CN114036461A - Crop total nitrogen content inversion method and system based on hyperspectral remote sensing - Google Patents
Crop total nitrogen content inversion method and system based on hyperspectral remote sensing Download PDFInfo
- Publication number
- CN114036461A CN114036461A CN202111312193.0A CN202111312193A CN114036461A CN 114036461 A CN114036461 A CN 114036461A CN 202111312193 A CN202111312193 A CN 202111312193A CN 114036461 A CN114036461 A CN 114036461A
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- nitrogen content
- total nitrogen
- training
- hyperspectral remote
- 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
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 title claims abstract description 204
- 229910052757 nitrogen Inorganic materials 0.000 title claims abstract description 102
- 238000000034 method Methods 0.000 title claims abstract description 49
- 239000011159 matrix material Substances 0.000 claims abstract description 69
- 238000002310 reflectometry Methods 0.000 claims abstract description 67
- 238000012216 screening Methods 0.000 claims abstract description 27
- 238000012549 training Methods 0.000 claims description 85
- 238000012360 testing method Methods 0.000 claims description 34
- 238000007781 pre-processing Methods 0.000 claims description 19
- 238000005259 measurement Methods 0.000 claims description 14
- 238000012937 correction Methods 0.000 claims description 7
- 230000005855 radiation Effects 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 3
- 241000209140 Triticum Species 0.000 description 17
- 235000021307 Triticum Nutrition 0.000 description 13
- 238000012544 monitoring process Methods 0.000 description 12
- 238000005070 sampling Methods 0.000 description 9
- 238000003066 decision tree Methods 0.000 description 7
- 241000196324 Embryophyta Species 0.000 description 5
- 230000003595 spectral effect Effects 0.000 description 5
- 244000038559 crop plants Species 0.000 description 3
- VEXZGXHMUGYJMC-UHFFFAOYSA-N Hydrochloric acid Chemical compound Cl VEXZGXHMUGYJMC-UHFFFAOYSA-N 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000001066 destructive effect Effects 0.000 description 2
- 238000001035 drying Methods 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 239000002689 soil Substances 0.000 description 2
- 238000007696 Kjeldahl method Methods 0.000 description 1
- 241000234314 Zingiber Species 0.000 description 1
- 235000006886 Zingiber officinale Nutrition 0.000 description 1
- 230000009418 agronomic effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000035558 fertility Effects 0.000 description 1
- 235000008397 ginger Nutrition 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000009659 non-destructive testing Methods 0.000 description 1
- 230000029553 photosynthesis Effects 0.000 description 1
- 238000010672 photosynthesis Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06F18/2148—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the process organisation or structure, e.g. boosting cascade
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Software Systems (AREA)
- Algebra (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Operations Research (AREA)
- Probability & Statistics with Applications (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Health & Medical 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 relates to a crop total nitrogen content inversion method and system based on hyperspectral remote sensing, wherein the method comprises the following steps: acquiring a plurality of hyperspectral remote sensing images of a field to be measured by using an unmanned aerial vehicle; screening characteristic wave bands of the hyperspectral remote sensing images to obtain a reflectivity average value of a plurality of fields to be detected on a set wave band; the set waveband comprises a plurality of wavebands; constructing a reflectivity mean characteristic matrix according to the reflectivity mean; and inputting the reflectivity mean value characteristic matrix into a trained inversion model to obtain the inversion value of the total nitrogen content of the field to be detected. The invention can realize nondestructive detection of the total nitrogen content of large-area crops.
Description
Technical Field
The invention relates to the technical field of processing and application of remote sensing information, in particular to a crop total nitrogen content inversion method and system based on hyperspectral remote sensing.
Background
In traditional agriculture, relevant agronomic indexes are obtained by monitoring growth information of crops mainly through manual destructive sampling and manual measurement, and the monitoring method is difficult to be applied to large-scale crop growth monitoring, and is large in workload and high in cost. During the growth and photosynthesis of crops, the spectral characteristics of the canopy are closely related to the nitrogen content of the plants. Therefore, the spectral information of the crops can be obtained in real time by using a remote sensing means, so that the nitrogen content information of the crop plants can be indirectly obtained. Although satellite remote sensing can be used for monitoring in a large range and is low in cost, the satellite remote sensing is easily influenced by cloud layers, the running period is long, and the space-time resolution required by precision agriculture is lacked. In recent years, unmanned aerial vehicles are widely applied to various fields of agricultural monitoring at home and abroad, are flexible in operation, high in data acquisition speed and high in space-time resolution, and can be applied to complex farmland environments. Therefore, a method for nitrogen content inversion based on unmanned aerial vehicles is needed.
Disclosure of Invention
The invention aims to provide a crop total nitrogen content inversion method and system based on hyperspectral remote sensing so as to realize nondestructive detection of the total nitrogen content of large-area crops.
In order to achieve the purpose, the invention provides the following scheme:
a crop total nitrogen content inversion method based on hyperspectral remote sensing comprises the following steps:
acquiring a plurality of hyperspectral remote sensing images of a field to be measured by using an unmanned aerial vehicle;
screening characteristic wave bands of the hyperspectral remote sensing images to obtain a reflectivity average value of a plurality of fields to be detected on a set wave band; the set waveband comprises a plurality of wavebands;
constructing a reflectivity mean characteristic matrix according to the reflectivity mean;
and inputting the reflectivity mean value characteristic matrix into a trained inversion model to obtain the inversion value of the total nitrogen content of the field to be detected.
Optionally, the characteristic wave band screening is performed on the hyperspectral remote sensing image, and before obtaining a plurality of reflectance mean values of the field to be measured on the set wave band, the method further includes:
splicing the plurality of hyperspectral remote sensing images to obtain a complete hyperspectral remote sensing image of the field to be measured;
preprocessing the complete hyperspectral remote sensing image; the preprocessing comprises radiation calibration and atmospheric correction.
Optionally, the training process of the inversion model specifically includes:
acquiring a plurality of training hyperspectral remote sensing images of a training field by using an unmanned aerial vehicle;
carrying out characteristic band screening on the training hyperspectral remote sensing image to obtain a reflectivity average value on a set band of a training field;
constructing a sample set feature matrix according to the reflectivity mean value on the set waveband of the training field;
measuring the total nitrogen content of the training fields to obtain the measured value of the total nitrogen content of the training fields;
constructing a label matrix according to the total nitrogen content measurement value;
longitudinally combining and dividing the sample set characteristic matrix and the label matrix to obtain a training data set and a testing data set;
according to the training data set, taking the sample set characteristic matrix as input, taking the total nitrogen content measurement value as output, and training a set number of base learners by using a Boosting method;
carrying out weighted combination on the base learners with the set number to obtain an XGboost model;
and inverting the XGboost model according to the test data set to obtain a trained inversion model.
Optionally, the expression of the reflectance mean is:
wherein,
is the average value of the reflectivity,
for the nth wave band lambdanThe reflectivity value of the a-th pixel point in the m-th field block area; a. themThe total number of pixel points in the region of interest of the mth plot area.
A crop total nitrogen content inversion system based on hyperspectral remote sensing comprises:
the acquisition module is used for acquiring a plurality of hyperspectral remote sensing images of a field to be detected by using the unmanned aerial vehicle;
the characteristic band screening module is used for screening characteristic bands of the hyperspectral remote sensing images to obtain a reflectivity average value of a plurality of fields to be tested on a set band; the set waveband comprises a plurality of wavebands;
the reflectivity mean characteristic matrix determining module is used for constructing a reflectivity mean characteristic matrix according to the reflectivity mean;
and the total nitrogen content inversion value determining module is used for inputting the reflectivity mean value characteristic matrix into the trained inversion model to obtain the total nitrogen content inversion value of the field to be detected.
Optionally, the method further comprises:
the splicing module is used for splicing a plurality of hyperspectral remote sensing images to obtain a complete hyperspectral remote sensing image of the field to be measured;
the preprocessing module is used for preprocessing the complete hyperspectral remote sensing image; the preprocessing comprises radiation calibration and atmospheric correction.
Optionally, the method further includes a training module of the inverse model, where the training module of the inverse model specifically includes:
the acquisition unit is used for acquiring a plurality of training hyperspectral remote sensing images of a training field by using the unmanned aerial vehicle;
the characteristic band screening and screening unit is used for screening the characteristic bands of the training hyperspectral remote sensing images to obtain a reflectivity average value on a set band of a training field;
the sample set feature matrix determining unit is used for constructing a sample set feature matrix according to the reflectivity mean value on the set waveband of the training field;
the measuring unit is used for measuring the total nitrogen content of the training field to obtain a measured value of the total nitrogen content of the training field;
the tag matrix construction unit is used for constructing a tag matrix according to the total nitrogen content measurement value;
the longitudinal merging and dividing unit is used for longitudinally merging and dividing the sample set characteristic matrix and the label matrix to obtain a training data set and a test data set;
the training unit is used for training a set number of base learners by a Boosting method according to the training data set by taking the sample set characteristic matrix as input and the total nitrogen content measurement value as output;
the weighted combination unit is used for carrying out weighted combination on the base learners with the set number to obtain an XGboost model;
and the inversion unit is used for inverting the XGboost model according to the test data set to obtain a trained inversion model.
Optionally, the expression of the reflectance mean is:
wherein,
is the average value of the reflectivity,
for the nth wave band lambdanThe reflectivity value of the a-th pixel point in the m-th field block area; a. themIs of interest in the mth plot areaThe total number of pixels in the region.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
according to the crop total nitrogen content inversion method and system based on hyperspectral remote sensing, an unmanned aerial vehicle is used for obtaining a plurality of hyperspectral remote sensing images of a field to be measured; screening characteristic wave bands of the hyperspectral remote sensing images to obtain a reflectivity average value of a plurality of fields to be measured on a set wave band; the set waveband comprises a plurality of wavebands; constructing a reflectivity mean characteristic matrix according to the reflectivity mean; and inputting the reflectivity mean characteristic matrix into the trained inversion model to obtain the inversion value of the total nitrogen content of the field to be detected. The speed of acquiring the hyperspectral remote sensing image through the unmanned aerial vehicle is high, the covered field range is wide, the inversion value of the total nitrogen content of the field to be detected can be directly obtained through the processing of the hyperspectral remote sensing image, and therefore the nondestructive testing of the total nitrogen content of large-area crops is realized.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a flow chart of a crop total nitrogen content inversion method based on hyperspectral remote sensing provided by the invention;
FIG. 2 is a high spectral remote sensing image of a wheat canopy unmanned aerial vehicle of a test field obtained according to the present invention;
FIG. 3 is a graph of the model prediction results of total nitrogen content of winter wheat in the jointing stage of a part of field block areas in a test field provided by the present invention;
FIG. 4 is a graph of the validation accuracy of an inversion model provided by the present invention;
FIG. 5 is a schematic diagram of a crop total nitrogen content inversion system based on hyperspectral remote sensing provided by the invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide a crop total nitrogen content inversion method and system based on hyperspectral remote sensing so as to realize nondestructive detection of the total nitrogen content of large-area crops.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
As shown in FIG. 1, the invention provides a crop total nitrogen content inversion method based on hyperspectral remote sensing, which comprises the following steps:
step 101: and acquiring a plurality of hyperspectral remote sensing images of the field to be detected by using the unmanned aerial vehicle.
Step 102: screening characteristic wave bands of the hyperspectral remote sensing images to obtain a reflectivity average value of a plurality of fields to be detected on a set wave band; the set band includes a plurality of bands.
Step 103: and constructing a reflectivity mean characteristic matrix according to the reflectivity mean.
Step 104: and inputting the reflectivity mean value characteristic matrix into a trained inversion model to obtain the inversion value of the total nitrogen content of the field to be detected.
In practical application, before the step 101 of performing characteristic band screening on the hyperspectral remote sensing image to obtain a plurality of reflectance mean values of the field to be measured on a set band, the method further comprises:
and splicing the plurality of hyperspectral remote sensing images to obtain the complete hyperspectral remote sensing image of the field to be detected.
Preprocessing the complete hyperspectral remote sensing image; the preprocessing comprises radiation calibration and atmospheric correction.
In practical applications, the training process of the inversion model specifically includes:
and acquiring a plurality of training hyperspectral remote sensing images of a training field by using the unmanned aerial vehicle.
And screening characteristic wave bands of the training hyperspectral remote sensing images to obtain a reflectivity average value on a set wave band of a training field.
And constructing a sample set feature matrix according to the reflectivity mean value on the set wave band of the training field.
And measuring the total nitrogen content of the training fields to obtain the total nitrogen content measured value of the training fields.
And constructing a label matrix according to the total nitrogen content measurement value.
And longitudinally combining and dividing the sample set characteristic matrix and the label matrix to obtain a training data set and a testing data set.
And according to the training data set, taking the sample set characteristic matrix as input, taking the total nitrogen content measurement value as output, and training a set number of base learners by using a Boosting method.
And carrying out weighted combination on the base learners with the set number to obtain the XGboost model.
And inverting the XGboost model according to the test data set to obtain a trained inversion model.
In practical applications, the expression of the reflectance mean value is as follows:
wherein,
is the average value of the reflectivity,
for the nth wave band lambdanThe reflectivity value of the a-th pixel point in the m-th field block area; a. themIs the mth field block areaThe total number of pixels in the region of interest.
The invention also provides a concrete work flow of the crop total nitrogen content inversion method based on hyperspectral remote sensing in practical application, which comprises the following steps:
s1, acquiring hyperspectral remote sensing images of the field by using an unmanned aerial vehicle, and obtaining a reflectivity average value of each plot area on each wave band on N wave bands; constructing a sample set characteristic matrix X according to the reflectivity mean value of each field area on each wave band,
m＝1,2,…,M，n＝1,2,…,N；
for the mth field region in the nth wave band lambdanReflectance average of (c). The number of bands N of the present invention is 176.
In step S1, a GaiaSky-mini2 hyperspectral imager is carried by using a Dajiang M600PRO six-winged unmanned aerial vehicle to obtain hyperspectral remote sensing images of the wheat in the jointing stage of the Mongolian horse-shop testing station long-term positioning test field (116 degrees 37 'E, 33 degrees 13' N). The unmanned aerial vehicle acquires a plurality of hyperspectral remote sensing images of the test field, and the plurality of hyperspectral remote sensing images of the test field are spliced by HiSpectraL Stitcher software to obtain a complete hyperspectral remote sensing image of the test field. Preprocessing hyperspectral remote sensing images of a test field using SPECVIEW software, the preprocessing comprising: and carrying out radiometric calibration and atmospheric correction, and preprocessing to obtain the reflectivity value of each pixel point in the hyperspectral remote sensing image.
In step S1, an ROI may be extracted from each field region by using an ROI tool in the ENVI software, reflectance values of each pixel point in the region of interest of each field region on each band are respectively derived on N bands, a reflectance mean value of the region of interest of each field region on each band is calculated, and the reflectance mean value of the region of interest of each field region on each band is used as the reflectance mean value of the field region on each band. That is, the mth block is in the nth band λnMean value of reflectivity of
Comprises the following steps:
wherein,
for the nth wave band lambdanThe reflectivity value of the a-th pixel point in the interested area of the m-th field block area; n is 1,2, …, N; m is 1,2, …, M; a is 1,2, …, Am，AmThe total number of pixel points in the region of interest of the mth plot area.
And S2, measuring the total nitrogen content of the plants sampled in the field, dividing the field into M field areas, and obtaining the total nitrogen content measured value of each field area.
Constructing a label matrix Y according to the total nitrogen content measured value sampled in each field area, wherein Y is { Y ═ Y }m}，m＝1,2,…,M，ymThe total nitrogen content of the mth field block area is measured.
In step S2, dividing the test field into M field areas, where M is 126 in the present invention; sampling is carried out on each field block area, 3-5 plant samples are collected in each field block area, and the field blocks are packaged and fresh-keeping is carried back to a laboratory. And (3) placing the plant sample in a drying oven at 70 ℃ for constant-temperature drying, crushing, and measuring the total nitrogen content of the wheat by using a Kjeldahl method. For each sampling point, the Total Nitrogen Content (TNC) at that point was taken as the mean of 3-5 wheat plants. The calculation formula is as follows:
C＝(V×0.05×14×100)/(1000×M)
in the formula: c is total nitrogen content (%); v is the volume variable (mL) of hydrochloric acid; and screening out the total nitrogen content of the plant of one sampling point in each field area as the total nitrogen content measured value of the field area.
And S3, longitudinally combining the feature matrix and the label matrix, and randomly dividing the feature matrix and the label matrix into a training data set and a testing data set.
In step S3, randomly dividing the training data set into a training data set and a testing data set, the specific method is as follows:
and S31, column merging the feature matrix and the label matrix to obtain a matrix Instance _ matrix.
And S32, randomly dividing the matrix Instance _ matrix in a row direction by using a train _ test _ split method of Python software, thereby generating a training data set and a testing data set.
And S4, repeatedly training the base learners by using a Boosting method according to the obtained training data set until the number of the regression decision trees reaches a preset value T, and performing weighted combination on the T base learners to establish an XGboost model.
S41, a number T of basis learners is preset.
And S42, randomly extracting an initial training set from the training data set, and training to obtain the base learner.
And S43, adjusting the distribution of the training samples according to the performance of the base learner, so that the training samples which are wrongly made by the base learner are paid more attention in the following process, and the adjusted distribution of the samples is obtained.
S44, training the next base learner based on the adjusted sample distribution.
And S45, repeating the steps S43 and S44 until the number of the base learners reaches a preset value T, and performing weighted combination on the T base learners.
S5, respectively corresponding the feature matrix of each cluster set to the total nitrogen content measured value of the corresponding field block area in the tag set Y, and training to obtain an inversion model; and inputting the test data set into an XGboost model, and inverting the total nitrogen content of the crops to obtain the distribution of the total nitrogen content of the field.
The input of the inversion model is a vector formed by the reflectivity mean values of a certain field area on each wave band, namely the input is the vector of the field area, and the output of the inversion model is the inversion value of the total nitrogen content of the field area.
When the inversion model is used for prediction, a test sample is input, the test sample is a vector formed by the reflectivity mean values of a certain field block area on all wave bands, each decision tree in the XGboost regression model respectively extracts the reflectivity mean values of the corresponding characteristic wave bands in the test sample, each decision tree performs prediction according to the reflectivity mean values on the corresponding characteristic wave bands, each decision tree outputs the total nitrogen content prediction value of the field block area, the total nitrogen content prediction values of the field block area output by the decision trees are subjected to weighted summation, and the value is used as the final inversion value of the total nitrogen content of the field block area.
In the invention, a test area is in an ecological environment station (116 degrees 37 'E and 33 degrees 13' N) of Mongolian sand ginger black soil in the rural area of agriculture, a hyperspectral remote sensing image of an unmanned aerial vehicle at a wheat canopy of an obtained test field is shown in figure 2, and the test field is in a white dotted line frame. FIG. 3 is a model prediction result of the total nitrogen content of winter wheat in the jointing stage in a part of field block areas in a test field, and the result shows that the total nitrogen content of winter wheat in the jointing stage ranges from 14% to 29% as a whole, and the model can realize accurate prediction of the total nitrogen content of winter wheat plants from point to surface. FIG. 4 is a graph of the verification accuracy of the inverse model, and as shown in FIG. 4, the total nitrogen content predicted by the model and the actually measured total nitrogen content have a strong positive correlation, and a coefficient R is determined2I.e. the goodness of fit reaches 0.76.
According to the method, based on the influence of hyperspectral remote sensing of the unmanned aerial vehicle, abundant hyperspectral information of the wheat field is fully utilized, the measurement values of the total nitrogen content of the wheat plants in each field block area are combined, an inversion model is established by adopting an integrated learning algorithm of XGboost regression, the total nitrogen content of the winter wheat plants in the jointing stage is inverted, a better inversion result is obtained, the spatial distribution and the characteristics of the total nitrogen content of the winter wheat in the field of long-term positioning tests can be visually and accurately displayed, a novel and reliable method is provided for monitoring the growth information of the wheat and judging the influence of soil fertility on the total nitrogen content of the winter wheat, and the method has high practical value.
As shown in fig. 5, the crop total nitrogen content inversion system based on hyperspectral remote sensing provided by the invention comprises:
an obtaining module 501, configured to obtain multiple hyperspectral remote sensing images of a field to be measured by using an unmanned aerial vehicle.
A characteristic band screening module 502, configured to perform characteristic band screening on the hyperspectral remote sensing images to obtain a reflectance mean value of a plurality of fields to be tested in a set band; the set band includes a plurality of bands.
The reflectivity mean characteristic matrix determining module 503 is configured to construct a reflectivity mean characteristic matrix according to the reflectivity mean.
And a total nitrogen content inversion value determination module 504, configured to input the reflectance mean characteristic matrix into a trained inversion model, so as to obtain a total nitrogen content inversion value of the field to be measured.
In practical application, the crop total nitrogen content inversion system based on hyperspectral remote sensing further comprises:
and the splicing module is used for splicing the hyperspectral remote sensing images to obtain the complete hyperspectral remote sensing image of the field to be detected.
The preprocessing module is used for preprocessing the complete hyperspectral remote sensing image; the preprocessing comprises radiation calibration and atmospheric correction.
In practical application, the crop total nitrogen content inversion system based on hyperspectral remote sensing further comprises an inversion model training module, wherein the inversion model training module specifically comprises:
and the acquisition unit is used for acquiring a plurality of training hyperspectral remote sensing images of the training fields by using the unmanned aerial vehicle.
And the characteristic wave band screening and screening unit is used for screening the characteristic wave bands of the training hyperspectral remote sensing images to obtain a reflectivity average value on a set wave band of a training field.
And the sample set feature matrix determining unit is used for constructing a sample set feature matrix according to the reflectivity mean value on the set waveband of the training field.
And the measuring unit is used for measuring the total nitrogen content of the training field to obtain the measured value of the total nitrogen content of the training field.
And the label matrix construction unit is used for constructing a label matrix according to the total nitrogen content measurement value.
And the longitudinal merging and dividing unit is used for longitudinally merging and dividing the sample set characteristic matrix and the label matrix to obtain a training data set and a test data set.
And the training unit is used for training a set number of base learners by using a Boosting method according to the training data set, taking the sample set characteristic matrix as input and the total nitrogen content measurement value as output.
And the weighted combination unit is used for carrying out weighted combination on the base learners with the set number to obtain the XGboost model.
And the inversion unit is used for inverting the XGboost model according to the test data set to obtain a trained inversion model.
In practical applications, the expression of the reflectance mean value is as follows:
wherein,
is the average value of the reflectivity,
for the nth wave band lambdanThe reflectivity value of the a-th pixel point in the m-th field block area; a. themThe total number of pixel points in the region of interest of the mth plot area.
According to the method and the system provided by the invention, the field is divided into a plurality of field areas, and the total nitrogen content measured value of each field area is obtained; acquiring a hyperspectral remote sensing image of a field wheat canopy unmanned aerial vehicle; preprocessing hyperspectral image data and actually measured buoy data of the unmanned aerial vehicle; establishing a corresponding matrix of the spectral characteristics and the measured data of the total nitrogen content; selecting spectral characteristics by using an RReliefF algorithm; selecting a certain number of sampling sets by adopting a bootstrap sampling method, and establishing a plurality of regression decision trees; giving different weights to each decision tree for combination to form a random forest; and acquiring the total nitrogen content of the crops based on the constructed random forest model. The invention provides a macroscopic, continuous and effective method for monitoring the total nitrogen content of winter wheat, which does not damage crops and has higher practical value. The method and the system provided by the invention have the following advantages:
(1) the traditional crop total nitrogen content monitoring method is characterized in that destructive sampling is carried out, monitoring of a single sampling point can be realized only, and the method is difficult to apply to large-area crop monitoring. The invention introduces the hyperspectral remote sensing technology of the unmanned aerial vehicle into the monitoring of the total nitrogen content of the crops, realizes the breakthrough from point to surface, namely from a single sampling point to a field area, is beneficial to monitoring the total nitrogen content of the crops in large area, and can not damage the crops.
(2) Based on the integrated learning idea, the invention establishes an intelligent remote sensing inversion model to perform remote sensing inversion on the total nitrogen content of crop plants in field scale by adopting an XGboost (extreme Gradientboosting) advanced machine learning method, obtains a good inversion result, and can intuitively and accurately display the spatial distribution and the characteristics of the total nitrogen content of the crop plants in the field.
(3) The method screens the characteristic wave bands, reduces the modeling dimension, reduces the redundancy and complexity of an inversion model, and avoids the over-fitting phenomenon.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (8)
1. A crop total nitrogen content inversion method based on hyperspectral remote sensing is characterized by comprising the following steps:
acquiring a plurality of hyperspectral remote sensing images of a field to be measured by using an unmanned aerial vehicle;
screening characteristic wave bands of the hyperspectral remote sensing images to obtain a reflectivity average value of a plurality of fields to be detected on a set wave band; the set waveband comprises a plurality of wavebands;
constructing a reflectivity mean characteristic matrix according to the reflectivity mean;
and inputting the reflectivity mean value characteristic matrix into a trained inversion model to obtain the inversion value of the total nitrogen content of the field to be detected.
2. The crop total nitrogen content inversion method based on hyperspectral remote sensing according to claim 1, wherein before the hyperspectral remote sensing images are subjected to characteristic band screening to obtain a reflectivity average value of a plurality of fields to be measured on a set band, the method further comprises:
splicing the plurality of hyperspectral remote sensing images to obtain a complete hyperspectral remote sensing image of the field to be measured;
preprocessing the complete hyperspectral remote sensing image; the preprocessing comprises radiation calibration and atmospheric correction.
3. The crop total nitrogen content inversion method based on hyperspectral remote sensing according to claim 1, characterized in that the training process of the inversion model specifically comprises:
acquiring a plurality of training hyperspectral remote sensing images of a training field by using an unmanned aerial vehicle;
carrying out characteristic band screening on the training hyperspectral remote sensing image to obtain a reflectivity average value on a set band of a training field;
constructing a sample set feature matrix according to the reflectivity mean value on the set waveband of the training field;
measuring the total nitrogen content of the training fields to obtain the measured value of the total nitrogen content of the training fields;
constructing a label matrix according to the total nitrogen content measurement value;
longitudinally combining and dividing the sample set characteristic matrix and the label matrix to obtain a training data set and a testing data set;
according to the training data set, taking the sample set characteristic matrix as input, taking the total nitrogen content measurement value as output, and training a set number of base learners by using a Boosting method;
carrying out weighted combination on the base learners with the set number to obtain an XGboost model;
and inverting the XGboost model according to the test data set to obtain a trained inversion model.
4. The crop total nitrogen content inversion method based on hyperspectral remote sensing according to claim 1, characterized in that the reflectance mean value is represented by the following formula:
wherein,
is the average value of the reflectivity,
for the nth wave band lambdanThe reflectivity value of the a-th pixel point in the m-th field block area; a. themThe total number of pixel points in the region of interest of the mth plot area.
5. A crop total nitrogen content inversion system based on hyperspectral remote sensing is characterized by comprising:
the acquisition module is used for acquiring a plurality of hyperspectral remote sensing images of a field to be detected by using the unmanned aerial vehicle;
the characteristic band screening module is used for screening characteristic bands of the hyperspectral remote sensing images to obtain a reflectivity average value of a plurality of fields to be tested on a set band; the set waveband comprises a plurality of wavebands;
the reflectivity mean characteristic matrix determining module is used for constructing a reflectivity mean characteristic matrix according to the reflectivity mean;
and the total nitrogen content inversion value determining module is used for inputting the reflectivity mean value characteristic matrix into the trained inversion model to obtain the total nitrogen content inversion value of the field to be detected.
6. The crop total nitrogen content inversion system based on hyperspectral remote sensing according to claim 5, characterized by further comprising:
the splicing module is used for splicing a plurality of hyperspectral remote sensing images to obtain a complete hyperspectral remote sensing image of the field to be measured;
the preprocessing module is used for preprocessing the complete hyperspectral remote sensing image; the preprocessing comprises radiation calibration and atmospheric correction.
7. The crop total nitrogen content inversion system based on hyperspectral remote sensing according to claim 5, characterized by further comprising a training module of an inversion model, wherein the training module of the inversion model specifically comprises:
the acquisition unit is used for acquiring a plurality of training hyperspectral remote sensing images of a training field by using the unmanned aerial vehicle;
the characteristic band screening and screening unit is used for screening the characteristic bands of the training hyperspectral remote sensing images to obtain a reflectivity average value on a set band of a training field;
the sample set feature matrix determining unit is used for constructing a sample set feature matrix according to the reflectivity mean value on the set waveband of the training field;
the measuring unit is used for measuring the total nitrogen content of the training field to obtain a measured value of the total nitrogen content of the training field;
the tag matrix construction unit is used for constructing a tag matrix according to the total nitrogen content measurement value;
the longitudinal merging and dividing unit is used for longitudinally merging and dividing the sample set characteristic matrix and the label matrix to obtain a training data set and a test data set;
the training unit is used for training a set number of base learners by a Boosting method according to the training data set by taking the sample set characteristic matrix as input and the total nitrogen content measurement value as output;
the weighted combination unit is used for carrying out weighted combination on the base learners with the set number to obtain an XGboost model;
and the inversion unit is used for inverting the XGboost model according to the test data set to obtain a trained inversion model.
8. The crop total nitrogen content inversion system based on hyperspectral remote sensing according to claim 5, wherein the expression of the reflectance mean is as follows:
wherein,
is the average value of the reflectivity,
for the nth wave band lambdanThe reflectivity value of the a-th pixel point in the m-th field block area; a. themThe total number of pixel points in the region of interest of the mth plot area.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111312193.0A CN114036461A (en) | 2021-11-08 | 2021-11-08 | Crop total nitrogen content inversion method and system based on hyperspectral remote sensing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111312193.0A CN114036461A (en) | 2021-11-08 | 2021-11-08 | Crop total nitrogen content inversion method and system based on hyperspectral remote sensing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114036461A true CN114036461A (en) | 2022-02-11 |
Family
ID=80143191
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111312193.0A Pending CN114036461A (en) | 2021-11-08 | 2021-11-08 | Crop total nitrogen content inversion method and system based on hyperspectral remote sensing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114036461A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109543768A (en) * | 2018-11-30 | 2019-03-29 | 福州大学 | Ocean interior thermohaline information intelligent extracting method based on multi-source satellite remote sensing |
| CN112632847A (en) * | 2020-11-26 | 2021-04-09 | 淮阴师范学院 | XGboost regression algorithm-based rice leaf starch content remote sensing inversion model and method |
| 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 |
-
2021
- 2021-11-08 CN CN202111312193.0A patent/CN114036461A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109543768A (en) * | 2018-11-30 | 2019-03-29 | 福州大学 | Ocean interior thermohaline information intelligent extracting method based on multi-source satellite remote sensing |
| CN112632847A (en) * | 2020-11-26 | 2021-04-09 | 淮阴师范学院 | XGboost regression algorithm-based rice leaf starch content remote sensing inversion model and method |
| 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 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112051222A (en) | River and lake water quality monitoring method based on high-resolution satellite image | |
| CN111678866B (en) | Soil water content inversion method for multi-model ensemble learning | |
| Whetton et al. | Nonlinear parametric modelling to study how soil properties affect crop yields and NDVI | |
| CN111007021A (en) | Hyperspectral water quality parameter inversion system and method based on one-dimensional convolution neural network | |
| Wang et al. | Integrating remotely sensed leaf area index and leaf nitrogen accumulation with RiceGrow model based on particle swarm optimization algorithm for rice grain yield assessment | |
| CN110927082A (en) | Winter wheat yield prediction method based on unmanned aerial vehicle imaging hyperspectral remote sensing | |
| CN113537108B (en) | Crop chlorophyll content inversion method based on unmanned aerial vehicle hyperspectral remote sensing | |
| Popescu et al. | Remote sensing in the analysis and characterization of spatial variability of the territory. a study case in Timis County, Romania. | |
| CN112749627A (en) | Method and device for dynamically monitoring tobacco based on multi-source remote sensing image | |
| Zhang et al. | Estimating wheat yield by integrating the WheatGrow and PROSAIL models | |
| CN101144861A (en) | High spectrum sub-pixel target detection method and device | |
| CN113466143B (en) | Soil nutrient inversion method, device, equipment and medium | |
| CN115860269B (en) | Crop yield prediction method based on triple attention mechanism | |
| Tumbo et al. | Hyperspectral–based neural network for predicting chlorophyll status in corn | |
| CN111738066A (en) | Gridding rice sheath blight habitat evaluation method integrating multi-source remote sensing information | |
| CN116403048B (en) | Crop growth estimation model construction method based on multi-mode data fusion | |
| CN112906537A (en) | Crop identification method and system based on convolutional neural network | |
| CN112149827A (en) | System and method for constructing soil quick-acting potassium analysis model based on satellite images | |
| CN110836870A (en) | GEE-based large-area lake transparency rapid drawing method | |
| CN115169728A (en) | Soil fertility prediction method based on simplified neural network | |
| Park et al. | Improvement of satellite-based estimation of gross primary production through optimization of meteorological parameters and high resolution land cover information at regional scale over East Asia | |
| Rossi et al. | Sentinel-2 estimation of CNC and LAI in rice cropping system through hybrid approach modelling | |
| CN115759524A (en) | Soil productivity grade identification method based on remote sensing image vegetation index | |
| CN112257531B (en) | Remote sensing monitoring method for forest land change based on diversity feature combination | |
| CN114417728A (en) | Near-surface air temperature inversion method based on temperature, emissivity and deep learning |
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 |