LU102667B1 - Method for selecting near-infrared (nir) spectral characteristic bands for soil nitrogen based on algorithm fusion - Google Patents
Method for selecting near-infrared (nir) spectral characteristic bands for soil nitrogen based on algorithm fusion Download PDFInfo
- Publication number
- LU102667B1 LU102667B1 LU102667A LU102667A LU102667B1 LU 102667 B1 LU102667 B1 LU 102667B1 LU 102667 A LU102667 A LU 102667A LU 102667 A LU102667 A LU 102667A LU 102667 B1 LU102667 B1 LU 102667B1
- Authority
- LU
- Luxembourg
- Prior art keywords
- soil
- nir
- spectral characteristic
- nitrogen
- selecting
- Prior art date
Links
- 239000002689 soil Substances 0.000 title claims abstract description 134
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 title claims abstract description 106
- 230000003595 spectral effect Effects 0.000 title claims abstract description 56
- 229910052757 nitrogen Inorganic materials 0.000 title claims abstract description 53
- 238000000034 method Methods 0.000 title claims abstract description 35
- 230000004927 fusion Effects 0.000 title claims description 17
- XSQUKJJJFZCRTK-UHFFFAOYSA-N Urea Chemical compound NC(N)=O XSQUKJJJFZCRTK-UHFFFAOYSA-N 0.000 claims abstract description 16
- 239000004202 carbamide Substances 0.000 claims abstract description 16
- 230000000694 effects Effects 0.000 claims abstract description 6
- 230000036961 partial effect Effects 0.000 claims abstract description 6
- 238000002156 mixing Methods 0.000 claims abstract description 5
- 230000003044 adaptive effect Effects 0.000 claims abstract description 3
- 230000002860 competitive effect Effects 0.000 claims abstract description 3
- 238000000643 oven drying Methods 0.000 claims abstract description 3
- 238000005070 sampling Methods 0.000 claims abstract description 3
- 238000001228 spectrum Methods 0.000 claims description 26
- 238000010238 partial least squares regression Methods 0.000 claims description 9
- 238000003825 pressing Methods 0.000 claims description 4
- 239000000203 mixture Substances 0.000 claims description 3
- 230000000717 retained effect Effects 0.000 claims description 3
- 239000002245 particle Substances 0.000 claims description 2
- 238000010183 spectrum analysis Methods 0.000 claims description 2
- 239000000243 solution Substances 0.000 description 7
- 238000004497 NIR spectroscopy Methods 0.000 description 6
- 238000001514 detection method Methods 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000010187 selection method Methods 0.000 description 2
- 238000002835 absorbance Methods 0.000 description 1
- 239000007864 aqueous solution Substances 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000001035 drying Methods 0.000 description 1
- 238000007499 fusion processing Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 235000015097 nutrients Nutrition 0.000 description 1
- 235000016709 nutrition Nutrition 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/359—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light using near infrared light
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/3563—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light for analysing solids; Preparation of samples therefor
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2201/00—Features of devices classified in G01N21/00
- G01N2201/12—Circuits of general importance; Signal processing
- G01N2201/129—Using chemometrical methods
- G01N2201/1293—Using chemometrical methods resolving multicomponent spectra
Landscapes
- Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
Method for selecting near-infrared (NIR) spectral characteristic bands for soil nitrogen, including the following steps: step 1: preparing soil samples: collecting four different types of soils; mixing one of urea solutions with different concentration gradients with a soil, and oven-drying all soil samples at 80°C for 24 h; step 2: using an NIR spectrometer to acquire spectral information of four soil samples; step 3: using partial least squares (PLS) to build prediction models for soil nitrogen content; step 4: using backward interval partial least squares (BIPLS) to select an NIR spectral characteristic interval, and using competitive adaptive reweighted sampling (CARS) to select an NIR spectral characteristic variable of a soil, and fusing results of the two algorithms to determine characteristic bands for the four soils; and step 5: using the PLS to build a prediction model for soil nitrogen content, and comparing effects of models built and characteristic bands, respectively.
Description
TECHNICAL FIELD The present disclosure relates to the technical field of soil component detection, and in particular to a method for selecting near-infrared (NIR) spectral characteristic bands for soil nitrogen based on algorithm fusion.
BACKGROUND Soil is the main matrix for crop nutrient sources, and spectral reflectance of soil is one of the basic properties of soil, which is closely related to the physical and chemical properties of soil. A nitrogen content in soil directly affects a nutritional level for crop growth and is an important reference factor for crop growth estimation. Near-infrared spectroscopy (NIRS) is an analysis method with characteristics of fast analysis, simple sample treatment, simple operation, and low cost. Compared with traditional methods, the NIRS, when used to accurately and quickly estimate a nitrogen content in soil, shows more advantages and brighter application prospects. China has a vast territory and a wide variety of soils that show large differences in properties. Exploring the differences among NIR spectra of different soils and the selection of characteristic bands is of great value for developing a universal testing instrument that can meet the requirements of rapid, accurate, and real-time detection in agriculture.
The NIRS can be used to quickly detect a nitrogen content in soil, which leads to a fast, accurate and pollution-free detection process. The selection of NIR spectral characteristic bands for soil nitrogen is affected by various factors such as soil type, detection method, and selection method. Lu Yanli et al. analyzed the spectral reflectance changes of the northeast black soil during a band range of 350 nm to 2,500 nm and determined an optimal prediction model to predict a total nitrogen content in the black soil using a normalized spectral index composed of visible light bands of 550 nm and 450 nm. Pan Tao et al. used the moving window partial least squares (MWPLS) and the Savitzky-Golay smoothing algorithm to select an optimized NIR spectral characteristic band for total nitrogen in soil as 1,692 nm to 2,138 nm, with a modeling-set correlation coefficient of 0.931 and a prediction-set correlation coefficient of 0.882. Zhang Yao et al. combined the results of wavelet analysis and successive elimination algorithm (SEA) to determine 6 sensitive bands for predicting soil total nitrogen, and results showed that predicting soil total nitrogen in real time with sensitive bands exhibited a high prediction accuracy.
Although some researchers have adopted different methods to select NIR spectral characteristic bands for soil nitrogen, determining a characteristic band selection method with high universality is of important reference value for developing an NIR real-time rapid detection device for soil nitrogen, because there are a large variety of soils and different soils have similar but different NIR spectra.
SUMMARY The present disclosure provides a method for selecting NIR spectral characteristic bands for soil nitrogen based on algorithm fusion to improve the efficiency of soil nitrogen detection.
The present disclosure provides a method for selecting NIR spectral characteristic bands for soil nitrogen based on algorithm fusion, including the following steps: step 1: preparing soil samples, including the following steps: step 1-1: according to original nitrogen contents of four soils (yellow soil, calcareous soil, black soil, and red soil), preparing urea solutions with different nitrogen concentration gradients; and thoroughly mixing 15 mL of a urea solution with 100 g of a soil sample each time, and pressing a resulting mixture into a thin slice, which is cut into a size convenient for spectral analysis; and step 1-2: oven-drying all soil samples in an oven at 80°C for 24 h; step 2: using a portable NIR spectrometer to acquire spectral information of all soil samples; step 3: using partial least squares (PLS) to build four prediction models for soil nitrogen content based on soil full spectral data; step 4: using backward interval partial least squares (BIPLS) and competitive adaptive reweighted sampling (CARS) to select an NIR spectral characteristic interval and an NIR spectral characteristic variable of a soil, respectively, and optimizing and fusing results of the two algorithms to determine an NIR spectral characteristic interval for the soil; and using the PLS once again to build a prediction model for soil nitrogen content based on a characteristic band spectrum, and comparing an effect of a prediction model built based on a full spectrum with that of a prediction model built based on characteristic bands.
Preferably, in step 1-1, the urea solutions with different nitrogen concentrations may be prepared from urea particles; and according to the original nitrogen contents in soils, the concentration gradients for the prepared urea solutions may be as follows: yellow soil: 0 g/kg to
2.0 g/kg, calcareous soil: 0 g/kg to 2.5 g/kg, black soil: 0 g/kg to 4.6 g/kg, and red soil: 0 g/kg to
4.5 g/kg.
In step 1-1, the urea aqueous solution needs to be fully mixed with the soil sample to ensure the uniform distribution of nitrogen in the soil sample, thus facilitating the accuracy of subsequent spectrum acquisition results.
In step 1-1, the thin slice obtained by mixing a urea solution with a soil sample and pressing may have a size of about 100 mm x 100 mm, which may be cut into soil sample blocks with a size of about 10 mm x 10 mm.
Preferably, in step 2, an instrument may be preheated for 15 min before the spectrum determination is conducted, and calibration may be conducted with a blackboard and a whiteboard.
In step 2, an NIR spectrum may be acquired at a band range of 900 nm to 1,700 nm; 400 points may be acquired for each spectrum; a spectral image may be obtained every 3 scans on average; and a total of 618 soil samples may be determined.
In step 2, in order to reduce the influence of experimental ambient light and fluorescent light on the acquisition of spectral information, spectrum acquisition is conducted in a dark environment.
Preferably, in step 4, the BIPLS may be used to select an NIR spectral characteristic interval of soil nitrogen, the CARS may be used to select an NIR spectral characteristic variable of soil nitrogen, and according to selection results of the two algorithms, an optimized NIR spectral characteristic band may be selected for the four soils, separately.
In step 4, during a process of gradually eliminating spectral intervals by BIPLS, an RMSECYV value of a model may change continuously, and the numbers of remaining intervals and variables in the model may be continuously reduced until the RMSECV value of the model reaches a minimum value, thus obtaining several characteristic intervals.
In step 4, during an operating process of the CARS algorithm, some samples may be selected for PLS regression modeling and iterative modeling may be repeated hundreds of times like this; only wavelength variables with PLS regression coefficients of large absolute values may be retained during a process of selecting characteristic variables; a PLS regression model may be built with selected wavelength variables; an RMSECV value may be calculated for the model; and a variable subset corresponding to the smallest RMSECV value may be selected as an optimal variable subset.
The present disclosure provides a method for selecting NIR spectral characteristic bands for soil nitrogen based on algorithm fusion. That is, on the basis of exploring a universal method for selecting NIR spectral characteristic bands for soil nitrogen, the two algorithms of BIPLS and CARS are fused to establish a universal characteristic band selection method for different types of soils. The method is more suitable to the need of real-time and on-line detection of soil nitrogen content in an actual environment, and using characteristic bands instead of a full spectrum for modeling improves an operating efficiency of an NIRS prediction model for soil nitrogen.
Compared with the prior art, the present disclosure has the following advantages: (1) The algorithm fusion process is simple, and the method for selecting NIR spectral characteristic bands for soil nitrogen has high universality.
(2) Using characteristic bands instead of a full spectrum for building a prediction model for soil nitrogen greatly improves an operating efficiency of the model.
(3) The present disclosure provides a theoretical support for developing efficient NIRS detection instruments.
BRIEF DESCRIPTION OF DRAWINGS FIG. 1 is a flowchart of the method for selecting NIR spectral characteristic bands for soil nitrogen based on algorithm fusion according to the present disclosure; FIG. 2 shows the average NIR spectral curves of the four soils according to the present disclosure; FIG. 3 shows the NIR full spectrum-based PLS modeling effects of the four soils according to the present disclosure; FIG. 4 shows the characteristic band-based PLS modeling effects of the four soils according to the present disclosure.
DETAILED DESCRIPTION The method for selecting NIR spectral characteristic bands for soil nitrogen based on algorithm fusion in the present disclosure will be described in detail below in conjunction with the accompanying drawings.
A method for selecting NIR spectral characteristic bands for soil nitrogen based on algorithm fusion as shown in FIG. 1 included the following steps: (1) Preparation of samples Soil samples used in this example included four different types of soils: yellow soil, calcareous soil, black soil, and red soil from Xi'an in Shaanxi, Jining in Shandong, Daxinganling in Inner Mongolia, and Lishui in Zhejiang, respectively.
A method of preparing samples included the following steps: According to original nitrogen contents of four soils, urea solutions with the following different nitrogen concentration gradients were prepared: yellow soil: 0 g/kg to 2.0 g/kg, calcareous soil: 0 g/kg to 2.5 g/kg, black soil: 0 g/kg to 4.6 g/kg, and red soil: 0 g/kg to 4.5 g/kg. mL of a urea solution was thoroughly mixed with 100 g of a soil sample each time, and a resulting mixture was pressed into a thin slice, which was cut into soil sample blocks with a size of about 10 mm x 10 mm.
All soil samples were oven-dried in an oven at 80°C for 24 h. (2) Spectrum acquisition In this experiment, an NIR spectrophotometer of ISUZU OPTICS was used, and a spectral band range, a resolution, and the number of scans could be set by oneself.
A spectrum was acquired at a band range of 900 nm to 1,700 nm, and an intensity, reflectance, and absorbance of light could be acquired; 400 points were acquired for each spectrum; and a spectral image was obtained every 3 scans on average.
The instrument was preheated for 15 min before the spectrum determination was conducted, and calibration was conducted with a blackboard and a whiteboard.
In order to reduce the influence of experimental ambient light and fluorescent light on the acquisition of spectral information, spectrum acquisition was conducted in a dark environment.
In the experiment, the determination was conducted on a smooth side of the soil sample block, and a total of 618 soil samples were determined. (3) Selection of characteristic bands In the present disclosure, BIPLS was used to select a spectral characteristic interval, and CARS was used to select an NIR spectral characteristic variable of the soil.
During a process of gradually eliminating spectral intervals by BIPLS, an RMSECV value of a model changed continuously, and the numbers of remaining intervals and variables in the model were continuously reduced until the RMSECV value of the model reached a minimum value, thus obtaining several characteristic intervals.
According to the characteristic variable and characteristic interval results, an optimized NIR spectral characteristic band was selected for soil nitrogen.
During an operating process of the CARS algorithm, some samples were selected for PLS regression modeling and iterative modeling was repeated hundreds of times like this; only wavelength variables with PLS regression coefficients of large absolute values were retained during a process of selecting characteristic variables; a PLS regression model was built with selected wavelength variables; an RMSECV value was calculated for the model; and a variable subset corresponding to the smallest RMSECV value was selected as an optimal variable subset.
With the red soil as an example, a selection process and results were shown in Tables 1 and 2 below: Table 1 A characteristic interval selection process of BIPLS intervals interval variables intervals interval variables 2 | 6 | 0587 | 400 | 10 | 3 | 0546 | 200 | 1 | 16 | 0578 | 38 [| 9 || | 0546 | 180 | 18 | 19 | 0571 | 360 | 8 | 20 | 0544 | 160 | |. 16 | KM | 0561 | 30 | 6 | 10 | 0516 | M0 |
6 U10266 |. 15 [| 8 | 0549 | 300 | 5 | um | 0506 | 100 | |. 14 | 5 | 0546 | 280 | 4 | |& | 0500 | 80 | |. 13 | 4 [| 0545 | 20 | 3 | 13 | 0514 | 60 | |. M2 | 2 | 0542 | 240 | 2 | 9 | 058 | 40 | Table 2 Results of characteristic interval and characteristic variable selection Characteristic Number of Characteristic Characteristic intervals Number of Soil type variables selected by characteristic intervals selected selected by BIPLS intervals CARS variables by BIPLS (divided bands) 1036-1078 1250-1329 Yellow soil 1152-1162 11 1411-1448 4,8, 9, 10, 13, 15, 17 7 1487-1523 1561-1596 cal 900-944 eus 1129-1159 31 1080-1290 1, 4, 5, 6, 7,8, 9, 16, 20 sol 1525-1559 1036-1078 Black soil | 1414-1429 1472-1493 40 1250-1590, 4, 9, 12, 13, 14, 15, 16 7 1411-1559 ; 1225-1290,1441-1 According to selection results of the two algorithms, the final characteristic bands were determined for the four soils as follows: 1,152 nm to 1,162 nm and 1,296 nm to 1,309 nm for the yellow soil; 1,036 nm to 1,055 nm and 1,129 nm to 1,156 nm for the calcareous soil; 1,055 nm, 1,281 nm, 1,414 nm to 1,428 nm, and 1,472 nm to 1,493 nm for the black soil; and 1,250 nm, 1,480 nm, and 1,680 nm for the red soil. (4) Building of models The present disclosure used the PLS to build prediction models for soil nitrogen content based on NIR full spectral data and characteristic bands, respectively.
The full spectral data subjected to baseline calibration and normalization were adopted as an independent variable X and the total nitrogen content was adopted as a dependent variable Y to build a total nitrogen content prediction model under each drying time.
A modeling set and a prediction set were obtained by dividing at a ratio of 2:1. A correlation coefficient R was used to reflect the closeness among variables, and a root-mean-square error (RMSE) was used to reflect the accuracy of measurement.
The closer the correlation coefficient is to 1 and the closer the RMSE is to 0, the better the performance of the prediction model and the higher the accuracy of the prediction model.
A calculation formula of the correlation coefficient R was as follows: R= Ix — ir, = 5) JEL {x ER EL On FF A calculation formula of the RMSE was as follows: n lee, T2 RMSE = Su - yi) n.
Results showed that the characteristic band-based prediction model generally exhibited a high accuracy; correlation coefficients for the yellow soil and black soil even exceeded that of the full spectrum-based model, reaching 0.9826 and 0.91, respectively; the correlation coefficient for the calcareous soil was 0.9561, which was almost the same as the correlation coefficient of the full spectrum-based model; and the correlation coefficient for the red soil was 0.9188, which was slightly inferior to that of the full spectrum-based model (0.9467) but still achieving a high accuracy.
In general, the characteristic band-based models for the four soils can completely replace the full spectrum-based models in terms of modeling accuracy, which greatly improves an operating efficiency of a prediction model.
Claims (8)
1. A method for selecting near-infrared (NIR) spectral characteristic bands for soil nitrogen based on algorithm fusion, comprising the following steps: step 1: preparing soil samples, comprising the following steps: step 1-1: according to original nitrogen contents of four soils (yellow soil, calcareous soil, black soil, and red soil), preparing urea solutions with different nitrogen concentration gradients; and thoroughly mixing 15 mL of a urea solution with 100 g of a soil sample each time, and pressing a resulting mixture into a thin slice, which is cut into a size convenient for spectral analysis; and step 1-2: oven-drying all soil samples in an oven at 80°C for 24 h; step 2: using a portable NIR spectrometer to acquire spectral information of all soil samples; step 3: using partial least squares (PLS) to build four prediction models for soil nitrogen content based on soil full spectral data; step 4: using backward interval partial least squares (BIPLS) and competitive adaptive reweighted sampling (CARS) to select an NIR spectral characteristic interval and an NIR spectral characteristic variable of a soil, respectively, and optimizing and fusing results of the two algorithms to determine an NIR spectral characteristic interval for the soil; and using the PLS once again to build a prediction model for soil nitrogen content based on a characteristic band spectrum, and comparing an effect of a prediction model built based on a full spectrum with that of a prediction model built based on characteristic bands.
2. The method for selecting NIR spectral characteristic bands for soil nitrogen based on algorithm fusion according to claim 1, wherein, in step 1-1, the urea solutions with different nitrogen concentrations are prepared from urea particles; and according to the original nitrogen contents in soils, the concentration gradients for the prepared urea solutions are as follows: yellow soil: 0 g/kg to 2.0 g/kg, calcareous soil: 0 g/kg to 2.5 g/kg, black soil: 0 g/kg to 4.6 g/kg, and red soil: 0 g/kg to 4.5 g/kg.
3. The method for selecting NIR spectral characteristic bands for soil nitrogen based on algorithm fusion according to claim 1, wherein, in step 1-1, the thin slice obtained by mixing a urea solution with a soil sample and pressing has a size of about 100 mm x 100 mm, which is cut into soil sample blocks with a size of about 10 mm x 10 mm.
4. The method for selecting NIR spectral characteristic bands for soil nitrogen based on algorithm fusion according to claim 1, wherein, in step 2, an NIR spectrum is acquired at a band range of 900 nm to 1,700 nm; 400 points are acquired for each spectrum; a spectral image is obtained every 3 scans on average; and a total of 618 soil samples are determined.
5. The method for selecting NIR spectral characteristic bands for soil nitrogen based on algorithm fusion according to claim 1, wherein, in step 4, the BIPLS is used to select the NIR spectral characteristic interval of soil nitrogen, and the CARS is used to select the NIR spectral characteristic variable of soil nitrogen; and according to selection results of the two algorithms, selecting an optimized NIR spectral characteristic band for the four soils, separately.
6. The method for selecting NIR spectral characteristic bands for soil nitrogen based on algorithm fusion according to claim 5, wherein, during a process of gradually eliminating spectral intervals by BIPLS, an RMSECV value of a model changes continuously, and the numbers of remaining intervals and variables in the model are continuously reduced until the RMSECYV value of the model reaches a minimum value, thus obtaining several characteristic intervals.
7. The method for selecting NIR spectral characteristic bands for soil nitrogen based on algorithm fusion according to claim 5, wherein, during an operating process of the CARS algorithm, some samples are selected for PLS regression modeling and iterative modeling is repeated hundreds of times like this; only wavelength variables with PLS regression coefficients of large absolute values are retained during a process of selecting characteristic variables; a PLS regression model is built with selected wavelength variables; an RMSECV value is calculated for the model; and a variable subset corresponding to the smallest RMSECV value is selected as an optimal variable subset.
8. The method for selecting NIR spectral characteristic bands for soil nitrogen based on algorithm fusion according to claim 1, wherein, in step 4, an effect of a model built based on an NIR full spectrum is compared with that of a model built based on selected characteristic bands to explore the feasibility of using characteristic bands instead of a full spectrum for modeling; and using characteristic bands for modeling can greatly improve an operating efficiency of a model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
LU102667A LU102667B1 (en) | 2021-03-18 | 2021-03-18 | Method for selecting near-infrared (nir) spectral characteristic bands for soil nitrogen based on algorithm fusion |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
LU102667A LU102667B1 (en) | 2021-03-18 | 2021-03-18 | Method for selecting near-infrared (nir) spectral characteristic bands for soil nitrogen based on algorithm fusion |
Publications (1)
Publication Number | Publication Date |
---|---|
LU102667B1 true LU102667B1 (en) | 2021-09-20 |
Family
ID=77800198
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
LU102667A LU102667B1 (en) | 2021-03-18 | 2021-03-18 | Method for selecting near-infrared (nir) spectral characteristic bands for soil nitrogen based on algorithm fusion |
Country Status (1)
Country | Link |
---|---|
LU (1) | LU102667B1 (en) |
-
2021
- 2021-03-18 LU LU102667A patent/LU102667B1/en active IP Right Grant
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101915744B (en) | Near infrared spectrum nondestructive testing method and device for material component content | |
Büning-Pfaue | Analysis of water in food by near infrared spectroscopy | |
CN107271382A (en) | A kind of different growing rape leaf SPAD value remote sensing estimation methods | |
Wang et al. | Independent component analysis-based baseline drift interference suppression of portable spectrometer for optical electronic nose of Internet of Things | |
CN111965140B (en) | Wavelength point recombination method based on characteristic peak | |
CN108982406A (en) | A kind of soil nitrogen near-infrared spectral characteristic band choosing method based on algorithm fusion | |
CN110726474A (en) | Hyperspectral high-frequency observation system and spectral observation method for ground objects | |
CN106018331B (en) | The method for estimating stability and pretreatment optimization method of multi-channel spectral system | |
Wang et al. | General model of multi-quality detection for apple from different origins by Vis/NIR transmittance spectroscopy | |
Wang et al. | Rapid detection of chlorophyll content and distribution in citrus orchards based on low-altitude remote sensing and bio-sensors | |
Wu et al. | Study on the optimal algorithm prediction of corn leaf component information based on hyperspectral imaging | |
Itoh et al. | Development of method for non-destructive measurement of nitrate concentration in vegetable leaves by near-infrared spectroscopy | |
CN114169165A (en) | Model method for estimating rice chlorophyll under cadmium stress by using three-band vegetation index | |
CN111220552A (en) | Chlorophyll hyperspectral inversion method considering illumination direction leaf radiation transmission model | |
CN107036999A (en) | A kind of five yuan of ready-mixed oil quantitative analysis methods based near infrared spectrum and Chemical Measurement | |
Qiong et al. | Field monitoring of wheat seedling stage with hyperspectral imaging | |
Cui et al. | Hyperspectral reflectance imaging for water content and firmness prediction of potatoes by optimum wavelengths | |
LU102667B1 (en) | Method for selecting near-infrared (nir) spectral characteristic bands for soil nitrogen based on algorithm fusion | |
CN117269106A (en) | Hyperspectral model for rapidly predicting content of capsicum protein | |
Moreira et al. | Calibration of a photometer for quantification of ethanol in gasoline: Absorbance determination | |
Ning et al. | Study on apple damage detecting method based on relaxation single-wavelength laser and convolutional neural network | |
Norris | Understanding and correcting the factors which affect diffuse transmittance spectra | |
Behera et al. | Detection of spent turmeric adulteration in powdered Curcuma longa using Vis-NIR spectroscopy and machine learning | |
US20210318280A1 (en) | System and computer-implemented method for extrapolating calibration spectra | |
Chen et al. | Feature selecting based on fourier series fitting |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
FG | Patent granted |
Effective date: 20210920 |