CN107037001A - A kind of corn monoploid seed discrimination method based on near-infrared spectrum technique - Google Patents
A kind of corn monoploid seed discrimination method based on near-infrared spectrum technique Download PDFInfo
- Publication number
- CN107037001A CN107037001A CN201710455303.6A CN201710455303A CN107037001A CN 107037001 A CN107037001 A CN 107037001A CN 201710455303 A CN201710455303 A CN 201710455303A CN 107037001 A CN107037001 A CN 107037001A
- Authority
- CN
- China
- Prior art keywords
- spectroscopic data
- data
- corn monoploid
- discrimination method
- neural network
- 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
- 238000000034 method Methods 0.000 title claims abstract description 36
- 235000005824 Zea mays ssp. parviglumis Nutrition 0.000 title claims abstract description 32
- 235000002017 Zea mays subsp mays Nutrition 0.000 title claims abstract description 32
- 235000005822 corn Nutrition 0.000 title claims abstract description 32
- 238000002329 infrared spectrum Methods 0.000 title claims abstract description 15
- 240000008042 Zea mays Species 0.000 title claims description 30
- 238000012850 discrimination method Methods 0.000 title claims description 18
- 238000004611 spectroscopical analysis Methods 0.000 claims abstract description 63
- 238000010606 normalization Methods 0.000 claims abstract description 31
- 238000001228 spectrum Methods 0.000 claims abstract description 25
- 238000013528 artificial neural network Methods 0.000 claims abstract description 19
- 238000011156 evaluation Methods 0.000 claims abstract description 6
- 238000012549 training Methods 0.000 claims description 40
- 239000011159 matrix material Substances 0.000 claims description 26
- 238000012360 testing method Methods 0.000 claims description 23
- 230000009467 reduction Effects 0.000 claims description 21
- 230000007935 neutral effect Effects 0.000 claims description 8
- 230000006870 function Effects 0.000 claims description 7
- 238000004458 analytical method Methods 0.000 claims description 6
- 238000003062 neural network model Methods 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 6
- 230000005540 biological transmission Effects 0.000 claims description 4
- 210000005036 nerve Anatomy 0.000 claims description 3
- 238000002834 transmittance Methods 0.000 claims description 3
- 230000008859 change Effects 0.000 claims description 2
- 239000004576 sand Substances 0.000 claims description 2
- 241000209149 Zea Species 0.000 abstract 2
- 230000003595 spectral effect Effects 0.000 description 18
- 238000005516 engineering process Methods 0.000 description 5
- 238000000354 decomposition reaction Methods 0.000 description 4
- 238000012216 screening Methods 0.000 description 3
- 240000007594 Oryza sativa Species 0.000 description 2
- 235000007164 Oryza sativa Nutrition 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000000205 computational method Methods 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000010977 jade Substances 0.000 description 2
- 239000003550 marker Substances 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 235000009566 rice Nutrition 0.000 description 2
- 244000287680 Garcinia dulcis Species 0.000 description 1
- 235000013339 cereals Nutrition 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000009828 non-uniform distribution Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 230000000149 penetrating effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000009738 saturating Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- 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
- 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
Landscapes
- Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention discloses a kind of corn monoploid seed mirror method for distinguishing based on near-infrared spectrum technique, including:Original spectrum file is read in, spectrum intensity data is obtained;Data set is divided;The feature normalization of spectroscopic data;Dimension is reduced using partial least-square regression method;Neural network classifier parameter regulation;Neural network classifier is finely tuned and performance evaluation;Preserve neural network parameter.This method can identify corn monoploid seed from heterozygote seed, while ensuring higher correct resolution and model stability.
Description
Technical field
The invention belongs to computer and optical technology application field, more particularly to a kind of jade based on near-infrared spectrum technique
Rice monoploid seed discrimination method.
Background technology
The discriminating of corn monoploid seed is a new technical field of computer and spectrum of use, due to corn monoploid
Seed generation rate during cultivating seeds is extremely low, and usual 10% less than artificial screening corn monoploid seed is than relatively time-consuming consumption
Power.Therefore, it is Land use models mirror method for distinguishing using near-infrared spectrum technique and computer technology purpose, to realize computer
Monoploid seed screening function is aided in, so as to realize automation seed selection, the effect of optimization agricultural is reached.
Corn monoploid authentication technique based on near-infrared spectrum technique, such as obtains seed spectrum using diffusing transmission method,
The seed inner material information of non-uniform Distribution can be obtained.But, because spectral component is complicated, diffusing transmission mechanism is not also complete
Kind theoretical explanation, differentiates seed seed by spectrum in actual applications, in addition it is also necessary to reference to CRT technology algorithm
To improve the correct resolution of monoploid seed as far as possible.
CRT technology algorithm is designed, and is typically pre-processed from feature normalization, Feature Dimension Reduction denoising, Yi Jite
Three aspects of classification are levied to account for.Center normalization or Standard normalization are conventional at present
Feature normalization processing method, but still there is strong correlation between feature after this method characteristic processing.
In summary there is provided rational pattern discrimination method, the discriminating performance of monoploid seed spectrum can be improved, so that
The need for meeting in real work, effectively to realize that the screening of automation corn monoploid seed provides computational methods.
The content of the invention
(1) technical problem to be solved
, can be by jade the invention provides a kind of corn monoploid seed mirror method for distinguishing based on near-infrared spectrum technique
Rice monoploid seed is identified from heterozygote seed, while ensuring higher correct resolution and model stability.
(2) technical scheme
The present invention is achieved by the following technical solutions:
A kind of corn monoploid seed discrimination method based on near-infrared spectrum technique, comprises the following steps:
Spectroscopic data is obtained, and spectroscopic data is divided;
Spectroscopic data after division is subjected to feature normalization;
Spectroscopic data after normalization is reduced into dimension;
Spectroscopic data after dimensionality reduction is classified.
Preferably, the acquisition spectroscopic data, and the step of spectroscopic data is divided in, using diffusing reflection or unrestrained saturating
The mode of penetrating gathers the spectroscopic data of corn monoploid and heterozygote seed, and obtains corresponding corn monoploid and heterozygote seed
The label data of grain.
Preferably, the acquisition spectroscopic data, and the step of spectroscopic data is divided in, spectroscopic data is divided into three
Individual data set:Training set, checking collection and test set;Training set is used for training pattern, and checking collection is used to adjust model parameter, tested
Collect for test model performance.
Preferably, in the step of spectroscopic data by after division carries out feature normalization, using Zero-phase
Components analysis methods carry out spectroscopic data intensities normalised, the spectroscopic data after being normalized.
Preferably, the calculation formula of spectroscopic data progress feature normalization is:
Wherein, X represents eigenmatrix,Represent that eigenmatrix X subtracts the zero-mean matrix after mean vector, U represents empty
Between base, S represents diagonal matrix, and eps is a small constant,Represent the eigenmatrix after normalization.
Preferably, in the step of spectroscopic data by after normalization reduces dimension, using PLS side
Method reduces the dimension of spectroscopic data, obtains score matrix XSAnd its transformation matrix ST.
Preferably, the dimension of the use partial least-square regression method reduction spectroscopic data includes:
Wherein,For the eigenmatrix after normalization, y is label data, and p is characterized matrixProjecting direction, q for mark
Data y projecting direction is signed, transformation matrix ST is calculated according to the p obtained every time and obtained, score matrix XsForAfter being projected through ST
Matrix.
Preferably, the step of spectroscopic data to after dimensionality reduction carries out tagsort includes:
Spectroscopic data after dimensionality reduction is inputted into neutral net, neural network parameter is adjusted;
Neural network parameter is finely tuned and performance evaluation;
Preserve neural network parameter.
Preferably, in the step of spectroscopic data by after dimensionality reduction inputs neutral net, regulation neural network parameter, adopt
The spectroscopic data after dimensionality reduction is classified with L2 norm regularizations reverse transmittance nerve network, the form of its cost function is:
Wherein, m represents sample size, xiRepresent i-th of sample, yiRepresent the true value of the label data of i-th of sample, hθ
() represents neural network model hypothesis, and λ represents punishment parameter, and θ represents model parameter to be learned.
Preferably, reverse neural network model learning parameter θ is initialized using random fashion, and its more new formula is:
Wherein, ε is learning rate parameter, JbFor cost function.
(3) beneficial effect
Compared with prior art, the invention has the advantages that:
(1) the corn monoploid seed discrimination method proposed by the present invention based on near-infrared spectrum technique, realizes single times
The discriminating of body seed, goes to design discrimination method, is effectively improved the identity of model from the angle of the vivid geometry of higher dimensional space
Energy.
(2) Zero-phase components analysis feature normalization methods of the present invention, Neng Gouyou
Effect ground removes the correlation between spectral signature, so that effectively enhanced spectrum feature.
(3) PLS of the present invention realizes the reduction of data characteristics dimension, can preferably retain
The weak characteristic component of data, and removal noise and redundancy are realized by dimension reduction, so that efficiency of algorithm is improved.
(4) neural network model grader of the present invention, can be realized in the embedded hardware systems such as FPGA
Parallel computation, so as to improve training and predetermined speed of model, collateral security differentiates operational efficiency.
Brief description of the drawings
Fig. 1 is the corn monoploid seed mirror method for distinguishing stream provided in an embodiment of the present invention based on near-infrared spectrum technique
Cheng Tu;
Fig. 2 is the corn monoploid seed mirror method for distinguishing stream provided in an embodiment of the present invention based on near-infrared spectrum technique
Journey total figure;
Fig. 3 is punishment parameter regulation comparison diagram provided in an embodiment of the present invention;
Fig. 4 is average correct recognition rata provided in an embodiment of the present invention and training set size relationship figure.
Embodiment
For the object, technical solutions and advantages of the present invention are more clearly understood, below in conjunction with specific embodiment, and reference
Accompanying drawing, the present invention is described in more detail.It should be appreciated that described herein is only a part of embodiment of the invention, ability
The every other embodiment that domain those of ordinary skill is obtained on the premise of creative work is not made, belongs to the present invention
The scope of protection.
Fig. 1 and Fig. 2 are the corn monoploid seed discriminating sides provided in an embodiment of the present invention based on near-infrared spectrum technique
The method flow diagram of method, for convenience of description, illustrate only the part related to the embodiment of the present invention, the discrimination method it is main by
The normalization of spectral signature, the reduction of spectral signature dimension and spectral signature, which are classified, to be constituted.As depicted in figs. 1 and 2, this method includes
Following steps:
Step S1:Spectroscopic data is obtained, and spectroscopic data is divided.
Step S1 includes:
Sub-step S11, reads in original spectrum file, obtains spectroscopic data, and spectroscopic data includes spectroscopic data collection and spectrum
Label data collection.
Specifically, reading in original spectrum file, the spectroscopic data of acquisition includes original spectrum intensity data, the label of spectrum
The dimension information of information and spectrum, the dimension information of spectrum includes dimension and bar number, is designated as n and mT。
The present embodiment gathers the near infrared spectrum data collection X of corn monoploid and heterozygote seed by diffusing transmission modeT
And the spectrum label data set y of corresponding corn monoploid and heterozygote.Spectroscopic data collection XTIncluding original spectrum intensity number
According to, the information such as spectral dimension information, spectral Dimensions n is 125 dimensions, and spectral wavelength ranges are floating number between 900nm-1700nm
Value is represented.In spectrum label data set y, label information is monoploid spectral marker 1, heterozygote spectral marker 0.
Sub-step S12, spectroscopic data collection is divided.
Spectroscopic data collection is divided, including using random without spectrum is extracted by the way of putting back to, m is arrived in generation 1TRandom manifold R, will
Spectroscopic data random manifold corresponding with label data, the m for proportionally concentrating spectroscopic dataTBar spectroscopic data is divided into three numbers
According to collection:Training set, checking collection and test set.Training set is that checking collection is that, for regulation parameter, test set is for training pattern
For test model performance.
It is preferred that, training set, checking collects the ratio with test set, may be configured as 2: 1: 2 or 3: 1: 1.
Step S2:Spectroscopic data after division is subjected to feature normalization.
Spectral signature is normalized, and is that the spectroscopic data for detecting near infrared spectrometer carries out intensities normalised, the present embodiment
The detecting instrument used is MicroNIR-1700 near infrared spectrometer.Utilize Zero-phase component analysis
(ZCA) correlation between feature can be removed very well.
Specifically, the present embodiment carries out Zero-phase components using the training set of gained in sub-step S12
Analysis (ZCA), obtains its normalization matrix, computational methods:
In formula, eigenmatrix X represents the original spectral data of training set, and size is n × m, and n represents wall scroll spectral signature
Dimension, m is training samples number.Mean (X) represents the average value of training set spectrum,Subtracted for training set eigenmatrix X
It is worth the spectroscopic data after vector mean (X), is zero-mean matrix.Svd () is singular value decomposition, by singular value decomposition
(singular value decomposition), U is by the space base vector that is obtained after decomposition, and S is is obtained to angular moment
Battle array.Eps represents a small constant, 1e-6 preferably.For the new spy of training set data gained after ZCA feature normalizations
Levy data.
Spectroscopic data in training set utilized above, according to Zero-phase components analysis (ZCA) sides
The calculating of method has obtained normalization matrix Z, ZThen checking collection and test set normalizing are calculated according to Z
Spectroscopic data after change.I.e. according to the average value for verifying collection and test set, and above-mentioned U and S collect and survey to calculate checking
The normalized data of examination collection, training set, checking collects the spectroscopic data after being normalized with test set, Tr, Va and Te is designated as respectively.Protect
Deposit the data of the gained after normalization.
Step S3:Spectroscopic data after normalization is reduced into dimension.
Spectral signature dimension is reduced, and is to extract effective spectral signature information.Partial least-square regression method dimensionality reduction denoising phase
Than the feature that PCA can retain some weak components, the present embodiment is using partial least-square regression method to spectrum number
According to progress dimensionality reduction.
Specifically, the present embodiment utilizes the training set data and corresponding number of tags after the normalization obtained in step S2
According to concurrently setting the dimension k of data, it is preferred that k is set { 3,10,20 } intermediate value, with reference to partial least-square regression method, meter
Calculation form has:
In formula,Represent that spectroscopic data Tr, p after training set normalization are training set dataProjecting direction, q is
The corresponding spectrum label data set y projecting directions of training set.Meanwhile, p in each iteration, q meets the limitation of mould a length of 1.So as to
Obtain training set score matrix Xs and its transformation matrix ST, score matrix Xs beMatrix after being projected through ST, ST foundations
The p obtained every time, which is calculated, to be obtained, and is recycled identical linear transformation to be calculated respectively by Va and Te and is verified collection and test set
Score matrix.Accordingly, as the low-dimensional characteristic after dimensionality reduction.
Step S4:Tagsort is carried out to the spectroscopic data after dimensionality reduction.
Step S4 includes:
Sub-step S41, neural network classifier parameter regulation.
Spectral signature is classified, and is to distinguish monoploid seed spectrum and heterozygote seed spectral matching factor, neural network classification
Device can extract the hiding feature in feature, it is adaptable to the complicated situation of some features.The present embodiment uses L2 norm regularizations
Reverse transmittance nerve network (back propagation neural network) method, to the score square obtained in step S3
Battle array Xs carries out two and classified, and reverse neural network has three layers, input layer, hidden layer and output layer, the god of input layer and hidden layer
All it is above-mentioned dimension k through first number, output layer neuron number is 1 or 2.
Specifically, by training set data feature (score matrix Xs obtained by step S3), accessing the neutral net of k-k-1 structures
In.The cost function of neutral net is:
In formula, m represents training set sample size, xiRepresent i-th of training sample of wall scroll, yiRepresent i-th of sample correspondence
Spectrum label data, hθ() is the model hypothesis of neutral net, and λ is punishment parameter, and θ represents reverse neural network learning
Parameter set, θ is initialized using random fashion, and its more new formula is:
In formula, ε is learning rate parameter, JbFor cost function.
It is preferred that, setting frequency of training N=800 times, λ can be collection { 1,10-1, 10-2, 10-3, 10-4, 10-5, 10-6Intermediate value,
ε is 0.05, by verifying that 50 Average Accuracies of collection are according to the optimum value for drawing λ, while according to fixed intrinsic dimensionality
Determine k probable ranges.It is illustrated in figure 3 λ in influences of the λ of drafting to checking ensemble average recognition accuracy, this specific embodiment
For 10-4, while determining k scope between [3,10].
Sub-step S42, neural network classifier fine setting and performance evaluation.
Punishment parameter is fixed, training set and test set is obtained, obtained neural network model, profit are trained using training set
Final mask Performance Evaluation is carried out with test set, optimal hidden layer neuron number k values are determined, exhaustion determines feature in scope
Dimension, repeats the flow n times, finally optimal characteristics dimension is determined using the distribution situation of test set accuracy rate, obtains simultaneously
Optimal models Performance Evaluation, exhaustion 3 to 10 in the range of k value, it is preferred that repeat 50 times acquisition test sets Average Accuracy and
Its standard deviation is used as model performance judging basis.It is training set Average Accuracy as shown in table 1 in this specific embodiment
The Average Accuracy (Average test accuracy) and standard deviation of (Average training accuracy), test set
Value (SD of test accuracy) is with the variation relation of k values, comprehensive Average Accuracy and standard deviation, institute in the present embodiment
It is 6 to obtain optimal k.Meanwhile, according to parameters obtained, we can further draw out shadow of the training set sample size to model performance
Ring, be illustrated in figure 4 the relation curve of training set size and test set Average Accuracy.
Table 1
Sub-step S43, preserves neural network parameter.
Specifically, neural network parameter θ, punishment parameter λ and optimal lower dimensional space dimension k obtained by fixing step S42,
Training set and test set are reacquired, using training set training pattern, model training is carried out according to above-mentioned identified parameter, protected
The training set average of gained, feature normalization matrix, offset minimum binary transformation matrix and neutral net network parameter are deposited, simultaneously
Generation model Parameter File.
The embodiment of the present invention by using unified calculating body structure, obtain near infrared spectrum different wave length point data and its
Dimension information, by the normalization to spectral signature, dimensionality reduction, a kind of corn monoploid seed mirror in conjunction with neural fusion
Method for distinguishing, is gone to design recognizer using the angle from higher-dimension bio-information geometry, is effectively improved Model Identification rate
And stability.
Through the above description of the embodiments, those skilled in the art can be understood that the present invention can be by
Software adds the mode of required general hardware platform to realize, naturally it is also possible to which by hardware, but in many cases, the former is most
Good embodiment.Understood based on such, what technical scheme substantially contributed to prior art in other words
Part can be embodied in the form of software product, be stored in computer software product in a storage medium, if including
Dry instruction is to cause a station terminal equipment (can be personal computer, server, or embedded device etc.) to perform this hair
Method described in each bright embodiment.The present embodiment test platform:CPU:Pentium(R)Dual-Core CPU E5800@
3.20GHz;Internal memory:4.00GB;The bit manipulation systems of system Windows 732;Software:Matlab R2013a.
Particular embodiments described above, has been carried out further in detail to the purpose of the present invention, technical scheme and beneficial effect
Describe in detail it is bright, should be understood that the foregoing is only the present invention specific embodiment, be not intended to limit the invention, it is all
Within the spirit and principles in the present invention, any modification, equivalent substitution and improvements done etc., should be included in the guarantor of the present invention
Within the scope of shield.
Claims (10)
1. a kind of corn monoploid seed discrimination method based on near-infrared spectrum technique, comprises the following steps:
Spectroscopic data is obtained, and spectroscopic data is divided;
Spectroscopic data after division is subjected to feature normalization;
Spectroscopic data after normalization is reduced into dimension;
Spectroscopic data after dimensionality reduction is classified.
2. corn monoploid seed discrimination method as claimed in claim 1, wherein, the acquisition spectroscopic data, and by spectrum
In the step of data are divided, the spectrum number of corn monoploid and heterozygote seed is gathered using diffusing reflection or diffusing transmission mode
According to, and obtain the label data of corresponding corn monoploid and heterozygote seed.
3. corn monoploid seed discrimination method as claimed in claim 1 or 2, wherein, the acquisition spectroscopic data, and by light
In the step of modal data is divided, spectroscopic data is divided into three data sets:Training set, checking collection and test set;Training set
For training pattern, checking collection is used to adjust model parameter, and test set is used for test model performance.
4. corn monoploid seed discrimination method as claimed in claim 1 or 2, wherein, the spectroscopic data by after division
In the step of carrying out feature normalization, spectroscopic data is carried out using Zero-phase components analysis methods strong
Degree standardization, the spectroscopic data after being normalized.
5. corn monoploid seed discrimination method as claimed in claim 4, wherein, spectroscopic data carries out the meter of feature normalization
Calculating formula is:
Wherein, X represents eigenmatrix,Represent that eigenmatrix X subtracts the zero-mean matrix after mean vector, U representation spaces
Base, S represents diagonal matrix, and eps is a small constant,Represent the eigenmatrix after normalization.
6. corn monoploid seed discrimination method as claimed in claim 1, wherein, the spectroscopic data by after normalization drops
In the step of low-dimensional number, the dimension of spectroscopic data is reduced using partial least-square regression method, score matrix X is obtainedSAnd its become
Change matrix ST.
7. corn monoploid seed discrimination method as claimed in claim 6, wherein, the use partial least-square regression method
The dimension of reduction spectroscopic data includes:
Wherein,For the eigenmatrix after normalization, y is label data, and p is characterized matrixProjecting direction, q is number of tags
According to y projecting direction, transformation matrix ST is calculated according to the p obtained every time and obtained, score matrix XsForSquare after being projected through ST
Battle array.
8. corn monoploid seed discrimination method as claimed in claim 1, wherein, the spectroscopic data to after dimensionality reduction is carried out
The step of tagsort, includes:
Spectroscopic data after dimensionality reduction is inputted into neutral net, neural network parameter is adjusted;
Neural network parameter is finely tuned and performance evaluation;
Preserve neural network parameter.
9. corn monoploid seed discrimination method as claimed in claim 8, wherein, the spectroscopic data by after dimensionality reduction is inputted
Neutral net, regulation neural network parameter the step of in, using L2 norm regularization reverse transmittance nerve networks to dimensionality reduction after
Spectroscopic data is classified, and the form of its cost function is:
Wherein, m represents sample size, xiRepresent i-th of sample, yiRepresent the true value of the label data of i-th of sample, hθ(·)
Neural network model hypothesis is represented, λ represents punishment parameter, and θ represents model parameter to be learned.
10. corn monoploid seed discrimination method as claimed in claim 9, wherein, reverse neural network model learning parameter θ
Initialized using random fashion, its more new formula is:
Wherein, ε is learning rate parameter, JbFor cost function.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710455303.6A CN107037001A (en) | 2017-06-15 | 2017-06-15 | A kind of corn monoploid seed discrimination method based on near-infrared spectrum technique |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710455303.6A CN107037001A (en) | 2017-06-15 | 2017-06-15 | A kind of corn monoploid seed discrimination method based on near-infrared spectrum technique |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107037001A true CN107037001A (en) | 2017-08-11 |
Family
ID=59541420
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710455303.6A Pending CN107037001A (en) | 2017-06-15 | 2017-06-15 | A kind of corn monoploid seed discrimination method based on near-infrared spectrum technique |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107037001A (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108509998A (en) * | 2018-03-30 | 2018-09-07 | 中国科学院半导体研究所 | A kind of transfer learning method differentiated on different devices for target |
CN109142269A (en) * | 2018-07-26 | 2019-01-04 | 江苏大学 | A kind of method for quick identification of chilled beef difference storage time |
CN109324188A (en) * | 2018-10-11 | 2019-02-12 | 珠海沃姆电子有限公司 | A kind of precision dynamic urine measurement method and system |
CN109540831A (en) * | 2019-01-25 | 2019-03-29 | 中国中医科学院中药研究所 | Fructus lycii variety ecotype method based on high light spectrum image-forming technology |
CN109766742A (en) * | 2018-11-20 | 2019-05-17 | 安徽农业大学 | A kind of Corn Seeds method for crack, device, system, equipment and storage medium |
CN110020679A (en) * | 2019-03-25 | 2019-07-16 | 中国科学院半导体研究所 | Classification method and device based on one-way analysis of variance selection bloom spectrum wavelength |
CN110090809A (en) * | 2018-01-31 | 2019-08-06 | 西派特(北京)科技有限公司 | A kind of device of online high-speed lossless automatic sorting monoploid corn kernel |
CN110208211A (en) * | 2019-07-03 | 2019-09-06 | 南京林业大学 | A kind of near infrared spectrum noise-reduction method for Detecting Pesticide |
CN114062305A (en) * | 2021-10-15 | 2022-02-18 | 中国科学院合肥物质科学研究院 | Single grain variety identification method and system based on near infrared spectrum and 1D-In-Resnet network |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101031789A (en) * | 2004-07-07 | 2007-09-05 | 孔布雷电子公司-Selco | Optical component for observing a nanometric sample, system comprising same, analysis method using same, and uses thereof |
CN102760197A (en) * | 2011-04-26 | 2012-10-31 | 电子科技大学 | Prediction method for spectroscopic test data of cancer patients by partial least square method based on Matlab |
CN103116804A (en) * | 2013-01-31 | 2013-05-22 | 苏州科技学院 | Fuzzy neural network model and intelligent prediction method for deep excavation deformation |
CN103234922A (en) * | 2013-03-29 | 2013-08-07 | 浙江大学 | Rapid soil organic matter detection method based on large sample soil visible-near infrared spectrum classification |
CN104990890A (en) * | 2015-06-24 | 2015-10-21 | 中国农业大学 | Nondestructive testing and automated sorting system for solid single-grains and solid single-grain sorting method |
CN105043998A (en) * | 2015-05-29 | 2015-11-11 | 中国农业大学 | Method for identifying corn haploid |
CN106251347A (en) * | 2016-07-27 | 2016-12-21 | 广东工业大学 | subway foreign matter detecting method, device, equipment and subway shield door system |
-
2017
- 2017-06-15 CN CN201710455303.6A patent/CN107037001A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101031789A (en) * | 2004-07-07 | 2007-09-05 | 孔布雷电子公司-Selco | Optical component for observing a nanometric sample, system comprising same, analysis method using same, and uses thereof |
CN102760197A (en) * | 2011-04-26 | 2012-10-31 | 电子科技大学 | Prediction method for spectroscopic test data of cancer patients by partial least square method based on Matlab |
CN103116804A (en) * | 2013-01-31 | 2013-05-22 | 苏州科技学院 | Fuzzy neural network model and intelligent prediction method for deep excavation deformation |
CN103234922A (en) * | 2013-03-29 | 2013-08-07 | 浙江大学 | Rapid soil organic matter detection method based on large sample soil visible-near infrared spectrum classification |
CN105043998A (en) * | 2015-05-29 | 2015-11-11 | 中国农业大学 | Method for identifying corn haploid |
CN104990890A (en) * | 2015-06-24 | 2015-10-21 | 中国农业大学 | Nondestructive testing and automated sorting system for solid single-grains and solid single-grain sorting method |
CN106251347A (en) * | 2016-07-27 | 2016-12-21 | 广东工业大学 | subway foreign matter detecting method, device, equipment and subway shield door system |
Non-Patent Citations (1)
Title |
---|
覃鸿 等: "光照强度和光阑孔径对近红外漫透射光谱鉴别单倍体玉米影响研究", 《红外技术》 * |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110090809A (en) * | 2018-01-31 | 2019-08-06 | 西派特(北京)科技有限公司 | A kind of device of online high-speed lossless automatic sorting monoploid corn kernel |
CN108509998A (en) * | 2018-03-30 | 2018-09-07 | 中国科学院半导体研究所 | A kind of transfer learning method differentiated on different devices for target |
CN109142269A (en) * | 2018-07-26 | 2019-01-04 | 江苏大学 | A kind of method for quick identification of chilled beef difference storage time |
CN109324188A (en) * | 2018-10-11 | 2019-02-12 | 珠海沃姆电子有限公司 | A kind of precision dynamic urine measurement method and system |
CN109324188B (en) * | 2018-10-11 | 2022-04-08 | 珠海沃姆电子有限公司 | Accurate dynamic urine measurement method and system |
CN109766742A (en) * | 2018-11-20 | 2019-05-17 | 安徽农业大学 | A kind of Corn Seeds method for crack, device, system, equipment and storage medium |
CN109540831A (en) * | 2019-01-25 | 2019-03-29 | 中国中医科学院中药研究所 | Fructus lycii variety ecotype method based on high light spectrum image-forming technology |
CN110020679A (en) * | 2019-03-25 | 2019-07-16 | 中国科学院半导体研究所 | Classification method and device based on one-way analysis of variance selection bloom spectrum wavelength |
CN110208211A (en) * | 2019-07-03 | 2019-09-06 | 南京林业大学 | A kind of near infrared spectrum noise-reduction method for Detecting Pesticide |
CN110208211B (en) * | 2019-07-03 | 2021-10-22 | 南京林业大学 | Near infrared spectrum noise reduction method for pesticide residue detection |
CN114062305A (en) * | 2021-10-15 | 2022-02-18 | 中国科学院合肥物质科学研究院 | Single grain variety identification method and system based on near infrared spectrum and 1D-In-Resnet network |
CN114062305B (en) * | 2021-10-15 | 2024-01-26 | 中国科学院合肥物质科学研究院 | Single grain variety identification method and system based on near infrared spectrum and 1D-In-Resnet network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107037001A (en) | A kind of corn monoploid seed discrimination method based on near-infrared spectrum technique | |
Feilhauer et al. | Multi-method ensemble selection of spectral bands related to leaf biochemistry | |
Ientilucci et al. | Advances in wide-area hyperspectral image simulation | |
Sohn et al. | Supervised and unsupervised spectral angle classifiers | |
CN105117688B (en) | Face identification method based on Texture Feature Fusion and SVM | |
Zhang et al. | Classification modeling method for near‐infrared spectroscopy of tobacco based on multimodal convolution neural networks | |
CN102324047B (en) | Hyper-spectral image ground object recognition method based on sparse kernel representation (SKR) | |
Kale et al. | A research review on hyperspectral data processing and analysis algorithms | |
CN105352895A (en) | Hyperspectral remote sensing data vegetation information extraction method | |
CN110502997A (en) | Method and relevant device are surveyed in agricultural insurance based on artificial intelligence | |
Wang et al. | A discriminative tensor representation model for feature extraction and classification of multispectral LiDAR data | |
CN104298999B (en) | EO-1 hyperion feature learning method based on recurrence autocoding | |
Pu et al. | An attention-driven convolutional neural network-based multi-level spectral–spatial feature learning for hyperspectral image classification | |
CN103235095A (en) | Water-injected meat detection method and device | |
CN104732244A (en) | Wavelet transform, multi-strategy PSO (particle swarm optimization) and SVM (support vector machine) integrated based remote sensing image classification method | |
Huang et al. | Detection of scab in wheat ears using in situ hyperspectral data and support vector machine optimized by genetic algorithm | |
Zhou et al. | Identification of the variety of maize seeds based on hyperspectral images coupled with convolutional neural networks and subregional voting | |
CN112766161B (en) | Hyperspectral target detection method based on integrated constraint multi-example learning | |
CN110276746A (en) | A kind of robustness method for detecting change of remote sensing image | |
CN111523587B (en) | Woody plant species spectrum identification method based on machine learning | |
CN103278467A (en) | Rapid nondestructive high-accuracy method with for identifying abundance degree of nitrogen element in plant leaf | |
Zhang et al. | Hyperspectral image classification using an unsupervised neuro-fuzzy system | |
CN110031414A (en) | A kind of multi-layer perception (MLP) high spectrum mineral classification method based on spectral absorption index | |
Gao et al. | Diagnosis of maize chlorophyll content based on hybrid preprocessing and wavelengths optimization | |
CN104463210A (en) | Polarization SAR image classification method based on object orienting and spectral clustering |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170811 |