CN105787275B - A kind of computer for colouring method of leather coloring - Google Patents
A kind of computer for colouring method of leather coloring Download PDFInfo
- Publication number
- CN105787275B CN105787275B CN201610110717.0A CN201610110717A CN105787275B CN 105787275 B CN105787275 B CN 105787275B CN 201610110717 A CN201610110717 A CN 201610110717A CN 105787275 B CN105787275 B CN 105787275B
- Authority
- CN
- China
- Prior art keywords
- computer
- model
- neural network
- formula
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16Z—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS, NOT OTHERWISE PROVIDED FOR
- G16Z99/00—Subject matter not provided for in other main groups of this subclass
Landscapes
- Spectrometry And Color Measurement (AREA)
Abstract
The present invention relates to a kind of computer for colouring methods of leather coloring, it is characterised in that the following steps are included: the reflectivity for 1) measuring sample to be matched colors is Rt(λ) converts K/S through SKM model formationt(λ).2) with K/St(λ) is input item, and the prediction of the equal reflectivity of weight is carried out by trained BP neural network, obtains K/Swt(λ).3) spectrum simulation algorithm is utilized, by K/Si(λ) is fitted K/Swt(λ) is calculated by least square method and is obtained dyeing recipe C (X1,X2...Xn).The computer for colouring method of a kind of leather coloring, which is characterized in that using integral ball-type dual-beam spectrophotometric color measurement instrument, λ wave-length coverage is visible light 400-700nm, wavelength interval 10nm for the test of reflectivity in the step 1).The front for selecting dermatotome is colour examining face, and respectively in left, center, right shoulder, left, center, right ridge, left, center, right stern, measurement is averaged.
Description
Technical field
The present invention relates to a kind of computer for colouring methods of leather coloring, and in particular to when a kind of leather or textile dyeing
Computer provides the method for formula forecast.
Background technique
Leather industry is an ancient, backward but capacity to earn foreign exchange through exports and stronger industry, although have passed through thousands of years
Development is not got rid of but always with situation handmade, based on experience control.The degree of automation of enterprise is very low, currently,
Domestic leather processing factory all uses Man-made Color Matching without exception, i.e., after dye-works is connected to client's sample, by dyeing master worker according to warp
The substantially formula for providing standard color sample is tested, then trial dyeing, then mixed colours, finally determines dyeing recipe.Obvious Man-made Color Matching pair
The skill requirement for dyeing master worker is very high and time-consuming and laborious, significantly limits the raising of leather processing factory production efficiency.And it answers
It replaces manually matching colors with computer, then can greatly improve the production efficiency of factory.Therefore, answered in leather processing enterprise
It will be a kind of new trend with computer color technology.
Computer for colouring technology has been widely applied in textile printing and dyeing industry, but does not answer substantially in leather coloring color matching
With.Computer for colouring technology is difficult to have in the major technique reason that leather industry is promoted: leather material difference is big, different seasons
Section, Different climate, different sources, the same kind skin base of different growing stages are very big to dyestuff absorption difference.It must be set up fitting thus
Close the computer for colouring method of leather color matching.
By historical sample library, solving conventional model by neural network algorithm with production actual sample can not answer the present invention
Pair specific production technology influence, improve an accuracy rate of color matching, push the automatization level of leather and textile industry
It is promoted, generates huge economic benefit.
Summary of the invention
The present invention is according to the above status, the characteristics of according to leather coloring, provides a kind of leather that primary color matching accuracy rate is high
Computer for colouring method.
The technical scheme adopted by the invention to solve the technical problem is that: a kind of computer for colouring method of leather coloring,
Characterized by the following steps:
1) reflectivity for measuring sample to be matched colors is Rt(λ) converts K/S through SKM model formationt(λ)。
2) with K/St(λ) is input item, and the prediction of the equal reflectivity of weight is carried out by trained BP neural network, obtains K/Swt
(λ)。
3) spectrum simulation algorithm is utilized, by K/Si(λ) is fitted K/Swt(λ) is calculated to obtain to dye by least square method and be matched
Square C (x1,x2...xn)。
A kind of computer for colouring method of leather coloring, which is characterized in that the survey of reflectivity in the step 1)
Integral ball-type dual-beam spectrophotometric color measurement instrument is used, λ wave-length coverage is visible light 400-700nm, wavelength interval 10nm.Select skin
The front of plate is colour examining face, and respectively in left, center, right shoulder, left, center, right ridge, left, center, right stern, measurement is averaged.
A kind of computer for colouring method of the leather coloring, which is characterized in that in the step 1, the SKM mould of use
Type is the mono- Constant Model of Kubeil-Munk, and expression formula is formula (1).
(formula 1)
The computer for colouring method of a kind of leather coloring, which is characterized in that using frequent in the step 2
With the BP neural network of database training.
The computer for colouring method of a kind of leather coloring, which is characterized in that BP nerve net is used in the step 2
Network is by K/Ss(λ) is input item, the equal K/S of weightw(λ) is that output item is built-up.By the sample stored in common prescription library into
Row albedo measurement obtains Rs(λ) is K/S through SKM model conversations(λ), the equal K/S value K/S of weightw(λ) is then matched by sample is corresponding
Side is transformed.
A kind of computer for colouring method of leather coloring, which is characterized in that the equal K/S value K/S of weightw(λ) is by converting
Formula (2) calculates, x1,x2...xnFor formula percentage.K/Si(λ) is reflection of i-th of composition dyestuff after SKM model conversation
Rate value.
K/Sw(λ)=x1K/S1(λ)+x2K/S2(λ)+…+xnK/Sn(λ) (formula 2)
The computer for colouring method of a kind of leather coloring, which is characterized in that the step 2) BP neural network
Training, intermediate hidden layer use logsig function, and output layer uses purelin function, using " Levenberg-Marquardt "
Algorithm.
A kind of computer for colouring method of leather coloring, which is characterized in that step 4) the digital simulation object
For K/Sw(λ), rather than K/St(λ), and be formulated using the spectral match algorithm calculating based on least square method.
Detailed description of the invention
The color matching flow chart of Fig. 1 embodiment of the present invention.
Fig. 2 is the training result schematic diagram of the embodiment of the present invention.
Fig. 3 is R described in the embodiment of the present invention 1t1(λ),K/St1(λ),K/Swt1(λ),Rc1The waveform diagram of (λ).
Fig. 4 is R described in the embodiment of the present invention 1t2(λ),K/St2(λ),K/Swt2(λ),Rc2The waveform diagram of (λ).
Fig. 5 is R described in the embodiment of the present invention 1t3(λ),K/St3(λ),K/Swt3(λ),Rc3The waveform diagram of (λ).
Fig. 6 is R described in the embodiment of the present invention 1t4(λ),K/St4(λ),K/Swt4(λ),Rc4The waveform diagram of (λ).
Specific embodiment
Embodiment 1
A simple sample database is used in embodiment of the present invention.The database share dyestuff in 3 (yellow E-3G,
Red S-BR, indigo plant E-BL).Sample group that it is stored is at being shown in Table 1.
1 data base set of table
Note: E-3G is Huang E-3G, and S-BR is red S-BR, and E-BL is indigo plant E-BL
Wherein, the building and training of BP neural network model: the training sample stored in common prescription library is measured
Obtain Rs(λ) is K/S by SKM model conversations(λ).The R of 3 dyestuffsi(λ) is through SKM model conversation K/Si(λ), according to formula and
K/S is calculated in formula 2w(λ).With K/Ss(λ) is input item, corresponding K/Sw(λ) is output item, and building BP neural network is matched
Color model.Here simplest 3-tier architecture is used, with sample pattern reflectivity K/Ss(λ) is input layer, and N number of hidden layer is (N number of
Number is arranged by experience Binding experiment, is set as here 38), with the equal model spectrum K/S of weightw(λ) is output layer, whereinEstablish K/Ss(λ) and K/SwNeural network model between (λ), the structure such as following figure
It is shown.Intermediate hidden layer uses logsig function, and output layer uses purelin function, using " Levenberg-Marquardt "
Algorithm.
The algorithm is run in the Matlab 2011b version with Neural Network Toolbox, and intermediate hidden layer uses
Logsig function, output layer are used purelin function, are calculated in this example using " trainlm (Levenberg-Marquardt) "
Method, main code are as follows :=net=newff (K/Ss(λ),K/Sw(λ),38,{'logsig','purelin'},'
trainlm')；% model construction [net, tr]=train (net, K/Ss(λ),K/Sw(λ),[],[],val,test)；% instruction
Practice, val, test are respectively the verifying formulated and training set.Training result is shown in Fig. 2.
When color matching, the reflectivity for measuring sample to be matched colors is Rt(λ) converts K/S through SKM model formationt(λ), with K/St
(λ) is input item, is predicted by BP neural network color matching model, and prediction obtains the equal K/S value K/S of weightwt(λ)。
K/Swt(λ)=sim (net, K/St(λ))；% predicted with trained neural network, input data K/
St(λ) obtains the equal K/S value K/S of pre- check weighingwt(λ)。
Spectrum simulation algorithm is finally utilized, by K/Si(λ) is fitted K/Swt(λ) is dyed by least square method calculating
It is formulated C (x1,x2...xn)。
Specifically, the equal K/S value K/S of the weight of acquisitionwt(λ), therefore the equal K/S value K/S of weightwt(λ) and color matching composition xiAnd K/
SiMeet following formula between (λ),
Expansion are as follows:
If:
Then: K/Sw=P × C,
Since above-mentioned equation is n unknown quantity of 31 equation solutions (usual n < 31), need to solve using least square method
Equation group obtains the spectral match algorithm are as follows:
C=(PTP)-1PTf(Rw)
Then following step is carried out, specifically includes that the reflectivity of (1) measurement standard specimen (red 1.2% Huang 0.8%), and is converted
For model reflectivity；Measurement uses Datacolor600 spectral luminosity instrument, 400~700nm of wave-length coverage, wavelength interval 10nm.
Use Rt1(λ) is indicated, by Rt1(λ) substitutes into color matching model formula (1), is converted into model reflectivity K/St1(λ)；(2) with K/St1(λ)
For input layer, the good BP neural network of application training is calculated, and obtains K/Swt1(λ)；(3) monochromatic model reflectivity K/S is usedt1
The equal model reflectivity K/S of (λ) fitting weightwt1(λ) obtains formula C with least square method1(red 1.06%, Huang is 0.82%)；
(4) formula C according to weather report1It draws a design, and measures the reflectivity R for color of drawing a designc1(λ)；Calculate standard specimen Rt1(λ) and color of drawing a design
Rc1Color difference between (λ), obtaining CMC color difference is 0.32.Color matching is completed.Rt1(λ),K/St1(λ),K/Swt1(λ),Rc1(λ) is shown in figure
3。
Embodiment 2:
In the embodiment of the present invention and comparative example, as illustratively and convenient for comparing, one is used simply
Sample database.The database shares 3 dyestuffs (yellow E-3G, red S-BR, blue E-BL).Sample group that it is stored is at being shown in Table 1.
1 data base set of table
Note: E-3G is Huang E-3G, and S-BR is red S-BR, and E-BL is indigo plant E-BL
Wherein, the building and training of BP neural network model: the training sample stored in common prescription library is measured
Obtain Rs(λ) is K/S by SKM model conversations(λ).The R of 3 dyestuffsi(λ) is through SKM model conversation K/Si(λ), according to formula and
K/S is calculated in formula 2w(λ).With K/Ss(λ) is input item, corresponding K/Sw(λ) is output item, and building BP neural network is matched
Color model.Here simplest 3-tier architecture is used, with sample pattern reflectivity K/Ss(λ) is input layer, and N number of hidden layer is (N number of
Number is arranged by experience Binding experiment, is set as here 38), with the equal model spectrum K/S of weightw(λ) is output layer, whereinEstablish K/Ss(λ) and K/SwNeural network model between (λ), structure are as follows
Shown in figure.Intermediate hidden layer uses logsig function, and output layer uses purelin function, using " Levenberg-
Marquardt " algorithm.
The algorithm is run in the Matlab 2011b version with Neural Network Toolbox, and intermediate hidden layer uses
Logsig function, output layer are used purelin function, are calculated in this example using " trainlm (Levenberg-Marquardt) "
Method, main code are as follows:
Net=newff (K/Ss(λ),K/Sw(λ),38,{'logsig','purelin'},'trainlm')；% model structure
It builds
[net, tr]=train (net, K/Ss(λ),K/Sw(λ),[],[],val,test)；% training, val, test points
The verifying that Wei do not formulate and training set.
Training solves to see Fig. 2.
When color matching, the reflectivity for measuring sample to be matched colors is Rt(λ) converts K/S through SKM model formationt(λ), with K/St
(λ) is input item, is predicted by BP neural network color matching model, and prediction obtains the equal K/S value K/S of weightwt(λ)。
K/Swt(λ)=sim (net, K/St(λ))；% predicted with trained neural network, input data K/St
(λ) obtains the equal K/S value K/S of pre- check weighingwt(λ)。
Spectrum simulation algorithm is finally utilized, by K/Si(λ) is fitted K/Swt(λ) is dyed by least square method calculating
It is formulated C (x1,x2...xn)。
Specifically, the equal K/S value K/S of the weight of acquisitionwt(λ), therefore the equal K/S value K/S of weightwt(λ) and color matching composition xiAnd K/
SiMeet following formula between (λ),
Expansion are as follows:
If:
Then: K/Sw=P × C,
Since above-mentioned equation is n unknown quantity of 31 equation solutions (usual n < 31), need to solve using least square method
Equation group obtains the spectral match algorithm are as follows:
C=(PTP)-1PTf(Rw)
Then the reflectivity of following key steps (1) measurement standard specimen (yellow 0.4% indigo plant 1.6%) is carried out, and is converted into model
Reflectivity；Measurement uses Datacolor600 spectral luminosity instrument, 400~700nm of wave-length coverage, wavelength interval 10nm.Use Rt2(λ)
It indicates, by Rt2(λ) substitutes into color matching model formula (1), is converted into model K/S value K/St2(λ)；(2) with K/St2(λ) is input layer,
The good BP neural network of application training is calculated, and K/S is obtainedwt2(λ)；(3) monochromatic model K/S value K/S is usedt2(λ) fitting weight is equal
Model K/S value K/Swt2(λ) obtains formula C with least square method2(red 0.12% Huang 0.22%, indigo plant is 1.58%)；(4) root
According to the formula C of forecast2It draws a design, and measures the reflectivity R for color of drawing a designc2(λ)；Calculate standard specimen Rt2(λ) and the color R that draws a designc2(λ)
Between color difference, obtain CMC color difference be 0.21.Color matching is completed.Rt2(λ),K/St2(λ),K/Swt2(λ),Rc2(λ) is shown in Fig. 4.
Embodiment 3:
In the embodiment of the present invention and comparative example, as illustratively and convenient for comparing, one is used simply
Sample database.The database shares 3 dyestuffs (yellow E-3G, red S-BR, blue E-BL).Sample group that it is stored is at being shown in Table 1.
1 data base set of table
Note: E-3G is Huang E-3G, and S-BR is red S-BR, and E-BL is indigo plant E-BL
Wherein, the building and training of BP neural network model: the training sample stored in common prescription library is measured
Obtain Rs(λ) is K/S by SKM model conversations(λ).The R of 3 dyestuffsi(λ) is through SKM model conversation K/Si(λ), according to formula and
K/S is calculated in formula 2w(λ).With K/Ss(λ) is input item, corresponding K/Sw(λ) is output item, and building BP neural network is matched
Color model.Here simplest 3-tier architecture is used, with sample pattern reflectivity K/Ss(λ) is input layer, and N number of hidden layer is (N number of
Number is arranged by experience Binding experiment, is set as here 38), with the equal model spectrum K/S of weightw(λ) is output layer, whereinEstablish K/Ss(λ) and K/SwNeural network model between (λ), structure is such as
Shown in the following figure.Intermediate hidden layer uses logsig function, and output layer uses purelin function, using " Levenberg-
Marquardt " algorithm.
The algorithm is run in the Matlab 2011b version with Neural Network Toolbox, and intermediate hidden layer uses
Logsig function, output layer are used purelin function, are calculated in this example using " trainlm (Levenberg-Marquardt) "
Method, main code are as follows:
Net=newff (K/Ss (λ), K/Sw(λ),38,{'logsig','purelin'},'trainlm')；% model
Building
[net, tr]=train (net, K/Ss(λ),K/Sw(λ),[],[],val,test)；% training, val, test
The verifying respectively formulated and training set.
Training solves to see Fig. 2.
When color matching, the reflectivity for measuring sample to be matched colors is Rt(λ) converts K/S through SKM model formationt(λ), with K/St
(λ) is input item, is predicted by BP neural network color matching model, and prediction obtains the equal K/S value K/S of weightwt(λ)。
K/Swt(λ)=sim (net, K/St(λ))；% predicted with trained neural network, input data K/St
(λ) obtains the equal K/S value K/S of pre- check weighingwt(λ)。
Spectrum simulation algorithm is finally utilized, by K/Si(λ) is fitted K/Swt(λ) is dyed by least square method calculating
It is formulated C (x1,x2...xn)。
Specifically, the equal K/S value K/S of the weight of acquisitionwt(λ), therefore the equal K/S value K/S of weightwt(λ) and color matching composition xiAnd K/
SiMeet following formula between (λ),
Expansion are as follows:
If:
Then: K/Sw=P × C,
Since above-mentioned equation is n unknown quantity of 31 equation solutions (usual n < 31), need to solve using least square method
Equation group obtains the spectral match algorithm are as follows:
C=(PTP)-1PTf(Rw)
Then following key steps are carried out: (1) measuring the reflectivity of standard specimen (red 0.8% indigo plant 1.2%), and is converted into model
Reflectivity；Measurement uses Datacolor600 spectral luminosity instrument, 400~700nm of wave-length coverage, wavelength interval 10nm.Use Rt3(λ)
It indicates, by Rt3(λ) substitutes into color matching model formula (1), is converted into model K/S value K/St3(λ)；(2) with K/St3(λ) is input layer,
The good BP neural network of application training is calculated, and K/S is obtainedwt3(λ)；(3) monochromatic model K/S value K/S is usedt3(λ) fitting weight is equal
Model K/S value K/Swt3(λ), with least square method, obtaining formula C3, (red 0.69%, Huang 0.08%, indigo plant is 1.25%)；(4) root
According to the formula C of forecast3It draws a design, and measures the reflectivity R for color of drawing a designc3(λ)；Calculate standard specimen Rt3(λ) and the color R that draws a designc3(λ)
Between color difference, obtain CMC color difference be 0.27.Color matching is completed.Rt3(λ),K/St3(λ),K/Swt3(λ),Rc3(λ) is shown in Fig. 5.
Embodiment 4:
In the embodiment of the present invention and comparative example, as illustratively and convenient for comparing, one is used simply
Sample database.The database shares 3 dyestuffs (yellow E-3G, red S-BR, blue E-BL).Sample group that it is stored is at being shown in Table 1.
1 data base set of table
Note: E-3G is Huang E-3G, and S-BR is red S-BR, and E-BL is indigo plant E-BL
Wherein, the building and training of BP neural network model: the training sample stored in common prescription library is measured
Obtain Rs(λ) is K/S by SKM model conversations(λ).The R of 3 dyestuffsi(λ) is through SKM model conversation K/Si(λ), according to formula and
K/S is calculated in formula 2w(λ).With K/Ss(λ) is input item, corresponding K/Sw(λ) is output item, and building BP neural network is matched
Color model.Here simplest 3-tier architecture is used, with sample pattern reflectivity K/Ss(λ) is input layer, and N number of hidden layer is (N number of
Number is arranged by experience Binding experiment, is set as here 38), with the equal model spectrum K/S of weightw(λ) is output layer, whereinEstablish K/Ss(λ) and K/SwNeural network model between (λ), structure are as follows
Shown in figure.Intermediate hidden layer uses logsig function, and output layer uses purelin function, using " Levenberg-
Marquardt " algorithm.
The algorithm is run in the Matlab 2011b version with Neural Network Toolbox, and intermediate hidden layer uses
Logsig function, output layer are used purelin function, are calculated in this example using " trainlm (Levenberg-Marquardt) "
Method, main code are as follows:
Net=newff (K/Ss (λ), K/Sw(λ),38,{'logsig','purelin'},'trainlm')；% model
Building
[net, tr]=train (net, K/Ss(λ),K/Sw(λ),[],[],val,test)；% training, val, test points
The verifying that Wei do not formulate and training set.
Training solves to see Fig. 2.
When color matching, the reflectivity for measuring sample to be matched colors is Rt(λ) converts K/S through SKM model formationt(λ), with K/St
(λ) is input item, is predicted by BP neural network color matching model, and prediction obtains the equal K/S value K/S of weightwt(λ)。
K/Swt(λ)=sim (net, K/St(λ))；% predicted with trained neural network, input data K/St
(λ) obtains the equal K/S value K/S of pre- check weighingwt(λ)。
Spectrum simulation algorithm is finally utilized, by K/Si(λ) is fitted K/Swt(λ) is dyed by least square method calculating
It is formulated C (x1,x2...xn)。
Specifically, the equal K/S value K/S of the weight of acquisitionwt(λ), therefore the equal K/S value K/S of weightwt(λ) and color matching composition xiAnd K/
SiMeet following formula between (λ),
Expansion are as follows:
If:
Then: K/Sw=P × C,
Since above-mentioned equation is n unknown quantity of 31 equation solutions (usual n < 31), need to solve using least square method
Equation group obtains the spectral match algorithm are as follows:
C=(PTP)-1PTf(Rw)
Then it carries out following step: (1) measuring the reflectivity of standard specimen (red 0.6% yellow 0.6% indigo plant 0.8%), and be converted into
Model reflectivity；Measurement uses Datacolor600 spectral luminosity instrument, 400~700nm of wave-length coverage, wavelength interval 10nm.With
Rt4(λ) is indicated, by Rt4(λ) substitutes into color matching model formula (1), is converted into model K/S value K/St4(λ)；(2) with K/St4(λ) is defeated
Enter layer, the good BP neural network of application training is calculated, and K/S is obtainedwt4(λ)；(3) monochromatic model reflectivity K/S is usedt4(λ) is quasi-
Close the equal model K/S value K/S of weightwt4(λ) obtains formula C with least square method4(red 0.57% yellow 0.69% indigo plant 0.87%)；
(4) formula C according to weather report4It draws a design, and measures the reflectivity R for color of drawing a designc4(λ)；Calculate standard specimen Rt4(λ) and color of drawing a design
Rc4Color difference between (λ), obtaining CMC color difference is 0.14.Color matching is completed.Rt4(λ),K/St4(λ),K/Swt4(λ),Rc4(λ) is shown in figure
6。
The computer for colouring method of dyeing provided by the present invention is described in detail above, tool used herein
Principle and implementation of the present invention are described for body example, the above embodiments are only used to help understand this hair
Bright method and core concept；At the same time, for those skilled in the art is being embodied according to the thought of the present invention
There will be changes in mode and application range, in conclusion the content of the present specification portion is interpreted as limitation of the present invention.
Claims (6)
1. a kind of computer for colouring method of leather coloring, it is characterised in that the following steps are included:
1) reflectivity for measuring sample to be matched colors is Rt(λ) converts K/S through SKM model formationt(λ)；
2) with K/St(λ) is input item, and the prediction of the equal K/S value of weight is carried out by trained BP neural network, obtains K/Swt(λ)；
3) spectrum simulation algorithm is utilized, by K/Si(λ) is fitted K/Swt(λ) is calculated by least square method and is obtained dyeing recipe C
(x1,x2...xn)；
The neural network model used in the step 2) is by the BP neural network after the training of frequently-used data library；
Using BP neural network in the step 2) is by K/Ss(λ) is input item, the equal K/S value K/S of weightw(λ) is output item structure
It builds；
The sample stored in common prescription library progress albedo measurement is obtained into Rs(λ) is K/S through SKM model conversations(λ), weight
Equal K/S value K/Sw(λ) is then transformed by the corresponding formula of sample.
2. a kind of computer for colouring method of leather coloring according to claim 1, which is characterized in that in the step 1)
The reflectivity of sample to be matched colors is measured using integral ball-type dual-beam spectrophotometric color measurement instrument, λ wave-length coverage is visible light 400-
700nm, wavelength interval 10nm；The front for selecting dermatotome is colour examining face, respectively in left, center, right shoulder, left, center, right ridge, it is left, in,
Right stern, measurement are averaged.
3. a kind of computer for colouring method of leather coloring according to claim 1, which is characterized in that the step 1)
In, for the SKM model used for the mono- Constant Model of Kubeil-Munk, expression formula is formula 1:
4. a kind of computer for colouring method of leather coloring according to claim 1, which is characterized in that in the step 2)
The equal K/S value K/S of weightw(λ) is calculated by conversion formula 2, x1,x2...xnFor formula percentage, K/S value K/Si(λ) is i-th group
At K/S value of the dyestuff after SKM model conversation:
K/Sw(λ)=x1K/S1(λ)+x2K/S2(λ)+…+xnK/Sn(λ) (formula 2).
5. a kind of computer for colouring method of leather coloring according to claim 1, which is characterized in that the step 2) BP
The training of neural network, intermediate hidden layer use logsig function, and output layer uses purelin function, using " Levenberg-
Marquardt " algorithm.
6. a kind of computer for colouring method of leather coloring according to claim 1, which is characterized in that the step 3) meter
Calculating fitting object is K/Swt(λ), rather than K/St(λ), and matched using the spectral match algorithm calculating based on least square method
Side.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610110717.0A CN105787275B (en) | 2016-02-29 | 2016-02-29 | A kind of computer for colouring method of leather coloring |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610110717.0A CN105787275B (en) | 2016-02-29 | 2016-02-29 | A kind of computer for colouring method of leather coloring |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105787275A CN105787275A (en) | 2016-07-20 |
| CN105787275B true CN105787275B (en) | 2019-03-22 |
Family
ID=56403895
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610110717.0A Active CN105787275B (en) | 2016-02-29 | 2016-02-29 | A kind of computer for colouring method of leather coloring |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105787275B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107944107A (en) * | 2017-11-15 | 2018-04-20 | 魔金真彩网络科技（长沙）有限公司 | A kind of single silver powder computer for colouring method and system for automobile metal paint |
| CN109946245B (en) * | 2019-02-01 | 2020-05-01 | 东华大学 | Computer color matching method of dye |
| CN109811493B (en) * | 2019-02-01 | 2020-05-01 | 东华大学 | Automatic liquid feeding method for multi-component dye dyeing process |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4914603A (en) * | 1988-12-14 | 1990-04-03 | Gte Laboratories Incorporated | Training neural networks |
| CN101719192A (en) * | 2009-11-17 | 2010-06-02 | 华侨大学 | Method and system for dynamically simulating dyeing process of fabrics |
| CN102750403A (en) * | 2012-05-28 | 2012-10-24 | 嘉兴学院 | Formula screening and correction method for spun-dyed yarn color matching |
| CN102819681A (en) * | 2012-08-02 | 2012-12-12 | 绍兴文理学院 | Method of predicting homochromatism formula of brocade cotton fabric by using computer color matching system |
| CN104239659A (en) * | 2013-06-07 | 2014-12-24 | 中国石油天然气集团公司 | Carbon steel corrosion rate prediction method of back propagation (BP) neural network |
-
2016
- 2016-02-29 CN CN201610110717.0A patent/CN105787275B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4914603A (en) * | 1988-12-14 | 1990-04-03 | Gte Laboratories Incorporated | Training neural networks |
| CN101719192A (en) * | 2009-11-17 | 2010-06-02 | 华侨大学 | Method and system for dynamically simulating dyeing process of fabrics |
| CN102750403A (en) * | 2012-05-28 | 2012-10-24 | 嘉兴学院 | Formula screening and correction method for spun-dyed yarn color matching |
| CN102819681A (en) * | 2012-08-02 | 2012-12-12 | 绍兴文理学院 | Method of predicting homochromatism formula of brocade cotton fabric by using computer color matching system |
| CN104239659A (en) * | 2013-06-07 | 2014-12-24 | 中国石油天然气集团公司 | Carbon steel corrosion rate prediction method of back propagation (BP) neural network |
Non-Patent Citations (3)
| Title |
|---|
| 基于BP神经网络的色纺纱配色;沈加加 等;《纺织学报》;20151130;第36卷(第11期);第34-38页 |
| 基于Kubelka－Munk单常数理论的皮革配色实践;杨峰 等;《中国皮革》;20120930;第41卷(第17期);第12-15页 |
| 基于约束最小二乘法的色纺纱配色方法;沈加加 等;《针织工业》;20150630(第6期);第31-33页 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105787275A (en) | 2016-07-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107766603B (en) | A kind of colour-spun yarns computer is measured color method | |
| CN105787275B (en) | A kind of computer for colouring method of leather coloring | |
| CN107103181A (en) | A kind of colour-spun yarns color matching method based on least square method supporting vector machine | |
| CN101100807B (en) | Textile color stability and color aberration grading method | |
| CN105787559B (en) | Colour-spun yarns color matching method based on neutral net | |
| CN102750403B (en) | The recipe determination of a kind of colour-spun yarns color matching and modification method | |
| CN101833593A (en) | Computer color-matching method of color-blended wool fabrics based on full-spectrum color matching | |
| CN109582994B (en) | Intelligent color selection and matching method for colored spun yarn | |
| CN106053024B (en) | A kind of LED light source preference degree prediction technique towards monochromatic system object | |
| CN101668109A (en) | Soft-proofing method for color-spinning yarn by tri-stimulus value matching | |
| CN102934988A (en) | Manufacture method of evaluation color board of digital system for inspection of traditional Chinese medicine | |
| CN109523600A (en) | A kind of historical relic's protection reparation recruits color method | |
| CN107194081A (en) | A kind of automobile plain color paint computer for colouring method | |
| CN109914120A (en) | A kind of design method of dye formulation | |
| CN109811493A (en) | A kind of multi-component dyes dyeing course gives liquid method automatically | |
| CN108872156A (en) | A kind of ink component scale prediction method and apparatus based on spectral reflectivity inverse | |
| CN103728023B (en) | Lovibond colour scale spectrometer color measuring device and method for measuring color thereof | |
| CN100334162C (en) | Making dye mixtures to produce a certain target colour | |
| CN107703064A (en) | A kind of method of natural color-cotton color grading | |
| CN105740951B (en) | A kind of leather finish computer for colouring method | |
| KR100914952B1 (en) | Color-matching method with computer color-matching apparatus | |
| CN109946245B (en) | Computer color matching method of dye | |
| CN105548038A (en) | Detection method of wine color | |
| CN105806483B (en) | A kind of non-linear recipe correction for leather coloring color matching | |
| CN113128104A (en) | Computer color measuring and matching method based on improved BP neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information |
Address after: 314000 2 building, Photovoltaic Science Park, 1288 Kang He road, Xiuzhou District, Jiaxing, Zhejiang. Applicant after: Jiaxing University Address before: 314001 No. 56 South Yuexiu Road, Zhejiang, Jiaxing Applicant before: Jiaxing University |
|
| CB02 | Change of applicant information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20220402 Address after: 276400 west section of Tengfei Road, Yishui Economic Development Zone, Linyi City, Shandong Province Patentee after: SHANDONG HENGTAI TEXTILE Co.,Ltd. Address before: 314000 2 building, Photovoltaic Science Park, 1288 Kang He road, Xiuzhou District, Jiaxing, Zhejiang. Patentee before: JIAXING University |
|
| TR01 | Transfer of patent right |