CN109685099A - A kind of apple variety discriminating conduct of the preferred fuzzy clustering of spectral band - Google Patents
A kind of apple variety discriminating conduct of the preferred fuzzy clustering of spectral band Download PDFInfo
- Publication number
- CN109685099A CN109685099A CN201811338607.5A CN201811338607A CN109685099A CN 109685099 A CN109685099 A CN 109685099A CN 201811338607 A CN201811338607 A CN 201811338607A CN 109685099 A CN109685099 A CN 109685099A
- Authority
- CN
- China
- Prior art keywords
- apple
- sample
- near infrared
- spectral
- 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.)
- Granted
Links
- 230000003595 spectral effect Effects 0.000 title claims abstract description 67
- 244000141359 Malus pumila Species 0.000 title 1
- 241000220225 Malus Species 0.000 claims abstract description 90
- 238000002329 infrared spectrum Methods 0.000 claims abstract description 39
- 238000012360 testing method Methods 0.000 claims abstract description 23
- 238000000034 method Methods 0.000 claims abstract description 20
- 238000000513 principal component analysis Methods 0.000 claims abstract description 12
- 238000004458 analytical method Methods 0.000 claims abstract description 11
- 238000010239 partial least squares discriminant analysis Methods 0.000 claims abstract description 8
- 230000008569 process Effects 0.000 claims abstract description 6
- 239000000284 extract Substances 0.000 claims abstract description 5
- 239000011159 matrix material Substances 0.000 claims description 22
- 238000012549 training Methods 0.000 claims description 22
- 238000004364 calculation method Methods 0.000 claims description 15
- 238000001228 spectrum Methods 0.000 claims description 15
- 239000013598 vector Substances 0.000 claims description 12
- 238000004611 spectroscopical analysis Methods 0.000 claims description 8
- 238000010238 partial least squares regression Methods 0.000 claims description 7
- 230000008859 change Effects 0.000 claims description 4
- 238000000354 decomposition reaction Methods 0.000 claims description 4
- 241001269238 Data Species 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims description 2
- 230000003287 optical effect Effects 0.000 claims description 2
- 239000002245 particle Substances 0.000 claims description 2
- 239000007787 solid Substances 0.000 claims description 2
- 238000012545 processing Methods 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 238000009659 non-destructive testing Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 241000196324 Embryophyta Species 0.000 description 1
- 238000004566 IR spectroscopy Methods 0.000 description 1
- 235000010724 Wisteria floribunda Nutrition 0.000 description 1
- 235000021016 apples Nutrition 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 235000013305 food Nutrition 0.000 description 1
- 238000003306 harvesting Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
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
- 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
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/3563—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light for analysing solids; Preparation of samples therefor
-
- 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
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/359—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light using near infrared light
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/68—Food, e.g. fruit or vegetables
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Spectroscopy & Molecular Physics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Immunology (AREA)
- General Health & Medical Sciences (AREA)
- Biochemistry (AREA)
- Analytical Chemistry (AREA)
- Chemical & Material Sciences (AREA)
- Health & Medical Sciences (AREA)
- Pathology (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Probability & Statistics with Applications (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention discloses a kind of apple variety discriminating conducts of the preferred fuzzy clustering of spectral band, include the following steps: S1, the Fourier transform near infrared of different cultivars apple sample acquires: for the apple sample of different cultivars, apple sample is detected with Fourier transform near infrared instrument, obtain apple sample Fourier transform near infrared data and is stored data in computer.S2 pre-processes the apple sample near infrared spectrum of S1 with standard normal variable variation (SNV).S3, to interval partial least squares discriminant analysis (BIPLSDA), to carry out wave band to the near infrared spectrum of S2 preferred after.S4 is carried out dimension-reduction treatment to apple near infrared spectrum and authentication information extracts: compressed using principal component analysis (PCA) to the apple near infrared spectrum data in S3;Then the authentication information of data is extracted using linear discriminant analysis (LDA).S5 distinguishes apple variety to a kind of improved fuzzy C-means clustering method of test sample in S4 including authentication information.
Description
Technical field
The present invention relates to a kind of apple variety discriminating conducts, and in particular to a kind of apple of the preferred fuzzy clustering of spectral band
Kind discriminating conduct.
Background technique
Nutriment rich in apple, it is one of fruit that people often eat.The classification of apple is apple
The important link of commercial treatment after harvesting.Manual sort is not only time-consuming but also is influenced by subjective factor;And to the tradition of apple
Physico-chemical analysis needs to detect pol, acidity, and experiment is complicated and time-consuming.Therefore a kind of simple, quick, nondestructive apple is studied
Kind discriminating conduct is very important.
Near-infrared spectrum technique is inhaled according to itself vibration of the organo-functional group (O-H, C-H, N-H, S-H) of sample interior
The energy of near infrared spectrum respective wavelength is received, to generate the performance of energy jump in spectrum.Near-infrared spectrum analysis skill
Art is a kind of efficient, quick modern analytical technique.With lossless, detection speed is fast, and high accuracy for examination is widely used in
Agricultural product/food field of non destructive testing.It can get the full spectroscopic data of detectable substance, full spectroscopic data when carrying out spectral detection
Amount is big, while also containing some redundant datas, and redundant data affects the accuracy rate of data processing, therefore needs to carry out spectrum wave
Duan Youxuan.
Fuzzy C-Means Clustering (FCM) is built upon the clustering algorithm on the basis of square error minimum criteria, passes through possibility
Property constraint condition, FCM make the sum of the degree of membership of data point in all classes be 1.Possibility Constraint condition avoids all be subordinate to
The trivial solution that degree is 0.But to result in FCM sensitive to noise or outlier for a possibility that FCM constraint condition.FCM is noisy in cluster
Cluster result is influenced by noise data when sound data.
Summary of the invention
The present invention be directed to existing Fuzzy C-Means Clusterings in the apple Fourier transform near infrared number for clustering Noise
According to when cluster result influenced by noise data the shortcomings that, and when the processing of full spectroscopic data, redundant data affects data processing
Accuracy rate the shortcomings that and the improvement project that proposes.Spectrum wave is carried out to interval partial least squares discriminant analysis after present invention use
Duan Youxuan solves the problems, such as that redundant data affects the accuracy rate of data processing;Using a kind of improved fuzzy C-means clustering
Method carries out the fuzzy cluster analysis of apple Fourier transform near infrared data to distinguish apple variety, solves Fuzzy C-means
Cluster the problem of cluster result is influenced by noise data when clustering the near infrared spectrum data of Noise.The present invention has detection
Speed is fast, non-destructive testing, distinguishes apple variety high accuracy for examination.
A kind of apple variety discriminating conduct of the preferred fuzzy clustering of spectral band, specifically includes the following steps:
The Fourier transform near infrared acquisition of different cultivars apple sample: S1 for the apple sample of different cultivars, uses Fu
In leaf near infrared spectrometer apple sample is detected, obtain apple sample Fourier transform near infrared data simultaneously data are deposited
Storage is in computer.
S2 pre-processes the apple sample near infrared spectrum of S1 with standard normal variable variation (SNV).
S3, to interval partial least squares discriminant analysis (BIPLSDA), to carry out wave band to the near infrared spectrum of S2 preferred after.
S4 carries out dimension-reduction treatment to apple near infrared spectrum and authentication information extracts: using principal component analysis (PCA) to S3
In apple near infrared spectrum data compressed;Then the authentication information of data is extracted using linear discriminant analysis (LDA).
S5 distinguishes apple to a kind of improved fuzzy C-means clustering method of test sample in S4 including authentication information
Kind.
Initialization: weighted index m, classification number c is arranged, wherein m > 1, c > 1 in S5.1;Be arranged cycle count r initial value and
Maximum number of iterations rmax;Iteration worst error parameter ε is set;The class central value v that operation fuzzy C-means clustering obtainsi,FCMAs
Initial class central value νi (0);Calculating parameter ηi:
In above formula, m (m > 1) is weighted index, uik,FCMWhat is obtained after terminating for fuzzy C-means clustering iterative calculation is fuzzy
It is subordinate to angle value, vi,FCMI-th (i=1,2,3 ..., c) a class central value obtained after being terminated for fuzzy C-means clustering iterative calculation,
xkFor k-th of test sample.
S5.2 calculates r (r=1,2 ..., rmax) secondary iteration when fuzzy membership angle value uik (r):
uikIt is sample xkIt is under the jurisdiction of the fuzzy membership angle value of classification i;viBe i-th (i=1,2,
3 ..., c) class class central value, νi (r-1)It is the class center v of the r-1 times iterative calculationiValue;vjIt is jth
The class central value of (j=1,2,3 ..., c) class, νj (r-1)It is the class center v of the r-1 times iterative calculationjValue, n be test sample number;
S5.3, calculate the r times iteration when the i-th class class central value νi (r)
νi (r)It is the class center v of the r times iterative calculationiValue, class center matrix V is formed by c class central value(r)=
[ν1 (r),ν2 (r),…,νc (r)];
S5.4, cycle count increase, i.e. r=r+1;
If meeting condition: | | V(r)-V(r-1)| | < ε or r > rmaxTermination is then calculated, otherwise continues S5.2, after iteration ends
Fuzzy membership angle value can be obtained and apple variety is distinguished according to fuzzy membership angle value.
Beneficial effects of the present invention:
1, a kind of apple variety discriminating conduct of the preferred fuzzy clustering of spectral band of the invention uses a kind of improvement Fuzzy C
Means clustering method clusters apple near infrared spectrum data, thus has cluster accuracy rate high, and cluster speed is fast;Using near-infrared
Spectral technique and have the advantages that non-destructive testing.
2, a modification of the present invention fuzzy C-means clustering method is near infrared spectrum data of the cluster comprising noise data
Aspect is better than fuzzy C-means clustering (FCM), can fast implement the quick and accurate discrimination of apple variety.
3, it is excellent to interval partial least squares discriminant analysis (BIPLSDA) progress near infrared spectrum wave band after the present invention uses
Choosing can remove redundancy spectrum, improve cluster accuracy rate.
Detailed description of the invention
Fig. 1 is flow chart of the invention;
Fig. 2 is the Fourier transform near infrared figure of apple sample;
Fig. 3 is the pretreated Fourier transform near infrared figure of SNV;
The sub- SPECTRAL REGION that Fig. 4 is selected after being to interval partial least squares discriminant analysis;
Fig. 5 is the three-dimensional test sample data figure obtained after LDA authentication information extracts;
Fig. 6 is the fuzzy membership figure of fuzzy C-means clustering;
Fig. 7 is a kind of fuzzy membership figure for improving fuzzy C-means clustering.
Specific embodiment
Present invention will be further explained below with reference to the attached drawings and examples.
Because for the apple of different cultivars, their Fourier transform near infrared is had differences, reality of the invention
It is as shown in Figure 1 to apply process.The present embodiment is illustrated with the apple sample of four kinds of kinds:
As shown in Figure 1, a kind of apple variety discriminating conduct of the preferred fuzzy clustering of spectral band, comprising the following steps:
The Fourier transform near infrared acquisition of different cultivars apple sample: S1 for the apple sample of different cultivars, uses Fu
In leaf near infrared spectrometer apple sample is detected, obtain apple sample Fourier transform near infrared data simultaneously data are deposited
Storage is in computer.
Acquire the near infrared spectrum of apple sample;Take Fuji apple, Hua Niu, yellow any of several broadleaf plants, four kinds of apple samples of Ghana, every kind of apple
50, sample.Apple sample temperature be 20~25 DEG C experiment indoor storage 12 hours, Antaris II near-infrared spectrum analysis
Instrument booting 1 hour of preheating.Using reflection integrating sphere type collection apple near infrared spectrum, near-infrared spectrometers scanning is every
A sample 32 times is to obtain the diffusing reflection spectrum mean value of sample.The wave number of spectral scan is
10000~4000cm-1, sweep spacing 3.856cm-1, the spectrum for collecting each sample is the number of 1557 dimensions
According to.To reduce error, each apple sample samples 3 times along equator track, takes its average value as final experimental data.Apple
The Fourier transform near infrared figure of sample is as shown in Figure 2.
S2 pre-processes the apple sample near infrared spectrum of S1 with standard normal variable variation (SNV).
Solid particle size is mainly eliminated in standard normal variable variation, and surface scattering and change in optical path length are close to diffusing reflection
The influence of infrared spectroscopy.The apple sample near infrared spectrum data of S1 is subtracted after the average value of spectroscopic data again divided by spectrum number
According to standard deviation.Spectrogram of the apple sample near infrared spectrum of S1 after standard normal variable change process is as shown in Figure 3.
S3, to interval partial least squares discriminant analysis (BIPLSDA), to carry out wave band to the near infrared spectrum of S2 preferred after.
200 samples are divided into two parts of training sample and test sample, taking training set sample number is 80 (i.e. every classes
20), test set sample number 120 (i.e. every class 30).
Establish training sample classification information matrix Ytrain
If Ytrain(i, j)=1 indicates that i-th of sample belongs to jth classification, conversely, if Ytrain(i, j)=0 indicates i-th
A sample is not belonging to jth classification.N1 is number of training, and N1=80, C are number of degrees, C=4.
The near infrared spectrum region of apple is fifty-fifty divided into L wide sub- SPECTRAL REGIONs, then removal 1 every time
Sub- SPECTRAL REGION carries out training sample spectral value and training sample classification information matrix in remaining SPECTRAL REGION partially minimum
Two multiply recurrence gained validation-cross root-mean-square error (RMSECV) minimum, until last 1 sub- SPECTRAL REGION.It is missed according to root mean square
Poor minimum principle chooses several sub- SPECTRAL REGIONs.The results are shown in Table 1 for sub- SPECTRAL REGION selection, and SPECTRAL REGION is averagely divided
For L=20 wide sub- SPECTRAL REGIONs, 1 sub- SPECTRAL REGION, the training in remaining 19 sub- SPECTRAL REGIONs are removed every time
Sample light spectrum and training sample classification information matrix carry out Partial Least Squares Regression, resulting when removing sub- SPECTRAL REGION 11
Validation-cross root-mean-square error (RMSECV) minimum 0.1930 obtained by Partial Least Squares Regression.Therefore, by 20 sub- spectral regions
11 sub- SPECTRAL REGION removals in domain, remaining 19 sub- SPECTRAL REGIONs remove 1 sub- SPECTRAL REGION again, make remaining 18 sons
Training sample spectral value in SPECTRAL REGION is interacted with training sample classification information matrix progress Partial Least Squares Regression gained to be tested
Root-mean-square error (RMSECV) minimum is demonstrate,proved, as shown in table 1, root-mean-square error is minimum after removing sub- SPECTRAL REGION 20, thus remains
Lower 18 sub- SPECTRAL REGIONs.The rest may be inferred for subsequent calculating.As shown in table 1, when removing sub- SPECTRAL REGION 11,20,8,9,5,16,
When 17,18,15,13, root-mean-square error reaches minimum value 0.0862, then remaining sub- SPECTRAL REGION 1,2,3,4,6,7,10,
12,14 and 19 be preferred sub- SPECTRAL REGION.
Table 1
The sub- SPECTRAL REGION that Fig. 4 is selected after being to interval partial least squares discriminant analysis.According to preferred sub- SPECTRAL REGION
Select the training sample and test sample spectrum of training sample and test sample SPECTRAL REGION and composition apple.
S4 carries out dimension-reduction treatment to apple near infrared spectrum and authentication information extracts: using principal component analysis (PCA) to S3
In apple near infrared spectrum data compressed, spectrum is compressed to 8 dimensions;Apple near infrared spectrum data in S3 is subtracted
Subtract the average value of spectroscopic data, then calculate covariance matrix and to covariance matrix carry out feature decomposition calculate characteristic value and
Feature vector arranges characteristic value from big to small, corresponding 8 feature vectors of preceding 8 maximum eigenvalue is taken, by 200 apples
The near infrared spectrum data of sample projects in this 8 feature vectors, so that near infrared spectrum is compressed to 8 dimensions from 1557 dimensions.
Then the authentication information of 8 dimension datas is extracted using linear discriminant analysis (LDA), treated that training sample is calculated with PCA
Collision matrix S in classWThe collision matrix S between classB, to matrixIt carries out feature decomposition and calculates characteristic value and feature vector, take
Corresponding 3 feature vectors of preceding 3 maximum eigenvalue (being respectively as follows: 12.241,9.979,2.290), test sample is projected to
It is obtained in this 3 feature vectors by linear discriminant analysis LDA treated three-dimensional test sample data, as shown in Figure 5.
S5 distinguishes apple to a kind of improved fuzzy C-means clustering method of test sample in S4 including authentication information
Kind.
Apple product are distinguished to a kind of improved fuzzy C-means clustering method of 3 dimension test sample shown in fig. 5 in S4
Kind.
Initialization: weighted index m=2, classification number c=4 is arranged, wherein m > 1, c > 1 in S5.1;N=120, setting circulation meter
The initial value of number r is 1 and maximum number of iterations rmax=100;Iteration worst error parameter ε=0.00001 is set;Run Fuzzy C
The class central value v that mean cluster obtainsi,FCMAs initial class central value νi (0): ν1 (0)=[14.666-7.452
12.781];ν2 (0)=[14.793-7.543 12.869];ν3 (0)=[14.87-7.309 12.742];ν4 (0)=[14.869
-7.555 12.74].Calculating parameter ηi:
Calculated result: η1=0.00403, η2=0.00409, η3=0.00421, η4=0.00305.
In above formula, m (m > 1) is weighted index, uik,FCMWhat is obtained after terminating for fuzzy C-means clustering iterative calculation is fuzzy
It is subordinate to angle value, uik,FCMAs shown in Figure 6;vi,FCMFor fuzzy C-means clustering iterative calculation terminate after obtain i-th (i=1,2,
3 ..., c) a class central value, xkFor k-th of test sample.
S5.2 calculates r (r=1,2 ..., rmax) secondary iteration when fuzzy membership angle value uik (r):
uikIt is sample xkIt is under the jurisdiction of the fuzzy membership angle value of classification i;viBe i-th (i=1,2,
3 ..., c) class class central value, νi (r-1)It is the class center v of the r-1 times iterative calculationiValue;vjIt is
The class central value of jth (j=1,2,3 ..., c) class, νj (r-1)It is the class center v of the r-1 times iterative calculationjValue, n be test
Sample number;
S5.3, calculate the r times iteration when the i-th class class central value νi (r)
νi (r)It is the class center v of the r times iterative calculationiValue, class center matrix V is formed by c class central value(r)=
[ν1 (r),ν2 (r),…,νc (r)];
S5.4, cycle count increase, i.e. r=r+1;
If meeting condition: | | V(r)-V(r-1)| | < ε or r > rmaxTermination is then calculated, otherwise continues S5.2, after iteration ends
Fuzzy membership angle value can be obtained and apple variety is distinguished according to fuzzy membership angle value.
Experimental result: r=31 when iteration ends, class center matrix V(r)For
Fuzzy membership after iteration ends as shown in fig. 7, can realize that four kinds of apple varieties distinguish according to fuzzy membership,
Distinguish that accuracy rate reaches 100%.
The series of detailed descriptions listed above only for feasible embodiment of the invention specifically
Protection scope bright, that they are not intended to limit the invention, it is all without departing from equivalent implementations made by technical spirit of the present invention
Or change should all be included in the protection scope of the present invention.
Claims (9)
1. a kind of apple variety discriminating conduct of the preferred fuzzy clustering of spectral band, which comprises the steps of:
The Fourier transform near infrared acquisition of different cultivars apple sample: S1 for the apple sample of different cultivars, uses Fourier
Near infrared spectrometer detects apple sample, obtains apple sample Fourier transform near infrared data and storing data;
S2 pre-processes the apple sample near infrared spectrum of step S1 with standard normal variable variation (SNV);
S3, to interval partial least squares discriminant analysis (BIPLSDA), to carry out wave band to the near infrared spectrum of step S2 preferred after;
S4 carries out dimension-reduction treatment to apple near infrared spectrum and authentication information extracts: using principal component analysis (PCA) to step S3
In apple near infrared spectrum data compressed;Then the authentication information of data is extracted using linear discriminant analysis (LDA);
S5 distinguishes apple product using improved fuzzy C-means clustering method to the test sample in step S4 including authentication information
Kind.
2. a kind of apple variety discriminating conduct of the preferred fuzzy clustering of spectral band according to claim 1, feature exist
In the specific implementation of step S1:
Apple sample is stored 12 hours under the conditions of temperature is 20~25 DEG C, the booting of Antaris II near-infrared spectrometers
Preheat 1 hour;Using reflection integrating sphere type collection apple near infrared spectrum, near-infrared spectrometers scan each sample
32 times to obtain the diffusing reflection spectrum mean value of sample, the wave number of spectral scan is set as 10000~4000cm-1, sweep spacing is set as
3.856cm-1, the spectrum for collecting each sample is the data of 1557 dimensions.
3. a kind of apple variety discriminating conduct of the preferred fuzzy clustering of spectral band according to claim 2, feature exist
In step S1 further include: each apple sample samples 3 times along equator track, takes its average value as final acquisition data.
4. a kind of apple variety discriminating conduct of the preferred fuzzy clustering of spectral band according to claim 1, feature exist
In pretreated specific implementation in step S2:
Changed using standard normal variable and eliminate solid particle size, surface scattering and change in optical path length are to diffusing reflection near infrared light
The influence of spectrum;The apple sample near infrared spectrum data of step S1 is subtracted after the average value of spectroscopic data again divided by spectroscopic data
Standard deviation.
5. a kind of apple variety discriminating conduct of the preferred fuzzy clustering of spectral band according to claim 1, feature exist
In the specific implementation of step S3:
Several apple samples are divided into two parts of training sample and test sample by S3.1;
S3.2 establishes training sample classification information matrix, and it is a wide that the near infrared spectrum region of apple is fifty-fifty divided into L
Then sub- SPECTRAL REGION removes 1 sub- SPECTRAL REGION every time, makes training sample spectral value and the training in remaining SPECTRAL REGION
It is minimum that sample class information matrix carries out validation-cross root-mean-square error (RMSECV) obtained by Partial Least Squares Regression, until finally
1 sub- SPECTRAL REGION;
S3.3 chooses several sub- SPECTRAL REGIONs according to root-mean-square error minimum principle;Regional choice is composed according to preferred sub-light
Test sample SPECTRAL REGION and composition apple test sample spectrum.
6. a kind of apple variety discriminating conduct of the preferred fuzzy clustering of spectral band according to claim 5, feature exist
In the method for choosing several sub- SPECTRAL REGIONs according to root-mean-square error minimum principle in the S3.3:
SPECTRAL REGION is averagely divided into L=20 wide sub- SPECTRAL REGIONs, removes 1 sub- SPECTRAL REGION every time, it is remaining
Training sample spectral value in 19 sub- SPECTRAL REGIONs and training sample classification information matrix carry out Partial Least Squares Regression, when going
Validation-cross root-mean-square error (RMSECV) obtained by resulting Partial Least Squares Regression is minimum when except sub- SPECTRAL REGION 11
0.1930;Therefore, the sub- SPECTRAL REGION 11 in 20 sub- SPECTRAL REGIONs is removed, remaining 19 sub- SPECTRAL REGIONs remove 1 again
A sub- SPECTRAL REGION, make training sample spectral value in remaining 18 sub- SPECTRAL REGIONs and training sample classification information matrix into
Validation-cross root-mean-square error (RMSECV) obtained by row Partial Least Squares Regression is minimum, and the rest may be inferred for subsequent calculating.
7. a kind of apple variety discriminating conduct of the preferred fuzzy clustering of spectral band according to claim 1, feature exist
In being compressed using principal component analysis (PCA) to the apple near infrared spectrum data in S3 in step S4, is by spectrum pressure
8 dimensions are reduced to, specific method:
The average value that apple near infrared spectrum data in step S3 is subtracted to spectroscopic data, then calculates covariance matrix
And feature decomposition is carried out to covariance matrix and calculates characteristic value and feature vector, characteristic value is arranged from big to small, take first 8 most
Corresponding 8 feature vectors of big characteristic value, the near infrared spectrum data of several samples is projected in this 8 feature vectors,
To which near infrared spectrum is compressed to 8 dimensions.
8. a kind of apple variety discriminating conduct of the preferred fuzzy clustering of spectral band according to claim 7, feature exist
In the specific method of the authentication information for extracting 8 dimension datas using linear discriminant analysis (LDA) in step S4:
With PCA, treated that collision matrix S in class is calculated in training sampleWThe collision matrix S between classB, to matrixIt carries out
Feature decomposition calculates characteristic value and feature vector, takes corresponding 3 feature vectors of preceding 3 maximum eigenvalue, test sample is thrown
It is obtained on shadow to this 3 feature vectors by linear discriminant analysis LDA treated three-dimensional test sample data.
9. a kind of apple variety discriminating conduct of the preferred fuzzy clustering of spectral band according to claim 1, feature exist
In the specific implementation of step S5:
Initialization: weighted index m, classification number c is arranged, wherein m > 1, c > 1 in S5.1;The initial value and maximum of cycle count r are set
The number of iterations rmax;Iteration worst error parameter ε is set;The class central value v that operation fuzzy C-means clustering obtainsi,FCMAs initial
Class central value νi (0);Calculating parameter ηi:
In above formula, m (m > 1) is weighted index, uik,FCMThe fuzzy membership obtained after being terminated for fuzzy C-means clustering iterative calculation
Angle value, vi,FCMI-th (i=1,2,3 ..., c) a class central value obtained after being terminated for fuzzy C-means clustering iterative calculation, xkFor
K-th of test sample.
S5.2 calculates r (r=1,2 ..., rmax) secondary iteration when fuzzy membership angle value uik (r):
uikIt is sample xkIt is under the jurisdiction of the fuzzy membership angle value of classification i;viBe i-th (i=1,2,3 ..., c)
The class central value of class, νi (r-1)It is the class center v of the r-1 times iterative calculationiValue;vjIt is jth (j=
1,2,3 ..., c) class class central value, νj (r-1)It is the class center v of the r-1 times iterative calculationjValue, n be test sample number;
S5.3, calculate the r times iteration when the i-th class class central value νi (r)
νi (r)It is the class center v of the r times iterative calculationiValue, class center matrix V is formed by c class central value(r)=[ν1 (r),
ν2 (r),…,νc (r)];
S5.4, cycle count increase, i.e. r=r+1;
If meeting condition: | | V(r)-V(r-1)| | < ε or r > rmaxTermination is then calculated, otherwise continues S5.2, can be obtained after iteration ends
Fuzzy membership angle value simultaneously distinguishes apple variety according to fuzzy membership angle value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811338607.5A CN109685099B (en) | 2018-11-12 | 2018-11-12 | Apple variety distinguishing method based on spectrum band optimization fuzzy clustering |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811338607.5A CN109685099B (en) | 2018-11-12 | 2018-11-12 | Apple variety distinguishing method based on spectrum band optimization fuzzy clustering |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109685099A true CN109685099A (en) | 2019-04-26 |
CN109685099B CN109685099B (en) | 2023-04-07 |
Family
ID=66185355
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811338607.5A Active CN109685099B (en) | 2018-11-12 | 2018-11-12 | Apple variety distinguishing method based on spectrum band optimization fuzzy clustering |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109685099B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110987866A (en) * | 2019-12-19 | 2020-04-10 | 汉谷云智(武汉)科技有限公司 | Gasoline property evaluation method and device |
CN111257242A (en) * | 2020-02-27 | 2020-06-09 | 西安交通大学 | High-spectrum identification method for pollutant components of insulator |
CN115015165A (en) * | 2022-06-06 | 2022-09-06 | 江苏大学 | Yogurt variety identification method based on fuzzy optimal identification vector |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109030407A (en) * | 2018-04-16 | 2018-12-18 | 江苏大学 | A kind of apple variety classification method of mixed fuzzy C mean cluster |
-
2018
- 2018-11-12 CN CN201811338607.5A patent/CN109685099B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109030407A (en) * | 2018-04-16 | 2018-12-18 | 江苏大学 | A kind of apple variety classification method of mixed fuzzy C mean cluster |
Non-Patent Citations (1)
Title |
---|
武斌等: "FUDT在苹果近红外光谱分类中的应用", 《计算机工程与应用》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110987866A (en) * | 2019-12-19 | 2020-04-10 | 汉谷云智(武汉)科技有限公司 | Gasoline property evaluation method and device |
CN111257242A (en) * | 2020-02-27 | 2020-06-09 | 西安交通大学 | High-spectrum identification method for pollutant components of insulator |
CN115015165A (en) * | 2022-06-06 | 2022-09-06 | 江苏大学 | Yogurt variety identification method based on fuzzy optimal identification vector |
Also Published As
Publication number | Publication date |
---|---|
CN109685099B (en) | 2023-04-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103048273B (en) | Fruit near infrared spectrum sorting method based on fuzzy clustering | |
Li et al. | Non-destructive discrimination of Chinese bayberry varieties using Vis/NIR spectroscopy | |
WO2019192433A1 (en) | Method for chemical pattern recognition of authenticity of traditional chinese medicine chinese honeylocust spine based on near-infrared spectroscopy | |
Cai et al. | Using FTIR spectra and pattern recognition for discrimination of tea varieties | |
CN104062263B (en) | The near-infrared universal model detection method of light physical property close fruit quality index | |
CN109685099A (en) | A kind of apple variety discriminating conduct of the preferred fuzzy clustering of spectral band | |
Cui et al. | Identification of maize seed varieties based on near infrared reflectance spectroscopy and chemometrics | |
Tian et al. | Measurement orientation compensation and comparison of transmission spectroscopy for online detection of moldy apple core | |
CN109030407A (en) | A kind of apple variety classification method of mixed fuzzy C mean cluster | |
CN105138834A (en) | Tobacco chemical value quantifying method based on near-infrared spectrum wave number K-means clustering | |
CN107192686A (en) | A kind of Possibility Fuzzy Clustering local tea variety discrimination method of fuzzy covariance matrix | |
CN109685098A (en) | The local tea variety classification method of cluster is separated between a kind of Fuzzy Cluster | |
Zhaoyong et al. | Detection of moldy core in apples and its symptom types using transmittance spectroscopy | |
Zhang et al. | Detection of pears with moldy core using online full-transmittance spectroscopy combined with supervised classifier comparison and variable optimization | |
CN110084227A (en) | Mode identification method based on near-infrared spectrum technique | |
CN104990891B (en) | A kind of seed near infrared spectrum and spectrum picture qualitative analysis model method for building up | |
CN110220863A (en) | A kind of discrimination method of honeysuckle and Honeysuckle flower Chinese materia medica preparation based on ATR-FTIR | |
CN117874609A (en) | Universal rapid method for rapidly identifying whether natural product is specific production place | |
CN103954582A (en) | Apple cultivar near-infrared-spectrum sorting method based on hybrid K-harmonic means clustering | |
CN107886115A (en) | A kind of tealeaves mid-infrared light profile classification method of adaptively possible C mean clusters | |
Fu et al. | Discrimination of pear varieties using three classification methods based on near-infrared spectroscopy | |
Yuan et al. | Application of hyperspectral imaging to discriminate waxy corn seed vigour after aging. | |
CN110231300A (en) | A kind of lossless method for quickly identifying true and false Aksu red fuji apple | |
Jiang et al. | The utility of Fourier transform near-infrared spectroscopy to identify geographical origins of Chinese pears | |
CN106568740A (en) | Method for rapid judging of varieties of fresh tea leaves by 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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |