CN109115701B - Intelligent color-selecting wool-splicing method for wool knitting yarns - Google Patents

Intelligent color-selecting wool-splicing method for wool knitting yarns Download PDF

Info

Publication number
CN109115701B
CN109115701B CN201810852848.5A CN201810852848A CN109115701B CN 109115701 B CN109115701 B CN 109115701B CN 201810852848 A CN201810852848 A CN 201810852848A CN 109115701 B CN109115701 B CN 109115701B
Authority
CN
China
Prior art keywords
wool
reflectivity
color
formula
sample
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
Application number
CN201810852848.5A
Other languages
Chinese (zh)
Other versions
CN109115701A (en
Inventor
沈加加
何铠君
杨颖�
张弛
徐国华
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiaxing University
Original Assignee
Jiaxing University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Jiaxing University filed Critical Jiaxing University
Priority to CN201810852848.5A priority Critical patent/CN109115701B/en
Publication of CN109115701A publication Critical patent/CN109115701A/en
Application granted granted Critical
Publication of CN109115701B publication Critical patent/CN109115701B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
    • G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
    • G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry

Landscapes

  • Physics & Mathematics (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Spectrometry And Color Measurement (AREA)

Abstract

The invention relates to an intelligent color-selecting wool-splicing method for wool knitting yarns, which comprises the following steps of measuring the reflectivity of a wool knitting yarn sample; measuring the reflectivity of the wool knitting yarn sample by using a spectrophotometer, and identifying the component color by using a support vector machine; firstly, training a support vector machine model, defining a wool knitting yarn sample with a known color composition formula as a standard sample and carrying out reflectivity measurement, carrying out data preprocessing on collected reflectivity data of the standard sample to obtain a first characteristic vector data set, and dividing the first characteristic vector data set into a training set and a verification set for training the support vector machine model; carrying out data preprocessing on collected wool knitting yarn sample reflectivity data to form a second feature vector data set; and thirdly, calculating the optimal combination proportion of all component colors of the wool knitting yarns by using a tristimulus value matching algorithm or a full spectrum color matching algorithm, and calculating the reflectivity of the wool knitting yarn sample measured in the first step by using a single-color reflectivity to obtain a wool splicing formula.

Description

Intelligent color-selecting wool-splicing method for wool knitting yarns
Technical Field
The invention relates to an intelligent color-selecting wool-splicing method for wool knitting yarns, and belongs to the technical field of textile color matching.
Background
In wool knitting products, loose fibers or wool tops are dyed firstly and then mixed with colors more and more, and the mixed color yarn is composed of 2 or more than 2 fibers with different colors and the same or different colors. In the production and processing process of the color-mixed wool knitting yarn, wool splicing is the most critical production link. The 'wool-piecing' is to match dyed wool tops or loose wool fibers with different colors with each other to achieve the required color and style. In the prior art, the color matching is generally performed by wool splicing personnel in a factory by means of experience and repeated trial spinning or by means of a computer arrangement and combination method, all possibilities are exhausted, a plurality of formulas are often given, and the real color of the wool splicing personnel cannot be quickly and accurately identified, so that the production period is long, the production efficiency is low, and the market requirements of small batch, multiple varieties and quick delivery are difficult to adapt.
A Support Vector Machine (SVM) model, as an already mature recognition model, has a great number of applications in a plurality of fields, and its principle is as follows: a Support Vector Machine (SVM) mainly carries out classification decision on different types of samples through support vectors. The decision basis is mainly to search an Optimal classification surface (Optimal Hyper-plane) through a nonlinear kernel function algorithm, so that the classification interval (Margin) of samples on two sides of the Optimal classification surface is maximum. Therefore, the SVM algorithm is a binary classification method, namely, the SVM algorithm belongs to another class of relation without belonging to the class. The kernel of the algorithm is a symmetric function φ mapping k X → F, so for all XiAnd x, both having k (x)i,x)＝{φ(xi) Phi (X), the input space X is transformed into a feature space F.
However, the application of the method in the field of color spinning is in the starting stage, for example, china discloses an invention of color spinning color matching based on a least square support vector machine, the application number is CN201710188008.9, the patent mentions that an SVM model is applied to color spinning color matching, the main technology is that the reflectance and the corresponding proportion relation of a standard sample are used as training samples for training the SVM model, the predicted composition color and the proportion of a formula can be directly obtained through the trained SVM model, and the existence is obviously insufficient:
(1) the method uses a least square support vector machine to directly calculate the proportion, the relation between the reflectivity of a standard sample and the corresponding proportion is established, an SVM model is a classification technology and is used for classifying according to the proportion, the proportion is equivalent to one classification, and the SVM model has application value and must exhaust all composition schemes and proportions to be used as training samples, so that the training samples are too large in quantity and low in feasibility;
(2) the invention also requires that the target sample is a sample composed of the reference color, such as a standard sample in the training in a patent, and can forecast a closer proportion. In practical application, the target color is usually not composed of the reference color sample, and the generalization performance of the target color sample to the unknown reference color sample is lacked.
Disclosure of Invention
The invention aims to overcome the problems and provides an intelligent color selecting and wool splicing method for knitting wool yarns, which can quickly and accurately match colors, and can simulate the experience identification capability of human eyes to colors in artificial color matching without exhaustively exhausting all composition schemes and proportions as training samples, reduce the quantity of the training samples and realize intelligent color matching.
The intelligent color selecting and wool splicing method for the wool knitting yarns comprises the following steps;
step one, measuring the sample reflectivity of the wool knitting yarn: measuring the reflectivity of a wool knitting yarn sample by using a spectrophotometer, wherein the test condition requires mirror reflection, complete data with the wavelength of 400-700nm are included, and the reflectivity wavelength measurement interval is 0-10 nm;
secondly, recognizing the component colors by the SVM; firstly, training an SVM model, defining a wool knitting yarn sample with a known color composition formula as a standard sample and carrying out reflectivity measurement, carrying out data preprocessing on collected reflectivity data of the standard sample to obtain a first characteristic vector data set, and dividing the first characteristic vector data set into a training set and a verification set for training a support vector machine model; carrying out data preprocessing on the collected incoming sample reflectivity data to form a second feature vector data set; identifying the second data set of the feature vectors by using a trained SVM model to obtain the color composition of the component colors of the knitting yarns of the wool of the coming sample;
and step three, calculating the optimal combination proportion of all the component colors of the wool knitting yarns by using a tristimulus value matching algorithm or a full spectrum color matching algorithm, namely obtaining the single-color reflectivity of all the component colors according to the component colors of the wool knitting yarns coming sample obtained in the step two, and calculating the reflectivity of the coming sample measured in the step one by using the single-color reflectivity to obtain the wool splicing formula.
In the color-selecting and wool-piecing method, the SVM model type is selected to be C-SVC, an RBF kernel function is adopted, the penalty coefficient C is 3, the kernel function parameter gamma is 0.0323, the RBF kernel function formula (1) is as follows,
K(xi，x)＝exp(-γ||xi-x||2)，γ>0 … … equation (1).
The color-selecting and wool-splicing method is characterized in that the SVM model converts multi-classification recognition into two-classification recognition and comprises the following steps,
1) identification y1When it contains y1Is one, the rest y2…ymIs one type, and whether y is present or not is determined1；
2) Identification y2When it contains y2As one group, y1，y3…ymRegarding as one type, determining whether there is y2；
……
m) identification of ymWhen it contains ymAs one group, y1…ym-1Regarding as one type, determining whether there is ym；
This converts the multiple output equations of the multi-class recognition formula (2) into a single output of the two-class recognition formula (3),
f(xi)＝{y1,y2,…ym}，xi＝{x1,x2,…,xly ± 1, i 1, …, n … … … formula (2);
f(xi)＝yj，xi＝{x1,x2,…,xl}，yj± 1, j ═ 1, …, m, … … … … formula (3)
Where n is the total number of samples, m is the total number of constituent colors, and l is the dimension of the input data, equation (2) indicates that x is inputiIs l-dimensional data, the output is { y1,y2,…ym+1 represents containing, -1 represents not containing, formula (3) represents the judgment of whether each output result corresponds to a single color, if m colors exist in the single color library, the output is obtained after m times of calculationColor-spun yarn xiThe color composition of (a).
The method for selecting color and splicing wool is characterized in that the data preprocessing adopts the steps of firstly enlarging the reflectivity by 100 times and then combining the values of La b on the basis of the reflectivity to form x [100 × (lambda) ] R (lambda)1),100*R(λ2)，…,100*R(λn),L*,a*,b*]TWherein R (λ)n) Represents lambdanThe value of n is determined by the measuring interval of the reflectivity wavelength, and when the interval is 10, the value of n is 31; l, a, b are CIE1976LAB values calculated from R (λ) according to CIELAB formula.
The color-selecting and wool-splicing method comprises the third step of further including a mixed color reflectivity model, wherein the mixed color reflectivity model satisfies the following two relational expressions:
Figure BDA0001747845560000031
Figure BDA0001747845560000032
wherein: rs(lambda) represents the reflectance of the wool yarn at a wavelength of lambda, Ri(λ) represents the reflectance of the i component in a single color at a wavelength λ, αiAnd (3) representing the mass proportion of the component i in single color, wherein M is a model parameter and is adjusted according to the type and the yarn count of wool yarns.
Compared with the prior art, the invention has the following positive effects:
the method has the outstanding advantages that a support vector machine-based component color identification model is constructed, the component colors are identified in advance, the color selection range is reduced, the calculated amount is reduced, and the coincidence rate of yarn effects is improved.
Drawings
FIG. 1 is a flow chart of the intelligent color selection and wool-splicing of the present invention.
Fig. 2 is a flowchart of the SVM recognizing component colors according to the present invention.
Fig. 3 is an architectural diagram of an SVM recognition model of the present invention.
FIG. 4 is a chart showing the forecast results of HY-70291 recipe in accordance with the present invention.
FIG. 5 is a diagram of the forecast results of the 5M0040 formulation of the invention.
FIG. 6 is a diagram showing the forecast results of HY-65511 recipe in accordance with the present invention.
FIG. 7 is a diagram showing the result of forecasting HY-65511 recipe without SVM recognition.
Detailed Description
The present invention will be described in detail with reference to specific examples.
The intelligent color-selecting and wool-splicing method of the invention needs a hardware part and a software part. Wherein, the hardware part mainly comprises a spectrophotometer and a computer; the software part mainly comprises procedures of incoming sample reflectivity acquisition, SVM identification and color matching operation. The working principle is as follows: as shown in fig. 1, the reflectivity of the incoming sample is measured by a spectrophotometer as a data input end, the trained SVM is used for carrying out component color recognition on the preprocessed reflectivity data of the incoming sample, the recognized component color is combined with a color mixing model to carry out full spectrum and tristimulus value matching operation on the reflectivity of the incoming sample, and a formula is output.
The detailed steps of this embodiment are as follows,
step one, measuring reflectivity;
and (3) measuring the reflectivity: collected by a Datacolor850 spectrophotometer under the test conditions of a D65 light source, a 10-degree field of view and a 20mm aperture, the wavelength range is 400-700nm, and the interval is 10 nm.
Secondly, identifying the component colors by a Support Vector Machine (SVM);
training an SVM;
data: 100 wool knitting yarn samples of known wool-pieced composition formula are selected from the sample card as standard samples, and 10 composition colors (marked as y) are totally selected02,y06,y08,y10,y27,y31,y42,y47,y57,y58)。
Construction and training of the SVM model: the SVM algorithm is implemented by an integrated software-supported vector classification LIBSVM under Matlab R2011b version, wherein the selected SVM type is C-SVC, because it not only has excellent performance, but also has only two parameters (C, γ), C is a penalty factor for setting the width of the edge separating different classes, γ is the structure of the decision surface determined by the kernel parameter, and the two values are determined by the self-contained grid search method of the tool. The best parameters obtained by the implementation case are as follows: an RBF function is adopted, the penalty coefficient C is 3, the kernel function parameter gamma is 0.0323, the kernel function formula (1) is as follows,
K(xi，x)＝exp(-γ||xi-x||2)，γ>0 … … … equation (1).
The training process is as shown in figure 2, firstly, defining the wool knitting yarn sample with known color composition formula as a standard sample, and collecting the reflectivity data of the standard sample to perform data preprocessing, wherein the data preprocessing adopts the steps of firstly enlarging the reflectivity by 100 times and then combining the values of L, a and b on the basis to form xi＝[100*R(λ1),100*R(λ2)，…,100*R(λn),L*,a*,b*]T. After preprocessing, a characteristic vector first data set x is formediThen, dividing the first data set into a training set and a verification set, and determining a weight and a support vector of a support vector machine classification recognition model by using the training set; identifying a verification set by using the determined classification identification support vector machine model in the verification process, inputting the reflectivity data of the verification set in the verification process, outputting the reflectivity data as the color composition of the verification set, and comparing the coincidence degree of an output value and an actual value; if the verification result is high in accuracy, the training is effective, otherwise, new parameters are selected again for training until a satisfactory result is achieved. The average training prediction accuracy rate of 10 colors of the wool knitting yarn standard sample obtained by training is 96%.
Preprocessing the reflectivity of the sample: namely, the extraction process of the spectral characteristics can adopt methods such as data value amplification, 1-order derivative, 2-order derivative, principal component analysis, characteristic value increase and the like. The data preprocessing of the invention adopts a method of firstly expanding the reflectivity by 100 times and then combining L, a and b values to form xi＝[100*R(λ1),100*R(λ2)，…,100*R(λn),L*,a*,b*]TWherein R (λ)n) Represents lambdanThe value of n is determined by the measuring interval of the reflectivity wavelength, and when the interval is 10, the value of n is 31; l, a, b are CIE1976LAB values, calculated from R (λ) according to CIELAB formula, which is a mature technology in the prior art and not described herein.
Collected wool knitting yarn sample reflectivity data RSSimilarly, the reflectance was scaled up by 100 times and then used in conjunction with the values of L, a, b, which were obtained from a Datacolor850 spectrophotometer and were preprocessed to form a second dataset of feature vectors, with the scaled-up data obtained after Rs processing being shown in table 1.
Table 1 data and la a b values for 100 fold increase in reflectance
Figure BDA0001747845560000051
③ SVM color selection
When the wool knitting yarn has only two component colors, the structure of the constructed SVM color two-classification recognition model is shown in FIG. 3, when the color component of the wool knitting yarn is more than two colors, the SVM model is a multi-classification recognition model, namely, a certain sample input is defined to correspond to a plurality of component color outputs, as shown in formula (2),
f(xi)＝{y1,y2,…ym}，xi＝{x1,x2,…,xly ± 1, i 1, …, n … … … formula (2);
where n is the total number of samples, m is the total number of constituent colors, and l is the dimension of the input data, equation (2) indicates that x is inputiIs l-dimensional data, the output is { y1,y2,…ymAnd +1 represents contained and-1 represents not contained, which are represented by symbols formed of corresponding m colors.
Because there is no real parallel algorithm in strict meaning, the method converts multi-classification recognition into two-classification recognition, and comprises the following steps,
1) identificationy1When it contains y1Is one, the rest y2…ymIs one type, and whether y is present or not is determined1；
2) Identification y2When it contains y2As one group, y1，y3…ymRegarding as one type, determining whether there is y2；
……
m) identification of ymWhen it contains ymAs one group, y1…ym-1Regarding as one type, determining whether there is ym；
This converts the multiple output equations of equation (2) into a single output of equation (3):
f(xi)＝yj，xi＝{x1,x2,…,xl}，yj± 1, j ═ 1, …, m, … … … … formula (3)
Formula (3) shows the judgment of whether each output result corresponds to a single color, if m colors exist in the single color library, the input colored spun yarn x is obtained after m times of calculationiThe color composition of (a).
Taking HY-70291 sample as an example, determining that y is contained02,y06,y08,y10,y27,y31,y42,y47,y57,y58Which of the 10 colors is calculated by using the trained SVM expression (3) to obtain an output result after 10 times of calculation,
y { -1, -1, -1, +1, -1, +1, +1, -1, -1, +1}, which indicates that the spun-dyed yarn is formed of a color y10，y31，y42And y and58。
table 2 shows the composition range of the component color of the wool yarn obtained by recognizing the reflectance data of the incoming sample using the trained SVM model.
TABLE 2 component colors of wool knitting yarn samples identified by SVM model
Sample coming Component colors
HY-70291 B-058,B-010,B-042,B-031
5M0040 B-42,B-047,B-027
HY-65511 B-010,B-057
And step three, calculating the optimal combination proportion of all the component colors of the wool knitting yarns by using a tristimulus value matching algorithm or a full-spectrum color matching algorithm, namely obtaining the single-color reflectivity of all the component colors according to the component colors of the wool knitting yarns obtained in the step two, and calculating the reflectivity of the incoming sample measured in the step one by using the single-color reflectivity to obtain a wool splicing formula, wherein the single-color reflectivity is the existing known data and is not described in detail herein. Wherein the color matching algorithm adopts an optimized mixed color reflectivity model, the mixed color reflectivity model satisfies the following two relational expressions,
Figure BDA0001747845560000061
Figure BDA0001747845560000062
wherein: rs(lambda) represents the reflectance of the wool yarn at a wavelength of lambda, Ri(λ) represents the reflectance of the i component in a single color at a wavelength λ, xiAnd (3) representing the mass proportion of the component i in single color, wherein M is a model parameter and is adjusted according to the type and the yarn count of wool yarns.
The specific processes of the tristimulus value matching algorithm and the full spectrum color matching algorithm are as follows;
(1) tristimulus value color matching algorithm
Monochrome (R) according to the composition selected in step twoα、Rβ…Rγ) The calculation adopts a tristimulus value algorithm to calculate the composition proportion of each component color of the wool knitting yarn, and the tristimulus value color matching only needs to simultaneously satisfy X(s)＝X(m),Y(s)＝Y(m),Z(s)＝Z(m)；
Figure BDA0001747845560000071
Compared with a full spectrum algorithm, the algorithm has only three equations, has higher flexibility and can obtain more accurate calculation results than full spectrum color matching;
where k is a normalization constant and Δ λ is the wavelength interval. X, Y, Z is the tristimulus value, RλIs the spectral reflectance, S, of the sampleλIs the relative spectral power of the standard illuminant;
Figure BDA0001747845560000072
is the tristimulus value of a standard chromaticity observer;
to facilitate the implementation of computer algorithms, the following matrices and vectors are defined:
Figure BDA0001747845560000073
the tristimulus value equation is expressed as kTSR by matrix(s)＝kTSR(m)(ii) a Because of the monochromatic spectral matching algorithm selection, there is no particularly severe spectral heterogeneity, i.e., the reflectance of the matched sample at any one wavelength is not too different from the corresponding value of the standard sample, and can be expressed quite accurately as,
Figure BDA0001747845560000074
definition of
Figure BDA0001747845560000075
Wherein
Figure BDA0001747845560000076
Can obtain R(s)-R(m)＝D[F(s)-F(m)]Thus TSDF(s)＝TSDF(m)Is provided with F(m)＝F×x，
Wherein
Figure BDA0001747845560000077
(f) (R) represents the formula (3),
to obtain TSDF(s)TSDFx, solved x ═ TSDF-1TSDF(s)，
Screening to obtain the final formula
Figure BDA0001747845560000081
(2) Spectral color matching algorithm
Using least square method, the color is formed by several single colors (R)1,R2…RnArbitrary combination) to match the customer sample (R)s) The difference of the spectral curves reaches the minimum;
namely, setting to satisfy matching colors and incoming sample colors
Figure BDA0001747845560000082
In the formula
Figure BDA0001747845560000083
Represents the reflectance of the sample at the wavelength λ;
Figure BDA0001747845560000084
representing the reflectivity of the matching sample at the wavelength lambda. The wavelength range is 400-700nm of visible light, and the interval is 10 nm;
thus, it is possible to provide
Figure BDA0001747845560000085
Namely, it is
Figure BDA0001747845560000086
Wherein f (R)λRepresents formula (3).
Sample for setting
Figure BDA0001747845560000087
Matching sample
Figure BDA0001747845560000088
Then F(S)＝F(t)，
I.e. F(S)＝F×X，
Wherein
Figure BDA0001747845560000089
Since the above equations solve for 3 unknowns for 31 equations, the least squares method is used to solve the system of equations
Figure BDA00017478455600000810
Then X is ═ FT×F)-1×FT×F(S)Corresponding to a single color of (R)x…Rn) Selecting 2-3 combinations (R) of which the differences are smallα、Rβ…Rγ)。
In the formula: the superscript "T" represents the matrix transposition, and the superscript "-1" represents the matrix inversion; rxRepresents R1-RnOf any one of the monochromatic colors, Rα、Rβ，RγRepresents R1-RnIs a specific one of the monochrome colors.
And (3) selecting one of the tristimulus value algorithm and the full spectrum color matching algorithm for operation, wherein the forecast formula obtained in the third step is shown in table 3, as can be seen from table 3, the forecast formula is nearly consistent with the actual formula, the color matching forecast result of the incoming sample is shown in fig. 4-6, only one of the formulas HY-70291, 5M0040 and HY-65511 is used for the incoming sample, and the accurate formula can be locked, so that the forecast accuracy of the method is high.
TABLE 3 forecast formula and actual formula of wool knitting yarn sample
Figure BDA0001747845560000091
Taking HY-65511 as an example, the formula of the color matching agent is predicted by the existing computer color matching technology. The existing computer color matching technology does not adopt SVM identification to form a color model in the color identification process, and other implementation conditions are the same as those in the first step and the third step. The formula prediction result is shown in fig. 7, and it can be seen from fig. 7 that the formulas predicted by the formula without the component colors identified by the SVM model are very many, the number of the formulas is 18, and the accurate formula cannot be locked quickly.
The foregoing description of specific embodiments of the present invention has been presented. It is to be understood that the present invention is not limited to the specific embodiments described above, and that various changes and modifications may be made by one skilled in the art within the scope of the appended claims without departing from the spirit of the invention.

Claims (4)

1. An intelligent color-selecting wool-splicing method for wool knitting yarns is characterized by comprising the following steps;
step one, measuring the reflectivity of a wool knitting yarn sample; measuring the reflectivity of a wool knitting yarn sample by using a spectrophotometer, wherein the test condition requires mirror reflection, the complete data of the reflectivity with the wavelength of 400-700nm is included, and the wavelength measurement interval of the reflectivity is 0-10 nm;
step two, the support vector machine identifies the component colors; firstly, training a support vector machine model, defining a wool knitting yarn sample with a known color composition formula as a standard sample and carrying out reflectivity measurement, carrying out data preprocessing on collected reflectivity data of the standard sample to obtain a first characteristic vector data set, and dividing the first characteristic vector data set into a training set and a verification set for training the support vector machine model; carrying out data preprocessing on collected wool knitting yarn sample reflectivity data to form a second feature vector data set; identifying the second data set of the characteristic vector by using a trained support vector machine model to obtain the color composition of the component colors of the wool knitting yarn sample;
and step three, calculating the optimal combination proportion of all the component colors of the wool knitting yarns by using a tristimulus value matching algorithm or a full spectrum color matching algorithm, namely obtaining the single-color reflectivity of all the component colors according to the component colors of the wool knitting yarns obtained in the step two, and calculating the reflectivity of the wool knitting yarns obtained in the step one by using the single-color reflectivity to obtain a wool splicing formula.
2. The intelligent color-selecting and wool-splicing method for knitting wool yarns according to claim 1, wherein the model type of the support vector machine is C-SVC, an RBF kernel function is adopted, a penalty coefficient C is 3, a kernel function parameter gamma is 0.0323, and an RBF kernel function formula (1) is as follows;
K(xi，x)＝exp(-γ||xi-x||2)，γ>0 … … … equation (1).
3. The intelligent color-selecting and wool-splicing method for knitting wool yarns according to claim 1, wherein the support vector machine model adopts a method for converting multi-classification recognition into two-classification recognition, and comprises the following steps;
1) identification y1When it contains y1Is one, the rest y2…ymIs one type, and whether y is present or not is determined1；
2) Identification y2When it contains y2As one group, y1，y3…ymRegarding as one type, determining whether there is y2；
……
m) identification of ymWhen it contains ymAs one group, y1…ym-1Regarding as one type, determining whether there is ym；
Thus, a plurality of output equations of the multi-classification recognition formula (2) can be converted into a single output of the two-classification recognition formula (3);
f(xi)＝{y1,y2,…ym}，xi＝{x1,x2,…,xly ± 1, i 1, …, n … formula (2);
f(xi)＝yj，xi＝{x1,x2,…,xl}，yj± 1, j ═ 1, …, m, … … … …, equation (3);
where n is the total number of samples, m is the total number of constituent colors, and l is the dimension of the input data, equation (2) indicates that x is inputiIs l-dimensional data, the output is { y1,y2,…ym+1 represents containing, -1 represents not containing, formula (3) represents the judgment of whether each output result corresponds to a single color, if m colors exist in the monochrome database of the support vector machine model, m times of calculation are carried out, and the input colored spun yarn x is obtainediThe color composition of (a).
4. The intelligent color-selecting and wool-splicing method for knitting wool yarn according to claim 1, characterized in that the data preprocessing adopts the method of firstly expanding the reflectivity by 100 times and then combining the value of La b on the basis of the reflectivity to form x-R (lambda) 1001),100*R(λ2)，…,100*R(λn),L*,a*,b*]TWherein R (λ)n) Represents lambdanThe value of n is determined by the measuring interval of the reflectivity wavelength, and when the interval is 10, the value of n is 31; l, a, b are CIE1976LAB values calculated from R (λ) according to CIELAB formula.
CN201810852848.5A 2018-07-30 2018-07-30 Intelligent color-selecting wool-splicing method for wool knitting yarns Active CN109115701B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810852848.5A CN109115701B (en) 2018-07-30 2018-07-30 Intelligent color-selecting wool-splicing method for wool knitting yarns

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810852848.5A CN109115701B (en) 2018-07-30 2018-07-30 Intelligent color-selecting wool-splicing method for wool knitting yarns

Publications (2)

Publication Number Publication Date
CN109115701A CN109115701A (en) 2019-01-01
CN109115701B true CN109115701B (en) 2021-01-15

Family

ID=64863656

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810852848.5A Active CN109115701B (en) 2018-07-30 2018-07-30 Intelligent color-selecting wool-splicing method for wool knitting yarns

Country Status (1)

Country Link
CN (1) CN109115701B (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102750403B (en) * 2012-05-28 2016-06-15 嘉兴学院 The recipe determination of a kind of colour-spun yarns color matching and modification method
CN107103181A (en) * 2017-03-27 2017-08-29 东华大学 A kind of colour-spun yarns color matching method based on least square method supporting vector machine
CN107766603A (en) * 2017-04-24 2018-03-06 东华大学 A kind of colour-spun yarns computer is measured color method

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102750403B (en) * 2012-05-28 2016-06-15 嘉兴学院 The recipe determination of a kind of colour-spun yarns color matching and modification method
CN107103181A (en) * 2017-03-27 2017-08-29 东华大学 A kind of colour-spun yarns color matching method based on least square method supporting vector machine
CN107766603A (en) * 2017-04-24 2018-03-06 东华大学 A kind of colour-spun yarns computer is measured color method

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
Mathematical and empirical evaluation of accuracy of the Kubelka–Munk model for color match prediction of opaque and translucent surface coatings;Bandpay, M. Gorji,et al;《J. Coat. Technol. Res.》;20180327;第1-15页 *
基于最小二乘拟合的数值分析方法在织物染色配色中的应用研究;张永昌;《中国优秀硕士学位论文全文数据库 工程科技Ⅰ辑》;20120615;第11-13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31-33页 *
数值分析和支持向量机在织物染色配色中的对比研究;李含春;《中国优秀硕士学位论文全文数据库 信息科技辑》;20091015;第32-41页 *

Also Published As

Publication number Publication date
CN109115701A (en) 2019-01-01

Similar Documents

Publication Publication Date Title
CN107766603B (en) A kind of colour-spun yarns computer is measured color method
CN109142317A (en) A kind of Raman spectrum substance recognition methods based on Random Forest model
US8190551B2 (en) Classification of fabrics by near-infrared spectroscopy
CN102509243A (en) Method and system for evaluating quality in process of manufacturing cigarette
CN110717368A (en) Qualitative classification method for textiles
WO2016000088A1 (en) Hyperspectral waveband extraction method based on optimal index factor-correlation coefficient method
CN109582994B (en) Intelligent color selection and matching method for colored spun yarn
US20210109140A1 (en) Method for identifying parameters of 10 kv static load model based on similar daily load curves
CN107767079A (en) A kind of objective integrated evaluating method of tobacco style feature
EP3066435A1 (en) Texture analysis of a coated surface using pivot-normalization
CN101833593A (en) Computer color-matching method of color-blended wool fabrics based on full-spectrum color matching
Amirshahi et al. Adaptive non-negative bases for reconstruction of spectral data from colorimetric information
CN103983359A (en) Method for classifying observer color matching functions
CN109115701B (en) Intelligent color-selecting wool-splicing method for wool knitting yarns
CN108460380A (en) A kind of bamboo cane method for sorting colors and system based on domain color
CN110887797A (en) Color difference analysis method of colored textile fabric
CN109447097B (en) Fabric principal component detection method based on convolutional neural network
CN114253232B (en) Configuration method and system for processing stations in production line
Hsu Developing accurate industrial standards to facilitate production in apparel manufacturing based on anthropometric data
CN109387484A (en) A kind of ramee variety recognition methods of combination EO-1 hyperion and support vector cassification
Hashim et al. Classification of rubber tree leaf disease using spectrometer
Hadjianfar et al. Fuzzy classification of luster of textile fabrics
Guifang et al. Identification of varieties of natural textile fiber based on Vis/NIR spectroscopy technology
Dhanuja et al. Areca nut disease detection using image processing technology
CN106326905A (en) Multispectral vision imaging big-leaf crop leaf classification method and system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant