CN107271372A - A kind of Apple Leaves chlorophyll remote sensing estimation method - Google Patents
A kind of Apple Leaves chlorophyll remote sensing estimation method Download PDFInfo
- Publication number
- CN107271372A CN107271372A CN201710408574.6A CN201710408574A CN107271372A CN 107271372 A CN107271372 A CN 107271372A CN 201710408574 A CN201710408574 A CN 201710408574A CN 107271372 A CN107271372 A CN 107271372A
- Authority
- CN
- China
- Prior art keywords
- chlorophyll
- red
- model
- coefficient
- apple leaves
- 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
- 229930002875 chlorophyll Natural products 0.000 title claims abstract description 100
- 235000019804 chlorophyll Nutrition 0.000 title claims abstract description 100
- ATNHDLDRLWWWCB-AENOIHSZSA-M chlorophyll a Chemical compound C1([C@@H](C(=O)OC)C(=O)C2=C3C)=C2N2C3=CC(C(CC)=C3C)=[N+]4C3=CC3=C(C=C)C(C)=C5N3[Mg-2]42[N+]2=C1[C@@H](CCC(=O)OC\C=C(/C)CCC[C@H](C)CCC[C@H](C)CCCC(C)C)[C@H](C)C2=C5 ATNHDLDRLWWWCB-AENOIHSZSA-M 0.000 title claims abstract description 100
- 238000000034 method Methods 0.000 title claims abstract description 24
- 238000001228 spectrum Methods 0.000 claims abstract description 49
- 238000013528 artificial neural network Methods 0.000 claims abstract description 43
- 230000003595 spectral effect Effects 0.000 claims abstract description 25
- 238000012545 processing Methods 0.000 claims abstract description 5
- 238000004611 spectroscopical analysis Methods 0.000 claims description 12
- 238000012360 testing method Methods 0.000 claims description 10
- 230000001537 neural effect Effects 0.000 claims description 7
- 238000010276 construction Methods 0.000 claims description 5
- 238000005259 measurement Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 5
- 210000005036 nerve Anatomy 0.000 claims description 4
- 238000012952 Resampling Methods 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 3
- 238000007405 data analysis Methods 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000007811 spectroscopic assay Methods 0.000 claims description 3
- 230000001360 synchronised effect Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 abstract description 9
- 238000004422 calculation algorithm Methods 0.000 abstract description 7
- 238000002310 reflectometry Methods 0.000 abstract description 7
- 238000010219 correlation analysis Methods 0.000 abstract description 4
- 238000003062 neural network model Methods 0.000 description 11
- 235000013399 edible fruits Nutrition 0.000 description 10
- 235000011430 Malus pumila Nutrition 0.000 description 6
- 235000015103 Malus silvestris Nutrition 0.000 description 6
- 241000196324 Embryophyta Species 0.000 description 5
- 244000141359 Malus pumila Species 0.000 description 5
- 238000003556 assay Methods 0.000 description 5
- 230000000875 corresponding effect Effects 0.000 description 5
- 238000011160 research Methods 0.000 description 5
- 238000011161 development Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000007935 neutral effect Effects 0.000 description 4
- 239000002689 soil Substances 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 230000002596 correlated effect Effects 0.000 description 3
- 238000009826 distribution Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 description 2
- 241000219998 Philenoptera violacea Species 0.000 description 2
- 241000209140 Triticum Species 0.000 description 2
- 235000021307 Triticum Nutrition 0.000 description 2
- 240000008042 Zea mays Species 0.000 description 2
- 235000005824 Zea mays ssp. parviglumis Nutrition 0.000 description 2
- 235000002017 Zea mays subsp mays Nutrition 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 229920002678 cellulose Polymers 0.000 description 2
- 239000001913 cellulose Substances 0.000 description 2
- 235000005822 corn Nutrition 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 239000002420 orchard Substances 0.000 description 2
- 239000004069 plant analysis Substances 0.000 description 2
- 230000007480 spreading Effects 0.000 description 2
- 238000003892 spreading Methods 0.000 description 2
- 238000007619 statistical method Methods 0.000 description 2
- 239000002028 Biomass Substances 0.000 description 1
- 208000003643 Callosities Diseases 0.000 description 1
- 206010020649 Hyperkeratosis Diseases 0.000 description 1
- 244000081841 Malus domestica Species 0.000 description 1
- 240000007594 Oryza sativa Species 0.000 description 1
- 235000007164 Oryza sativa Nutrition 0.000 description 1
- 241001464837 Viridiplantae Species 0.000 description 1
- 241000607479 Yersinia pestis Species 0.000 description 1
- 238000010521 absorption reaction Methods 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 239000010242 baoji Substances 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 210000003608 fece Anatomy 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003455 independent Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 239000010871 livestock manure Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000007620 mathematical function Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 210000004218 nerve net Anatomy 0.000 description 1
- 229910052757 nitrogen Inorganic materials 0.000 description 1
- 238000010899 nucleation Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000000243 photosynthetic effect Effects 0.000 description 1
- 239000000049 pigment Substances 0.000 description 1
- 238000001556 precipitation Methods 0.000 description 1
- 238000000985 reflectance spectrum Methods 0.000 description 1
- 235000009566 rice Nutrition 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000013519 translation 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
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
- G01J3/00—Spectrometry; Spectrophotometry; Monochromators; Measuring colours
- G01J3/28—Investigating the spectrum
- G01J3/2823—Imaging spectrometer
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
- G01J3/00—Spectrometry; Spectrophotometry; Monochromators; Measuring colours
- G01J3/28—Investigating the spectrum
- G01J3/2823—Imaging spectrometer
- G01J2003/2826—Multispectral imaging, e.g. filter imaging
Landscapes
- Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention discloses a kind of Apple Leaves chlorophyll remote sensing estimation method, Apple Leaves hyper spectral reflectance and corresponding chlorophyll relative content are synchronously obtained using SVC HR 1024i type bloom spectrometers and the chlorophyll meters of SPAD 502, correlation analysis is carried out to original spectrum reflectivity and first derivative spectrum, extract the Red-edge parameter of Apple Leaves spectrum, use traditional single argument regression algorithm, BP neural network and radial primary function network, and processing is optimized to artificial neural network, set up chlorophyll content inverse model.Work network model is compared traditional univariate model inversion accuracy and significantly improved, and radial primary function network pace of learning is fast, precision is high, and fitting result is relatively reliable, is a kind of chlorophyll content inverse model being worthy to be popularized.
Description
Technical field
The invention belongs to agricultural technology field, it is related to a kind of Apple Leaves chlorophyll remote sensing estimation method.
Background technology
Vegetation chlorophyll content has preferable correlation with photosynthetic capacity, growth and development stage and nitrogen level,
Have become a kind of means of effective evaluation vegetation growing way.Because the spectral reflectivity of green plants is green by leaf in visible light wave range
Cellulose content influences, and is dominated by blade construction and cellulose etc. near infrared band is main, thus can with the reflectance spectrum of plant come
Estimate pigment content.In recent years, high-spectrum remote-sensing is so that its spectral resolution is high, simple and effective and the advantages of non-destructive, into
To monitor a developing direction of vegetation chlorophyll content.
Red side is due to that vegetation is multiple in blade interior near infrared band light in red spectral band absorption strong to chlorophyll
Strong reflection formed by scattering, makes spectral reflectivity be steeply risen in 680~760nm intervals, the plant spectral of formation it is most aobvious
Principal coordinates will.Correlation of the scholars on the one hand between chlorophyll and Red edge position is started with, and proposes " red side " to long wave direction
Displacement reflects the increase of vegetation chlorophyll concentration；On the other hand the chlorophyll based on Red-edge parameter is built using statistical analysis to estimate
Model is calculated, and precision evaluation is carried out to it.Influenceed by conditions such as field management, fertilising and weathers, scholars recognize tradition
Statistical analysis technique estimation chlorophyll content precision of prediction is not high, it is necessary to build the estimation mould that a Stability and veracity has both
Type；The plant chlorophyll based on Red-edge parameter, which is estimated, simultaneously carried out correlative study on the crops such as some corns, wheat, but in fruit
Application on tree is relatively fewer.
It is a kind of non-linear relation between chlorophyll and spectral reflectance parameter, two kinds of parameters are with what kind of mathematical function relationship
It is the important content for building Chlorophyll inversion model and its precision of prediction to set up contact, and scholars continuously attempt to various letters in recent years
Number relation, wherein artificial neural network are a kind of effective means for handling complex nonlinear problem.With high spectrum resolution remote sensing technique
Development, artificial neural network more and more be applied to high-spectrum remote-sensing research in.Such as utilize artificial nerve network model base
Soil parameters is predicted in soil spectrum, preferable effect is achieved；Artificial nerve network model is set up based on EO-1 hyperion to predict
Wheat, paddy rice, seeding corn and other crops biochemical parameter.Apple is the important industrial crops in the Northwest, and its yield accounts for national apple
More than half of yield.Apple Leaves chlorophyll content is the important channel for monitoring apple production and liquid manure situation, using artificial
Neutral net, which carries out the prediction of apple chlorophyll, has important theory and application value.
The content of the invention
It is an object of the invention to provide a kind of Apple Leaves chlorophyll remote sensing estimation method, this method utilizes SVC HR-
1024i type bloom spectrometers and SPAD-502 chlorophyll meters synchronously obtain Apple Leaves hyper spectral reflectance and corresponding chlorophyll phase
To content, correlation analysis is carried out to original spectrum reflectivity and first derivative spectrum, the Red-edge parameter of Apple Leaves spectrum is extracted,
Use traditional single argument regression algorithm, BP (Back Propagation) neutral nets and RBF (Radial Basis
Function, RBF) network, and processing is optimized to artificial neural network, set up chlorophyll content inverse model.
Its concrete technical scheme is：
A kind of Apple Leaves chlorophyll remote sensing estimation method, comprises the following steps：
Step 1, spectroscopic data are determined
Blade is carried out using the portable bloom spectrometer of U.S.'s SCVHR 1024i types to diffuse the measure of modal data.Spectrum
The wave-length coverage that instrument is determined is 350~2500nm, and port number is 1024, wherein 350~1000nm intervals spectral resolution is
1.4nm, 1000~1850nm interval spectral resolution are 3.8nm, and 1850~2500nm intervals spectral resolution is 2.4nm.Survey
Fixed work is carried out indoors, is carried out every time before sample spectra measure, and instrumental correction, every group of blade are carried out using diffusing reflection reference plate
1 leaf is selected, first blade face is cleaned.Then spectroscopic assay is carried out.Each point, which is determined 2 times, averages as the light
Measured value is composed, each blade takes 3~5 points, finally takes the average value of each point as the spectral value of the blade.
Step 2, measuring chlorophyll content
SPAD (the Soil Plant Analysis Development produced using Japanese KONICA MINOLTA companies
Unit) the synchronous relative chlorophyll content to Apple Leaves of 502 chlorophyll meters carries out nondestructive measurement.In every leaf diverse location
10 SPAD values of upper measure, the SPAD values averaged as the leaf represent the chlorophyll content of the blade, and 500 are obtained altogether
The SPAD values of blade.
Step 3, data processing
The processing software ViewSpec Pro 6.0 carried before data analysis using spectrometer are by the Apple Leaves of measure
Spectroscopic data carries out resampling, and the setting sampling interval is 1nm.Spectroscopic data is carried out using Savitzky-Golay smothing filterings
Pretreatment, it is 5 to set smooth points, and the first derivative spectra is tried to achieve by original spectrum, and blade is extracted from the first derivative spectra
Red-edge parameter；
Step 4, model construction and accuracy test
Using unitary is linear, index, logarithm, multinomial and power function build chlorophyll content in leaf blades and every Red-edge parameter
Common regression model, selection correlation is good, precision is high Red-edge parameter is used as independent variable, builds chlorophyll content estimation mould
Type；The Red-edge parameter selected out builds the artificial neural network based on Red-edge parameter as the input vector of artificial neural network
Model.
Further, in step 3, the Red-edge parameter of blade, including Red edge position, it is red in amplitude, it is red while area, coefficient of kurtosis
And the coefficient of skewness.
Further, in step 3,286 groups of spectroscopic datas therein are have selected altogether and SPAD values are analyzed and researched.Modeled
Cheng Zhong, 236 groups of random screening is used to model in 286 groups of data, and another 50 groups are used to verify.
Further, in step 4, two kinds of artificial neural network selection BP neural network and RBF neural.
Further, it is testing model precision in step 4, model predication value and measured value is subjected to regression fit, using logical
The coefficient of determination (R2), root-mean-square error (RMSE), relative error (RE) verify model accuracy.
Compared with prior art, beneficial effects of the present invention：
What is behaved oneself best in the univariate model that the present invention is set up is Red edge position, coefficient of kurtosis and the coefficient of skewness, correlation
Coefficient is all higher than 0.7；These three variables are chosen as the input variable of artificial neural network, BP neural network optimal models is set up
Predicted value and measured value between the linear regression coefficient of determination to be linear between 0.916, RBF neural network forecasts value and measured value
It is 0.939 to return the coefficient of determination, and precision of prediction is up to 96.57%.Artificial network's model compares traditional univariate model inverting essence
Degree is significantly improved, and radial primary function network pace of learning is fast, precision is high, and fitting result is relatively reliable, is that one kind is worthy to be popularized
Chlorophyll content inverse model.
Brief description of the drawings
Fig. 1 is the correlation of chlorophyll content and original spectrum and first derivative spectrum；
Fig. 2 is the Apple Leaves chlorophyll content inverse model based on the coefficient of skewness；
Fig. 3 is the comparison of Apple Leaves inverse model measured value based on the coefficient of skewness and predicted value；
Fig. 4 is the comparison of optimal BP neural network model actual measurement value and predicted value；
Fig. 5 is the comparison of optimal RBF neural network model measured value and predicted value.
Embodiment
Technical scheme is described in more detail with specific embodiment below in conjunction with the accompanying drawings.
1 materials and methods
1.1 research area's overviews
Selection Baoji, Shaanxi province city Fufeng County Xing Lin towns orchard is research object, and the ground is located at 34 ° 19 ' 02 "~34 ° 23 '
33 " between N, 107 ° 54 ' 45 "~108 ° 02 ' 40 " E, 470~570m of height above sea level, continental semi-moist monsoon climate, the four seasons point are belonged to
It is bright, annual sunshine 2134.3h, 12.4 DEG C of average temperature of the whole year, mean annual precipitation 591.9mm.It is richness that orchard, which is located at fruit variety,
Scholar.It is in April, 2015-September to gather the date, and the florescence (April 27) of corresponding fruit tree, young fruit period (May 30), fruit are swollen respectively
Big phase (July 6), fruit color phase (August 5 days), fructescence (September 11 days).According to research area's distribution situation, in garden
Selection is representative, growth is uniform, the age of tree is close and the apple tree without pest and disease damage is sampled point, and 25 fruit trees are chosen altogether and are made
For research sample.In random method when taking blade sample, every fruit tree takes leaf color to have 4 groups of blades of notable difference, every group of collection 4
Piece leaf, each issue of collection, 100 groups of leaves, gathers 500 groups of blades, carries out spectrum and measuring chlorophyll content altogether.
1.2 determine project and method
1.2.1 spectroscopic data is determined
Blade is carried out using the portable bloom spectrometer of U.S.'s SCVHR 1024i types to diffuse the measure of modal data.Spectrum
The wave-length coverage that instrument is determined is 350~2500nm, and port number is 1024, wherein 350~1000nm intervals spectral resolution is
1.4nm, 1000~1850nm interval spectral resolution are 3.8nm, and 1850~2500nm intervals spectral resolution is 2.4nm.Survey
Fixed work is carried out indoors, is carried out every time before sample spectra measure, and instrumental correction, every group of blade are carried out using diffusing reflection reference plate
1 leaf is selected, first blade face is cleaned.Then spectroscopic assay is carried out.Each point, which is determined 2 times, averages as the light
Measured value is composed, each blade takes 3~5 points, finally takes the average value of each point as the spectral value of the blade.
1.2.2 measuring chlorophyll content
SPAD (the Soil Plant Analysis Development produced using Japanese KONICA MINOLTA companies
Unit) the synchronous relative chlorophyll content to Apple Leaves of 502 chlorophyll meters carries out nondestructive measurement.Numerous studies show, blade
Greenness (SPAD) value has significant correlation with chlorophyll content, and SPAD values can be directly as sign chlorophyll concentration
Relative value.10 SPAD values are determined on every leaf diverse location, the SPAD values averaged as the leaf represent the blade
Chlorophyll content, altogether obtain 500 blades SPAD values.
1.3 data processing
The processing software ViewSpec Pro 6.0 carried before data analysis using spectrometer are by the Apple Leaves of measure
Spectroscopic data carries out resampling, and the setting sampling interval is 1nm.Using Savitzky-Golay (SG) smothing filtering to spectroscopic data
Pre-processed, it is 5 to set smooth points.The first derivative spectra is tried to achieve by original spectrum, leaf is extracted from the first derivative spectra
The Red-edge parameter of piece, including Red edge position, it is red in amplitude, it is red while area, coefficient of kurtosis and the coefficient of skewness (table 1).
The definition of the Red-edge parameter of table 1 and algorithm
Table1 Definition and algorithm of red edge parameter
R (λ) is the spectral reflectivity of any wavelength；R ' (λ) is the corresponding derivative spectrum of any wavelength；E (X) is vector X
Desired value, μ be vector X average value, σ for vector X standard deviation.
R(λ)is the spectral reflectance of any wavelength,R′(λ)is the
corresponding first derivative spectra.E(X)is the expected value of vector X,
μis the average value of vector X,σis the standard deviation of vector X.
Due to the measurement result error of some samples is larger or spectroscopic data is weak value, the present invention have selected it altogether
In 286 groups of spectroscopic datas and SPAD values analyzed and researched.In modeling process, 236 groups of use of random screening in 286 groups of data
In modeling, another 50 groups are used to verify.
1.4 model constructions and accuracy test
Using unitary is linear, index, logarithm, multinomial and power function build chlorophyll content in leaf blades and every Red-edge parameter
Common regression model, selection correlation is good, precision is high Red-edge parameter is used as independent variable, builds chlorophyll content estimation mould
Type；The Red-edge parameter selected out builds the artificial neural network based on Red-edge parameter as the input vector of artificial neural network
Model.
Artificial neural network selects two kinds of BP neural network and RBF neural.BP neural network is a kind of multilayer feedforward
Neutral net, the network is mainly characterized by before signal to transmission, error back propagation.Forward direction transmission in, input signal from
Input layer is successively handled through hidden layer, until one layer of neuron state under the influence of output layer, each layer of neuron state.Such as
Fruit output layer cannot get desired output, then be transferred to backpropagation, network weight and threshold value be adjusted according to predicated error, so that BP
Neural network prediction output constantly approaches desired output.BP neural network is made up of input layer, hidden layer and output layer, hidden layer
The number of plies and nodes have large effect to BP neural network precision of prediction；RBF (RBF, Radical Basic
Function) it is a kind of hyperspace interpolation technique, the same BP neural network of structure, basic thought is as hidden unit with RBF
" base " constitutes implicit sheaf space, and hidden layer enters line translation to input vector, and it is empty that the pattern input data of low-dimensional is transformed into higher-dimension
In, make the linear separability in higher dimensional space of the nonlinear problem in lower dimensional space, spreading coefficient SPREAD selection is to knot
Fruit has certain influence.
For testing model precision, model predication value and measured value are subjected to regression fit, using the general coefficient of determination
(R2), root-mean-square error (RMSE), relative error (RE) verify model accuracy.
2 results and analysis
The correlation of 2.1 Apple Leaves chlorophyll contents and original spectrum and first derivative spectrum
Correlation analysis is carried out with Apple Leaves chlorophyll content to original spectrum and first derivative spectrum at each wavelength
(Fig. 1), it can be seen that Apple Leaves chlorophyll and original spectrum reflectivity are in 0.01 pole in 400nm~760nm wave-length coverages
It is significantly correlated, since coefficient correlation after 700nm reduce, to 810nm at coefficient correlation close to 0, i.e., from ripple later 810nm
Section hardly reflects chlorophyll information；Coefficient correlation is between -0.8 in 520nm~620nm, 695nm~725nm wave-length coverages
Between~-0.9, maximum is reached at 554nm, is 0.899.Apple Leaves chlorophyll content exists with the first derivative spectra
Extremely notable negative correlation is reached in 435nm~550nm, 675nm~710nm wave-length coverages, 555nm~670nm, 715nm~
Reach extremely notable positive correlation in 890nm wavelength bands, after 890nm in wavelength band coefficient correlation be gradually reduced and fluctuate compared with
Greatly, wherein 575nm~585nm, coefficient correlation is higher than 0.9 in 725nm~765nm wavelength bands, and maximum is reached at 733nm
Value, is 0.921.Compared with primary light spectrum, first derivative spectrum value and chlorophyll content correlation are higher.
The 2.2 Apple Leaves chlorophyll content single argument appraising models based on Red-edge parameter
Corresponding Red-edge parameter is extracted from the first derivative spectra of original spectrum.These Red-edge parameters and plant it is various
Physical and chemical parameter is closely related, and the factor such as chlorophyll content, vegetation coverage, canopy structure, leaf area index, biomass is all
The change of Red edge position and amplitude can be caused, the present invention only does related divide to the chlorophyll content of blade aspect with Red-edge parameter
Analysis, as a result as shown in table 2.
Coefficient correlation between the Red-edge parameter of table 2 and Apple Leaves chlorophyll content
Table2 Correlation coefficients between red edge parameter and
chlorophyll content of apple leaf
* represents that correlation reaches the pole level of signifiance
indicates high significance at 0.01 level
From table 2, the correlation of Apple Leaves chlorophyll content and five kinds of Red-edge parameters reaches the pole level of signifiance.
The more commonly used Red edge position, it is red in amplitude, it is red while three parameters of area in, Red edge position with chlorophyll content horizontal relationship most
To be close, coefficient correlation reaches 0.799, and it is red in amplitude and it is red while area coefficient correlation it is relatively low, this and previous karyotype studies one
Cause.It is that, in order to preferably describe the distribution of crest, the degree of bias is reflection data distribution pair to introduce coefficient of kurtosis and the coefficient of skewness
The statistic of title property, kurtosis is the statistic for the intensity for reflecting data near average, and coefficient of kurtosis is same with the coefficient of skewness
Chlorophyll content in leaf blades is in extremely notable positive correlation, and it is feasible that chlorophyll content appraising model is set up with this.By comparing, with
Correlation level is foundation, selection Red edge position, coefficient of kurtosis and coefficient of skewness inverting chlorophyll content in leaf blades, sets up single argument
Linear processes chlorophyll content appraising model.
Apple Leaves chlorophyll content single argument estimation models and assay of the table 3 based on Red-edge parameter
Table 3 Estimation models and test results of single variable for
chlorophyll content in apple leaf based on red edge parameter
Institute's established model and its precision such as table 3, all model coefficient correlations of foundation are all higher than 0.7, correlation considerably beyond
The confidential interval upper limit, reaches extremely significantly correlated level.The model set up based on Red edge position, coefficient correlation is led to more than 0.8
The error and the result of comparison model are crossed, it is larger and root-mean-square error and relative error are smaller for standard with coefficient correlation, it is many
Although a formula model coefficient correlation is maximum, error is also very big, so the advantage of unitary linear model and logarithmic model is bigger；And
The result error is more than modeling error, illustrates that the suitability of model has much room for improvement.Set up based on coefficient of kurtosis and the coefficient of skewness
Model (because coefficient of kurtosis and coefficient of skewness part trend data contain null value, therefore logarithm and power function model can not be set up), phase
Relation number reached between 0.7-0.8, it is extremely significantly correlated, wherein, model and the result of the multinomial model in foundation
In all behave oneself best, and the result coefficient correlation and error level all close to modeling accuracy, illustrate that model stability is preferable.
Overall apparently the model set up based on the coefficient of skewness is behaved oneself best, and accuracy is higher than other two Red-edge parameters, and the result
Coefficient correlation is more than 0.86, and error is also minimum, and Model suitability is good.From coefficient of skewness inverting Apple Leaves leaf
The multinomial inverse model (Fig. 2) of chlorophyll contents, is verified to model, comparison such as Fig. 3 of measured value and predicted value, equation
Regression coefficient reaches 0.8598, and the coefficient of determination is 0.8715, and both correlations are high.
The 2.3 Apple Leaves chlorophyll content estimations based on BP neural network
The present invention sets up BP neural network using MATLAB 11.0, and output layer is Apple Leaves chlorophyll content, input layer
For Red edge position, coefficient of kurtosis and the coefficient of skewness.BP neural network is trained using training data, makes network that there is predictive ability,
The plan of BP neural network is analyzed with the model prediction output valve trained, and by comparison model prediction output and desired output
Conjunction ability.Influence of the node in hidden layer, network type of BP neural network to model accuracy is very big, and table 4 and table 5 are respectively not
With the precision and assay of nodes and network classification estimation models.
The different node in hidden layer BP neural network estimation models of table 4 and assay
Table 4 Estimation models and test results of BP neural network model
with different node of hidden layer
As can be seen from Table 4, the precision of BP neural network increases afterwards as the increase of node in hidden layer presents first to reduce
Trend, when nodes are 5, the coefficient of determination of model is maximum, and error is minimum, and the result accuracy is high, and model performance is steady
It is fixed.
The heterogeneous networks classification BP neural network estimation models of table 5 and assay
Table 4 Estimation models and test results of BP neural network model
with different network type
As can be seen from Table 5, the more single hidden layer BP neural network of double hidden layer BP neural networks, the coefficient of determination is improved,
Relative error and root-mean-square error reduce, and precision of prediction increases, but amplitude is little.Contrast nodes for 4 and 5 it is double implicit
Layer estimation models, consider from neural network accuracy and on the training time, and nodes behave oneself best for 4 double hidden layer models, and
The result Stability and veracity is all very high, is optimal BP neural network Apple Leaves chlorophyll content estimation models.With
This model is estimated that the relation (Fig. 4) between predicted value and measured value is fine in each value area prediction effect, is a kind of leaf
The effective ways of green plain lossless estimation.
The 1.5 Apple Leaves chlorophyll content appraising models based on RBF neural
RBF neural is built using the newrbe functions in MATLAB 11.0, Apple Leaves chlorophyll content is as defeated
Outgoing vector, Red edge position, coefficient of kurtosis, the coefficient of skewness are as output vector, and spreading coefficient SPREAD values are set to acquiescence, use RBF
Neural network model studies the inverting relation of Red-edge parameter and chlorophyll content.Drawn by model construction, training and inspection
As a result, the training pattern coefficient of determination is 0.943, and relative error is 4.41%, and root-mean-square error is 2.825, and testing model is determined
Coefficient is 0.909, and root-mean-square error is 2.500, and relative error is 4.25%, has higher stability than BP neural network.
Be less than the general principle of the typical range between input vector according to SPREAD values, determine SPREAD values should be set to [0,
1] between, the RBF neural network model estimation result such as table 6 of different SPREAD values.
The lower RBF neural estimation models of the difference SPREAD of table 6 values and assay
Table 4 Estimation models and test results of RBF neural network
model with different SPREAD value
As can be seen from Table 6, SPREAD values are smaller, coefficient of determination increase, and root-mean-square error and relative error gradually subtract
Small, the model of foundation is more accurate；But when SPREAD values are less than after 0.6, the result precise decreasing illustrates that network performance becomes
Difference, occurred in that adaptation, so determining when SPREAD values are 0.6, the RBF neural of foundation is estimation Apple Leaves
The best model of chlorophyll content, the coefficient of determination reaches 0.955, and side is 2.517 with error, and relative error is 3.69%, and
Prediction effect accuracy is high, and model stability is good.Fig. 5 is that the optimal Apple Leaves chlorophyll content based on RBF drawn out is estimated
Survey the 1 of model predication value and measured value:1 graph of a relation, comparison diagram 3 and Fig. 4 are visible, compared with conventional model and BP neural network model,
RBF network degrees of fitting are higher, and with speed faster, the characteristics of algorithm is easy.
2 discuss
By analyzing related pass between Apple Leaves original spectrum and first derivative spectrum and chlorophyll content in the present invention
System, first derivative spectrum correlation is higher than original spectrum, consistent with previous karyotype studies, analyzes its reason, it may be possible to utilize micro-
Point technology reduces the influence of background noise, therefore the correlation of differential smoothing and vegetation Physiological And Biochemical Parameters is more preferable.From differential light
The Red edge position that is extracted in spectrum, it is red in amplitude, it is red while three parameters of area in, Red edge position with chlorophyll content horizontal relationship most
To be close, red side is the most distinctive marks of plant spectral, when chlorophyll content increase, and Red edge position can be inclined to long wave direction
Move, when chlorophyll content is reduced, Red edge position then changes round about, Red edge position is red side regional change most fast ripple
Section, thus Red edge position it is redder in amplitude and it is red while area it is more sensitive to chlorophyll content reacting condition, be a kind of effective
Chlorophyll content indicates parameter.
Introduce the chlorophyll content appraising model stability and precision of the two parameters foundation of coefficient of kurtosis and the coefficient of skewness all
Higher than the model set up based on Red edge position, it may be possible to by the information content that coefficient of kurtosis and the coefficient of skewness are included is bigger.Cause
The overall variation of spectral shape in the range of red side can be embodied in itself for coefficient of kurtosis and the coefficient of skewness, and spectral shape changes
Just because of caused by the change of chlorophyll content, so coefficient of kurtosis and the coefficient of skewness can also be used as estimation leaf chlorophyll
The parameter of content.Multinomial model is better than other models in institute's established model, and this is probably due to every Red-edge parameter and chlorophyll
More complicated non-linear relation is there is between content, linear relationship is difficult to reflection, and it is essential.
Artificial neural network is a kind of effective means for handling complex nonlinear problem, high, non-linear with concurrency
Mapping ability is strong, self-learning capability and the features such as strong adaptive ability so that neutral net has very strong unascertained information
Disposal ability.It is of the invention to be found by analyzing two kinds of artificial nerve network models of BP and RBF compared with traditional univariate model,
Neural network model has higher Stability and veracity, and can solve the problem that while there are multiple independents variable and multiple because becoming
The problem of amount, a kind of feasible method for estimating chlorophyll content in leaf blades can be used as.BP neural network algorithm is substantially control
The difference of reality output and desired output, is repeated using input sample forward-propagating and error back propagation this process, right
A kind of supervised learning algorithm that each layer connection weight of network and the threshold value of each node are successively adjusted from the front to the back.So BP is neural
Network performance depends on primary condition, and convergence rate is slow, and less stable, learning process is easily absorbed in local minimum, and randomness
By force, BP neural network precision is made to be affected.And RBF networks overcome the shortcoming of feedforward neural network (such as BP neural network),
It should determine that, exported independent of the good characteristic such as initial weight and pace of learning be fast with Adaptation of structure, can be at utmost
Ground approaches actual measured value.But at the same time it is noted that adjusting SPREAD values and network structure, in order to avoid cause over-fitting.
3 conclusions
The present invention using apple in Shaanxi province as subjects, to Apple Leaves original spectrum reflectivity, first derivative spectrum and
Its Red-edge parameter has carried out correlation analysis with chlorophyll content, sets up traditional single argument regression model based on Red-edge parameter, uses
Red edge position, coefficient of kurtosis, the coefficient of skewness set up chlorophyll content BP neural network model and RBF network appraising models, and compare
The precision of more several models, draws to draw a conclusion：
(1) original spectrum is compared, first derivative spectrum and chlorophyll content correlation are higher, and maximum is reached at 733nm
Value 0.921；In Apple Leaves Red-edge parameter, Red edge position, coefficient of kurtosis and the coefficient of skewness and chlorophyll content correlation compared with
Good, the coefficient of determination is above 0.7.
(2) simulation precision of multinomial model is most in the single argument regression model that Red-edge parameter and chlorophyll content are set up
Height, by comparing, the multinomial model set up based on the coefficient of skewness is the best-estimated model, and the result coefficient of determination is
0.872, root-mean-square error is 4.631, and relative error is 8.81%.
(3) BP neural network and RBF nerve nets are built using Red edge position, coefficient of kurtosis and the coefficient of skewness as input variable
Two kinds of models of network carry out inverting to Apple Leaves chlorophyll content, and model accuracy is higher than traditional single argument regression model；After optimization
The RBF neural network model the result coefficient of determination be 0.939, root-mean-square error is 2.009, and relative error is 3.44%,
Pace of learning is fast, precision is high, and fitting result is reliable, is the best model for estimating Apple Leaves chlorophyll content.
The foregoing is only a preferred embodiment of the present invention, protection scope of the present invention not limited to this, any ripe
Those skilled in the art are known in the technical scope of present disclosure, the letter for the technical scheme that can be become apparent to
Altered or equivalence replacement are each fallen within protection scope of the present invention.
Claims (5)
1. a kind of Apple Leaves chlorophyll remote sensing estimation method, it is characterised in that comprise the following steps：
Step 1, spectroscopic data are determined
Blade is carried out using the portable bloom spectrometer of U.S.'s SCVHR 1024i types to diffuse the measure of modal data；Spectrometer is surveyed
Fixed wave-length coverage is 350~2500nm, and port number is 1024, wherein 350~1000nm intervals spectral resolution is
1.4nm, 1000~1850nm interval spectral resolution are 3.8nm, and 1850~2500nm intervals spectral resolution is 2.4nm；Survey
Fixed work is carried out indoors, is carried out every time before sample spectra measure, and instrumental correction, every group of blade are carried out using diffusing reflection reference plate
1 leaf is selected, first blade face is cleaned；Then spectroscopic assay is carried out；Each point, which is determined 2 times, averages as the light
Measured value is composed, each blade takes 3~5 points, finally takes the average value of each point as the spectral value of the blade；
Step 2, measuring chlorophyll content
The synchronous relative chlorophyll to Apple Leaves of SPAD502 chlorophyll meters produced using Japanese KONICA MINOLTA companies
Content carries out nondestructive measurement；10 SPAD values of measure on every leaf diverse location, the SPAD values averaged as the leaf,
The chlorophyll content of the blade is represented, the SPAD values of 500 blades are obtained altogether；
Step 3, data processing
The processing software ViewSpec Pro 6.0 carried before data analysis using spectrometer are by the Apple Leaves spectrum of measure
Data carry out resampling, and the setting sampling interval is 1nm；Spectroscopic data is located in advance using Savitzky-Golay smothing filterings
Reason, it is 5 to set smooth points, and the first derivative spectra is tried to achieve by original spectrum, and the red side of blade is extracted from the first derivative spectra
Parameter；
Step 4, model construction and accuracy test
Using unitary is linear, index, logarithm, multinomial and power function build the general of chlorophyll content in leaf blades and items Red-edge parameter
Logical regression model, the selection Red-edge parameter that correlation is good, precision is high builds chlorophyll content appraising model as independent variable；Discriminate
The Red-edge parameter selected builds the artificial nerve network model based on Red-edge parameter as the input vector of artificial neural network.
2. Apple Leaves chlorophyll remote sensing estimation method according to claim 1, it is characterised in that in step 3, blade
Red-edge parameter, including Red edge position, it is red in amplitude, it is red while area, coefficient of kurtosis and the coefficient of skewness.
3. Apple Leaves chlorophyll remote sensing estimation method according to claim 1, it is characterised in that in step 3, altogether selection
286 groups of spectroscopic datas therein and SPAD values are analyzed and researched；In modeling process, the random screening 236 in 286 groups of data
Group is used to model, and another 50 groups are used to verify.
4. Apple Leaves chlorophyll remote sensing estimation method according to claim 1, it is characterised in that in step 4, artificial god
Two kinds of BP neural network and RBF neural are selected through network.
5. Apple Leaves chlorophyll remote sensing estimation method according to claim 1, it is characterised in that in step 4, to examine
Model accuracy, carries out regression fit, using general coefficient of determination R by model predication value and measured value2, root-mean-square error
RMSE, relative error RE verify the accuracy of model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710408574.6A CN107271372A (en) | 2017-06-02 | 2017-06-02 | A kind of Apple Leaves chlorophyll remote sensing estimation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710408574.6A CN107271372A (en) | 2017-06-02 | 2017-06-02 | A kind of Apple Leaves chlorophyll remote sensing estimation method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107271372A true CN107271372A (en) | 2017-10-20 |
Family
ID=60064439
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710408574.6A Pending CN107271372A (en) | 2017-06-02 | 2017-06-02 | A kind of Apple Leaves chlorophyll remote sensing estimation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107271372A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108896176A (en) * | 2018-05-14 | 2018-11-27 | 浙江大学 | A kind of Space Consistency bearing calibration of multi-optical spectrum imaging system |
| CN109564155A (en) * | 2016-08-17 | 2019-04-02 | 索尼公司 | Signal processing apparatus, signal processing method and program |
| CN110658174A (en) * | 2019-08-27 | 2020-01-07 | 厦门谱识科仪有限公司 | Intelligent identification method and system based on surface enhanced Raman spectrum detection |
| CN111044516A (en) * | 2019-12-26 | 2020-04-21 | 沈阳农业大学 | Remote sensing estimation method for chlorophyll content of rice |
| CN111678599A (en) * | 2020-07-07 | 2020-09-18 | 安徽大学 | Laser spectrum noise reduction method and device based on deep learning optimization S-G filtering |
| CN112070234A (en) * | 2020-09-04 | 2020-12-11 | 中国科学院南京地理与湖泊研究所 | Ground-based remote sensing machine learning algorithm for chlorophyll and phycocyanin in water body under complex scene |
| CN116660206A (en) * | 2023-05-31 | 2023-08-29 | 浙江省农业科学院 | Crop yield estimation method and system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105277491A (en) * | 2015-09-24 | 2016-01-27 | 中国农业科学院农业资源与农业区划研究所 | Chlorophyll content measurement method and apparatus thereof |
| CN106442338A (en) * | 2016-09-06 | 2017-02-22 | 西北农林科技大学 | Hyperspectral inversion method for content of chlorophyll in apple leaves based on SVR (support vector regression) algorithm |
| CN106469240A (en) * | 2016-09-06 | 2017-03-01 | 西北农林科技大学 | Rape leaf SPAD estimation based on spectral index and estimation models construction method |
-
2017
- 2017-06-02 CN CN201710408574.6A patent/CN107271372A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105277491A (en) * | 2015-09-24 | 2016-01-27 | 中国农业科学院农业资源与农业区划研究所 | Chlorophyll content measurement method and apparatus thereof |
| CN106442338A (en) * | 2016-09-06 | 2017-02-22 | 西北农林科技大学 | Hyperspectral inversion method for content of chlorophyll in apple leaves based on SVR (support vector regression) algorithm |
| CN106469240A (en) * | 2016-09-06 | 2017-03-01 | 西北农林科技大学 | Rape leaf SPAD estimation based on spectral index and estimation models construction method |
Non-Patent Citations (3)
| Title |
|---|
| 姚付启等: "基于红边参数的植被叶绿素含量高光谱估算模型", 《农业工程学报》 * |
| 朱西存等: "基于高光谱红边参数的不同物候期苹果叶片SPAD的值估测", 《红外》 * |
| 李媛媛等: "基于高光谱和BP 神经网络的玉米叶片SPAD 值遥感估算", 《农业工程学报》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109564155A (en) * | 2016-08-17 | 2019-04-02 | 索尼公司 | Signal processing apparatus, signal processing method and program |
| CN109564155B (en) * | 2016-08-17 | 2022-03-11 | 索尼公司 | Signal processing device, signal processing method, and program |
| CN108896176A (en) * | 2018-05-14 | 2018-11-27 | 浙江大学 | A kind of Space Consistency bearing calibration of multi-optical spectrum imaging system |
| CN110658174A (en) * | 2019-08-27 | 2020-01-07 | 厦门谱识科仪有限公司 | Intelligent identification method and system based on surface enhanced Raman spectrum detection |
| CN111044516A (en) * | 2019-12-26 | 2020-04-21 | 沈阳农业大学 | Remote sensing estimation method for chlorophyll content of rice |
| CN111678599A (en) * | 2020-07-07 | 2020-09-18 | 安徽大学 | Laser spectrum noise reduction method and device based on deep learning optimization S-G filtering |
| CN112070234A (en) * | 2020-09-04 | 2020-12-11 | 中国科学院南京地理与湖泊研究所 | Ground-based remote sensing machine learning algorithm for chlorophyll and phycocyanin in water body under complex scene |
| CN112070234B (en) * | 2020-09-04 | 2024-01-30 | 中国科学院南京地理与湖泊研究所 | Water chlorophyll and phycocyanin land-based remote sensing machine learning algorithm under complex scene |
| CN116660206A (en) * | 2023-05-31 | 2023-08-29 | 浙江省农业科学院 | Crop yield estimation method and system |
| CN116660206B (en) * | 2023-05-31 | 2024-05-28 | 浙江省农业科学院 | Crop yield estimation method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107271372A (en) | A kind of Apple Leaves chlorophyll remote sensing estimation method | |
| Yang et al. | Winter wheat SPAD estimation from UAV hyperspectral data using cluster-regression methods | |
| CN107103306B (en) | Winter wheat powdery mildew remote sensing monitoring method based on wavelet analysis and support vector machine | |
| Zhang et al. | A cloud computing-based approach using the visible near-infrared spectrum to classify greenhouse tomato plants under water stress | |
| Zhang et al. | Allocate soil individuals to soil classes with topsoil spectral characteristics and decision trees | |
| CN107271382A (en) | A kind of different growing rape leaf SPAD value remote sensing estimation methods | |
| Guo et al. | Estimating leaf chlorophyll content in tobacco based on various canopy hyperspectral parameters | |
| CN111044516B (en) | Remote sensing estimation method for chlorophyll content of rice | |
| CN110119169A (en) | A kind of tomato greenhouse temperature intelligent early warning system based on minimum vector machine | |
| CN104778349B (en) | One kind is used for rice table soil nitrogen application Classified Protection | |
| CN110569605A (en) | Non-glutinous rice leaf nitrogen content inversion model method based on NSGA2-ELM | |
| CN103278467A (en) | Rapid nondestructive high-accuracy method with for identifying abundance degree of nitrogen element in plant leaf | |
| CN110779875B (en) | Method for detecting moisture content of winter wheat ear based on hyperspectral technology | |
| CN110320164A (en) | A kind of method for building up of romaine lettuce total nitrogen content EO-1 hyperion inverse model and its application | |
| Zhang et al. | Hyperspectral model based on genetic algorithm and SA-1DCNN for predicting Chinese cabbage chlorophyll content | |
| CN106568722A (en) | Spectrum technology-based facility cucumber disease early warning method, and device | |
| Gai et al. | Spectroscopic determination of chlorophyll content in sugarcane leaves for drought stress detection | |
| Sharabiani et al. | Application of soft computing methods and spectral reflectance data for wheat growth monitoring | |
| CN113065230A (en) | High-spectrum inversion model for establishing rice leaf SPAD based on optimized spectral index | |
| CN116151454A (en) | Method and system for predicting yield of short-forest linalool essential oil by multispectral unmanned aerial vehicle | |
| CN116883874A (en) | Evaluation method and system for comprehensive growth vigor of cinnamomum camphora dwarf forest | |
| CN101726742A (en) | Remote sensing measurement method for level of crops pollution stress | |
| CN107064069A (en) | A kind of actinidia tree chlorophyll content in leaf blades EO-1 hyperion evaluation method | |
| Li et al. | HSI combined with CNN model detection of heavy metal Cu stress levels in apple rootstocks | |
| Zhai et al. | Stability evaluation of the PROSPECT model for leaf chlorophyll content retrieval |
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: 20171020 |
|
| RJ01 | Rejection of invention patent application after publication |