CN106228562A - Printed on line product chromaticity evaluation methodology based on probabilistic neural network algorithm - Google Patents
Printed on line product chromaticity evaluation methodology based on probabilistic neural network algorithm Download PDFInfo
- Publication number
- CN106228562A CN106228562A CN201610615069.4A CN201610615069A CN106228562A CN 106228562 A CN106228562 A CN 106228562A CN 201610615069 A CN201610615069 A CN 201610615069A CN 106228562 A CN106228562 A CN 106228562A
- Authority
- CN
- China
- Prior art keywords
- neural network
- probabilistic neural
- quality evaluation
- leaflet
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
- G06F18/24137—Distances to cluster centroïds
- G06F18/2414—Smoothing the distance, e.g. radial basis function networks [RBFN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30144—Printing quality
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Probability & Statistics with Applications (AREA)
- Spectrometry And Color Measurement (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of printed on line product chromaticity evaluation methodology based on probabilistic neural network algorithm.Including leaflet color quality evaluation index selection, the foundation of probabilistic neural network algorithm model, utilize supporting online detection instrument to obtain test print product chromaticity evaluation index the achievement data of acquisition inputs to algorithm model, such probabilistic neural network algorithm model just can export leaflet color quality evaluation result through computing.The probabilistic neural network modeling time of the method only needs 0.041313 second, can obtain new training sample the most in real time, it is also possible to revise quality evaluation index, and rapid build algorithm model as required；The fault-tolerance of the method is high, obtains more training samples in real time as long as online and input algorithm model and can optimize leaflet color quality evaluation result；The method grading is stable, and evaluation result is the highest with human eye overall merit effect goodness of fit.
Description
Technical field
The present invention relates to a kind of printed on line product chromaticity evaluation methodology, especially relate to a kind of based on probabilistic neural net
The printed on line product chromaticity evaluation methodology of network algorithm.
Background technology
At present, during print production, printing machine platform personnel need to inspect chromatograp at random by random samples, and according to
Its quality is evaluated by respective quality evaluation index, multiple has the quality evaluation index of the relation of influencing each other to print utilizing
When brush product chromaticity makes overall merit, it is necessary to board personnel make comprehensive descision according to the experience of oneself, and this results in
The repeatability of comprehensive descision result is poor, in order to solve this problem, occurs in that and utilizes special algorithm comprehensively to comment to simulate human eye
The evaluation methodology of valency leaflet color Quality Process, final ensure leaflet chromaticity can particular up to optimization criteria, and
The essence that human eye carries out overall merit to leaflet chromaticity is, first obtains each leaflet comprehensive evaluation index, then ties
The rating scale closing each evaluation index carries out the calculating of complexity by human brain, finally exports leaflet color quality evaluation result,
This process belongs to the process of pattern recognition, and artificial neural network is by the best approach of pattern recognition, is currently used for mould
The algorithm of anthropomorphic eye overall merit has BP artificial neural network algorithm and Fuzzy Artificial Neural Networks algorithm, but all has convergence speed
Degree is relatively slow, convergence result non-optimal shortcoming.
Summary of the invention
For the deficiency in background technology, it is an object of the invention to provide a kind of based on probabilistic neural network algorithm
Line leaflet color quality evaluating method, the method may be implemented in line accurate evaluation leaflet chromaticity, based on probabilistic neural
The fireballing characteristic of network convergence, is greatly enhanced leaflet color quality evaluation efficiency, and evaluation result is comprehensive with human eye simultaneously
Evaluation effect goodness of fit is the highest.
In order to solve above-mentioned technical problem, the technical solution used in the present invention is:
Printed on line product chromaticity evaluation methodology based on probabilistic neural network algorithm, the step of the method is as follows:
1) the choosing of leaflet color quality evaluation index；
2) foundation of probabilistic neural network algorithm model, including the determination of input layer, hidden layer neuron really
Calmly, the determination of factor sigma is slided in the summation determination of layer neuron, the determination peace of output layer neuron；
3) supporting online detection instrument is utilized to obtain leaflet color quality evaluation index to be evaluated the index that will obtain
Data input to probabilistic neural network algorithm model；
4) probabilistic neural network algorithm model exports leaflet color quality evaluation result, the most each tested print through computing
The opinion rating of brush product.
The selection of described leaflet color quality evaluation index, it is to select in numerous printing quality evaluation indexes and print
The most related index of brush product chromaticity, i.e. selects field density, site increase, print contrast, aberration, hue difference and gray scale
The chromaticity Testing index of totally 6 national regulations.
The determination of described input layer, it determines that the quantity of input layer, when probabilistic neural network is set up
6 the chromaticity Testing index collected are done statistical procedures, forms 18 dimensional feature vectors, and input layer
Quantity consistent with the dimension of input feature value, therefore, the quantity of input layer is 18.
The determination of described hidden layer neuron, it determines that the quantity of hidden layer neuron, and hidden layer is radially basic unit, hidden
Identical containing the quantity of layer neuron and the number of training sample, training sample i.e. comprises different color quality through what expert judging went out
The master sample of opinion rating, training sample is the most in theory, and the output result of probabilistic neural network algorithm is the most accurate, tests table
Bright 15 training samples can reach satisfied output result, and therefore, the number of hidden layer neuron is 15.
The determination of described summation layer neuron, it determines that the quantity of summation layer neuron, to leaflet chromaticity
When being evaluated, different leaflet color quality evaluation grades is the different classification modes of probabilistic neural network, will classification
Pattern is defined as 1,2,3,4 totally four kinds, the difference of the most corresponding leaflet color quality evaluation grade, in, good, excellent, summation layer god
Identical with selected classification mode kind number through the number of unit, therefore, the quantity of summation layer neuron is 4.
The determination of described output layer neuron, it determines that the output valve of output layer neuron, output layer neuron number
Being 1, it is made up of competition neurons, receives the output of summation layer, by having maximum a posteriori probability density in contrast summation layer
Neuron, it is output as 1, and remaining is output as 0, thus exports optimal pattern class.
The determination of described smoothing factor σ, it determines that the value of smoothing factor σ, smoothing factor σ are in probabilistic neural network
A most important part, plays vital impact to network performance, and σ value is the least the most only plays separation effect to single sample
Really, σ value is too high, and be not easily distinguishable the both of which being closer to, and distinguishes effect for details accurate, the most how to σ not
Value is a most important ring in probabilistic neural network；Conventional method is experience estimation method, and now σ value is directly obtained by experience
Go out, or be taken as in same group the half of distance average between characteristic vector.
The foundation of described probabilistic neural network algorithm model, is to use Matlab programming realization.
Described utilize supporting online detection instrument obtain leaflet color quality evaluation index to be evaluated, it be utilize supporting
Online detection instrument measure respectively the CMYK color lump of test on leaflet to be evaluated in colour code 100% reflection density value and
The reflection density value at non-printing position and the CMYK color of 75% on LAB value, the reflection density value of CMYK color lump of 50%, leaflet
The reflection density value of block, the reflection density value of CMYK color lump, the reflection density value of the CMYK color lump of 100% and the printing by 50%
On product, the reflection density value at non-printing position calculates the net bag method of the CMYK color lump of 50%, the CMYK color lump by 75%
Reflection density value and 100% the reflection density value of CMYK color lump calculate the print contrast value of leaflet, by the CMYK of 100%
The LAB value of color lump and the LAB value of standard CMYK color lump calculate value of chromatism and form and aspect difference, the CMYK color lump by 100% anti-
Penetrate density value and calculate gray value, i.e. obtain 6 leaflet color quality evaluation indexs.
The invention have the advantages that:
The probabilistic neural network modeling time of the present invention only needs 0.041313 second, can be real in printing actual production process
Time obtain new training sample, it is also possible to revise quality evaluation index, and rapid build algorithm model as required；The method
Fault-tolerance is high, and hidden layer have employed the nonlinear mapping function of radially base, as long as there being enough training samples, it determines just can in interface
Approaching Bayes Optimum classifying face progressively, experiment shows that 15 training samples can meet needs, therefore, in print production mistake
Journey obtains in real time more training samples and inputs algorithm model and can optimize the recognition effect of probabilistic neural network, it is thus achieved that be optimal
Printed on line product chromaticity evaluation result, accordingly adjust printing process state is respectively set, print quality can be effectively improved,
Improve Business Economic Benefit；The method can be prevented effectively from the subjective impact during conventional manual evaluates, and grading is stable, evaluation result with
Human eye overall merit effect goodness of fit is the highest.
Accompanying drawing explanation
Fig. 1 is probabilistic neural network model.
Fig. 2 is printed on line product chromaticity overhaul flow chart.
Detailed description of the invention
The invention will be further described with embodiment below in conjunction with the accompanying drawings.
As in figure 2 it is shown, the step of the inventive method is as follows:
1) the choosing of leaflet color quality evaluation index；
2) foundation of probabilistic neural network algorithm model, including the determination of input layer, hidden layer neuron really
Calmly, the determination of factor sigma is slided in the summation determination of layer neuron, the determination peace of output layer neuron；
3) supporting online detection instrument is utilized to obtain leaflet color quality evaluation index to be evaluated the index that will obtain
Data input to probabilistic neural network algorithm model；
4) probabilistic neural network algorithm model exports leaflet color quality evaluation result, the most each tested print through computing
The opinion rating of brush product.
The selection of described leaflet color quality evaluation index, it is to select in numerous printing quality evaluation indexes and print
The most related index of brush product chromaticity, i.e. selects field density, site increase, print contrast, aberration, hue difference and gray scale
The chromaticity Testing index of totally 6 national regulations.Described probabilistic neural network algorithm is that doctor D.F.Specht was in 1989
A kind of artificial neural network proposed, can be considered as a kind of RBF (Radial Basis Function, RBF) network,
But merge again estimation of density function and Bayesian decision theory, be made up of input layer, hidden layer, summation layer and output layer,
Its model is as shown in Figure 1
As it is shown in figure 1, the determination of described input layer, it determines that the quantity of input layer, probabilistic neural
When network is set up, 6 the chromaticity Testing index collected are done statistical procedures, form 18 dimensional feature vectors, and defeated
The quantity entering layer neuron is consistent with the dimension of input feature value, and therefore, the quantity of input layer is 18.
The determination of described hidden layer neuron, it determines that the quantity of hidden layer neuron, and hidden layer is radially basic unit, hidden
Identical containing the quantity of layer neuron and the number of training sample, training sample i.e. comprises different color quality through what expert judging went out
The master sample of opinion rating, training sample is the most in theory, and the output result of probabilistic neural network algorithm is the most accurate, tests table
Bright 15 training samples can reach satisfied output result, so the training sample number that this method finally selects is 15,
Therefore, the number of hidden layer neuron is 15.
The determination of described summation layer neuron, it determines that the quantity of summation layer neuron, to leaflet chromaticity
When being evaluated, different leaflet color quality evaluation grades is the different classification modes of probabilistic neural network, this method
Classification mode is defined as 1,2,3,4 totally four kinds, the difference of the most corresponding leaflet color quality evaluation grade, in, good, excellent, ask
Identical with selected classification mode kind number with the number of layer neuron, therefore, the quantity of summation layer neuron is 4.
The determination of described output layer neuron, it determines that the output valve of output layer neuron, output layer neuron number
Being 1, it is made up of competition neurons, receives the output of summation layer, by having maximum a posteriori probability density in contrast summation layer
Neuron, it is output as 1, and remaining is output as 0, thus exports optimal pattern class.
The determination of described smoothing factor σ, it determines that the value of smoothing factor σ, smoothing factor σ are in probabilistic neural network
A most important part, plays vital impact to network performance, and σ value is the least the most only plays separation effect to single sample
Really, σ value is too high, and be not easily distinguishable the both of which being closer to, and distinguishes effect for details accurate, the most how to σ not
Value is a most important ring in probabilistic neural network；Conventional method is experience estimation method, and now σ value is directly obtained by experience
Go out, or be taken as in same group the half of distance average between characteristic vector.
The foundation of described probabilistic neural network algorithm model, is to use Matlab programming realization.
Described utilize supporting online detection instrument obtain leaflet color quality evaluation index to be evaluated, it be utilize supporting
Online detection instrument measure respectively the CMYK color lump of test on leaflet to be evaluated in colour code 100% reflection density value and
The reflection density value at non-printing position and the CMYK color of 75% on LAB value, the reflection density value of CMYK color lump of 50%, leaflet
The reflection density value of block,
On the reflection density value of CMYK color lump, the reflection density value of the CMYK color lump of 100% and leaflet by 50% non-
The reflection density value at printing position calculates the net bag method of the CMYK color lump of 50%, and the reflection by the CMYK color lump of 75% is close
Angle value and 100% the reflection density value of CMYK color lump calculate the print contrast value of leaflet, the CMYK color lump by 100%
The LAB value of LAB value and standard CMYK color lump calculates value of chromatism and form and aspect difference, by the reflection density of the CMYK color lump of 100%
Value calculates gray value, i.e. obtains 6 leaflet color quality evaluation indexs.
Embodiment:
Printed on line product chromaticity evaluation methodology based on probabilistic neural network algorithm, comments including leaflet chromaticity
The selection of valency index；The foundation of probabilistic neural network algorithm model；Supporting online detection instrument is utilized to obtain leaflet color matter
The achievement data of acquisition is also inputed to probabilistic neural network algorithm model by amount evaluation index；So probabilistic neural network algorithm mould
Type just can export leaflet color quality evaluation result through computing.
When user prepares to realize the detection of printed on line quality with evaluation,
First, select leaflet color quality evaluation index, i.e. field density, site increase, print contrast, aberration, color
Difference and gray scale totally 6；
Second, set up probabilistic neural network algorithm model, including the determination of input layer, hidden layer neuron really
Fixed, the summation determination of layer neuron, the determination of output layer neuron, the determination of smoothing factor σ.Determine the number of input layer
6 the chromaticity evaluation indexes collected are done statistical procedures when probabilistic neural network is set up, are formed 6 Wei Te by amount
Levying vector, and the quantity of input layer is consistent with the dimension of input feature value, therefore, the quantity of input layer is
6.Determining the quantity of hidden layer neuron, hidden layer is radial direction basic unit, the quantity of this layer of neuron and the number of training sample
Identical, that training sample i.e. goes out through the expert judging master sample comprising different color quality evaluation grade, trains sample in theory
These are the most, and the output result of PNN algorithm is the most accurate, and experiment shows that 15 training samples i.e. can reach satisfied output result, institute
The training sample number finally selected with the present invention is 15, and the color aesthetics achievement data of training sample is shown in Table 1, therefore, hidden
Number containing layer neuron is 15.Determine the number of summation layer neuron, when leaflet chromaticity is evaluated, no
Same leaflet color quality evaluation grade is the different classification modes of probabilistic neural network, and classification mode is determined by the present invention
Be 1,2,3,4 totally four kinds, the difference of the most corresponding leaflet color quality evaluation grade, in, good, excellent, the number of summation layer neuron
Mesh is identical with selected classification mode kind number, and this, the number of summation layer neuron is 4.Determine output layer neuron
Output valve, output layer neuron number is 1, and it is made up of competition neurons, receives the output of summation layer, by contrast summation layer
In have the neuron of maximum a posteriori probability density, it is output as 1, and remaining is output as 0, thus can export optimal pattern class
Not.Determine that the value of smoothing factor σ, smoothing factor σ are most important parts in probabilistic neural network, network performance is played
Vital impact, σ value is the least the most only plays separating effect to single sample, and σ value is too high, is not easily distinguishable and more connects
Near both of which, distinguishes effect for details accurate not, to be the most how most important in probabilistic neural network to σ value
One ring, conventional method is experience estimation method, and now σ value can directly be drawn by experience, it is possible to be taken as in same group feature to
The half of distance average between amount.So far the foundation of probabilistic neural network algorithm is completed, available Matlab programming realization；
3rd, described utilize supporting online detection instrument obtain leaflet color quality evaluation index to be evaluated, it be profit
Measure respectively with supporting online detection instrument (the colour examining equipment that this example uses is Ai Seli 528) and test on leaflet to be evaluated
The reflection density value of the CMYK color lump of 100% in colour code and LAB value, the reflection density value of CMYK color lump of 50%, leaflet
The reflection density value at upper non-printing position and the reflection density value of the CMYK color lump of 75%, the reflection by the CMYK color lump of 50% is close
On angle value, the reflection density value of CMYK color lump of 100% and leaflet, the reflection density value at non-printing position calculates 50%
The net bag method of CMYK color lump, the reflection density value of the CMYK color lump by 75% and the reflection density of the CMYK color lump of 100%
Value calculates the print contrast value of leaflet, and the LAB value of the CMYK color lump by 100% and the LAB value of standard CMYK color lump calculate
Go out value of chromatism and form and aspect difference, the reflection density value of the CMYK color lump of 100% calculate gray value, i.e. obtain 6 leaflets
The color aesthetics achievement data of chromaticity evaluation index, i.e. sample to be tested, is shown in Table 2, and 6 achievement datas is inputed to probability
Neural network algorithm model, such probabilistic neural network algorithm model just can be through computing output leaflet color quality evaluation knot
Really, the simulation evaluation grade of the most each tested leaflet, output the results are shown in Table 3, as shown in Table 3 in the result simulated, has
13 results are excellent, sample number B01, B02, B04, B05, B06, B08, B09, B10, B12, B13, B14, B15, B16 respectively,
This type of leaflet effect is preferable；Result is good to have 5, and sample number is respectively B03, B11, B17, B18, B19；Result is for poor
Have 2, sample number is respectively B07, B20, this type of leaflet weak effect.Printing evaluation result is excellent above totally 18
, such printing use for printing qualified products, evaluation result is the printer that the specimen page of B07 and B20 of difference then uses
The waste product produced during inking, it shows as shortcomings such as being off color, contrast is unintelligible, it can thus be seen that probabilistic neural net
Network is capable of the accurate evaluation to print quality.
Above-mentioned detailed description of the invention is used for illustrating the present invention rather than limiting the invention, the present invention's
In spirit and scope of the claims, any modifications and changes that the present invention is made, both fall within the protection model of the present invention
Enclose.
Table 1 probabilistic neural network training sample color aesthetics achievement data and evaluation result thereof
The color aesthetics achievement data of table 2 probabilistic neural network sample to be tested
Table 3 probabilistic neural network sample to be tested simulation evaluation result
Claims (9)
1. printed on line product chromaticity evaluation methodology based on probabilistic neural network algorithm, it is characterised in that the step of the method
Rapid as follows:
1) the choosing of leaflet color quality evaluation index；
2) foundation of probabilistic neural network algorithm model, including the determination of input layer, the determination of hidden layer neuron,
The determination of factor sigma is slided in the summation determination of layer neuron, the determination peace of output layer neuron；
3) supporting online detection instrument is utilized to obtain leaflet color quality evaluation index to be evaluated the achievement data that will obtain
Input to probabilistic neural network algorithm model；
4) probabilistic neural network algorithm model exports leaflet color quality evaluation result, the most each tested printing through computing
The opinion rating of product.
Printed on line product chromaticity evaluation methodology based on probabilistic neural network algorithm the most according to claim 1, its
Be characterised by: the selection of described leaflet color quality evaluation index, it be in numerous printing quality evaluation indexes select with
The most related index of leaflet chromaticity, i.e. selects field density, site increase, print contrast, aberration, hue difference and ash
The chromaticity Testing index of degree totally 6 national regulations.
Online color Presswork Quality Evaluation method based on probabilistic neural network algorithm the most according to claim 1, its
Being characterised by: the determination of described input layer, it determines that the quantity of input layer, when probabilistic neural network is set up
6 the chromaticity Testing index collected are done statistical procedures, forms 18 dimensional feature vectors, and input layer
Quantity consistent with the dimension of input feature value, therefore, the quantity of input layer is 18.
Online color Presswork Quality Evaluation method based on probabilistic neural network algorithm the most according to claim 1, its
Being characterised by: the determination of described hidden layer neuron, it determines that the quantity of hidden layer neuron, and hidden layer is radial direction basic unit,
The quantity of hidden layer neuron is identical with the number of training sample, and training sample i.e. comprises different color matter through what expert judging went out
The master sample of amount opinion rating, training sample is the most in theory, and the output result of probabilistic neural network algorithm is the most accurate, experiment
Showing that 15 training samples can reach satisfied output result, therefore, the number of hidden layer neuron is 15.
Online color Presswork Quality Evaluation method based on probabilistic neural network algorithm the most according to claim 1, its
Being characterised by: the determination of described summation layer neuron, it determines that the quantity of summation layer neuron, to leaflet chromaticity
When being evaluated, different leaflet color quality evaluation grades is the different classification modes of probabilistic neural network, will classification
Pattern is defined as 1,2,3,4 totally four kinds, the difference of the most corresponding leaflet color quality evaluation grade, in, good, excellent, summation layer god
Identical with selected classification mode kind number through the number of unit, therefore, the quantity of summation layer neuron is 4.
Online color Presswork Quality Evaluation method based on probabilistic neural network algorithm the most according to claim 1, its
Being characterised by: the determination of described output layer neuron, it determines that the output valve of output layer neuron, output layer neuron number
Being 1, it is made up of competition neurons, receives the output of summation layer, by having maximum a posteriori probability density in contrast summation layer
Neuron, it is output as 1, and remaining is output as 0, thus exports optimal pattern class.
Online color Presswork Quality Evaluation method based on probabilistic neural network algorithm the most according to claim 1, its
Being characterised by: the determination of described smoothing factor σ, it determines that the value of smoothing factor σ, smoothing factor σ are probabilistic neural networks
In a most important part, network performance is played vital impact, σ value is the least the most only plays separation to single sample
Effect, σ value is too high, and be not easily distinguishable the both of which being closer to, and the most how distinguishes effect for details accurate not
It it is a most important ring in probabilistic neural network to σ value；Conventional method is experience estimation method, and now σ value directly passes through warp
Test and draw, or be taken as in same group the half of distance average between characteristic vector.
Online color Presswork Quality Evaluation method based on probabilistic neural network algorithm the most according to claim 1, its
It is characterised by: the foundation of described probabilistic neural network algorithm model, is to use Matlab programming realization.
Online color Presswork Quality Evaluation method based on probabilistic neural network algorithm the most according to claim 1, its
Be characterised by: described utilize supporting online detection instrument obtain leaflet color quality evaluation index to be evaluated, it be utilize join
Set online detection instrument measure respectively the CMYK color lump of test on leaflet to be evaluated in colour code 100% reflection density value and
The reflection density value at non-printing position and the CMYK color lump of 75% on LAB value, the reflection density value of CMYK color lump of 50%, leaflet
Reflection density value, on the reflection density value of CMYK color lump, the reflection density value of the CMYK color lump of 100% and the leaflet by 50%
The reflection density value at non-printing position calculates the net bag method of the CMYK color lump of 50%, and the reflection by the CMYK color lump of 75% is close
Angle value and 100% the reflection density value of CMYK color lump calculate the print contrast value of leaflet, the CMYK color lump by 100%
The LAB value of LAB value and standard CMYK color lump calculates value of chromatism and form and aspect difference, by the reflection density value of the CMYK color lump of 100%
Calculate gray value, i.e. obtain 6 leaflet color quality evaluation indexs.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610615069.4A CN106228562A (en) | 2016-08-01 | 2016-08-01 | Printed on line product chromaticity evaluation methodology based on probabilistic neural network algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610615069.4A CN106228562A (en) | 2016-08-01 | 2016-08-01 | Printed on line product chromaticity evaluation methodology based on probabilistic neural network algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106228562A true CN106228562A (en) | 2016-12-14 |
Family
ID=57535768
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610615069.4A Pending CN106228562A (en) | 2016-08-01 | 2016-08-01 | Printed on line product chromaticity evaluation methodology based on probabilistic neural network algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106228562A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109255389A (en) * | 2018-09-28 | 2019-01-22 | 中国科学院长春光学精密机械与物理研究所 | A kind of equipment evaluation method, device, equipment and readable storage medium storing program for executing |
| CN109992928A (en) * | 2019-04-30 | 2019-07-09 | 南京林业大学 | A method of establishing china picture-character paper output quality prediction model |
| CN110598973A (en) * | 2019-07-30 | 2019-12-20 | 北京信息科技大学 | IAP-based risk evaluation method for authentication process of green furniture product |
| CN112630375A (en) * | 2020-07-31 | 2021-04-09 | 南通贝得彩色印刷有限公司 | Environment-friendly detection method and device for color printing device |
| CN112767306A (en) * | 2020-12-24 | 2021-05-07 | 凌云光技术股份有限公司 | Printed matter quality detection and receiving method and system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2000070310A1 (en) * | 1999-05-12 | 2000-11-23 | Kyushu Kyohan Co., Ltd. | Signal identification device using genetic algorithm and on-line identification system |
| CN103076334A (en) * | 2013-01-25 | 2013-05-01 | 上海理工大学 | Method for quantitatively evaluating perceived quality of digital printed lines and texts |
| CN104408717A (en) * | 2014-11-24 | 2015-03-11 | 北京航空航天大学 | Pathological image color quality comprehensive evaluation method based on color separation |
-
2016
- 2016-08-01 CN CN201610615069.4A patent/CN106228562A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2000070310A1 (en) * | 1999-05-12 | 2000-11-23 | Kyushu Kyohan Co., Ltd. | Signal identification device using genetic algorithm and on-line identification system |
| CN103076334A (en) * | 2013-01-25 | 2013-05-01 | 上海理工大学 | Method for quantitatively evaluating perceived quality of digital printed lines and texts |
| CN104408717A (en) * | 2014-11-24 | 2015-03-11 | 北京航空航天大学 | Pathological image color quality comprehensive evaluation method based on color separation |
Non-Patent Citations (2)
| Title |
|---|
| 初红艳等: "基于BP神经网络的印刷色彩质量综合评价", 《包装工程》 * |
| 陈雯柏等: "《人工神经网络原理与实践》", 31 January 2016 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109255389A (en) * | 2018-09-28 | 2019-01-22 | 中国科学院长春光学精密机械与物理研究所 | A kind of equipment evaluation method, device, equipment and readable storage medium storing program for executing |
| CN109255389B (en) * | 2018-09-28 | 2022-03-25 | 中国科学院长春光学精密机械与物理研究所 | Equipment evaluation method, device, equipment and readable storage medium |
| CN109992928A (en) * | 2019-04-30 | 2019-07-09 | 南京林业大学 | A method of establishing china picture-character paper output quality prediction model |
| CN110598973A (en) * | 2019-07-30 | 2019-12-20 | 北京信息科技大学 | IAP-based risk evaluation method for authentication process of green furniture product |
| CN112630375A (en) * | 2020-07-31 | 2021-04-09 | 南通贝得彩色印刷有限公司 | Environment-friendly detection method and device for color printing device |
| CN112630375B (en) * | 2020-07-31 | 2022-03-29 | 南通贝得彩色印刷有限公司 | Environment-friendly detection method and device for color printing device |
| CN112767306A (en) * | 2020-12-24 | 2021-05-07 | 凌云光技术股份有限公司 | Printed matter quality detection and receiving method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106228562A (en) | Printed on line product chromaticity evaluation methodology based on probabilistic neural network algorithm | |
| Tian et al. | An integrated approach for failure mode and effects analysis based on fuzzy best-worst, relative entropy, and VIKOR methods | |
| Geirhos et al. | ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness | |
| Ghazinoory et al. | Fuzzy SWOT analysis | |
| CN105117602B (en) | A kind of metering device running status method for early warning | |
| CN110110887A (en) | To the prediction technique of low-voltage platform area line loss per unit | |
| CN109389171B (en) | Medical image classification method based on multi-granularity convolution noise reduction automatic encoder technology | |
| Zhou et al. | Utilizing dictionary learning and machine learning for blind quality assessment of 3-D images | |
| CN107704925A (en) | The visual analysis system and method for deep neural network training process | |
| CN109948522A (en) | A kind of X-ray hand osseous maturation degree interpretation method based on deep neural network | |
| CN105427309A (en) | Multiscale hierarchical processing method for extracting object-oriented high-spatial resolution remote sensing information | |
| CN109684922A (en) | A kind of recognition methods based on the multi-model of convolutional neural networks to finished product dish | |
| Kervadec et al. | How transferable are reasoning patterns in vqa? | |
| CN108764298A (en) | Electric power image-context based on single classifier influences recognition methods | |
| CN108280490A (en) | A kind of fine granularity model recognizing method based on convolutional neural networks | |
| CN110189305A (en) | A kind of multitask tongue picture automatic analysis method | |
| CN108345898A (en) | A kind of novel line insulator Condition assessment of insulation method | |
| CN109934221A (en) | Power equipment based on attention mechanism automatically analyzes identification monitoring method and system | |
| CN109523514A (en) | To the batch imaging quality assessment method of Inverse Synthetic Aperture Radar ISAR | |
| CN101539517A (en) | Method for evaluating high-gloss coating surfaces by fuzzy logic model | |
| CN109858832A (en) | A kind of more attribute green Index grading Coordination Evaluation methods | |
| CN108647485A (en) | Catalyst carbon deposition measurement method, system, medium and equipment in fluid catalytic cracking | |
| Akay et al. | A neuro-fuzzy based approach to affective design | |
| CN111738870A (en) | Method and platform for identifying insurance risk of engineering performance guarantee based on characteristic engineering | |
| Kuo et al. | Computerized color separation system for printed fabrics by using backward-propagation neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161214 |