CN110398466A - Crop growth state monitoring method based on remote-sensing inversion - Google Patents
Crop growth state monitoring method based on remote-sensing inversion Download PDFInfo
- Publication number
- CN110398466A CN110398466A CN201910717679.9A CN201910717679A CN110398466A CN 110398466 A CN110398466 A CN 110398466A CN 201910717679 A CN201910717679 A CN 201910717679A CN 110398466 A CN110398466 A CN 110398466A
- Authority
- CN
- China
- Prior art keywords
- vegetation index
- variable
- remote
- crop growth
- state monitoring
- 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 26
- 238000012544 monitoring process Methods 0.000 title claims abstract description 22
- 230000003595 spectral effect Effects 0.000 claims abstract description 48
- 238000001228 spectrum Methods 0.000 claims abstract description 38
- 238000007637 random forest analysis Methods 0.000 claims abstract description 20
- 229930002875 chlorophyll Natural products 0.000 claims abstract description 19
- 235000019804 chlorophyll Nutrition 0.000 claims abstract description 19
- 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 claims abstract description 19
- 235000007164 Oryza sativa Nutrition 0.000 claims abstract description 17
- 235000009566 rice Nutrition 0.000 claims abstract description 17
- 238000003066 decision tree Methods 0.000 claims abstract description 12
- 238000005259 measurement Methods 0.000 claims abstract description 10
- 238000010219 correlation analysis Methods 0.000 claims abstract description 8
- 238000000605 extraction Methods 0.000 claims abstract description 7
- 238000000611 regression analysis Methods 0.000 claims abstract description 7
- 239000000284 extract Substances 0.000 claims abstract description 6
- 238000012549 training Methods 0.000 claims abstract description 4
- 240000007594 Oryza sativa Species 0.000 claims abstract 4
- 238000004364 calculation method Methods 0.000 claims description 21
- 238000012545 processing Methods 0.000 claims description 16
- 238000012937 correction Methods 0.000 claims description 15
- 238000002310 reflectometry Methods 0.000 claims description 11
- 241001062009 Indigofera Species 0.000 claims description 6
- 230000009286 beneficial effect Effects 0.000 claims description 3
- 239000012535 impurity Substances 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 244000283207 Indigofera tinctoria Species 0.000 abstract 1
- 239000012141 concentrate Substances 0.000 abstract 1
- 241000209094 Oryza Species 0.000 description 13
- 230000001133 acceleration Effects 0.000 description 4
- 238000007726 management method Methods 0.000 description 4
- 239000002689 soil Substances 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000006872 improvement Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 239000002028 Biomass Substances 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 235000013339 cereals Nutrition 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000035558 fertility Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000010899 nucleation Methods 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000002271 resection Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000003971 tillage Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 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
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/3563—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light for analysing solids; Preparation of samples therefor
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/359—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light using near infrared light
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- 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
- G01N2021/1793—Remote sensing
Abstract
The invention discloses the crop growth state monitoring methods based on remote-sensing inversion, comprising the following steps: obtains the Airborne Hyperspectral data in area to be monitored;Calculate the spectrum parameter of the indigo plant side Huang Hongsan respective positions, slope and area;Calculate EO-1 hyperion vegetation index;Rice leaf chlorophyll content and leaf area are measured, and the chlorophyll content of actual measurement and leaf area and spectrum parameter and EO-1 hyperion vegetation index are subjected to correlation analysis;Random forest based on study carries out rice chlorophyll regression analysis and modeling;S spectrum samples are extracted out at random with putting back to from EO-1 hyperion spectrum training sample data concentration, concentrates from spectral signature variable data and randomly extracts t variable out, and s sample of extraction and t variable are subjected to operation, form an individual decision tree;It repeats this step X times, constructs the Random Forest model set with X;Prediction classification is carried out to new data respectively using X tree, comprehensive final vote result obtains prediction result.High-efficient when the present invention is used for crop growth monitoring, precision is high.
Description
Technical field
The present invention relates to crops monitoring technology, specifically the crop growth state monitoring method based on remote-sensing inversion.
Background technique
Since crop seeding range is wide, wide in variety, yield is influenced by extraneous factor, and (such as: tillage method, farmland are managed
The factors such as reason, weather, soil fertility can all influence the yield of crops), in order to improve the yield of crops, people are usually right
The growth conditions of crops are monitored management, in order to make regulation in time.Traditional crop growth status monitoring management
The main experience accumulated over a long period by people is realized, is not easy to grasp Grain Growth Situation in real time, is carrying out large area and monitor to manage
When reason, traditional monitoring management method there is low efficiency, low precision, it is costly the deficiencies of, be unable to satisfy modern agricultural development
Requirement, do not adapt to the society of current high speed development.
Summary of the invention
It is an object of the invention to solve the problems, such as conventional crop Growing state survey, there are low efficiencys and low precision, provide
A kind of crop growth state monitoring method based on remote-sensing inversion can promote precision compared with traditional monitoring management method when applying
And efficiency.
The purpose of the present invention is achieved through the following technical solutions: the crop growth status monitoring based on remote-sensing inversion
Method, including the following steps successively carried out:
Step 1, the Airborne Hyperspectral data for obtaining area to be monitored;
Step 2, the spectrum parameter for calculating the indigo plant side Huang Hongsan respective positions, slope and area;
Step 3 calculates EO-1 hyperion vegetation index;
Step 4, measurement rice full heading time chlorophyll content in leaf blades and leaf area, and by the chlorophyll content of actual measurement and blade face
The EO-1 hyperion vegetation index that the long-pending spectrum parameter being calculated with step 2 and step 3 are calculated carries out correlation analysis;
Step 5, the random forest based on study carry out rice chlorophyll regression analysis and modeling;
Step 6 extracts s spectrum samples out at random from EO-1 hyperion spectrum training sample data concentration with putting back to, from spectrum
It randomly extracts t variable in characteristic variable data set out, s sample of extraction and t variable is subjected to operation, form a list
Only decision tree;It repeats this step X times, constructs the Random Forest model set with X;
Step 7 carries out prediction classification to new data respectively using X tree, and comprehensive final vote result obtains prediction result.
The present invention is in application, using Airborne Data Classification, and using the different spectral signatures of crops blade, calculating is not shared the same light
Compose parameter, according to spectrum parameter it is blue while, it is yellow while and phase between Red edge position difference, with the growing ways index such as chlorophyll content of rice
Guan Xing determines sensitive band and effective spectral index, and utilizes the random forest tree method inverting crop growth shape based on study
State.Vegetation index is used to qualitative and quantitative assessment vegetative coverage and its growth vigor, and the index is rapid with the increase of biomass
Increase.Vegetation index mainly reflects the index of vegetation difference between visible light, near infrared band reflection and Soil Background, each
Vegetation index can be used to quantitatively illustrate the upgrowth situation of vegetation under certain condition.
Further, the step 1 further includes pre-processing to the Airborne Hyperspectral data of acquisition, wherein pretreatment
Including radiant correction, sensor attitude data processing, GPS positioning data processing, attitude data and location data time synchronization with
Integrated, geometric correction.
Further, position, slope and the area on indigo plant side are determined in the step 2 in 432-465nm wave-length coverage;In
Position, slope and the area on Huang side are determined in 565-582nm wave-length coverage;Red side is determined in 680-750nm wave-length coverage
Position, slope and area.
Further, the EO-1 hyperion vegetation index includes ratio vegetation index, difference vegetation index and normalization vegetation
Index;
The calculation formula of ratio vegetation index are as follows:
Rλ1/Rλ2 (1)
Wherein, Rλ1For the spectral reflectance values at 800nm, Rλ2Spectral reflectance values between 620nm-680nm;
The calculation formula of difference vegetation index are as follows:
Rλ3-Rλ4 (2)
Wherein, Rλ3Spectral reflectance values between 760nm-1200nm, Rλ4Spectrum between 620nm-680nm is anti-
Radiance rate value;
The calculation formula of normalized differential vegetation index are as follows:
(Rλ5-Rλ6)/(Rλ1+Rλ2) (3)
Wherein, Rλ5Spectral reflectance values between 760nm-800nm, Rλ6Spectral reflectance between 620nm-680nm
Rate value, Rλ1For the spectral reflectance values at 800nm, Rλ2Spectral reflectance values between 620nm-680nm.
Further, in the step 4 correlation analysis related coefficient calculation formula are as follows:
Wherein, for R (λ) in the related coefficient of af at wavelength lambda between X and Y, i is sample size, the value of i is 1,2,
3 ..., n, n are total sample number, XiFor vegetation index,For vegetation index arithmetic mean of instantaneous value, YλIt is original anti-for the spectrum of wavelength X
Rate is penetrated,For the arithmetic mean of instantaneous value of spectral reflectivity first derivative.The present invention is in application, using rice object reflective spectrum and agriculture
Crop growthing state carries out correlation analysis, wherein R (λ) is the digital representation for describing correlativity between two groups of data, and value is situated between
In [- 1.0,1.0], the absolute value of R (λ) value shows that the degree being fitted with linear model is higher, shows when close to 0 closer to 1
The degree of linear fit is very poor.
Further, in the Geordie formula of node v in the model that the step 5 is established are as follows:
Wherein, p is the sum of node,It is observation of j-th of variable in node v.Random forest based on study is
The combination of more decision trees, the present invention solve rice chlorophyll and Airborne Hyperspectral data using the random forest based on study
Return inversion problem.
The Spectral Characteristic extracted from decision tree has redundancy phenomena, in order to solve this problem, further, the step
Carry out regression analysis in rapid 5 the following steps are included: judge next group of variable subset information delta and upper one group of characteristic variable
Whether information delta gap is less than given threshold, if more than or be equal to, then be selected into;If being less than, new spectral signature is not selected;
Its calculation formula is as follows:
Wherein, G (Zt, v) and it is regularization information gain, ZtIt is the son of node sum in the Geordie information delta of split vertexes v
The difference of the impurity level of node;D is the set of the aspect indexing for a upper node split, is in the root node of one tree
A empty set;K ∈ (0,1] it is penalty coefficient;WhenWhen, coefficient punishment is used for t-th of feature of split vertexes v;K is smaller,
Punishment dynamics are bigger.The present invention by by the information delta of the information delta of next group of variable subset and upper one group of characteristic variable into
Row comparison is carried out pair by the selection of penalty term and the similar characteristic variable of spectral signature that last time decision tree divides generation
Than when gap is smaller, random forest tree would not repeat to select new spectral signature, to effectively prevent the superfluous of spectral signature
Remaining phenomenon.When each node recursively carries out the division of a spectral signature variable, to the strategy of random forest Applied Learning,
To select the spectral signature subset of compression.
Further, s sample of extraction and t variable operation is carried out in the step 6 to transport using formula (6)
It calculates.
Further, the step 6 is further comprising the steps of after forming an individual decision tree: by each
Node utilizes regularization information gain G (Zt, v), when spectral signature variable is that existing variable increases enough predictive information
When, the index of the new variables is added in set D;Wherein, ZtImportance scores are as follows:
Wherein, n is the quantity of decision tree in random forest tree,It is in the random forest set at n by ZtThe section of division
Point set.
Further, comprehensive final vote result acquisition prediction result passes through calibration set coefficient of determination R in the step 72
Model of fit is evaluated with calibration set root-mean-square error RMSE two indices to choose, wherein calibration set coefficient of determination R2Calculating
Formula are as follows:
The calculation formula of calibration set root-mean-square error RMSE are as follows:
Wherein, yiIt is measured value, y* iIt is predicted value,It is the average value of measured value.One good model means more
High calibration set coefficient of determination R2, lower calibration set root-mean-square error RMSEc.In this way, prediction result of the invention is by commenting
Determine calibration set coefficient of determination R2The stabilization of model accuracy and model is confirmed with calibration set root-mean-square error RMSEc two indices
Property, final vote is determined by screening model as a result,
In conclusion compared with the prior art, the invention has the following beneficial effects: the present invention using based on study with
Machine forest number calculates the importance of single spectral signature, can screen to the importance spectrum parameter variable in research process,
Entire research process is controlled by important spectrum parameter variable.Random Forest model based on study is the group of more forest trees
It closes, compared to single tree model, it, which only needs to establish a model, can be thus achieved characteristic wave bands selection, can greatly improve fortune
Line efficiency and precision.
Detailed description of the invention
Attached drawing described herein is used to provide to further understand the embodiment of the present invention, constitutes one of the application
Point, do not constitute the restriction to the embodiment of the present invention.In the accompanying drawings:
Fig. 1 is the flow chart of a specific embodiment of the invention.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention clearer, below with reference to embodiment and attached drawing, to this
Invention is described in further detail, and exemplary embodiment of the invention and its explanation for explaining only the invention, are not made
For limitation of the invention.
Embodiment:
As shown in Figure 1, the crop growth state monitoring method based on remote-sensing inversion, including the following step successively carried out
It is rapid: step 1, the Airborne Hyperspectral data for obtaining area to be monitored;Step 2 calculates the indigo plant side Huang Hongsan respective positions, slope and face
Long-pending spectrum parameter;Step 3 calculates EO-1 hyperion vegetation index;Step 4, measurement rice full heading time chlorophyll content in leaf blades and leaf
Area, and the height that spectrum parameter and step 3 that the chlorophyll content of actual measurement and leaf area are calculated with step 2 are calculated
Spectral vegetation indexes carry out correlation analysis;Step 5, the random forest based on study carry out rice chlorophyll regression analysis and build
Mould;Step 6 extracts s spectrum samples out at random from EO-1 hyperion spectrum training sample data concentration with putting back to, and becomes from spectral signature
T variable is randomly extracted out in amount data set, and s sample of extraction and t variable are subjected to operation, form one individually certainly
Plan tree;It repeats this step X times, constructs the Random Forest model set with X;Step 7 is set using X respectively to new data
Prediction classification is carried out, comprehensive final vote result obtains prediction result.
The step 1 of the present embodiment further includes pre-processing to the Airborne Hyperspectral data of acquisition, Airborne Hyperspectral data
Pretreatment mainly includes radiant correction and geometric correction two large divisions, and specific flow chart of data processing mainly includes five steps
It is rapid: radiant correction, sensor attitude data processing, GPS positioning data processing, attitude data and location data time synchronization and collection
At, geometric correction.Wherein, radiant correction, sensor attitude data processing, GPS positioning data processing, attitude data and positioning number
The prior art is all made of according to time synchronization and integrated and geometric correction to realize.Radiant correction: it rejects and respectively visits member by detector
Noise and band existing for subband caused by non_uniform response.When machine upscaling system cannot work very well or cannot be complete
When totally disappeared except various bands, need to realize relative detector calibration based on the statistical method of image using some.To aviation bloom
The lower wave band of modal data signal-to-noise ratio does radiation intensification processing, that is, includes Histogram Matching, histogram stretches, bad track is filled up, gone
Except the methods of Banded improvement processing.Sensor attitude data processing: due to acceleration (such as determinand move when can generate acceleration
Degree, vibration can generate acceleration etc. when motor operation) and some shake etc. can generate noise jamming, sensor attitude data processing
By single order complement arithmetic, acceleration and angular speed is come together to do, merges out an angle value by six number of axle evidences, pass through four
The fusion of first number (quaternary number includes the variation of cartesian coordinate system, and there are three angle values for the inside), to remove sensor itself
The noise of generation.GPS positioning data processing: the basic principle of GPS positioning is the satellite instantaneous position conduct according to high-speed motion
Known known date determines the position of tested point using the method for space length resection.GPS data processing will be from original
The observation of beginning set out to obtain final measurement and positioning as a result, data handling procedure be roughly divided into the baseline of GPS measurement data to
Amount resolves, GPS basic lineal vector net adjusted data and GPS network adjustment or with several stages such as terrestrial network simultaneous adjustment.Geometric correction: disappear
Except the geometry deformation in Airborne Hyperspectral image, the space geometry process of imaging has been avoided in fine geometric correction, it is believed that aviation is high
The overall geometry distortion of spectrum picture is to squeeze, and distortion scales, the result of offset and other deformation comprehensive functions.
Position, slope and the area on indigo plant side are determined in the step 2 of the present embodiment in 432-465nm wave-length coverage;In
Position, slope and the area on Huang side are determined in 565-582nm wave-length coverage;Red side is determined in 680-750nm wave-length coverage
Position, slope and area.Red edge position is that wavelength corresponding to first derivative spectrum maximum value, red side slope are in red range
The maximum value of reflectivity first derivative spectrum, the area that red side area is surrounded by reflectivity first derivative spectrum line;Lan Bianwei
Be set to wavelength corresponding to first derivative spectrum maximum value in blue light range, blue side slope be reflectivity first derivative spectrum most
Big value, the area that blue side area is surrounded by reflectivity first derivative spectrum line;Yellow side position is first derivative in red range
Wavelength corresponding to spectral maximum, yellow side slope are the maximum value of reflectivity first derivative spectrum, and yellow side area is reflectivity
The area that first derivative spectrum line is surrounded.
The EO-1 hyperion vegetation index of the present embodiment includes ratio vegetation index (RVI), difference vegetation index (DVI) and normalizing
Change vegetation index (NDVI) and be used as EO-1 hyperion vegetation index, index calculation method is as follows:
The calculation formula of ratio vegetation index are as follows:
Rλ1/Rλ2 (1)
Wherein, Rλ1For the spectral reflectance values at 800nm, Rλ2Spectral reflectance values between 620nm-680nm;
The calculation formula of difference vegetation index are as follows:
Rλ3-Rλ4 (2)
Wherein, Rλ3Spectral reflectance values between 760nm-1200nm, Rλ4Spectrum between 620nm-680nm is anti-
Radiance rate value;
The calculation formula of normalized differential vegetation index are as follows:
(Rλ5-Rλ6)/(Rλ1+Rλ2) (3)
Wherein, Rλ5Spectral reflectance values between 760nm-800nm, Rλ6Spectral reflectance between 620nm-680nm
Rate value, Rλ1For the spectral reflectance values at 800nm, Rλ2Spectral reflectance values between 620nm-680nm.
Satellite visible and near infrared band are combined by vegetation index according to the spectral characteristic of vegetation, are formd each
Kind vegetation index.Vegetation index is simple, effective and experience measurement to earth's surface vegetation state.Ratio vegetation index is RVI,
Vegetation index mainly reflects the index of vegetation difference between visible light, near infrared band reflection and Soil Background, vegetative coverage
Degree influences RVI, and when vegetation coverage is higher, RVI is very sensitive to vegetation;When vegetation coverage < 50%, this sensibility
It significantly reduces;When vegetation is dense, sensitivity can also be reduced.Wherein, Rλ1For reflected value (the near-infrared wave spectrum model of near infrared band
Enclose: 760nm-3000nm, the spectral range of Airborne Hyperspectral: 380nm-2450nm, the preferred Airborne Hyperspectral of the present embodiment it is close
Stablize spectral range: R in infrared sectionλ1For 800nm), Rλ2For reflected value (the feux rouges spectral range: 620nm- of red spectral band
760nm, the present embodiment select the interior stable spectral range near infrared region of Airborne Hyperspectral: Rλ2For 620nm-680nm).
Difference vegetation index is DVI, extremely sensitive to the variation of Soil Background, be suitable for vegetation development early, mid-term or
The vegetation of low middle coverage monitors.Wherein, Rλ3For near infrared band reflected value (near-infrared spectral range: 760nm-3000nm,
The spectral range of Airborne Hyperspectral: 380nm-2450nm, the interior neutral wave near infrared region of the preferred Airborne Hyperspectral of the present embodiment
Spectral limit: Rλ3For 760nm-1200nm), Rλ4For reflected value (the feux rouges spectral range: 620nm-760nm, this reality of red spectral band
Apply the interior stable spectral range near infrared region of example selection Airborne Hyperspectral: Rλ4For 620nm-680nm).
Normalized differential vegetation index is NDVI, in remote sensing image, the reflected value of near infrared band and the reflected value of red spectral band
Difference than upper sum of the two, especially suitable for the whole world or the vegetation dynamic and variation of ecology and environment of each department large scale, but
NDVI has lower sensitivity to high vegetation region.Wherein, Rλ3For near infrared band reflected value (near-infrared spectral range:
760nm-3000nm, the spectral range of Airborne Hyperspectral: 380nm-2450nm, the present embodiment select the near-infrared of Airborne Hyperspectral
Stablize spectral range: R in sectionλ3For 760nm-800nm), Rλ6For reflected value (the feux rouges spectral range: 620nm- of red spectral band
760nm, the present embodiment select the interior stable spectral range near infrared region of Airborne Hyperspectral: Rλ6For 620nm-680nm).
The present embodiment is in application, the chlorophyll content in leaf blades of rice full heading time is detected using SPAD-502 chlorophyll meter, water
The leaf area of rice full heading time holds the detection of Leaf area determination instrument using YMJ-B.In the rice spectrum that full heading time is tested in 350nm-
It is changed greatly between 800nm in the content correlation of different-waveband and chlorophyll, the wave band of correlation maximum is 400nm-450nm
Between, related coefficient is 0.25 or so, and related coefficient is 0.2-0.25 between 350nm-400nm, between 500nm-800nm
It is gradually reduced, in 700nm minimum 0.05.To rice spectrum carry out first differential processing, it can be found that rice first differential with
The correlation of chlorophyll is obviously higher than the correlation of reflectivity, and wherein first differential is in the section 400-450nm highest, maximum phase
Relationship number reaches 0.4 or more, lower in 650-750nm related coefficient, and up to 0.15.Preferably, the present embodiment uses one
Rank differential carries out the inversion reckoning of the rice biochemical quantity (chlorophyll content, leaf area) of full heading time.
The related coefficient calculation formula of correlation analysis in the present embodiment step 4 are as follows:
Wherein, for R (λ) in the related coefficient of af at wavelength lambda between X and Y, i is sample size, the value of i is 1,2,
3 ..., n, n are total sample number, XiFor vegetation index,For vegetation index arithmetic mean of instantaneous value, YλIt is original anti-for the spectrum of wavelength X
Rate is penetrated,For the arithmetic mean of instantaneous value of spectral reflectivity first derivative.
In the Geordie formula of node v in the model that the present embodiment step 5 is established are as follows:
Wherein, p is the sum of node,It is observation of j-th of variable in node v.At each node, at p
M (m ≈ p is randomly selected in variable1/2) a variable, m is randomly drawing sample number, the final feature for obtaining maximum information gain
Variable is used for the division of node v.
Regression analysis is carried out in the present embodiment step 5 the following steps are included: judging the information delta of next group of variable subset
Whether be less than given threshold with the information delta gap of upper one group of characteristic variable, if more than or be equal to, then be selected into;If being less than,
New spectral signature is not selected;Its calculation formula is as follows:
Wherein, G (Zt, v) and it is regularization information gain, ZtIt is the son of node sum in the Geordie information delta of split vertexes v
The difference of the impurity level of node;D is the set of the aspect indexing for a upper node split, is in the root node of one tree
A empty set;K ∈ (0,1] it is penalty coefficient;WhenWhen, coefficient punishment is used for t-th of feature of split vertexes v;K is smaller,
Punishment dynamics are bigger.S sample of extraction and t variable are subjected to operation in step 6 and carry out operation using formula (6).
The step 6 of the present embodiment is further comprising the steps of after forming an individual decision tree: by each section
Point utilizes regularization information gain G (Zt, v), when spectral signature variable is that existing variable increases enough predictive information,
The index of the new variables is added in set D;Wherein, ZtImportance scores are as follows:
Wherein, n is the quantity of decision tree in random forest tree,It is in the random forest set at n by ZtThe section of division
Point set.
Comprehensive final vote result acquisition prediction result passes through calibration set coefficient of determination R in the present embodiment step 72And correction
Collection root-mean-square error RMSE two indices evaluate model of fit to choose, wherein calibration set coefficient of determination R2Calculation formula are as follows:
The calculation formula of calibration set root-mean-square error RMSE are as follows:
Wherein, yiIt is measured value, y* iIt is predicted value,It is the average value of measured value.
Above-described specific embodiment has carried out further the purpose of the present invention, technical scheme and beneficial effects
It is described in detail, it should be understood that being not intended to limit the present invention the foregoing is merely a specific embodiment of the invention
Protection scope, all within the spirits and principles of the present invention, any modification, equivalent substitution, improvement and etc. done should all include
Within protection scope of the present invention.
Claims (10)
1. the crop growth state monitoring method based on remote-sensing inversion, which is characterized in that including the following steps successively carried out:
Step 1, the Airborne Hyperspectral data for obtaining area to be monitored;
Step 2, the spectrum parameter for calculating the indigo plant side Huang Hongsan respective positions, slope and area;
Step 3 calculates EO-1 hyperion vegetation index;
Step 4, measurement rice full heading time chlorophyll content in leaf blades and leaf area, and by the chlorophyll content of actual measurement and leaf area with
The EO-1 hyperion vegetation index that the spectrum parameter and step 3 that step 2 is calculated are calculated carries out correlation analysis;
Step 5, the random forest based on study carry out rice chlorophyll regression analysis and modeling;
Step 6 extracts s spectrum samples out at random from EO-1 hyperion spectrum training sample data concentration with putting back to, from spectral signature
Variable data, which is concentrated, randomly extracts t variable out, and s sample of extraction and t variable are carried out operation, form one individually
Decision tree;It repeats this step X times, constructs the Random Forest model set with X;
Step 7 carries out prediction classification to new data respectively using X tree, and comprehensive final vote result obtains prediction result.
2. the crop growth state monitoring method according to claim 1 based on remote-sensing inversion, which is characterized in that described
Step 1 further includes pre-processing to the Airborne Hyperspectral data of acquisition, wherein pretreatment includes radiant correction, sensor appearance
State data processing, GPS positioning data processing, attitude data and location data time synchronization and integrated, geometric correction.
3. the crop growth state monitoring method according to claim 1 based on remote-sensing inversion, which is characterized in that described
Position, slope and the area on indigo plant side are determined in step 2 in 432-465nm wave-length coverage;In 565-582nm wave-length coverage really
Position, slope and the area on fixed Huang side;Position, slope and the area on red side are determined in 680-750nm wave-length coverage.
4. the crop growth state monitoring method according to claim 1 based on remote-sensing inversion, which is characterized in that described
EO-1 hyperion vegetation index includes ratio vegetation index, difference vegetation index and normalized differential vegetation index;
The calculation formula of ratio vegetation index are as follows:
Rλ1/Rλ2 (1)
Wherein, Rλ1For the spectral reflectance values at 800nm, Rλ2Spectral reflectance values between 620nm-680nm;
The calculation formula of difference vegetation index are as follows:
Rλ3-Rλ4 (2)
Wherein, Rλ3Spectral reflectance values between 760nm-1200nm, Rλ4Spectral reflectivity between 620nm-680nm
Value;
The calculation formula of normalized differential vegetation index are as follows:
(Rλ5-Rλ6)/(Rλ1+Rλ2) (3)
Wherein, Rλ5Spectral reflectance values between 760nm-800nm, Rλ6Spectral reflectance values between 620nm-680nm,
Rλ1For the spectral reflectance values at 800nm, Rλ2Spectral reflectance values between 620nm-680nm.
5. the crop growth state monitoring method according to claim 1 based on remote-sensing inversion, which is characterized in that described
The related coefficient calculation formula of correlation analysis in step 4 are as follows:
Wherein, for R (λ) in the related coefficient of af at wavelength lambda between X and Y, i is sample size, the value of i is 1,2,3 ..., n,
N is total sample number, XiFor vegetation index,For vegetation index arithmetic mean of instantaneous value, YλFor the spectrum primary reflection rate of wavelength X,
For the arithmetic mean of instantaneous value of spectral reflectivity first derivative.
6. the crop growth state monitoring method according to claim 1 based on remote-sensing inversion, which is characterized in that described
In the Geordie formula of node v in the model that step 5 is established are as follows:
Wherein, p is the sum of node,It is observation of j-th of variable in node v.
7. the crop growth state monitoring method according to claim 1 based on remote-sensing inversion, which is characterized in that described
In step 5 carry out regression analysis the following steps are included: judge next group of variable subset information delta and upper one group of characteristic variable
Information delta gap whether be less than given threshold, if more than or be equal to, then be selected into;If being less than, new Spectral Properties are not selected
Sign;Its calculation formula is as follows:
Wherein, G (Zt, v) and it is regularization information gain, ZtIt is the child node of node sum in the Geordie information delta of split vertexes v
Impurity level difference;D is the set of the aspect indexing for a upper node split, is a sky in the root node of one tree
Collection;K ∈ (0,1] it is penalty coefficient;WhenWhen, coefficient punishment is used for t-th of feature of split vertexes v;K is smaller, punishment
Dynamics is bigger.
8. the crop growth state monitoring method according to claim 7 based on remote-sensing inversion, which is characterized in that described
S sample of extraction and t variable are subjected to operation in step 6 and carry out operation using formula (6).
9. the crop growth state monitoring method according to claim 7 based on remote-sensing inversion, which is characterized in that described
Step 6 is further comprising the steps of after forming an individual decision tree: by being increased in each node using regularization information
Beneficial G (Zt, v), when spectral signature variable is that existing variable increases enough predictive information, the index of the new variables is added
Enter into set D;Wherein, ZtImportance scores are as follows:
Wherein, n is the quantity of decision tree in random forest tree,It is in the random forest set at n by ZtThe node collection of division
It closes.
10. the crop growth state monitoring method described according to claim 1~any one of 9 based on remote-sensing inversion,
It is characterized in that, comprehensive final vote result acquisition prediction result passes through calibration set coefficient of determination R in the step 72And correction
Collection root-mean-square error RMSE two indices evaluate model of fit to choose, wherein calibration set coefficient of determination R2Calculation formula are as follows:
The calculation formula of calibration set root-mean-square error RMSE are as follows:
Wherein, yiIt is measured value, y* iIt is predicted value,It is the average value of measured value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910717679.9A CN110398466A (en) | 2019-08-05 | 2019-08-05 | Crop growth state monitoring method based on remote-sensing inversion |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910717679.9A CN110398466A (en) | 2019-08-05 | 2019-08-05 | Crop growth state monitoring method based on remote-sensing inversion |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110398466A true CN110398466A (en) | 2019-11-01 |
Family
ID=68327519
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910717679.9A Pending CN110398466A (en) | 2019-08-05 | 2019-08-05 | Crop growth state monitoring method based on remote-sensing inversion |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110398466A (en) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111079672A (en) * | 2019-12-20 | 2020-04-28 | 中国科学院遥感与数字地球研究所 | Grape classification extraction method based on maximum entropy features |
CN111175783A (en) * | 2019-12-31 | 2020-05-19 | 塔里木大学 | Satellite remote sensing monitoring method for cotton canopy chlorophyll b content |
CN111398178A (en) * | 2020-04-07 | 2020-07-10 | 中国科学院空天信息创新研究院 | Leaf chlorophyll content inversion method and device, electronic equipment and storage medium |
CN111798327A (en) * | 2020-06-24 | 2020-10-20 | 安徽大学 | Construction method and application of wheat yield calculation model based on hyperspectral image |
CN112881353A (en) * | 2021-01-11 | 2021-06-01 | 江西师范大学 | Method and device for measuring concentration of soluble organic carbon in water body |
CN113063742A (en) * | 2021-03-24 | 2021-07-02 | 和数科技(浙江)有限公司 | Method and system for measuring vegetation biomass, storage medium and terminal |
CN113063739A (en) * | 2021-02-25 | 2021-07-02 | 北京麦飞科技有限公司 | Rice canopy nitrogen content monitoring method based on airborne hyperspectral sensor |
CN113111799A (en) * | 2021-04-19 | 2021-07-13 | 北华航天工业学院 | Cultivated land soil fertility level monitoring method based on ensemble empirical mode decomposition |
CN113514402A (en) * | 2021-04-12 | 2021-10-19 | 新疆农业大学 | System and method for predicting chlorophyll content of winter wheat |
CN113537108A (en) * | 2021-07-23 | 2021-10-22 | 安徽省农业科学院土壤肥料研究所 | Crop chlorophyll content inversion method based on unmanned aerial vehicle hyperspectral remote sensing |
CN113866102A (en) * | 2021-09-24 | 2021-12-31 | 中国地质调查局西安地质调查中心(西北地质科技创新中心) | Soil health investigation monitoring method based on spectrum |
CN117853931A (en) * | 2024-03-04 | 2024-04-09 | 中国铁路设计集团有限公司 | Hyperspectral image isolated anomaly detection method based on nearest neighbor distance |
CN117853931B (en) * | 2024-03-04 | 2024-05-14 | 中国铁路设计集团有限公司 | Hyperspectral image isolated anomaly detection method based on nearest neighbor distance |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108614992A (en) * | 2018-03-19 | 2018-10-02 | 中国地质大学(武汉) | A kind of sorting technique of high-spectrum remote sensing, equipment and storage device |
-
2019
- 2019-08-05 CN CN201910717679.9A patent/CN110398466A/en active Pending
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108614992A (en) * | 2018-03-19 | 2018-10-02 | 中国地质大学(武汉) | A kind of sorting technique of high-spectrum remote sensing, equipment and storage device |
Non-Patent Citations (8)
Title |
---|
HOUTAO DENG等: "Gene Selection With Guided Regularized Random Forest", 《PATTERN RECOGNITION》 * |
吴俊: "基于基因表达水平的分子网络研究", 《万方数据学位论文》 * |
周纪芗等: "《质量管理统计方法》", 31 October 2008, 中国统计出版社 * |
张影等: "《预测与评价》", 31 May 2015, 天津大学出版社 * |
武旭梅等: "水稻冠层叶绿素含量高光谱估算模型", 《干旱地区农业研究》 * |
潘岑岑: "基于随机森林的土壤重金属高光谱遥感反演研究", 《中国优秀博硕士学位论文全文数据库(硕士) 工程科技Ⅰ辑》 * |
申时凯等: "《基于云计算的大数据处理技术发展与应用》", 31 March 2019, 电子科技大学出版社 * |
章曼: "基于高光谱遥感的水稻生长监测研究", 《中国优秀硕士学位论文全文数据库 农业科技辑》 * |
Cited By (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111079672A (en) * | 2019-12-20 | 2020-04-28 | 中国科学院遥感与数字地球研究所 | Grape classification extraction method based on maximum entropy features |
CN111079672B (en) * | 2019-12-20 | 2023-10-13 | 中国科学院空天信息创新研究院 | Grape classification extraction method based on maximum entropy characteristics |
CN111175783A (en) * | 2019-12-31 | 2020-05-19 | 塔里木大学 | Satellite remote sensing monitoring method for cotton canopy chlorophyll b content |
CN111398178B (en) * | 2020-04-07 | 2023-05-16 | 中国科学院空天信息创新研究院 | Leaf chlorophyll content inversion method and device, electronic equipment and storage medium |
CN111398178A (en) * | 2020-04-07 | 2020-07-10 | 中国科学院空天信息创新研究院 | Leaf chlorophyll content inversion method and device, electronic equipment and storage medium |
CN111798327A (en) * | 2020-06-24 | 2020-10-20 | 安徽大学 | Construction method and application of wheat yield calculation model based on hyperspectral image |
CN112881353A (en) * | 2021-01-11 | 2021-06-01 | 江西师范大学 | Method and device for measuring concentration of soluble organic carbon in water body |
CN112881353B (en) * | 2021-01-11 | 2022-11-15 | 江西师范大学 | Method and device for measuring concentration of soluble organic carbon in water body |
CN113063739A (en) * | 2021-02-25 | 2021-07-02 | 北京麦飞科技有限公司 | Rice canopy nitrogen content monitoring method based on airborne hyperspectral sensor |
CN113063742A (en) * | 2021-03-24 | 2021-07-02 | 和数科技(浙江)有限公司 | Method and system for measuring vegetation biomass, storage medium and terminal |
CN113514402A (en) * | 2021-04-12 | 2021-10-19 | 新疆农业大学 | System and method for predicting chlorophyll content of winter wheat |
CN113111799A (en) * | 2021-04-19 | 2021-07-13 | 北华航天工业学院 | Cultivated land soil fertility level monitoring method based on ensemble empirical mode decomposition |
CN113111799B (en) * | 2021-04-19 | 2024-01-30 | 北华航天工业学院 | Cultivated land soil fertility level monitoring method based on ensemble empirical mode decomposition |
CN113537108A (en) * | 2021-07-23 | 2021-10-22 | 安徽省农业科学院土壤肥料研究所 | Crop chlorophyll content inversion method based on unmanned aerial vehicle hyperspectral remote sensing |
CN113537108B (en) * | 2021-07-23 | 2024-02-23 | 安徽省农业科学院土壤肥料研究所 | Crop chlorophyll content inversion method based on unmanned aerial vehicle hyperspectral remote sensing |
CN113866102A (en) * | 2021-09-24 | 2021-12-31 | 中国地质调查局西安地质调查中心(西北地质科技创新中心) | Soil health investigation monitoring method based on spectrum |
CN113866102B (en) * | 2021-09-24 | 2023-12-22 | 中国地质调查局西安地质调查中心(西北地质科技创新中心) | Soil health investigation monitoring method based on spectrum |
CN117853931A (en) * | 2024-03-04 | 2024-04-09 | 中国铁路设计集团有限公司 | Hyperspectral image isolated anomaly detection method based on nearest neighbor distance |
CN117853931B (en) * | 2024-03-04 | 2024-05-14 | 中国铁路设计集团有限公司 | Hyperspectral image isolated anomaly detection method based on nearest neighbor distance |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110398466A (en) | Crop growth state monitoring method based on remote-sensing inversion | |
Li et al. | Multi-LUTs method for canopy nitrogen density estimation in winter wheat by field and UAV hyperspectral | |
Dobrowski et al. | Grapevine dormant pruning weight prediction using remotely sensed data | |
Liu et al. | Estimating leaf area index using unmanned aerial vehicle data: shallow vs. deep machine learning algorithms | |
Zheng et al. | Remote sensing and machine learning in crop phenotyping and management, with an emphasis on applications in strawberry farming | |
Clevers et al. | A framework for monitoring crop growth by combining directional and spectral remote sensing information | |
Moriya et al. | Mapping mosaic virus in sugarcane based on hyperspectral images | |
LI et al. | Estimating grassland LAI using the Random Forests approach and Landsat imagery in the meadow steppe of Hulunber, China | |
Popescu et al. | Remote sensing in the analysis and characterization of spatial variability of the territory. a study case in Timis County, Romania. | |
Zhu et al. | Estimation of forest leaf water content through inversion of a radiative transfer model from LiDAR and hyperspectral data | |
Hall | Remote sensing applications for viticultural terroir analysis | |
CN112861810B (en) | Artificial forest planting time automatic detection method based on time sequence remote sensing observation data | |
Lin et al. | Using the 3D model RAPID to invert the shoot dieback ratio of vertically heterogeneous Yunnan pine forests to detect beetle damage | |
Liu et al. | Estimating potato above-ground biomass by using integrated unmanned aerial system-based optical, structural, and textural canopy measurements | |
Song et al. | Estimating reed loss caused by Locusta migratoria manilensis using UAV-based hyperspectral data | |
CN113252583A (en) | Method for calculating alpine hay coverage based on hay vegetation index | |
Villacrés et al. | Construction of 3D maps of vegetation indices retrieved from UAV multispectral imagery in forested areas | |
Poudyal et al. | Prediction of morpho-physiological traits in sugarcane using aerial imagery and machine learning | |
Qin et al. | Estimation of FPAR and FPAR profile for maize canopies using airborne LiDAR | |
Solymosi et al. | The development of vegetation indices: a short overview | |
Danoedoro et al. | Combining pan-sharpening and forest cover density transformation methods for vegetation mapping using Landsat-8 Satellite Imagery | |
Zhang et al. | Aboveground wheat biomass estimation from a low-altitude UAV platform based on multimodal remote sensing data fusion with the introduction of terrain factors | |
Tamiminia et al. | Random forest outperformed convolutional neural networks for shrub willow above ground biomass estimation using multi-spectral UAS imagery | |
Dutta | Assessment of tea bush health and yield using geospatial techniques | |
Jiao et al. | The Potential of hue angle calculated based on multispectral reflectance for leaf chlorophyll content estimation |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191101 |