Deep learning self-calibration method and system for high-precision color management

Abstract

The invention belongs to the technical field of printing color management, and provides a deep learning self-calibration method and system for high-precision color management, and the method comprises the steps: carrying out the comparison analysis of the ink layer thickness and the color difference, and determining the ink layer critical thickness of non-linear saturation of different ink types in a low-ink-amount printing scene; according to the ink layer critical thickness and the ink layer thickness data above the ink layer critical thickness and the corresponding color difference data, in combination with a nonlinear fitting mode, a nonlinear correlation model of different ink types about the ink layer thickness and the color difference under the printing low-ink-amount scene is constructed; according to the color difference between the current printing color difference and the target set color difference and in combination with an existing deep learning color calibration model, whether the ink layer thickness when the target set color difference is reached exceeds the ink layer critical thickness or not is predicted and judged; and the final ink layer increased thickness is determined by selectively combining the non-linear correlation models of different ink types about the ink layer thickness and the chromatic aberration under the printing low-ink-amount scene, so that efficient and high-precision color calibration is realized.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of printing color management, and particularly relates to a deep learning self-calibration method and system for high-precision color management. BACKGROUND

[0002] In color calibration management in the printing field, a deep learning model can be used to realize color calibration, but there are problems: a common model makes a large number of “simplified assumptions” (such as linear mapping and single color rendering mechanism) on the physical mechanism of printing color in order to reduce the training difficulty, but in a high-precision scenario, these assumptions are seriously inconsistent with the actual physical process, specifically: The modeling logic of the common model: assuming that the ink amount-reflection rate is in a linear relationship, a simple CNN or fully connected network is used to learn the linear mapping of input-output, and the nonlinear coupling of multiple factors is ignored, but in high-precision color management, color rendering is the result of nonlinear coupling of multiple factors, and the linear assumption of the common model leads to: When the ink amount (ink layer thickness <0.5 mu m) is low, the spectral reflectivity of the ink shows “nonlinear saturation” (for example, when the ink layer thickness increases by 0.1 mu m, the initial delta E increases by 0.3, and the later increase is only 0.1), and if the common model is calibrated according to the linear relationship, the problem of “over-calibration” (such as light red turning into dark red) may occur in the light color region.

[0003] Therefore, the application provides a deep learning self-calibration method and system for high-precision color management. SUMMARY

[0004] In order to make up for the deficiencies of the prior art and solve at least one technical problem in the background art.

[0005] The technical scheme adopted by the application to solve the technical problems is: a deep learning self-calibration method for high-precision color management, comprising the following steps: Step one: through historical continuous printing image data under different ink types, the change comparison and analysis of ink layer thickness and color difference are performed to determine the ink layer critical thickness at which the different ink types show nonlinear saturation in the printing low ink amount scenario; Step two: according to the ink layer thickness data and the corresponding color difference data above the ink layer critical thickness, and in combination with a nonlinear fitting mode, a nonlinear correlation model about the ink layer thickness and the color difference of different ink types in the printing low ink amount scenario is constructed; Step three: when the current printing color difference does not reach the target set color difference, the ink layer default thickness is output according to the color difference difference between the current printing color difference and the target set color difference and in combination with the existing deep learning color calibration model, and whether the ink layer thickness when the target set color difference is reached exceeds the ink layer critical thickness is predicted and judged in combination with the current printing ink layer thickness. Step four: according to the prediction judgment result of whether the ink layer thickness exceeds the critical thickness of the ink layer when the target set color difference is reached, selectively combine different ink types in the printing low ink amount scene to determine the final ink layer thickness increase.

[0006] As a further technical solution of the application: the process of determining the critical thickness of the ink layer of different ink types under the printing low ink amount scene is: Based on the historical continuous printing images of any ink type, the ink layer thickness and color difference of different historical continuous printing images are obtained, the color difference deviation and ink layer thickness deviation between adjacent two printing images are calculated, and the proportion is calculated to obtain the ink layer-color difference correlation proportion between adjacent printing images, and the critical thickness of the ink layer of the nonlinear saturation is determined. The ink layer thickness corresponding to the printing image in the later printing order in the adjacent printing image meeting the judgment condition for the first time is extracted as the critical thickness of the ink layer of the ink type, and the critical thickness of the ink layer of different ink types under the printing low ink amount scene is obtained by traversing all ink types.

[0007] As a further technical solution of the application: the critical thickness of the linear saturation of the ink layer includes: Judgment condition one: the ink layer-color difference correlation proportion between adjacent printing images is less than the default linear correlation proportion of the ink layer-color difference, and the correlation proportion difference meets the preset threshold condition. Judgment condition two: taking the adjacent printing image meeting the judgment condition one as the starting point, the correlation proportion difference between all adjacent printing images after the starting point meets the judgment condition one.

[0008] As a further technical solution of the application: the process of constructing the nonlinear correlation model of different ink types in the printing low ink amount scene about the ink layer thickness and the color difference is: For any ink type, the ink layer thickness data and the corresponding color difference data of the critical thickness of the ink layer and above are integrated into the ink layer thickness sequence and the color difference sequence according to the printing order of the printing image. The ink layer thickness sequence and the color difference sequence are fitted by using a nonlinear fitting method, and after fitting, the fitting model with the highest fitting goodness is screened out as the nonlinear correlation model of the ink layer thickness and the color difference, and the nonlinear correlation model of different ink types in the printing low ink amount scene about the ink layer thickness and the color difference is obtained by traversing all ink types.

[0009] As a further technical solution of the application: the output mode of the default ink layer thickness increase is: The color difference deviation value is obtained by calculating the deviation between the current printing color difference and the target set color difference, and the color difference deviation value is taken as the input of the existing deep learning color calibration model to output the default ink layer thickness increase. The current printing color difference refers to a color difference corresponding to a printed image in the last printing in the current printing process.

[0010] As a further technical solution of the present application, the process of predicting and judging whether the ink layer thickness when the target set color difference is reached exceeds the ink layer critical thickness comprises: Summing the default ink layer thickness increase and the current printing ink layer thickness to obtain the predicted ink layer thickness when the target set color difference is reached; The current printing ink layer thickness is the ink layer thickness of the last printed image in the current printing process. The predicted ink layer thickness is compared with the ink layer critical thickness corresponding to the current printing ink type to predict and judge whether the ink layer thickness when the target set color difference is reached exceeds the ink layer critical thickness.

[0011] As a further technical solution of the present application, the process of predicting and judging whether the ink layer thickness when the target set color difference is reached exceeds the ink layer critical thickness further comprises: If the ink layer thickness when the target set color difference is reached is greater than the ink layer critical thickness, it is indicated that the prediction exceeds. If the ink layer thickness when the target set color difference is reached is less than or equal to the ink layer critical thickness, it is indicated that the prediction does not exceed.

[0012] As a further technical solution of the present application, the process of determining the final ink layer thickness increase comprises: If the predicted ink layer thickness when the target set color difference is reached does not exceed the ink layer critical thickness, the default ink layer thickness increase is taken as the final ink layer thickness increase.

[0013] As a further technical solution of the present application, the process of determining the final ink layer thickness increase further comprises: If the predicted ink layer thickness when the target set color difference is reached exceeds the ink layer critical thickness, the non-linear correlation model of the ink layer thickness and the color difference of the current printing ink type in the printing low-ink-amount scene is extracted, and the color difference deviation value is taken as the input to output the final ink layer thickness increase.

[0014] A deep learning self-calibration system for high-precision color management comprises the following modules: An ink layer critical thickness determination module: through historical continuous printed image data under different ink types, the change comparison and analysis of the ink layer thickness and the color difference are performed to determine the ink layer critical thickness of different ink types in the printing low-ink-amount scene in which non-linear saturation occurs. An ink layer thickness-color difference correlation module: according to the ink layer thickness data and the corresponding color difference data of the ink layer critical thickness and above, a non-linear fitting method is combined to construct a non-linear correlation model of the ink layer thickness and the color difference of different ink types in the printing low-ink-amount scene. The ink layer thickness super-bound analysis module: when the current printing color difference does not reach the target set color difference, according to the color difference difference between the current printing color difference and the target set color difference and combining the existing deep learning color calibration model, the default ink layer thickness is output, and the ink layer thickness when the target set color difference is reached is predicted and judged whether it exceeds the ink layer critical thickness in combination with the current printing ink layer thickness; The printing ink layer adjustment module: according to the prediction and judgment result of whether the ink layer thickness when the target set color difference is reached exceeds the ink layer critical thickness, the final ink layer thickness is determined by selectively combining the nonlinear correlation model of different ink types about ink layer thickness and color difference in the printing low ink amount scene.

[0015] The beneficial effects of the present application are as follows: By using the historical continuous printing image data of different ink types, combining the CIE Lab color space quantization color difference and the laser interferometer measurement ink thickness, the correlation proportion judgment condition is established, the nonlinear saturated ink layer critical thickness in the low ink amount scene is accurately positioned, the problem that the relationship between the ink layer and the color difference is easy to deviate from the linearity in the low ink amount is solved, and the foundation for high-precision calibration is laid. Based on the critical thickness, the corresponding ink layer thickness and color difference data are integrated, the optimal model is selected by nonlinear fitting such as quadratic function, and the nonlinear correlation model of each ink type is constructed, which makes up for the precision defects of the traditional linear model in the nonlinear saturation interval. In actual calibration, the default increment of the existing deep learning model is output, and it is predicted whether the target ink thickness exceeds the critical value. If not, the default value is used to ensure efficiency, and if it does, the nonlinear model is used to calculate the increment. This differentiated strategy avoids color deviation caused by blind adjustment, accurately controls the ink layer increment, and realizes efficient and high-precision color calibration. BRIEF DESCRIPTION OF DRAWINGS

[0016] The present application will be further described below with reference to the accompanying drawings.

[0017] Figure 1 is a step flow chart of the deep learning self-calibration method for high-precision color management according to an embodiment of the present application; Figure 2 is a logic judgment diagram of the deep learning self-calibration method for high-precision color management according to an embodiment of the present application; Figure 3 is a system block diagram of the deep learning self-calibration system for high-precision color management according to an embodiment of the present application. DETAILED DESCRIPTION

[0018] In order to make the technical means, creative features, purposes and effects realized by the present application easy to understand, the present application will be further described below with reference to the specific embodiments.

[0019] Embodiment 1: Please refer to Figures 1-2As shown, the deep learning self-calibration method for high-precision color management comprises the following steps: Step one: through the historical continuous printing image data under different ink types, the change comparison analysis of ink layer thickness and color difference is performed to determine the critical thickness of the ink layer under the non-linear saturation of different ink types in the low ink printing scene; In step one, the ink type represents the type of ink used for printing; Wherein, the ink type includes but is not limited to: offset ink, flexographic ink and digital printing ink, etc.; In step one, the historical continuous printing image data includes multiple printing images printed continuously using different ink types, wherein the printing images are all in a low ink printing scene when printing; In step one, the low ink printing scene means that the ink layer thickness is <0.5μm; In step one, the process of determining the critical thickness of the ink layer under the non-linear saturation of different ink types in the low ink printing scene is as follows: Based on any one of the multiple printing of any ink type, the ink layer thickness and color difference of different historical continuous printing images are obtained, wherein the color difference refers to the deviation between the actual color of the printing image and the target standard color, and the CIE Lab color space commonly used in the printing industry is used for quantitative calculation. The Lab value difference between the printing image and the standard target color sample is calculated to obtain the color difference. The ink layer thickness can be calculated by emitting a laser beam to the ink layer surface using a laser interferometer, and the interference fringes of the reflected light and the incident light are used to calculate the ink layer thickness. It should be noted that the ink layer thickness in the printing image is uniform; In the historical continuous printing image, the color difference deviation and the ink layer thickness deviation between the adjacent two printing images are calculated, and the proportion is calculated to obtain the ink layer-color difference correlation proportion between the adjacent printing images; According to the ink layer-color difference correlation proportion between the adjacent printing images and the default linear correlation proportion of the ink layer-color difference, the critical thickness determination condition of the non-linear saturation of the ink layer is set, which is as follows: Condition one: the ink layer-color difference correlation proportion between the adjacent printing images is less than the default linear correlation proportion of the ink layer-color difference, and the correlation proportion difference meets the preset threshold condition; Specifically: | ink layer-color difference correlation ratio-ink layer-color difference default linear correlation ratio | / ink layer-color difference default linear correlation ratio ≥ preset threshold, wherein the preset threshold can be 20%. The setting is mainly based on the balance requirement of color control accuracy and process stability in the printing industry. The mainstream commercial printing requires color difference ΔE ≤ 2.0 as the acceptance standard, and the natural fluctuation of the printing system will cause the "ink layer-color difference correlation ratio" to have a measurement variation of about 3%-5%. By setting the threshold to 20%, it can ensure that when the correlation ratio decreases to 80% of the linear value, it will be timely warned, leaving enough buffer space for operation (at this time, the actual ΔE is usually still within the safe range of 1.5-1.8), and it can effectively avoid false alarms caused by normal fluctuations of the system (20% is much larger than the upper limit of 5% fluctuation). It can be understood that due to the non-linear saturation phenomenon, the color difference becomes smaller, so when determining the critical thickness of the ink layer, the ink layer-color difference correlation ratio between adjacent printed images is less than the ink layer-color difference default linear correlation ratio as the determination condition; Determination condition two: taking the adjacent printed image meeting the determination condition one as the starting point, the correlation ratio difference between all adjacent printed images after the starting point meets the determination condition one; Among all adjacent printed images meeting the critical thickness determination condition of the ink layer, the first appearing adjacent printed image is extracted according to the printing sequence, and the ink layer thickness of the printed image later in the printing sequence in the adjacent printed image is taken as the critical thickness of the ink layer appearing non-linear saturation in the low ink amount scene; By traversing all ink types, the critical thickness of the ink layer appearing non-linear saturation in the low ink amount scene can be obtained for different ink types; For example, all adjacent printed image sequences meeting the critical thickness determination condition of the ink layer are {Fig. 5, Fig. 6, Fig. 7, Fig. n}, wherein Fig. n is the image at the nth printing, and n is the printing sequence. Then the ink layer thickness of the printed image 6 is the critical thickness of the ink layer, which can be shown in Table 1 as follows. Table 1: Data display table for determining the critical thickness of the ink layer appearing non-linear saturation; It should be noted that the ink layer-color difference default linear correlation ratio is the ink layer-color difference correlation ratio set by default in the existing deep learning color calibration model; It can be understood that the significance of step one is to determine the critical thickness of the ink layer under the low ink printing scenario of different ink types, to establish the ink layer-color difference correlation ratio judgment condition by analyzing the historical printing data under low ink, combining the CIE Lab color space quantization color difference and the laser interferometer measurement ink thickness, to accurately locate the starting point of nonlinear saturation, to provide basis for subsequent differentiation of linear and nonlinear action intervals of ink layer thickness, to solve the problem that the relationship between ink layer and color difference is easy to deviate from linearity under low ink scenario, and to realize the basis prerequisite of high-precision calibration; Step two: according to the ink layer thickness data and the corresponding color difference data of the ink layer critical thickness and above, and combining the nonlinear fitting method, a nonlinear correlation model of ink layer thickness and color difference of different ink types under the printing low ink scenario is constructed; In step two, the ink layer thickness data of the ink layer critical thickness and above means that the ink layer thickness is at least 0.30 μm and above 0.30 μm, for example, the ink layer critical thickness is 0.30 μm, and the ink layer thickness data contains ink layer thickness of at least 0.30 μm and above 0.30 μm; In step two, the process of constructing the nonlinear correlation model of ink layer thickness and color difference of different ink types under the printing low ink scenario is as follows: Based on any ink type; Respectively, the ink layer thickness data of the ink layer critical thickness and above and the corresponding color difference data are integrated into ink layer thickness sequence and color difference sequence according to the printing order of the printed image; The ink layer thickness sequence and the color difference sequence are fitted by using a nonlinear fitting method, and after fitting, the fitting model with the highest goodness of fit is selected as the nonlinear correlation model of ink layer thickness and color difference; Wherein, the nonlinear fitting method includes but is not limited to quadratic function fitting, exponential function fitting and power function fitting, etc.; Iterate all ink types to obtain the nonlinear correlation model of ink layer thickness and color difference of different ink types under the printing low ink scenario; It can be understood that the significance of step two is to construct an ink layer-color difference correlation model that adapts to the low ink nonlinear scenario, to integrate the ink layer thickness and color difference data above the thickness determined in step one, to select the optimal model through various nonlinear fitting methods, to make up for the precision defects of traditional linear model in the nonlinear saturation interval of ink layer, to provide support for subsequent ink layer adjustment in line with actual printing rules, and to ensure the accuracy of color correlation relationship calibration; Step three: when the current printing color difference does not reach the target set color difference, according to the color difference difference between the current printing color difference and the target set color difference and combining the existing deep learning color calibration model, the default increased thickness of the ink layer is output, and the ink layer thickness when the target set color difference is reached is predicted and judged whether it exceeds the ink layer critical thickness in combination with the current printing ink layer thickness; In step three, the output mode of the default increased thickness of the ink layer is specifically: The current printing color difference is calculated with the target set color difference to obtain a color difference deviation value, and the color difference deviation value is taken as an input of an existing deep learning color calibration model to output the default increased thickness of the ink layer; The current printing color difference refers to the color difference corresponding to the printed image at the last printing in the current printing process, the existing deep learning color calibration model (such as a CNN or a fully connected network learning model) is an existing color calibration model originally set by the printing system, and the target set color difference refers to a target value of the color difference set in the printing process. In step three, the process of predicting whether the ink layer thickness when reaching the target set color difference exceeds the ink layer critical thickness is: The default increased thickness of the ink layer is summed with the current printing ink layer thickness to predict the ink layer thickness when reaching the target set color difference. It should be noted that the current printing ink layer thickness refers to the ink layer thickness corresponding to the printed image at the last printing in the current printing process. Based on the ink layer critical thickness of different ink types in the printing low ink amount scene, the ink layer critical thickness of the current printing ink type that appears nonlinear saturation is extracted. The ink layer thickness when reaching the target set color difference is compared with the ink layer critical thickness to determine whether the ink layer thickness when reaching the target set color difference exceeds the ink layer critical thickness, which is specifically: If the ink layer thickness when reaching the target set color difference is greater than the ink layer critical thickness, it means that the prediction exceeds. If the ink layer thickness when reaching the target set color difference is less than or equal to the ink layer critical thickness, it means that the prediction does not exceed. It can be understood that the significance of step three is to combine the output of the existing deep learning model with the default increased amount of the ink layer to predict whether the ink layer thickness when reaching the target set color difference exceeds the ink layer critical thickness, to make an early judgment of the consequences of ink layer adjustment, to provide decision support for selecting appropriate ink layer adjustment basis in the future by clearly defining the relationship between the target ink thickness and the ink layer critical thickness, and to avoid color deviation caused by blind adjustment. Step four: according to the prediction result of whether the ink layer thickness when reaching the target set color difference exceeds the ink layer critical thickness, selectively combine the nonlinear correlation model of the ink layer thickness and the color difference of different ink types in the printing low ink amount scene to determine the final ink layer increased thickness. In step four, the process of determining the final ink layer increased thickness is: If the prediction determines that the ink layer thickness when reaching the target set color difference does not exceed the ink layer critical thickness, the default increased thickness of the ink layer is taken as the final ink layer increased thickness. If the ink layer thickness when the prediction judgment reaches the target set color difference exceeds the ink layer critical thickness, a nonlinear correlation model of the current printing ink type about ink layer thickness and color difference in a low ink amount printing scene is extracted, and the color difference deviation value is taken as input, and the final ink layer increase thickness is output; The ink layer thickness is increased in the current printing process according to the final ink layer increase thickness; It can be understood that the significance of step four is that, according to the prediction result of step three, the adjustment basis is selected accordingly - if not exceeding the critical value, the default value (ink layer default increase thickness) is used, and if exceeding the critical value, the nonlinear model corresponding to the ink type is used to calculate the ink layer increment. This differentiated adjustment strategy not only ensures the adjustment efficiency in the linear interval, but also solves the calibration accuracy problem in the nonlinear interval. Finally, by accurately controlling the ink layer increase amount, the printing color difference reaches the target requirement, and high-precision color management is completed.

[0020] Embodiment 2: Please refer to Figure 3 As shown in the figure, the deep learning self-calibration system for high-precision color management comprises the following modules: An ink layer critical thickness determination module: by comparing and analyzing the changes of ink layer thickness and color difference under different ink types, the ink layer critical thickness of different ink types in a low ink amount printing scene is determined. An ink layer thickness-color difference correlation module: according to the ink layer thickness data and the corresponding color difference data above the ink layer critical thickness, and combining a nonlinear fitting method, a nonlinear correlation model of different ink types about ink layer thickness and color difference in a low ink amount printing scene is constructed. An ink layer thickness over-limit analysis module: when the current printing color difference does not reach the target set color difference, according to the color difference difference between the current printing color difference and the target set color difference and combining the existing deep learning color calibration model, the ink layer default increase thickness is output, and the current printing ink layer thickness is combined to predict whether the ink layer thickness when the target set color difference is reached exceeds the ink layer critical thickness. A printing ink layer adjustment module: according to the prediction judgment result of whether the ink layer thickness when the target set color difference is reached exceeds the ink layer critical thickness, the final ink layer increase thickness is determined by selectively combining the nonlinear correlation model of different ink types about ink layer thickness and color difference in a low ink amount printing scene.

[0021] The basic principles, main features and advantages of the present application are shown and described above. Those skilled in the art should understand that the present application is not limited by the above examples, and the above examples and descriptions in the specification are only to illustrate the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. A deep learning self-calibration method for high-precision color management, characterized in that: Includes the following steps: Step 1: By comparing and analyzing the changes in ink layer thickness and color difference through historical continuous printing image data under different ink types, the critical thickness of the ink layer that causes nonlinear saturation under low ink volume printing scenarios is determined. Step 2: Based on the ink layer thickness data at and above the critical ink layer thickness and the corresponding color difference data, and combined with nonlinear fitting, construct a nonlinear correlation model of ink layer thickness and color difference for different ink types in low ink volume printing scenarios; Step 3: When the current printing color difference does not reach the target color difference, based on the color difference between the current printing color difference and the target color difference and combined with the existing deep learning color calibration model, the ink layer thickness is increased by default. Combined with the current printing ink layer thickness, it is predicted whether the ink layer thickness exceeds the critical ink layer thickness when the target color difference is reached. Step 4: Based on the prediction results of whether the ink layer thickness exceeds the critical ink layer thickness when the target color difference is achieved, selectively combine the nonlinear correlation model of ink layer thickness and color difference for different ink types in low ink volume printing scenarios to determine the final increase in ink layer thickness.

2. The deep learning self-calibration method for high-precision color management according to claim 1, characterized in that: The process of determining the critical thickness of the ink layer that exhibits nonlinear saturation under low ink volume printing scenarios for different ink types is as follows: Based on historical continuous printing images of any ink type, the ink layer thickness and color difference of different historical continuous printing images are obtained. The color difference deviation and ink layer thickness deviation between two adjacent printing images are calculated and the ratio is calculated to obtain the ink layer-color difference correlation ratio between adjacent printing images. The non-linear saturation ink layer critical thickness judgment condition is set. The ink layer thickness corresponding to the later printing sequence in the adjacent printed images that first meet the judgment conditions is extracted as the critical ink layer thickness of the ink type. By traversing all ink types, the critical ink layer thickness of different ink types that exhibit nonlinear saturation in low ink volume printing scenarios is obtained.

3. The deep learning self-calibration method for high-precision color management according to claim 2, characterized in that: The criteria for determining the critical thickness of linearly saturated ink layers include: Judgment condition 1: The ink layer-color difference correlation ratio between adjacent printed images is less than the default linear correlation ratio of ink layer-color difference, and the difference in correlation ratio meets the preset threshold condition. Judgment Condition 2: Taking the adjacent printed images that meet Judgment Condition 1 as the starting point, the correlation ratio difference between all adjacent printed images after the starting point also meets Judgment Condition 1.

4. The deep learning self-calibration method for high-precision color management according to claim 1, characterized in that: The process of constructing a nonlinear correlation model of ink layer thickness and color difference for different ink types in low-ink-volume printing scenarios is as follows: For any ink type, the ink layer thickness data above the critical ink layer thickness and the corresponding color difference data are integrated into an ink layer thickness sequence and a color difference sequence according to the printing order of the printed image. The ink layer thickness sequence and color difference sequence were fitted using a nonlinear fitting method. After fitting, the fitting model with the highest goodness of fit was selected as the nonlinear correlation model between ink layer thickness and color difference. By traversing all ink types, nonlinear correlation models between ink layer thickness and color difference for different ink types in low ink volume printing scenarios were obtained.

5. The deep learning self-calibration method for high-precision color management according to claim 1, characterized in that: The default output method for increasing ink layer thickness is as follows: The deviation between the current printed color difference and the target color difference is calculated to obtain the color difference deviation value. The color difference deviation value is used as the input of the existing deep learning color calibration model, and the output ink layer thickness is increased by default. Among them, the current printing color difference refers to the color difference corresponding to the printed image in the most recent printing process in the current printing process.

6. The deep learning self-calibration method for high-precision color management according to claim 5, characterized in that: The process of predicting whether the ink layer thickness exceeds the critical ink layer thickness when the target color difference is achieved includes: The default increase in ink layer thickness is summed with the current printed ink layer thickness to obtain the predicted ink layer thickness when the target color difference is achieved. The current ink layer thickness refers to the ink layer thickness of the most recently printed image in the current printing process. Extract the critical thickness of the ink layer corresponding to the current printing ink type, compare the predicted ink layer thickness with the critical thickness of the ink layer, and predict whether the ink layer thickness when the target color difference is achieved exceeds the critical thickness of the ink layer.

7. A deep learning self-calibration method for high-precision color management according to claim 6, characterized in that: The process of predicting whether the ink layer thickness exceeds the critical ink layer thickness when the target color difference is achieved also includes: If the ink layer thickness when the target color difference is achieved is greater than the critical ink layer thickness, it indicates that the prediction has exceeded the limit. If the ink layer thickness when the target color difference is achieved is less than or equal to the critical ink layer thickness, it means that the prediction has not been exceeded.

8. A deep learning self-calibration method for high-precision color management according to claim 7, characterized in that: The process of determining the final ink layer thickness includes: If the predicted ink layer thickness does not exceed the critical ink layer thickness when the target color difference is achieved, the ink layer thickness will be increased by default as the final ink layer thickness increase.

9. A deep learning self-calibration method for high-precision color management according to claim 8, characterized in that: The process of determining the final ink layer thickness also includes: If the predicted ink layer thickness exceeds the critical ink layer thickness when the target color difference is reached, then the nonlinear correlation model of ink layer thickness and color difference of the current printing ink type under the low ink volume scenario is extracted, and the color difference deviation value is used as input to output the final increase in ink layer thickness.

10. A deep learning self-calibration system for high-precision color management, characterized in that: This system is used to perform the deep learning self-calibration method as described in any one of claims 1-9, and the system includes the following modules: Ink layer critical thickness determination module: By comparing and analyzing the changes in ink layer thickness and color difference through historical continuous printing image data under different ink types, the critical thickness of ink layer that causes nonlinear saturation under low ink volume printing scenarios is determined. Ink layer thickness-color difference correlation module: Based on the ink layer thickness data above the critical ink layer thickness and the corresponding color difference data, and combined with nonlinear fitting method, a nonlinear correlation model of ink layer thickness and color difference for different ink types in low ink volume printing scenarios is constructed. Ink layer thickness exceeding the limit analysis module: When the current printing color difference does not reach the target set color difference, based on the color difference between the current printing color difference and the target set color difference and combined with the existing deep learning color calibration model, the module outputs the default increase in ink layer thickness, and combined with the current printing ink layer thickness, predicts whether the ink layer thickness exceeds the critical ink layer thickness when the target set color difference is reached. Printing Ink Layer Adjustment Module: Based on the prediction results of whether the ink layer thickness exceeds the critical ink layer thickness when the target color difference is achieved, the module selectively combines the nonlinear correlation model of ink layer thickness and color difference for different ink types in low ink volume printing scenarios to determine the final increase in ink layer thickness.

Images