Spectrum color matching method based on KM model and intelligent optimization
By integrating the nonlinear relationship between the K/S value and concentration of the base color and the optimization of genetic algorithms, the problem of low color matching accuracy of KM theory in the printing field is solved, efficient and accurate ink color matching is achieved, and the performance of the ink automatic color matching system is improved.
Patent Information
- Application Number
- CN202411196436.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-29
- Publication Date
- 2024-12-06
AI Technical Summary
When the existing spectral color matching method based on KM theory is applied in the printing field, the single constant K/S value of the base color does not conform to the linear relationship, resulting in poor color matching accuracy. The KM theory assumes that the refractive index of the film layer interface is continuous. Changes, but there are actually mutations, and the spectral reflectance cannot be accurately predicted.
By using the product of the base color K/S value and concentration as a whole, the nonlinear relationship between the K/S value and concentration is integrated to construct a constrained nonlinear optimization problem, and using the genetic algorithm to obtain the optimal parameters to non-spectroreflectivity. Linear correction, establish an intelligent optimized spectral color matching method based on KM model.
It improves the color matching accuracy and efficiency, reduces artificial errors, realizes high-precision matching of automatic ink color matching, reduces the amount of color samples and measurement data required for model training, and improves the generalization ability of the KM model.
Smart Images

Figure CN119092006A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of printing technology, and in particular to a spectral color matching method based on a KM model and intelligent optimization. Background Art
[0002] Color matching is the process of obtaining the target color by mixing different colors of base inks. Traditional artificial experience color matching is often affected by the color matcher's physiology, psychology and other subjective and objective factors, resulting in the difficulty of stabilizing product quality and often failing to meet the requirements of modern printing and packaging production. Computer color matching technology is based on color theory and provides base ink formulas of any mixed colors through computer automatic calculation, which simplifies the color matching process in specific practice. Computer color matching technology stores the basic color database of ink in the computer in advance, and uses color matching software to calculate the mixing ratio of the base ink that matches the target color, so as to obtain an accurate ink formula, thereby improving the accuracy and stability of color matching and reducing human errors. In order to achieve efficient and accurate automatic color matching, multiple color matching models have been proposed at home and abroad, including models based on single constant and double constant Kubelka-Munk theory, Friele model, Stearns-Noechel model, and color matching models based on neural networks. The KM (Kubelka-Munk) theoretical model is a theoretical derivation model based on the modeling of light propagation behavior. It does not require too much basic experimental data for modeling, and its generalization ability is usually stronger than that of experimental models that rely on a small number of training samples. Some models that rely on training samples, such as neural network models, often require thousands or even tens of thousands of training samples to ensure the generalization ability of the model. This requires the production and measurement of a large number of color samples and model training, which consumes a lot of time, manpower and material resources.
[0003] KM theory has two algorithms: single constant and double constant. The double constant algorithm is generally used in industries such as paints, plastics and coatings, while the single constant algorithm is mainly used in the textile printing and dyeing industry. Based on this theory, industries such as printing and coatings have established their own computer color matching systems. KM theory proposes that the absorption and scattering coefficients of light in multiple non-transparent mixed substances are linearly related to the absorption and scattering coefficients of each non-transparent substance, and this absorption and scattering coefficient is only related to its reflectivity coefficient. In the invention patent "A color matching method for colored spinning based on a modified single constant KM theory" (CN 112668174 B), a linear function for correcting the spectral reflectance is proposed, and the expression is G(x) = k 1 x+k 2 , and the variable parameter k 1 and k 2 It is the value that achieves the best linear relationship between the single constant K / S value corrected by the base color fiber sample and the concentration, and satisfies the constraint condition k 1 +k2 =1. The invention patent "A printing ink color matching method based on KM theory" (CN 113910796B) proposes a method for obtaining the corrected absorption coefficient and scattering coefficient of the base color ink, which still uses the linear relationship between the single constant K / S value and the concentration for color matching. However, the KM theory is based on a series of assumptions and is inconsistent with the actual optical behavior. In practical applications, it is found that the single constant K / S value of the base color does not conform to the linear relationship with the concentration, which may lead to poor color matching accuracy of the model.
[0004] In addition, the KM theory assumes that the refractive index of the film interface changes continuously in color matching; however, the actual air-film interface usually has a sudden change in refractive index. This discontinuity makes it impossible for the KM spectral color matching model to obtain a spectral reflectance that is approximately consistent with the actual measured value. One solution is to correct the spectral reflectance of the target color, base color, and substrate before color matching. For example, the Saunderson correction method can be used, and its correction model is:
[0005]
[0006] Where R p (λ), R corr,p (λ) are the spectral reflectance coefficients before and after correction, λ is the wavelength, k 1 is the Fresnel reflection coefficient of parallel light, which is about 0.04; k 2 It is the Fresnel reflection coefficient of diffuse light emitted from the inside to the surface, which is about 0.4 to 0.6. These two parameters are often tried according to specific circumstances, and it is not easy to find accurate values. In addition, in the printing industry, in order to obtain a color lighter than the base color, diluents are often used instead of white ink, but the use of diluents will cause the absorption and scattering coefficients of the base color ink in the color matching system to shift, resulting in color matching errors. This situation can also be corrected by correcting the spectral reflectance, but the correction parameters need to be adjusted repeatedly. Therefore, the existing color matching method based on the traditional KM theory has low color matching accuracy when applied in the printing field. For this reason, it is urgent to develop new color matching methods and technologies to improve color matching accuracy and efficiency. Summary of the invention
[0007] In order to make up for the deficiencies in the prior art, the present invention provides a spectral color matching method, which calculates the product of the base color K / S value and the concentration as a whole to integrate the nonlinear relationship between the K / S value and the concentration, solves the optimal parameters to perform nonlinear correction on the spectral reflectance, and on this basis constructs a constrained nonlinear optimization problem, and uses a genetic algorithm to obtain the optimal formula.
[0008] The present invention is implemented by the following technical scheme: a spectral color matching method based on KM model and intelligent optimization, comprising the following steps:
[0009] Phase 1: Establish a basic color matching data set and a color matching model, and keep the correction coefficient k in the color matching model 1 , k 2 To be determined parameter.
[0010] Step S1: Select the base color ink and diluent, prepare coated paper with high gloss and smoothness; mix the base color ink and diluent in a mixture of c i :(1-c i )(0<c i ≤1, i=1,2,…,n) are uniformly mixed to obtain n groups of mixtures of base color ink and diluent in different ratios, and the mixtures are spread on coated paper to prepare n samples; preferably, n is 20, c 1 =5%,c 2 =10%,…,c 20 =100%, step is 5%;
[0011] Step S2: Measuring the spectral reflectance coefficient R of coated paper p (λ), the spectral reflectance coefficients of the n samples measured in step S1 Where i = 1, 2, ..., n; wavelength λ = 400, 410, ..., 700, in nm, with a wavelength interval of 10 nm;
[0012] Spectral reflectance coefficient R of coated paper p (λ) is corrected, and the formula is:
[0013]
[0014] Spectral reflectance coefficient for n samples To make corrections, the formula is:
[0015]
[0016] k 1 , k 2 is the correction factor, where k 1 Represents the Fresnel reflection coefficient of parallel light, which is about 0.04; k 2 Represents the Fresnel reflection coefficient of diffuse light from the inside to the surface, which is about 0.4 to 0.6; by default, k can be set 1 =0.04, k 2 =0.05;
[0017] Step S3: Calculate the ratio of the absorption coefficient and the scattering coefficient of the coated paper The wavelength λ=400,410,…,700; the formula is: Calculate the ratio of the absorption coefficient to the scattering coefficient of n samples Where i = 1, 2, ..., n, wavelength λ = 400, 410, ..., 700; the formula is:
[0018]
[0019] According to the single constant additivity principle of KM theory, the ratio of absorption coefficient to scattering coefficient K / S has a linear relationship with concentration; is the K / S value of the primary ink of unit concentration at wavelength λ, then the ratio of the absorption coefficient to the scattering coefficient of n samples satisfies the following formula:
[0020]
[0021] Considering that the K / S value and concentration usually do not conform to a linear relationship, a nonlinear function f(λ,c i ) replaces get:
[0022]
[0023] where f(λ,c i ) indicates that the concentration at wavelength λ is c i The K / S value of the base color ink; further obtain:
[0024]
[0025] Thus, for the wavelength λ, we can obtain the sequence [0,f(λ,c 1 ),f(λ,c 2 ),…,f(λ,c n )]; through the interpolation algorithm, the K / S value f(λ,c) function of the primary color ink corresponding to the wavelength λ and any concentration c can be obtained:
[0026]
[0027] here represents an interpolation function; preferably, the interpolation algorithm adopts cubic spline interpolation;
[0028] Step S4: Set the primary color inks to cyan ink, magenta ink, and yellow ink respectively, and repeat steps S1 to S3 to obtain the K / S value sequence [0, f c (λ,c 1 ),f c (λ,c 2 ),…,f c (λ,c n )]、[0,f m (λ,c 1 ),f m (λ,c2 ),…,f m (λ,c n )]、[0,f y (λ,c 1 ),f y (λ,c 2 ),…,f y (λ,c n )], wavelength λ=400,410,…,700; through the interpolation algorithm, the K / S value f of the base color ink corresponding to the calculated wavelength λ and arbitrary concentration c can be obtained c (λ,c),f m (λ,c),f y function of (λ,c);
[0029] According to the determined K / S value function f of cyan ink, magenta ink and yellow ink c (λ,c),f m (λ,c),f y (λ,c), we get the color matching model,
[0030]
[0031] Where c c 、c m 、c y The proportions of cyan ink, magenta ink and yellow ink are respectively; is the predicted K / S value of the formulated ink; According to step S3, f c (λ,c c ), f m (λ,c m ), f y (λ,c y ) is obtained according to the interpolation algorithm; since the correction coefficient k in step S2 1 , k 2 The value must be clearly determined before steps S3 and S4 can be calculated. Therefore, k in the color matching model 1 , k 2 is a parameter to be determined;
[0032] Phase 2: Perform a small amount of color matching experiments by mixing base color inks and use the grid search method to determine the correction factor k 1 , k 2 The optimal value of ; specifically includes the following steps:
[0033] Step S5: Cyan ink, magenta ink, and yellow ink are mixed at c c :c m :c y Mix evenly, 0 <cc ,c m ,c y ≤1, m groups of mixed inks with different proportions are obtained; if c c +c m +c y <1, the percentage of diluent is 1-(c c +c m +c y ); respectively spreading the mixed ink on coated paper to prepare m samples; m ≥ 2; and simultaneously measuring the spectral reflectance coefficients of the m samples;
[0034] Step S6: Divide k into intervals of 0.005 1 The range of k is 0.035~0.045, divided into intervals of 0.01 2 The range of k is 0.4~0.6, and k 1 ∈{0.035,0.04,0.045},k 2 ∈{0.4,0.41,0.42,…,0.6}, construct a search grid (k 1 ,k 2 ), contains 3×21=63 search points in total;
[0035] Step S7: Determine the undetermined parameter k 1 , k 2 The optimal value of is as follows, according to steps S7.1 to S7.3:
[0036] Step S7.1: For each search grid point k corresponding to the search grid point determined in step S6 1 , k 2 The color matching model obtained in step S4 is used to calculate the K / S prediction values of the prepared m samples.
[0037] Step S7.2: For each search grid point k corresponding to the search grid point determined in step S6 1 , k 2 The spectral reflectance coefficients of the m samples measured by S5 are used to calculate the K / S measurement values of the m samples using the formulas in steps S2 and S3.
[0038] Step S7.3: For each search grid point determined in step S6, k 1 , k 2 Value, using the results of step S7.1 and step S7.2, calculate the prediction accuracy index acc pred , the formula is as follows:
[0039]
[0040] Where Ω = {400, 410, ..., 700} is the wavelength range; a total of 63 accuracy index values acc are obtained pred , take the k corresponding to the smallest precision index value 1 , k 2 The value is taken as the undetermined parameter k 1 , k 2 The optimal value of .
[0041] The third stage: according to the color matching model and the undetermined parameter k 1 , k 2 After determining the target color sample, an optimization problem is constructed and solved to obtain the ink formula, including the following steps:
[0042] Step S8: Measure the spectral reflectance R of the target color sample T (λ) and its paper reflectivity R TP (λ), and make corrections:
[0043]
[0044] Calculate the ratio of the absorption coefficient to the scattering coefficient of the target color sample and the target paper. The formula is:
[0045]
[0046] Step S9: construct an optimization problem for solving the optimal ratio of each primary color ink; the optimization problem is expressed as:
[0047] C(c y ,c m ,c c ) = arg min C β(c y ,c m ,c c ),
[0048] where c c 、c m 、c y are the proportions of cyan ink, magenta ink and yellow ink respectively; C(c y ,c m ,c c ) is the optimal ink ratio solution set; the objective function β(c y ,c m ,c c ) ensures the color matching prediction accuracy and is defined as
[0049]
[0050] in is the K / S value predicted by the color matching model, Ω = {400, 410, …, 700} is the wavelength range;
[0051] The equality constraints for the optimization problem are:
[0052]
[0053] where f c (λ,c c ), f m (λ,c m ), f y (λ,c y ) is the interpolation function obtained in step S4, where the undetermined parameter k 1 , k 2 Determined by step S7;
[0054] The inequality constraints for the optimization problem are:
[0055]
[0056] Step S10: Use a genetic algorithm to solve the optimization problem described in step S9 to obtain the ratio c of cyan, magenta, and yellow inks. c 、c m 、c y , and obtain the predicted value of the corrected spectral reflectance of the color matching ink; Genetic Algorithm (GA) is a search algorithm that finds approximate solutions to optimization problems by simulating natural selection and genetic mechanisms; the specific process and steps are as follows:
[0057] Step S10.1, initialize the population and select an appropriate population size, such as 50 or 100; each individual is represented by a chromosome, and the chromosome consists of three genes, representing c c 、c m 、c y The value of the gene value is in the range of [0,1]; individuals in the population are randomly generated to ensure that c c +c m +c y ≤1;
[0058] Step S10.2, the objective function of the optimization problem is used as the fitness function, that is, Set the gene value of the chromosome to satisfy the inequality constraint: c c +c m +c y ≤1,0≤c c ≤1,0≤c m ≤1,0≤c y≤1; the selection strategy of the genetic algorithm adopts roulette selection, and individuals are selected according to the fitness value. The smaller the fitness value, the greater the probability of being selected;
[0059] Step S10.3, perform crossover operation, mutation operation and fitness calculation; set the crossover probability p c , such as 0.8; use single-point crossover or double-point crossover to exchange the genes of two parent individuals to generate new offspring individuals; set the mutation probability p m , such as 0.01; randomly select a gene in the chromosome and randomly change its value to ensure that the constraint conditions described in step S10.2 are still met after the mutation; for each individual in the newly generated population, calculate its fitness value β(c y ,c m ,c c );
[0060] Step S10.4: Repeat the crossover operation, mutation operation and fitness calculation described in step S10.3 until the preset maximum number of generations, such as 100 generations, is reached, and the algorithm is stopped; or the fitness value is less than a certain threshold, and the algorithm is stopped; after the algorithm is stopped, the gene value c of the optimal individual is obtained. c 、c m 、c y , that is, the optimal ratio c of cyan, magenta and yellow inks is obtained c :c m :c y .
[0061] Step S10.5: If the solution is c c 、c m 、c y After c c +c m +c y <1, the percentage of diluent is 1-(c c +c m +c y );get c c 、c m 、c y Then, the predicted K / S value of the prepared ink can be obtained by the equation constraint expression described in step S9. The predicted corrected spectral reflectance can be further obtained, and the calculation formula is:
[0062] The beneficial effects of the present invention are:
[0063] The present invention integrates the nonlinear relationship between K / S value and concentration, establishes a basic color matching data set, and obtains the correction coefficient of spectral reflectance through grid optimization; further, a nonlinear constraint optimization problem is constructed, and a genetic algorithm is used to solve the formula, which greatly improves the color matching accuracy of the single constant KM model. The present invention is of great significance to the development and application of the automatic color matching system of inks. Only a small amount of color matching experiments and less color matching sample data are needed to obtain better model generalization ability and improve the performance of the KM color matching model. BRIEF DESCRIPTION OF THE DRAWINGS
[0064] Figure 1 It is a flow chart of the spectral color matching method based on KM model and intelligent optimization of the present invention;
[0065] Figure 2 It is a comparison diagram of the spectral reflectance curves before and after the target color correction and the spectral reflectance curve of the predicted sample in the embodiment of the present invention. DETAILED DESCRIPTION
[0066] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Generally, the components of the embodiments of the present invention described and shown in the accompanying drawings here can be arranged and designed in various different configurations.
[0067] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present invention.
[0068] Figure 1-Figure 2 This is a specific embodiment of the present invention, which is a spectral color matching method based on the KM model and intelligent optimization, which is divided into three stages. The first stage is to build a color matching model based on the establishment of a basic color matching data set; the second stage is to optimize and obtain the correction coefficient k in the color matching model through a small amount of measurement data of color matching experiments. 1 , k 2 ; The third stage is to obtain the optimal ink formula by constructing and solving the optimization problem based on the color matching model after the target color sample is given.
[0069] In the first stage, a color matching basic data set is established, and a color matching model is established, maintaining the correction coefficient k in the color matching model. 1 , k 2 To be determined parameter.
[0070] Step S1 of the method of the present invention is to conduct a primary color ink preparation experiment. Before the experiment, the necessary instruments and materials need to be prepared, including a printability tester, an electronic analytical balance (accurate to 0.0001g) and a spectrophotometer (used when measuring spectral reflectance), as well as a glass bottle, an ink knife, an ink injector (with a range of 2mL), three primary color inks of yellow, magenta and cyan, a diluent and coated paper with a size of 70mm×320mm.
[0071] Steps S1 to S3 in the method of the present invention need to be repeated three times; according to step S4 in the method of the present invention, the primary color inks are set to cyan ink, magenta ink, and yellow ink respectively, and steps S1 to S3 are repeated to obtain the K / S value f of the primary color ink corresponding to the calculated wavelength λ and any concentration c. c (λ,c),f m (λ,c),f y Specifically, according to step S1 of the method of the present invention, the base color ink and the diluent are mixed at a ratio of c i :(1-c i )(0<c i ≤1, i=1,2,…,20) uniformly mixed, c 1 =5%,c 2 =10%,…,c 20 =100%, the step is 5%; for each base color ink, 20 groups of base color ink and diluent mixtures with different ratios are obtained, and the color is developed on the coated paper using a printability meter to prepare 20 samples. The printability meter pressure is adjusted to 300N, the ink injection amount each time is 0.3mL, and the ink uniformity time is controlled to be 2 to 3 minutes. Then, according to step S2 of the method of the present invention, the spectral reflectance coefficient R of the coated paper is measured using a spectrophotometer. p (λ), the spectral reflectance coefficient of 20 samples of each primary color ink measured in step S1 Where i=1,2,…,20, wavelength λ=400,410,…,700, unit is nm; then the spectral reflectance coefficients of coated paper and test sample are corrected and converted into K / S value; then according to step S3, the nonlinear relationship between K / S value and concentration is integrated to obtain the K / S value f(λ,c) function of the base color ink corresponding to the wavelength λ and any concentration c. This function can be regarded as a nonlinear interpolation function. here represents an interpolation function; preferably, the interpolation algorithm adopts cubic spline interpolation.
[0072] After executing the entire step S4, the color matching model is obtained, and the expression is: Where c c 、c m 、cy The proportions of cyan ink, magenta ink and yellow ink are respectively; is the predicted K / S value of the formulated ink; According to step S3, f c (λ,c c ), f m (λ,c m ), f y (λ,c y ) is obtained according to the interpolation algorithm described in step S4; since the correction coefficient k in step S2 1 , k 2 The subsequent steps need to determine the numerical value before the calculation can be carried out, so k in the color matching model 1 , k 2 is a parameter to be determined. The obtained color matching model can be written into a computer program function, the input parameter of which is k 1 , k 2 , the function outputs the predicted K / S value
[0073] In the second stage, a small amount of color matching experiments were conducted by mixing the base color inks, and the correction coefficient k was determined by using the grid search method. 1 , k 2 The optimal value of .
[0074] According to step S5 of the method of the present invention, a color matching experiment is performed, cyan ink, magenta ink, and yellow ink are uniformly mixed in a certain ratio to obtain mixed inks of different ratios; the mixed inks are respectively developed on coated paper to prepare sample sheets. This is performed m times, m ≥ 2, and the ratios of cyan ink, magenta ink, and yellow ink are respectively recorded. The spectral reflectance coefficients of the obtained m sample sheets are measured.
[0075] According to step S6 of the method of the present invention, k is divided into 1 The range of k is 0.035~0.045, divided into intervals of 0.01 2 The range of k is 0.4~0.6, and k 1 ∈{0.035,0.04,0.045},k 2 ∈{0.4,0.41,0.42,…,0.6}, construct a search grid (k 1 ,k 2 ), which contains 3×21=63 search points in total. Then, according to step S7 of the method of the present invention, 63 accuracy index values are calculated. Take the k corresponding to the smallest precision index value 1 , k 2 The value is taken as the undetermined parameter k 1 , k 2In this embodiment, we get k 1 =0.035, k 2 =0.43.
[0076] The third stage first determines the target color and then constructs an optimization problem to solve the ink formula.
[0077] According to step S8 of the method of the present invention, the spectral reflectance R of the target color sample is measured. T (λ) and its paper reflectivity R TP (λ) and make corrections; calculate the ratio of the absorption coefficient to the scattering coefficient of the target color sample and the target paper. Figure 2 The spectral reflectance of the target color sample and the corrected reflectance examples are given.
[0078] According to step S9 of the method of the present invention, an optimization problem for solving the optimal ratio of each primary color ink is constructed; the objective function, equality constraints, and inequality constraints are clarified. According to step S10 of the method of the present invention, a genetic algorithm is used to solve the optimization problem described in step S9 to obtain the ratio of cyan, magenta, and yellow inks, and to obtain the corrected spectral reflectance prediction value of the color matching ink. The genetic algorithm solution can be performed by writing a program according to the methods described in S10.1 to S10.3 of the method of the present invention. When initializing the population, a population size of 100 is selected; each individual is represented by a chromosome, and the chromosome is composed of three genes, representing c c 、c m 、c y The value of the gene value is in the range of [0,1]; individuals in the population are randomly generated to ensure that c c +c m +c y ≤1; the fitness function is Set the gene value of the chromosome to satisfy the inequality constraint: c c +c m +c y ≤1,0≤c c ≤1,0≤c m ≤1,0≤c y ≤1; the selection strategy uses roulette, and individuals are selected according to the fitness value. The smaller the fitness value, the greater the probability of being selected. Set the crossover probability p c = 0.8, using double-point crossover to exchange the genes of two parent individuals to generate new offspring individuals; setting the mutation probability p m =0.01; randomly select a gene in the chromosome and randomly change its value to ensure that the constraint conditions described in step S10.2 are still met after the mutation; for each individual in the newly generated population, calculate its fitness value β(c y ,c m ,c c). Repeat the crossover operation, mutation operation and fitness calculation until the preset maximum number of generations is 100, and then stop the algorithm. After stopping the algorithm, the gene value c of the optimal individual is obtained. c 、c m 、c y This is the optimal ratio of cyan, magenta and yellow inks. Figure 2 For the target color given in the present embodiment shown, the optimal ratio of cyan, magenta and yellow inks is 9.8%:11.9%:77.6% obtained by using genetic algorithm.
[0079] According to step S10.5 of the method of the present invention, c is obtained. c 、c m 、c y Then, the predicted K / S value of the prepared ink can be obtained by the equation constraint expression described in step S9. The predicted corrected spectral reflectance can be further obtained. Figure 2 The corrected spectral reflectance curve predicted in this embodiment is given, and it can be intuitively observed that the target color sample reflectance curve and the predicted spectral reflectance curve match well. Furthermore, if necessary, color matching proofing can be performed based on the obtained formula, and the deviation of the spectral reflectance or Lab color value of the color matching proofing sample and the target sample can be compared, and the color matching personnel can perform more precise formula correction based on experience.
[0080] The spectral color matching method of this embodiment obtains the spectral correction coefficient through a small amount of color matching experiments, and integrates the nonlinear relationship between the K / S value and the concentration, which is more accurate than the prediction formula obtained based on the existing KM theory. In the actual production process, after the two stages of establishing the color matching data set and determining the correction coefficient are completed, the third stage of color matching can be performed multiple times; the first two stages only need to be performed once, and there is no need to repeat them every time the color matching target sample is received. After determining the color matching target sample, the third stage steps and methods of the present invention can be used to accurately predict the optimal ratio of each primary color ink. The present invention can match the color of the target sample more accurately, greatly improve the color matching accuracy of the single constant KM theory, and has important guiding significance for the development and application of ink color matching systems.
[0081] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Other modifications or equivalent substitutions made to the technical solution of the present invention by ordinary technicians in the field should be included in the scope of the claims of the present invention as long as they do not depart from the spirit and scope of the technical solution of the present invention.
Claims
1. A spectral color matching method based on KM model and intelligent optimization, characterized in that: The following steps are involved: Phase 1: Establish a color matching model, including the following S1 to S4: S1. Prepare a sample using base color ink and diluent. The specific process is: select base color ink and diluent, prepare coated paper with high gloss and smoothness; mix the base color ink and diluent in a mixture of c i :(1-c i )(0<c i ≤1, i = 1, 2, ..., n) are mixed evenly to obtain n groups of mixtures of base color ink and diluent in different proportions, and the mixtures are spread on coated paper to prepare n samples; n is 20, c1 = 5%, c2 = 10%, ..., c 20 =100%, with a step of 5%; S2. Measure the spectral reflectance of the substrate and the sample and correct it, keeping the undetermined parameters in the correction formula. The specific process is: measure the spectral reflectance coefficient R of the coated paper p (λ), measure the spectral reflectance coefficients of n samples in S1 Where i = 1, 2, ..., n; wavelength λ = 400, 410, ..., 700, unit is nm, wavelength interval is 10nm; spectral reflectance coefficient R for coated paper p (λ) is corrected, the formula for: Spectral reflectance coefficient for n samples To make corrections, the formula is: k1 and k2 are correction coefficients; S3. Calculate the K / S value of the substrate and the sample, and obtain the K / S value sequence of the base color ink according to the KM theory, and establish an interpolation function for calculating the K / S value of the base color ink at any concentration. The specific process is: calculate the ratio of the absorption coefficient of the coated paper to the scattering coefficient The wavelength λ=400,410,…,700; the formula is: Calculate the ratio of the absorption coefficient to the scattering coefficient of n samples Where i = 1, 2, ..., n, wavelength λ = 400, 410, ..., 700; the formula is: Further we get: f(λ,c i ) indicates that the concentration at wavelength λ is c i The K / S value of the base color ink; thus, for the wavelength λ, the sequence [0,f(λ,c1),f(λ,c2),…,f(λ,c n )]; Through the interpolation algorithm, the K / S value f(λ,c) function of the primary color ink corresponding to the wavelength λ and any concentration c can be obtained: here represents the interpolation function; the interpolation algorithm uses cubic spline interpolation; S4, set the primary inks to cyan ink, magenta ink, and yellow ink, repeat S1 to S3, obtain the interpolation functions corresponding to different primary inks, and establish a color matching model. The specific process is: set the primary inks to cyan ink, magenta ink, and yellow ink, repeat S1 to S3, and obtain the K / S value sequence [0, f c (λ,c1),f c (λ,c2),…,f c (λ,c n )]、[0,f m (λ,c1),f m (λ,c2),…,f m (λ,c n )]、[0,f y (λ,c1),f y (λ,c2),…,f y (λ,c n )], wavelength λ=400,410,…,700; through the interpolation algorithm, the K / S value f of the base color ink corresponding to the calculated wavelength λ and arbitrary concentration c is obtained. c (λ,c),f m (λ,c),f y (λ, c); according to the determined K / S value function f of the cyan ink, magenta ink and yellow ink c (λ,c),f m (λ,c),f y (λ,c), we get the color matching model, Where c c 、c m 、c y The proportions of cyan ink, magenta ink and yellow ink are respectively; is the predicted K / S value of the formulated ink; According to step S3, f c (λ,c c ), f m (λ,c m ), f y (λ,c y ) is obtained according to the interpolation algorithm; since the correction coefficients k1 and k2 in S2 need to be clearly determined, S3 and S4 can be calculated, so k1 and k2 in the color matching model are undetermined parameters; Phase II: Determine the parameters to be determined, including S5 to S7 below: S5. Mix the base color inks to conduct color matching experiments, prepare a small amount of color matching samples, and measure the spectral reflectance. The specific process is: mix cyan ink, magenta ink, and yellow ink at c c :c m :c y Mix evenly, 0 <c c ,c m ,c y ≤1, m groups of mixed inks with different proportions are obtained; if c c +c m +c y <1, the percentage of diluent is 1-(c c +c m +c y ); respectively spreading the mixed ink on coated paper to prepare m samples; m ≥ 2; and simultaneously measuring the spectral reflectance coefficients of the m samples; S6. Divide the range of the undetermined parameter into certain intervals to construct a search grid. The specific process is: divide the range of k1 into 0.035-0.045 with an interval of 0.005, and divide the range of k2 into 0.4-0.6 with an interval of 0.01, and obtain k1∈{0.035,0.04,0.045}, k2∈{0.4,0.41,0.42,…,0.6} respectively, and construct a search grid (k1, k2), which contains 3×21=63 search points in total; S7, define the accuracy index composed of the model prediction value and the experimental measurement value, calculate and find the highest accuracy for each grid point, and determine the optimal value of the unknown parameter; The third stage: solving the optimal formula, the specific contents include the following S8 to S10: S8, measure the spectral emissivity of the target color sample and the target paper, and further calculate the K / S value after correction. The specific process is: measure the spectral reflectivity R of the target color sample T (λ) and its paper reflectivity R TP (λ), and make corrections: Calculate the ratio of the absorption coefficient to the scattering coefficient of the target color sample and the target paper. The formula is: S9. Construct an optimization problem for solving the optimal ratio of each primary color ink, clarify the objective function and constraints. The specific process is: construct an optimization problem for solving the optimal ratio of each primary color ink; the optimization problem is expressed as: C(c y ,c m ,c c )=arg min C β(c y ,c m ,c c ), where c c 、c m 、c y are the proportions of cyan ink, magenta ink and yellow ink respectively; C(c y ,c m ,c c ) is the optimal ink ratio solution set; the objective function β(c y ,c m ,c c ) ensures the color matching prediction accuracy and is defined as in is the K / S value predicted by the color matching model, Ω={400,410,…,700} is the wavelength range; the equality constraints of the optimization problem are: where f c (λ,c c ), f m (λ,c m ), f y (λ,c y ) is the interpolation function obtained in step S4, wherein the undetermined parameters k1 and k2 are determined by S7; The inequality constraints for the optimization problem are: S10, using genetic algorithm to solve the optimization problem and obtain the optimal ink formula. The specific process is: using genetic algorithm to solve the optimization problem described in S9 and obtain the ratio c of cyan, magenta and yellow inks. c 、c m 、c y , and obtain the corrected spectral reflectance prediction value of the color matching ink.
2. The spectral color matching method based on KM model and intelligent optimization according to claim 1, characterized in that: The S7 is specifically: S7.
1. For each k1 and k2 value corresponding to each search grid point determined in S6, use the color matching model obtained in S4 to calculate the K / S prediction value of the prepared m samples respectively. S7.2, for each k1 and k2 value corresponding to each search grid point determined in S6, the spectral reflectance coefficients of the m samples measured in S5 are used to calculate the K / S measurement value of the m samples using the formulas in S2 and S3 S7.3, for each search grid point corresponding to the k1, k2 values determined in S6, using the results of S7.1 and S7.2, calculate the prediction accuracy index acc pred , the formula is as follows: Where Ω = {400, 410, ..., 700} is the wavelength range; a total of 63 accuracy index values acc are obtained pred , take the k1 and k2 values corresponding to the smallest accuracy index value as the optimal values of the unknown parameters k1 and k2.
3. The spectral color matching method based on KM model and intelligent optimization according to claim 1, characterized in that: The S10 is specifically: S10.
1. Initialize the population and select an appropriate population size, ranging from 50 to 100. Each individual is represented by a chromosome, which consists of three genes, representing c c 、c m 、c y The value of the gene value is in the range of [0,1]; individuals in the population are randomly generated to ensure that c c +c m +c y ≤1; S10.2, the objective function of the optimization problem is used as the fitness function, that is, Set the gene values of the chromosome to satisfy the inequality constraints: c c +c m +c y ≤1,0≤c c ≤1,0≤c m ≤1,0≤c y ≤1; the selection strategy of the genetic algorithm adopts roulette, and individuals are selected according to the fitness value. The smaller the fitness value, the greater the probability of being selected; S10.3, perform crossover operation, mutation operation and fitness calculation; set crossover probability p c , ranging from 0.8 to 0.9; using single-point crossover or double-point crossover, the genes of two parent individuals are partially exchanged to generate new offspring individuals; setting the mutation probability p m , ranging from 0.01 to 0.02; randomly select a gene in the chromosome and randomly change its value to ensure that the constraint conditions described in S10.2 are still met after the mutation; for each individual in the newly generated population, calculate its fitness value β(c y ,c m ,c c ); S10.4, repeat the crossover operation, mutation operation and fitness calculation described in S10.3, and stop the algorithm when the preset maximum number of generations is reached, which is set to 100 to 200 generations; or when the fitness value is less than a certain threshold, stop the algorithm; after stopping the algorithm, obtain the gene value c of the optimal individual c 、c m 、c y , that is, the optimal ratio c of cyan, magenta and yellow inks is obtained c :c m :c y ; S10.
5. If the solution is c c 、c m 、c y After c c +c m +c y <1, the percentage of diluent is 1-(c c +c m +c y );get c c 、c m 、c y Then, the predicted K / S value of the prepared ink can be obtained by the equation constraint expression described in S9. The predicted corrected spectral reflectance can be further obtained, and the calculation formula is:
Citation Information
Cited By
Digital ink-jet printing color analysis method and system based on image processing
CN122023545A
