CN106570520A - Infrared spectroscopy tea quality identification method mixed with GK clustering - Google Patents
Infrared spectroscopy tea quality identification method mixed with GK clustering Download PDFInfo
- Publication number
- CN106570520A CN106570520A CN201610919763.5A CN201610919763A CN106570520A CN 106570520 A CN106570520 A CN 106570520A CN 201610919763 A CN201610919763 A CN 201610919763A CN 106570520 A CN106570520 A CN 106570520A
- Authority
- CN
- China
- Prior art keywords
- camelliae sinensis
- folium camelliae
- sample
- test sample
- fuzzy
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/30—Noise filtering
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Probability & Statistics with Applications (AREA)
- Multimedia (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention discloses an infrared spectroscopy tea quality identification method mixed with GK clustering in a tea detection technology. A linear discriminant analysis method is used to learn a compressed training sample to acquire a training sample with identification information and a test sample with identification information. Fuzzy C average value clustering is carried out on the training sample with identification information to acquire the initial fuzzy membership degree and an initial clustering center. The fuzzy scattering matrix and the fuzzy membership degree value are calculated, and then a typical value is calculated. The clustering center is calculated according to the typical value. The Euclidean distance from the average value of the training sample with identification information to the clustering center of the test sample is calculated. If the Euclidean distance from the clustering center to the average value of training tea is the minimum, the tea variety of the clustering center and the tea variety of the training sample are the same. The tea and the category of the test sample are determined according to the fuzzy membership degree value. According to the invention, the typical value is added into a function, which can significantly reduce the probability of noise data processing errors.
Description
Technical field
The present invention relates to Folium Camelliae sinensis detection technique, and in particular to based on GK clusters and the Folium Camelliae sinensis quality evaluation of infrared spectrum technology
Method.
Background technology
In Folium Camelliae sinensis detection, infrared spectrum detection is a kind of Fast nondestructive evaluation analytical technology, commonly uses mid-infrared light spectrometer
Detection Folium Camelliae sinensis, mid-infrared spectral wave-number range is in 4000cm-1～400cm-1Between, most inorganic compound and organic
The fundamental frequency of the chemical bond oscillations of compound is in this region.Functional group, the classification of compound and compound in different molecules
Stereochemical structure, its infrared absorption spectroscopy is not quite similar.Mid-infrared light spectral technology with its easily and fast, efficient, lossless, low cost
The features such as become detection food and medicine effective detection technology.
Common clustering method has two kinds：Hard clustering method and soft (fuzzy) clustering method, hard clustering method is applied to gather
The obvious situation in class border；For cluster boundary is not to adopt fuzzy clustering method then more particularly suitable when being apparent from, for example
Conventional fuzzy C-means clustering.GK clustering methods are (see document by a kind of clustering method of Gustafson and Kessel propositions
Gustafson D E,Kessel W C.Fuzzy clustering with fuzzy covariance matrix[C]//
Proceedings of the IEEE CDC,San Diego,1979:761～766), it is therefore an objective to by data set according to certain phase
Several subsets are divided into like criterion, bulk data is categorized as by clustering method for the cluster of many essential connections；For Fuzzy C
Mean cluster does not consider that the structure of data set this defect is improved, using distance of the fuzzy covariance matrix to cluster shape
Estimate with local auto-adaptive, the data set of various cluster shapes can be clustered.But traditional GK clustering methods contain in cluster
During noise data, cluster accuracy rate can be greatly affected because of noise data.And detecting Folium Camelliae sinensis mistake with mid-infrared light spectrometer
Cheng Zhonghui produces noise signal, and the middle infrared spectrum for being collected contains noise signal, due to GK clustering methods it is quick to noise
Sense, thus it is error-prone when the Folium Camelliae sinensis mid-infrared light modal data of Noise is processed.
The content of the invention
It is an object of the invention to solve existing GK clustering methods exist to noise data when Folium Camelliae sinensis infrared spectrum is clustered
Error-prone problem, proposes the infrared spectrum that a kind of mixing GK for being improved on the basis of GK clustering methods and optimizing is clustered
Folium Camelliae sinensis quality evaluation method, can well cluster the Folium Camelliae sinensis mid-infrared light modal data of Noise, improve to Folium Camelliae sinensis quality evaluation
Accuracy rate.
A kind of technical scheme of the infrared spectrum Folium Camelliae sinensis quality evaluation method employing of present invention mixing GK cluster is：Collection tea
Leaf sample infrared spectrum, by Folium Camelliae sinensis sample training sample and test sample are divided into, and then pretreatment Folium Camelliae sinensis sample infrared spectrum presses
Contracting ir data, the training sample comprising authentication information is obtained with Fisher face to the training sample study after compression
Sheet and test sample, to the operation fuzzy C-means clustering of the test sample comprising authentication information, obtain initial fuzzy membership uik,0
With initial cluster center v0,0, also sequentially comprise the following steps：
A, when first calculating the r time iteration the i-th class fuzzy collision matrix Sfi,rWith fuzzy membership angle value during the r time iteration
uik,r, then k-th test sample is under the jurisdiction of the representative value of the i-th class when calculating the r time iteration
For the r-1 time iteration when test sample xkTo cluster centre vi,r-1Apart from norm, d for test sample dimension, r is iteration time
Number, c is Folium Camelliae sinensis classification number, and 1≤i≤c, 1≤k≤n, n are test sample number；
B, according to representative value tik,rThe cluster centre of the i-th class when calculating the r time iteration
M is weighted index value；
C, the meansigma methodss for calculating the training sample comprising authentication information, then meansigma methodss are calculated respectively to test sample
Cluster centre νi,rEuclidean distance, if cluster centre to training Folium Camelliae sinensis meansigma methodss Euclidean distance minimum if judge the cluster
The local tea variety of the affiliated local tea variety in center and this training sample is same breed；
D, according to fuzzy membership angle value uik,rDiscriminating test sample xkAffiliated Folium Camelliae sinensis and classification, if uik,21＞ 0.5 then judges xk
Affiliated Folium Camelliae sinensis are high-quality tea.
It is infrared corresponding to good and bad Folium Camelliae sinensis because the infrared diffusing reflection spectrum of Folium Camelliae sinensis contains the component information inside Folium Camelliae sinensis
Diffusing reflection spectrum is different, and the present invention first compresses the ir data of good and bad Folium Camelliae sinensis with principal component analysiss, using linear discriminant
The authentication information of infrared spectrum is extracted in analysis, is finally differentiated between good and evil Folium Camelliae sinensis with mixing GK clustering methods, is compared traditional GK and is clustered
Method, Folium Camelliae sinensis sample needed for present invention mixing GK clustering methods is few, can effectively cluster the ir data of Folium Camelliae sinensis, clusters
Accuracy rate is high, clusters the advantages of speed is fast, and detection speed is fast, classification effectiveness is high, discrimination is high.Compare traditional GK cluster targets
Function, the present invention increases representative value in function, can be greatly lowered and process the probability malfunctioned during noise data, logarithm
Noise according in has good treatment effect.Substantially reduce impact of the noise to the Folium Camelliae sinensis result that differentiates between good and evil.
Description of the drawings
Fig. 1 is a kind of flow chart of the infrared spectrum Folium Camelliae sinensis quality evaluation method of mixing GK clusters of the present invention；
Fig. 2 is high-quality Folium Bambusae QINGCHAYE infrared spectrogram in embodiment；
Fig. 3 is Folium Bambusae QINGCHAYE infrared spectrogram inferior in embodiment；
Fig. 4 is pretreated Folium Camelliae sinensis infrared spectrogram in embodiment；
Fig. 5 is the training sample data figure that the infrared spectrum of Folium Camelliae sinensis in embodiment is obtained Jing after LDA Extraction and discrimination information；
Fig. 6 is the test sample datagram that the infrared spectrum of Folium Camelliae sinensis in embodiment is obtained Jing after LDA Extraction and discrimination information；
Fig. 7 and Fig. 8 are respectively the initial fuzzy memberships that two class Folium Camelliae sinensis samples operation fuzzy C-means clustering is produced in embodiment
Degree figure；
Fig. 9 and Figure 10 be respectively in embodiment the 2nd test sample of two class Folium Camelliae sinensis obtain after 21 iteration it is fuzzy
Degree of membership figure.
Specific embodiment
Referring to Fig. 1, good and bad Folium Camelliae sinensis sample is collected, with infrared spectrometer the infrared spectrum of Folium Camelliae sinensis sample is gathered, obtain Folium Camelliae sinensis
The infrared spectrum information that diffuses of sample, spectral information is stored in computer.Collection it is infrared diffuse spectrum information when, as far as possible
Keep the temperature and humidity of interior basically identical.The spectrum wave-number range of the infrared spectrum information that diffuses of collection is
4001.569cm-1～401.1211cm-1, the spectrum of each the Folium Camelliae sinensis sample for collecting is the data of 1868 dimensions.It is collected information
Afterwards, Folium Camelliae sinensis sample is divided into into training sample and test sample, number of training nrWith test sample number n, Folium Camelliae sinensis classification number c=2.
It is first infrared to Folium Camelliae sinensis sample with conventional multiplicative scatter correction method (MSC) and standard normal variable converter technique (SNV)
Spectroscopic data carries out pretreatment.Then pretreated Folium Camelliae sinensis sample ir data is entered using principal component analysiss (PCA)
Row dimensionality reduction, obtains the compressed data of Folium Camelliae sinensis sample infrared spectrum.Again to compressed data linear discriminant analysiss (LDA) Extraction and discrimination
Information, obtains the training sample comprising authentication information and test sample data.Finally the test sample comprising authentication information is used
Good and bad Folium Camelliae sinensis in mixing GK clustering methods with differential test sample.Mixing GK clustering methods are specific as follows：
Initialize installation is first carried out, the value of weighted index m is set and m ∈ (1 ,+∞) are met, iterationses initial value r=
0, maximum iteration time is rmax, iteration maximum error parameter ε；The Fuzzy C conventional to the operation of the test sample comprising authentication information
Mean cluster, fuzzy C-means clustering run abort after fuzzy membership and class center respectively as initial fuzzy membership
uik,0With initial cluster center vi,0。
According to initial fuzzy membership uik,0With initial cluster center vi,0, calculate r (r=1,2 ..., rmax) secondary iteration
When the i-th class fuzzy collision matrix Sfi,r：
In above formula, xkFor k-th Folium Camelliae sinensis examination of infrared spectrum sample comprising authentication information, vi,r-1For the r-1 time iteration
When i-th
The cluster centre of class, i=1,2, uik,r-1For the r-1 time iteration when test sample xkBelong to the fuzzy membership of the i-th class
Degree, Sfi,r
The fuzzy collision matrix of the i-th class when being the r time iteration.
And fuzzy membership angle value u when calculating the r time iterationik,r：
In above formulaFor the r-1 time iteration when test sample xkTo cluster centre vi,r-1Apart from norm,
For the r-1 time iteration when test sample xkTo cluster centre vj,r-1Apart from norm, j=1,2, j ≠ i, vj,r-1For the r-1 time repeatedly
For when jth class cluster centre.Wherein：
In above formula, Ai,rThe norm matrix at ith cluster center when being the r time iteration, d is test sample xkDimension.
Then according to apart from normWith fuzzy collision matrix Sfi,rK-th test sample x when calculating the r time iterationk
It is under the jurisdiction of the representative value t of the i-th classik,r：
According to representative value tik,rThe cluster centre ν of the i-th class when calculating the r time iterationi,r：
Judge iterationses or | | νi,r-νi,r-1| | value, when | | νi,r-νi,r-1| | ＜ ε or r>rmaxWhen, then calculate
Terminate, the fuzzy collision matrix S of the i-th class otherwise when the r time iteration is calculatedfi,rStart to recalculate, such iteration is until end
Only.
After iteration ends, the meansigma methodss of the training sample comprising authentication information are calculated respectively, and meansigma methodss are in cluster
Heart νi,rEuclidean distance, wherein, if cluster centre νi,rEuclidean distance to the meansigma methodss of training sample is minimum, then judge that this gathers
The local tea variety of the affiliated local tea variety in class center and the training sample is same breed.Further according to test sample xkFuzzy membership
Degree uik,rThe local tea variety belonging to test sample is differentiated, if uik,r＞ 0.5, then judge test sample xkFor high-quality tea, instead
It, then judge test sample xkFor Folium Camelliae sinensis inferior.
One embodiment of the present of invention presented below,
Embodiment
It is object to choose high-quality Trimeresurus stejnegeri and Trimeresurus stejnegeri inferior.FTIR-7600 type FTIR spectrum analysers are opened
Machine preheats 1 hour, and scanning times are 32, wave number 4001.569cm of spectral scan-1～401.1211cm-1, sweep spacing is
1.928cm-1, resolution is 4cm-1.High-quality Trimeresurus stejnegeri and both Folium Camelliae sinensis samples of Trimeresurus stejnegeri inferior are taken, two kinds of Folium Camelliae sinensis Jing are ground
Pulverizing is pure, then after being filtered with 40 mesh sieves, respectively take 0.5g respectively with potassium bromide 1:100 uniform mixing.Each sample takes mixing
Thing 1g carries out press mold, is then scanned 3 times with spectrogrph, takes the meansigma methodss of 3 times as sample spectrum data.Gathering ambient temperature is
25.5 DEG C, relative humidity is 49.2%, and voltage is 220V.Every kind of Folium Camelliae sinensis gather 32 samples, and 64 samples are obtained altogether.Each sample
This is one 1868 data tieed up.It is test set that every kind of sample chooses 18, then test sample number n is 36.Remaining 14 samples
For training set, then number of training nrFor 28.Test set is Folium Camelliae sinensis sample to be identified, and training set is known good and bad Folium Camelliae sinensis sample
This.The infrared spectrum of the Folium Camelliae sinensis sample of high-quality Trimeresurus stejnegeri and Trimeresurus stejnegeri inferior is respectively as shown in Figure 2 and Figure 3.
Pretreatment is carried out to Folium Camelliae sinensis sample infrared spectrum with multiplicative scatter correction (MSC) standard normal variable conversion (SNV),
Pretreated Folium Camelliae sinensis infrared spectrogram is as shown in Figure 4.
Pretreated Folium Camelliae sinensis sample ir data is compressed using principal component analysiss (PCA).Because first 9 it is main into
Accumulative credibility is divided to be more than 98%, so Folium Camelliae sinensis sample infrared spectrum is carried out into feature decomposition using principal component analytical method obtaining
Front 9 characteristic vectors v1,v2…v9With 9 eigenvalue λs1,λ2…λ9, each characteristic vector is the data of 1868 dimensions, eigenvalue
It is specific as follows：
λ1=13314, λ2=3473, λ3=1364.3,
λ4=704.07, λ5=464.64, λ6=283.73,
λ7=201.78, λ8=117.02, λ9=75.807.
Folium Camelliae sinensis sample infrared spectrum is projected to the data that 9 dimensions are obtained in 9 characteristic vectors, i.e., is compressed to 9 from 1868 dimensions
Dimension.
Again by the Folium Camelliae sinensis sample infrared spectrum compressed data after dimensionality reduction with after linear discriminant analysiss (LDA) Extraction and discrimination information
Obtain the training sample comprising authentication information and test sample data.Discriminant vectorses number is 1, and using linear discriminant analysiss 9 are extracted
The training sample comprising authentication information and test sample data are obtained after the authentication information of dimension data, respectively as shown in Figure 5, Figure 6,
The data of Fig. 5 and Fig. 6 are two groups of one-dimensional datas.
Good and bad Folium Camelliae sinensis during mixing GK clustering methods are used the test sample comprising authentication information with differential test sample, tool
Body is implemented according to the following steps：
A, initialization：The value of weighted index m is set, and meets m ∈ (1 ,+∞), d is the dimension of test sample, iteration is secondary
Number initial value r=0, maximum iteration time is rmax；Iteration maximum error parameter ε is set；Fuzzy C-mean algorithm is run to test sample
Cluster (FCM), FCM run abort after fuzzy membership and class center be subordinate to respectively as the initial fuzzy of mixing GK clustering methods
Category degree and initial cluster center；
Initialized numerical value is arranged：Classification number c=2 (i.e. two classifications), test sample number n=36, arrange weighted index m
=2, iterationses initial value r=0 and greatest iteration number rmax=100, error higher limit ε=0.00001, the dimension of test sample
Number d is 1.Two groups of one-dimensional test datas in Fig. 5 and Fig. 6 carry out fuzzy C-means clustering (FCM), FCM run abort after it is poly-
Class center is as initial cluster center, then initial cluster center：v1,0=0.80874, v2,0=-1.1268；FCM runs abort
Fuzzy membership afterwards is distinguished as shown in Figure 7 and Figure 8, i.e., initial fuzzy membership uik,0, i=1,2；, k is k-th test specimens
This, k=1,2 ..., 36.
B, calculate r (r=1,2 ..., rmax) secondary iteration when fuzzy collision matrix Sfi,r：
In above formula, xkFor k-th Folium Camelliae sinensis examination of infrared spectrum sample, vi,r-1For the r-1 time iteration when the i-th Lei Lei centers
(i=1,2), uik,r-1For the r-1 time iteration when test sample xkBelong to the fuzzy membership of the i-th class, Sfi,rIt is r
It is secondary to change
For when the i-th class fuzzy collision matrix.
Fuzzy membership angle value u when C, the r time iteration of calculatingik,r:
In above formulaFor the r-1 time iteration when test sample xkTo class center vi,r-1Apart from norm,For
Sample x during the r-1 time iterationkTo cluster centre vj,r-1Apart from norm (j=1,2), vj,r-1For the r-1 time iteration when jth class
Class center (j=1,2).
In above formula, Ai,rThe norm matrix at ith cluster center when being the r time iteration；D is the dimension of test sample.
Representative value t when D, the r time iteration of calculatingik,r：
tik,rFor k-th test sample xkIt is under the jurisdiction of the representative value of the i-th class.
The cluster centre ν of the i-th class when E, the r time iteration of calculatingi,r：
F, when | | νi,r-νi,r-1| | ＜ ε or r>rmaxWhen, then calculate and terminate, otherwise from calculating r (r=1,2 ..., rmax
Fuzzy collision matrix S during secondary iterationfi,rRestart to calculate.νi,rFor the r time iteration when the i-th class class central value, νi,r-1
For the r-1 time iteration when the i-th class cluster centre value.
As a result it is：R=21 time during iteration ends, two during the 21st iteration birds of the same feather flock together class central value νi,21Respectively：ν1,21
=0.68171, ν2,21=-1.1294.Fuzzy membership u during the 21st iteration endsik,21Respectively as shown in Figure 9 and Figure 10.
G, the training sample obtained after LDA Extraction and discrimination information are known two kinds Folium Camelliae sinensis sample (i.e. high-quality
Trimeresurus stejnegeri and Trimeresurus stejnegeri inferior), the meansigma methodss that the training sample of every kind of Folium Camelliae sinensis is calculated respectively are：High-quality Trimeresurus stejnegeri meansigma methodss are：Trimeresurus stejnegeri meansigma methodss inferior areWherein：z1kFor high-quality Trimeresurus stejnegeri
K-th training sample, z2kIt is k-th training sample of Trimeresurus stejnegeri inferior.Obtain with ν after 21 iterative calculation1,21With
ν2,21For the test sample data acquisition system of two classifications of cluster centre.
Calculate cluster centre ν1,65ArriveEuclidean distance be：
Calculate cluster centre ν1,65ArriveEuclidean distance be：BecauseValue be less thanValue, so, then judge with ν1,21Folium Camelliae sinensis for cluster centre belong to high-quality Trimeresurus stejnegeri.
Calculate cluster centre ν2,21ArriveEuclidean distance be：Calculate cluster centre ν2,21Arrive
Euclidean distance be：Due toValue be less thanValue, then judge with ν2,21For
The Folium Camelliae sinensis of cluster centre belong to Trimeresurus stejnegeri inferior.
Can differentiate Folium Camelliae sinensis classification belonging to 36 test samples further according to the fuzzy membership angle value of Fig. 9 and Figure 10：K-th
Test sample xkThe degree of membership obtained after 21 iteration is uik,21(i=1,2), if uik,21The then discriminating test samples of ＞ 0.5
xkAffiliated Folium Camelliae sinensis and cluster centre νi,21Affiliated local tea variety is identical.In the present embodiment, the 2nd test sample is through 21 iteration
The degree of membership for obtaining afterwards is u12,21=0.86518, u22,21=0.13482, due to u12,21＞ 0.5, then discriminating test sample x2Institute
Category Folium Camelliae sinensis and cluster centre ν1,21Affiliated local tea variety is identical.Judge test sample x2For high-quality Trimeresurus stejnegeri, because ν1,21Category
In high-quality Trimeresurus stejnegeri).The differentiation rate of accuracy reached 94.4% of the test sample in the present embodiment.
Claims (4)
1. a kind of infrared spectrum Folium Camelliae sinensis quality evaluation method of mixing GK clusters, gathers Folium Camelliae sinensis sample infrared spectrum, by Folium Camelliae sinensis sample
Originally it is divided into training sample and test sample, pretreatment Folium Camelliae sinensis sample infrared spectrum, then compressed ir modal data, with linearly sentencing
Other analytic process obtains the training sample comprising authentication information and test sample to the training sample study after compression, to comprising discriminating
The test sample operation fuzzy C-means clustering of information, obtains initial fuzzy membership uik,0With initial cluster center v0,0, it is special
Levy is also sequentially to comprise the following steps：
A, when first calculating the r time iteration the i-th class fuzzy collision matrix Sfi,rWith fuzzy membership angle value u during the r time iterationik,r,
K-th test sample is under the jurisdiction of the representative value of the i-th class when calculating the r time iteration again For
Test sample x during the r-1 time iterationkTo cluster centre vi,r-1Apart from norm, d for test sample dimension, r is iteration time
Number, c is Folium Camelliae sinensis classification number, and 1≤i≤c, 1≤k≤n, n are test sample number；
B, according to representative value tik,rThe cluster centre of the i-th class when calculating the r time iterationM is power
Weight exponential quantity；
C, the meansigma methodss for calculating the training sample comprising authentication information, then meansigma methodss are calculated respectively to the cluster of test sample
Center νi,rEuclidean distance, if cluster centre to training Folium Camelliae sinensis meansigma methodss Euclidean distance minimum if judge the cluster centre
The local tea variety of affiliated local tea variety and this training sample is same breed；
D, according to fuzzy membership angle value uik,rDiscriminating test sample xkAffiliated Folium Camelliae sinensis and classification, if uik,21＞ 0.5 then judges xkIt is affiliated
Folium Camelliae sinensis are high-quality tea.
2. a kind of infrared spectrum Folium Camelliae sinensis quality evaluation method that according to claim 1 mixing GK is clustered, is characterized in that：It is described
Fuzzy collision matrixxkFor k-th Folium Camelliae sinensis infrared spectrum comprising authentication information
Test sample, vi,r-1For the r-1 time iteration when the i-th class cluster centre, i=1,2, uik,r-1For the r-1 time iteration when test specimens
This xkBelong to the fuzzy membership of the i-th class.
3. a kind of infrared spectrum Folium Camelliae sinensis quality evaluation method that according to claim 2 mixing GK is clustered, is characterized in that：It is described
Fuzzy membership angle value For the r-1 time iteration when test sample xkTo cluster centre
vi,r-1Apart from norm,For the r-1 time iteration when test sample xkTo cluster centre vj,r-1Apart from norm, j=1,
2, j ≠ i, vj,r-1For the r-1 time iteration when jth class cluster centre, wherein： The norm matrix at ith cluster center during the r time iteration
4. a kind of infrared spectrum Folium Camelliae sinensis quality evaluation method that according to claim 1 mixing GK is clustered, is characterized in that：With many
First scatter correction method and standard normal variable converter technique carry out pretreatment to Folium Camelliae sinensis sample ir data, to pretreated
Folium Camelliae sinensis sample ir data carries out dimensionality reduction compressed data using principal component analysiss.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610919763.5A CN106570520A (en) | 2016-10-21 | 2016-10-21 | Infrared spectroscopy tea quality identification method mixed with GK clustering |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610919763.5A CN106570520A (en) | 2016-10-21 | 2016-10-21 | Infrared spectroscopy tea quality identification method mixed with GK clustering |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106570520A true CN106570520A (en) | 2017-04-19 |
Family
ID=58534137
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610919763.5A Pending CN106570520A (en) | 2016-10-21 | 2016-10-21 | Infrared spectroscopy tea quality identification method mixed with GK clustering |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106570520A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107194314A (en) * | 2017-04-20 | 2017-09-22 | 江苏大学 | The fuzzy 2DPCA and fuzzy 2DLDA of fusion face identification method |
| CN107860739A (en) * | 2017-11-27 | 2018-03-30 | 江苏大学 | A kind of tealeaves mid-infrared light profile classification method of fuzzy K mediations network clustering |
| CN107917896A (en) * | 2017-11-29 | 2018-04-17 | 宁夏医科大学 | Radix glycyrrhizae method for quick identification based near infrared spectrum and Clustering Analysis Technology |
| CN108764288A (en) * | 2018-04-25 | 2018-11-06 | 江苏大学 | A kind of GK differentiates the local tea variety sorting technique of cluster |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130077837A1 (en) * | 2010-03-29 | 2013-03-28 | Galderma Research & Development Snc | Fuzzy clustering algorithm and its application on carcinoma tissue |
| CN103048273A (en) * | 2012-11-09 | 2013-04-17 | 江苏大学 | Fruit near infrared spectrum sorting method based on fuzzy clustering |
| CN104123563A (en) * | 2014-07-17 | 2014-10-29 | 电子科技大学 | Cloude characteristic decomposition based polarimetric SAR (Synthetic Aperture Radar) image non-supervision classification method |
-
2016
- 2016-10-21 CN CN201610919763.5A patent/CN106570520A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130077837A1 (en) * | 2010-03-29 | 2013-03-28 | Galderma Research & Development Snc | Fuzzy clustering algorithm and its application on carcinoma tissue |
| CN103048273A (en) * | 2012-11-09 | 2013-04-17 | 江苏大学 | Fruit near infrared spectrum sorting method based on fuzzy clustering |
| CN104123563A (en) * | 2014-07-17 | 2014-10-29 | 电子科技大学 | Cloude characteristic decomposition based polarimetric SAR (Synthetic Aperture Radar) image non-supervision classification method |
Non-Patent Citations (1)
| Title |
|---|
| 武斌: "《苹果近红外光谱的聚类分析》", 《食品科技》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107194314A (en) * | 2017-04-20 | 2017-09-22 | 江苏大学 | The fuzzy 2DPCA and fuzzy 2DLDA of fusion face identification method |
| CN107194314B (en) * | 2017-04-20 | 2020-09-25 | 江苏大学 | Face recognition method fusing fuzzy 2DPCA and fuzzy 2DLDA |
| CN107860739A (en) * | 2017-11-27 | 2018-03-30 | 江苏大学 | A kind of tealeaves mid-infrared light profile classification method of fuzzy K mediations network clustering |
| CN107917896A (en) * | 2017-11-29 | 2018-04-17 | 宁夏医科大学 | Radix glycyrrhizae method for quick identification based near infrared spectrum and Clustering Analysis Technology |
| CN108764288A (en) * | 2018-04-25 | 2018-11-06 | 江苏大学 | A kind of GK differentiates the local tea variety sorting technique of cluster |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106408012A (en) | Tea infrared spectrum classification method of fuzzy discrimination clustering | |
| CN106570520A (en) | Infrared spectroscopy tea quality identification method mixed with GK clustering | |
| CN109030407B (en) | Apple variety classification method based on mixed fuzzy C-means clustering | |
| CN101819141B (en) | Maize variety identification method based on near infrared spectrum and information processing | |
| CN110068544B (en) | Substance identification network model training method and terahertz spectrum substance identification method | |
| CN110378374B (en) | Tea near infrared spectrum classification method for extracting fuzzy identification information | |
| CN109765194B (en) | Wolfberry producing area identification method based on hyperspectral imaging technology | |
| CN105181650B (en) | A method of quickly differentiating local tea variety using near-infrared spectrum technique | |
| CN109685098B (en) | Tea variety classification method for fuzzy inter-cluster separation and clustering | |
| CN107192686B (en) | Method for identifying possible fuzzy clustering tea varieties by fuzzy covariance matrix | |
| CN108072626A (en) | A kind of pitch brand identification method | |
| CN114399674A (en) | Hyperspectral image technology-based shellfish toxin nondestructive rapid detection method and system | |
| CN108764288A (en) | A kind of GK differentiates the local tea variety sorting technique of cluster | |
| CN104239895B (en) | SAR target identification method based on feature dimension reduction | |
| CN107886115A (en) | A kind of tealeaves mid-infrared light profile classification method of adaptively possible C mean clusters | |
| CN113310934A (en) | Method for quickly identifying milk cow milk mixed in camel milk and mixing proportion thereof | |
| CN108491894B (en) | Tea leaf classification method capable of fuzzy identification of C-means clustering | |
| CN110108661B (en) | Tea near infrared spectrum classification method based on fuzzy maximum entropy clustering | |
| CN109886296A (en) | A kind of authentication information extracts the local tea variety classification method of formula noise cluster | |
| Zeng et al. | Spectral Data Classification by One-Dimensional Convolutional Neural Networks | |
| Cai et al. | Deep metric learning framework combined with Gramian angular difference field image generation for Raman spectra classification based on a handheld Raman spectrometer | |
| Li et al. | SVM-based apple classification of soluble solids content by near-infrared spectroscopy | |
| CN112801172A (en) | Chinese cabbage pesticide residue qualitative analysis method based on fuzzy pattern recognition | |
| CN112801173A (en) | Lettuce near infrared spectrum classification method based on QR fuzzy discrimination analysis | |
| CN111595802A (en) | Construction method and application of Clinacanthus nutans seed source place classification model based on NIR (near infrared spectroscopy) |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170419 |
|
| RJ01 | Rejection of invention patent application after publication |