolutional neural network, and generates a probability distribution of multiple candidate solutions; and

[0037] An adjuster, the adjuster includes a printing control module and a feedback calibration module, the printing control module is used for a frequency modulation network printing device to dynamically adjust the parameters of the ink ratio solution, and the feedback calibration module iteratively corrects the parameters through a Bayesian algorithm and synchronously updates the correction result to the color matching prediction module.

[0038] Therefore, this application breaks through the limitations of traditional single-factor color matching by introducing multi-modal data fusion (spectrum + substrate + ambient light), and solves the non-linear mapping problem of high-dimensional spectral data through the cooperation of manifold learning dimensionality reduction and deep neural network prediction.

[0039] Specifically, this application lists the technical route for breaking through the color matching limitations of the prior art by adopting a Bayesian optimization feedback mechanism, including

[0040] This application realizes the purpose of synergistic effect through multi-modal data fusion, and solves the limitations of traditional single-factor color matching. Specifically, the traditional color matching system only relies on spectral reflectance or a single chromaticity parameter (such as Lab* value), ignoring actual influencing factors such as the ink absorption characteristics of the substrate and changes in ambient light, resulting in a significant deviation between the color matching results in the laboratory and the printing effects in the industrial field. Therefore, this system constructs a color matching model that couples multiple physical fields by fusing spectral data (accurately characterizing the essence of the color of the original), substrate characteristics (surface roughness, ink absorption rate, optical scattering model), and ambient light parameters (light source color temperature, color rendering index, multi-angle irradiation conditions), significantly improving the color reproduction consistency across different media (such as paper, plastic, metal). The measured color difference ΔE (CIE 2000 standard) is reduced by 40%-60% compared with the single-factor method.

[0041] And this application can also dynamically adapt to complex working conditions. Because in variable ambient light (such as fluctuations in workshop lighting) or mixed substrate printing scenarios, the system automatically adjusts the ink ratio weight by real-time collecting multi-modal data, avoiding batch color difference problems caused by external interference in traditional static color matching solutions.

[0042] A method for using an efficient and accurate frequency modulation screen printing color management color matching auxiliary system, including

[0043] Step S1: Obtain high-resolution spectral cube data of the printed original through a multispectral imaging device, with a sampling interval not exceeding 10 nm;

[0044] Step S2: Denoise and normalize the original spectral data, and extract the light scattering characteristics caused by the micro-structure on the substrate surface;

[0045] Step S3: Input the preprocessed data into a multi-task deep learning model to simultaneously predict ink ratio, dot distribution, and drying time parameters;

[0046] Step S31: Construct a heterogeneous data set, including spectral data under laboratory standard conditions and noise data collected from industrial sites;

[0047] Step S32: Adopt an attention mechanism to fuse spectral features and device state features to enhance the model's adaptability to dynamic working conditions;

[0048] Step S33: Train the model in stages through a curriculum learning strategy, first optimizing the color difference index, and then gradually introducing cost and efficiency constraints;

[0049] Step S4: Dynamically adjust the dot generation algorithm according to the real-time sensor data of the printing device (including ink viscosity, environmental humidity);

[0050] Step S41: Establish a physical model of ink diffusion - substrate penetration, and solve the partial differential equation to predict the dot edge blur;

[0051] Step S42: Adopt a meta-learning framework to quickly adapt the dot distribution parameters based on the test data of a small amount of new substrates;

[0052] Step S43: In the high-speed printing mode, enable a lightweight convolutional neural network to optimize the dot spatial arrangement in real time;

[0053] Step S5: During continuous printing, perform time series analysis on the color difference data based on the sliding window mechanism to trigger local or global color matching parameter calibration.

[0054] As a further solution of the present invention: An efficient and accurate frequency modulation screen printing color management color matching auxiliary system, the spectral analysis module includes a spectral dimensionality reduction unit and a color gamut matching unit. The spectral dimensionality reduction unit maps the high-dimensional spectral data to a three-dimensional feature space through a locally linear embedding algorithm and retains the local non-linear structure of the spectral reflection curve; the color gamut matching unit uses a non-dominated sorting genetic algorithm to solve multi-objective optimization problems within the target color gamut, and optimizes printing indicators including color difference, ink consumption, and printing efficiency.

[0055] As a further solution of the present invention: an efficient and accurate color management color matching assistance system for FM screening printing, the color matching prediction module includes a neural network prediction model, a generative adversarial network, and a solution optimization unit. The neural network prediction model is provided with an input layer that fuses spectral feature vectors, substrate surface roughness parameters, and ambient light color temperature data, and the output layer generates an ink ratio probability matrix through the Softmax function. The generative adversarial network generates synthetic spectral data and expands the training data set under the condition of limited samples; the solution optimization unit screens out Pareto optimal solutions from candidate solutions according to the preset gamut coverage threshold and cost constraint conditions.

[0056] As a further solution of the present invention: an efficient and accurate color management color matching assistance system for FM screening printing, the feedback calibration module includes a color difference detection unit and a dynamic compensation unit. The color difference detection unit is configured to collect multi-point reflection spectra on the surface of the printed matter through an optical fiber spectrometer and calculate the root mean square error between it and the target spectrum. The dynamic compensation unit establishes a response surface model of color difference-ink ratio, updates the color matching weight coefficient through the gradient descent algorithm, and simultaneously introduces the coupling constraint condition of printing speed-drying time.

[0057] As a further solution of the present invention: an efficient and accurate color management color matching assistance system for FM screening printing, further comprising a cross-media database. The cross-media database includes a substrate optical property library, a device property library, and an environmental light library. The substrate optical property library stores absorption / reflection characteristic curves of different substrates in the ultraviolet-visible-near-infrared band. The device property library records the maximum dot frequency of multiple FM screening printers, the ink viscosity-temperature response curve, and historical calibration data. The environmental light library associates the spectral power distribution data of standard light sources D50 / D65 and user-defined light sources.

[0058] As a further solution of the present invention: an efficient and accurate color management color matching assistance system for FM screening printing, the cross-media database establishes a substrate-ink-device ternary relationship network through knowledge graph technology, supports multi-condition retrieval based on semantic similarity, and the cross-media database adopts a federated learning framework to synchronously update the weight parameters of the neural network model across devices while protecting local data privacy.

[0059] As a further solution of the present invention: An efficient and accurate color management color matching assistance system for frequency modulation network printing further includes a user interaction interface, which includes a three-dimensional color gamut visualization unit, a dynamic simulation unit, and a decision support unit. The three-dimensional color gamut visualization unit displays the color space coverage ranges of the target color gamut and the candidate color matching schemes through volume rendering technology. The dynamic simulation unit generates a color rendering preview of the printed matter under different lighting conditions based on the ray tracing algorithm. The decision support unit provides a cost-quality Pareto frontier analysis of the color matching scheme and an interpretable feature importance ranking.

[0060] Working principle: This application breaks through the limitations of traditional single-factor color matching by introducing multi-modal data fusion (spectrum + substrate + ambient light) and a Bayesian optimization feedback mechanism, and solves the non-linear mapping problem of high-dimensional spectral data through the collaboration of manifold learning dimensionality reduction and deep neural network prediction.

[0061] Specifically, this application lists the technical routes for breaking through the color matching limitations of the existing technology by using the Bayesian optimization feedback mechanism, including

[0062] This application achieves the purpose of synergistic effect through multi-modal data fusion and solves the limitations of traditional single-factor color matching. Specifically, the traditional color matching system only relies on spectral reflectance or a single chromaticity parameter (such as Lab* value), ignoring actual influencing factors such as the ink absorption characteristics of the substrate and the change of ambient light, resulting in a significant deviation between the color matching results in the laboratory and the printing effect in the industrial field. Therefore, this system constructs a color matching model with multi-physical field coupling by fusing spectral data (accurately characterizing the essence of the original manuscript color), substrate characteristics (surface roughness, ink absorption rate, optical scattering model), and ambient light parameters (light source color temperature, color rendering index, multi-angle illumination conditions), significantly improving the color reproduction consistency across different media (such as paper, plastic, metal). The measured color difference ΔE (CIE 2000 standard) is reduced by 40%-60% compared with the single-factor method.

[0063] Moreover, this application can also dynamically adapt to complex working conditions. In the case of variable ambient light (such as workshop lighting fluctuations) or mixed substrate printing scenarios, the system automatically adjusts the ink ratio weight by real-time collecting multi-modal data, avoiding batch color difference problems caused by external interference in traditional static color matching schemes.

[0064] The technical scope of the present invention is not limited to the content described above. Those skilled in the art can make various deformations and modifications to the above embodiments without departing from the technical idea of the present invention, and these deformations and modifications should all fall within the protection scope of the present invention.

Claims

1. An efficient and accurate FM screen printing color management and color matching auxiliary system, characterized in that: It includes a collector, the collector includes an image acquisition module, and the image acquisition module can capture a multi-channel filter array of a printed manuscript through a multi-spectral imaging device; An analyzer, the analyzer comprising a spectral analysis module and a color matching prediction module, the spectral analysis module being in communication with the image acquisition module, generating a low-dimensional feature vector and then listing candidate solutions, the color matching prediction module predicting an ink ratio solution based on the low-dimensional feature vector and a deep convolutional neural network, and generating a probability distribution of multiple candidate solutions; and The adjuster includes a printing control module and a feedback calibration module. The printing control module is used for the FM screen printing device to dynamically adjust the ink ratio scheme parameters. The feedback calibration module iteratively corrects the parameters through the Bayesian algorithm and synchronously updates the correction results to the color matching prediction module.

2. According to claim 1, a highly efficient and accurate FM screen printing color management and color matching auxiliary system is characterized in that: The spectral analysis module includes a spectral dimension reduction unit and a color gamut matching unit. The spectral dimension reduction unit maps high-dimensional spectral data into a three-dimensional feature space through a local linear embedding algorithm and retains the local nonlinear structure of the spectral reflectance curve; the color gamut matching unit uses a non-dominated sorting genetic algorithm to solve a multi-objective optimization problem in a target color gamut to optimize printing indicators including color difference, ink consumption and printing efficiency.

3. According to claim 2, a highly efficient and accurate FM screen printing color management and color matching auxiliary system is characterized in that: The color matching prediction module includes a neural network prediction model, a generative adversarial network and a scheme optimization unit. The neural network prediction model is provided with an input layer that fuses spectral feature vectors, substrate surface roughness parameters and ambient light color temperature data, and the output layer generates an ink ratio probability matrix through a Softmax function. The generative adversarial network generates synthetic spectral data under finite sample conditions and expands the training data set; the scheme optimization unit selects the Pareto optimal solution from the candidate schemes according to a preset color gamut coverage threshold and cost constraints.

4. According to claim 3, a highly efficient and accurate FM screen printing color management and color matching auxiliary system is characterized in that: The feedback calibration module includes a color difference detection unit and a dynamic compensation unit. The color difference detection unit is configured to collect multi-point reflection spectra on the surface of the printed product through a fiber optic spectrometer and calculate the root mean square error between the color difference and the target spectrum. The dynamic compensation unit establishes a response surface model of color difference-ink ratio, updates the color matching weight coefficient through a gradient descent algorithm, and introduces a coupling constraint condition of printing speed-drying time.

5. The efficient and accurate FM screen printing color management and color matching auxiliary system according to claim 4, characterized in that: It also includes a cross-media database, which includes a substrate optical property library, a device property library and an ambient lighting library. The substrate optical property library stores the absorption / reflection characteristic curves of different substrates in the ultraviolet-visible-near infrared bands, the device property library records the maximum dot frequency, ink viscosity-temperature response curves and historical calibration data of multiple frequency-modulated screen printers, and the ambient lighting library associates the spectral power distribution data of standard light sources D50 / D65 and user-defined light sources.

6. The efficient and accurate FM screen printing color management and color matching auxiliary system according to claim 5, characterized in that: The cross-media database establishes a substrate-ink-equipment ternary relationship network through knowledge graph technology, supports multi-condition retrieval based on semantic similarity, and adopts a federated learning framework to synchronously update the weight parameters of the neural network model across devices while protecting local data privacy.

7. The efficient and accurate FM screen printing color management and color matching auxiliary system according to claim 6, characterized in that: It also includes a user interaction interface, which includes a three-dimensional color gamut visualization unit, a dynamic simulation unit and a decision support unit. The three-dimensional color gamut visualization unit displays the color space coverage of the target color gamut and the candidate color schemes through volume rendering technology. The dynamic simulation unit generates a color rendering preview of the printed product under different lighting conditions based on a ray tracing algorithm. The decision support unit provides a cost-quality Pareto front analysis of the color scheme and an importance ranking of explainable features.

8. A method for using the efficient and accurate FM screen printing color management and color matching auxiliary system according to claims 1-7, characterized in that: include Step S1: obtaining high-resolution spectral cube data of the printed manuscript through a multispectral imaging device, with a sampling interval not exceeding 10 nm; Step S2: De-noising and normalizing the original spectral data to extract the light scattering characteristics caused by the microstructure of the substrate surface; Step S3: input the preprocessed data into a multi-task deep learning model to simultaneously predict ink ratio, dot distribution and drying time parameters; Step S4: dynamically adjusting the dot generation algorithm according to the real-time sensor data of the printing device (including ink viscosity and ambient humidity); Step S5: During the continuous printing process, the color difference data is analyzed in time series based on the sliding window mechanism to trigger local or global color matching parameter calibration.

9. The method for using the efficient and accurate FM screen printing color management and color matching auxiliary system according to claim 8, characterized in that: The training process of the multi-task deep learning model in step S3 includes: Step S31: construct a heterogeneous data set, including spectral data under a standard laboratory environment and noise data collected at an industrial site; Step S32: using the attention mechanism to fuse the spectral features and the equipment status features to enhance the adaptability of the model to dynamic working conditions; Step S33: The model is trained in stages through a course learning strategy, first optimizing the color difference index, and then gradually introducing cost and efficiency constraints.

10. The method for using the efficient and accurate FM screen printing color management and color matching auxiliary system according to claim 9, characterized in that: The dynamic adjustment of the dot generation algorithm in step S4 includes: Step S41: establishing a physical model of ink diffusion-substrate penetration, solving a partial differential equation to predict dot edge blur; Step S42: using a meta-learning framework to quickly adapt the dot distribution parameters based on a small amount of test data of new substrates; Step S43: In high-speed printing mode, a lightweight convolutional neural network is enabled to optimize the dot space arrangement in real time.

Images