CN113607669A - Soil nutrient spectrum detection method based on transfer learning - Google Patents

Soil nutrient spectrum detection method based on transfer learning Download PDF

Info

Publication number
CN113607669A
CN113607669A CN202110871418.XA CN202110871418A CN113607669A CN 113607669 A CN113607669 A CN 113607669A CN 202110871418 A CN202110871418 A CN 202110871418A CN 113607669 A CN113607669 A CN 113607669A
Authority
CN
China
Prior art keywords
soil
data
spectrum
spectral
nutrient
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
Application number
CN202110871418.XA
Other languages
Chinese (zh)
Inventor
金�秀
李绍稳
郑文瑞
张筱丹
韩亚鲁
丁梦雅
王良龙
宣金祥
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Anhui Agricultural University AHAU
Original Assignee
Anhui Agricultural University AHAU
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Anhui Agricultural University AHAU filed Critical Anhui Agricultural University AHAU
Priority to CN202110871418.XA priority Critical patent/CN113607669A/en
Publication of CN113607669A publication Critical patent/CN113607669A/en
Pending legal-status Critical Current

Links

Images

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
    • 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
    • G01N21/01—Arrangements or apparatus for facilitating the optical investigation
    • 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
    • G01N21/01—Arrangements or apparatus for facilitating the optical investigation
    • G01N2021/0106—General arrangement of respective parts
    • G01N2021/0112—Apparatus in one mechanical, optical or electronic block

Abstract

The invention discloses a soil nutrient spectrum detection method based on transfer learning, which relates to the technical field of soil analysis and improves the accuracy of field in-situ soil spectrum data detection by reducing the influence of environmental factors; the method is combined with a migration learning method, the spectral data of the dry soil in the existing laboratory are utilized, the data distribution difference of the spectrums of the in-situ soil and the dry soil is reduced through spectral transformation, and the accuracy of a soil nutrient regression prediction model is improved; thereby improving the accuracy of the real-time detection of the nutrient content of the field in-situ soil. The detection system who sets up has reduced the instrument volume, has simplified the instrument structure and has constituteed, conveniently carries, operates in the field, gathers soil sample through the drill bit device of fetching earth, obtains soil spectral data and carries out spectral transformation through miniature spectrum appearance chip, detects soil nutrient content through regression model.

Description

Soil nutrient spectrum detection method based on transfer learning
Technical Field
The invention relates to the technical field of soil analysis, in particular to a soil nutrient spectrum detection method based on transfer learning.
Background
The soil provides nutrient substances and water for the growth of crops, detects the content of nutrients such as nitrogen, phosphorus, potassium, boron and the like in the soil, and has important significance for accurate fertilization and environmental protection of the crops. The traditional method is used for measuring the soil nutrient content by a chemical analysis method, but the method has high cost and low efficiency, has long measuring period, wastes time and labor, can generate chemical wastes polluting the environment, and cannot realize the rapid detection of the soil nutrient content.
Near-infrared spectroscopy (NIRS) is an environment-friendly, rapid and nondestructive analysis technology, and has the advantages of no use of chemical reagents, no pollution, simple operation and high stability. The combination of the chemometric method and the near infrared spectrum technology can realize the qualitative and quantitative analysis of the sample, so that the method is widely applied to the field of rapid detection of soil nutrients in recent years and gradually replaces the traditional chemical analysis method.
At present, soil spectra are measured by using a near-infrared spectrometer, soil is mainly transported to a laboratory for air drying, sieving and grinding, then spectra are acquired, and modeling prediction is performed on soil nutrients by combining a machine learning method, but when an established model is used for predicting in-situ soil, the problems of low prediction precision and model failure can occur. Because the traditional machine learning method assumes that the training data and the test data obey the same data distribution, and for the soil and the in-situ soil sample after the laboratory treatment, the reflectivity of the spectral data of the in-situ soil and the in-situ soil sample is different because the in-situ soil spectral detection is influenced by water, plant rhizome impurities and the like in the soil, the assumption of the same distribution is often not satisfied.
In order to solve the problems, the application provides a soil nutrient spectrum detection method based on migration learning, which is combined with a migration learning method, utilizes spectrum data of dry soil, reduces the spectrum data distribution difference of in-situ soil and dry soil through spectrum transformation, and improves the accuracy of a soil nutrient regression prediction model, so that the accuracy of real-time detection of the nutrient content of the in-situ soil in the field is improved.
Disclosure of Invention
The invention aims to provide a soil nutrient spectrum detection method based on migration learning, which is combined with a migration learning method, reduces the spectral data distribution difference of in-situ soil and dry soil through spectral transformation by using dry soil spectral data and improves the accuracy of a soil nutrient regression prediction model, thereby improving the accuracy of the real-time detection of the nutrient content of the in-situ soil in the field.
The invention provides a soil nutrient spectrum detection method based on transfer learning, which comprises the following steps:
obtaining soil to be detected;
acquiring soil spectrum data of soil to be detected;
based on a migration component analysis algorithm in migration learning, spectrum transformation is carried out on the spectrum data of the soil to be detected by combining with the spectrum data of the dry soil in a laboratory, new waveband data is obtained through linear transformation of a multispectral remote sensing image, and the purposes of reducing data redundancy and data compression, retaining main information and enhancing useful information are achieved;
importing the spectral data of the soil to be detected after spectral transformation into a regression model of soil nutrient and spectral reflectivity based on a TCA algorithm to perform real-time detection on the content of the soil nutrient;
the construction of the regression model comprises the following steps:
constructing a source domain and a target domain of soil spectral data;
calculating data after TCA spectrum transformation of a source domain and a target domain;
and establishing a soil nutrient and spectral reflectivity regression model by using a regression modeling algorithm.
Further, the real-time detection of the content of the soil nutrients is performed by the regression model of the soil nutrients and the spectral reflectance based on the TCA algorithm, which specifically comprises the following steps:
constructing a source domain and a target domain of soil spectral data;
inputting spectral data of a source domain and a target domain
Figure BDA0003188971880000031
And
Figure BDA0003188971880000032
obtaining a parameter L and a central matrix H in the regression model;
obtaining a kernel matrix K of a source domain and a target domain by utilizing kernel function mapping;
according to the parameter L, the central matrix H and the kernel matrix K, solving (KLK + mu I)-1Obtaining data after TCA spectrum transformation of a source domain and a target domain by using eigenvectors corresponding to the first m eigenvalues of KHK, wherein mu is a balance parameter;
and establishing a soil nutrient and spectral reflectivity regression model for the source domain data and the target domain data after the spectral transformation by using a regression modeling algorithm, and using the regression model for detecting the soil nutrient content.
Further, the calculation formula of the parameter L is:
Figure BDA0003188971880000033
wherein I is an all 1 matrix.
Further, the calculation formula of the central matrix H is:
Figure BDA0003188971880000034
wherein the content of the first and second substances,
Figure BDA0003188971880000035
is an all 1 matrix.
Further, the source domain is laboratory dry soil spectral data, and the target domain is in-situ soil spectral data.
Further, still include:
the method comprises the following steps of pretreating soil to be detected, wherein the pretreatment comprises the following steps: impurities in the soil to be detected are removed through screening by a screening device, and water in the soil to be detected is removed through heating of the USB heating sheet.
Compared with the prior art, the invention has the following remarkable advantages:
the invention provides a soil nutrient spectrum detection method based on transfer learning, which improves the accuracy of field in-situ soil spectrum data detection by reducing the influence of environmental factors; by combining a migration learning method, spectral data of dry soil in a laboratory are utilized, spectral data distribution difference of in-situ soil and dry soil is reduced through spectral transformation, and accuracy of a soil nutrient regression prediction model is improved; thereby improving the accuracy of the real-time detection of the nutrient content of the field in-situ soil. The detection system has the advantages that the volume of the device is reduced, the structure and the composition of the device are simplified, the device is convenient to carry and operate in the field, a soil sample is collected through the soil sampling device, soil spectrum data are obtained through the micro spectrometer chip and subjected to spectrum transformation, and the regression model is used for detecting the content of soil nutrients.
Drawings
FIG. 1 is a block diagram of a detection system according to an embodiment of the present invention;
FIG. 2 is a connection structure diagram of an earth measuring device according to an embodiment of the present invention;
fig. 3 is a diagram of a screen structure according to an embodiment of the present invention.
Description of reference numerals: the method comprises the following steps of 1-soil taking drill bit, 2-pipe fitting, 3-screen device, 4-soil measuring box, 5-cross shaft handle, 6-soil measuring device, 7-USB interface, 8-optical fiber, 9-reflection probe, 10-screw rod, 11-helicoid, 12-gasket and 13-screen.
Detailed Description
The technical solutions of the embodiments of the present invention are clearly and completely described below with reference to the drawings in the present invention, and it is obvious that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, shall fall within the scope of protection of the present invention.
Predicting the nutrient content of the soil by the in-situ soil spectrum: firstly, in-situ soil spectrum data acquisition is influenced by impurities such as water, plant roots and stems, and the accuracy of directly using the spectrum for modeling prediction is not high; secondly, the model established by the laboratory dry soil spectrum has high prediction precision, but the model is directly used for in-situ soil, so that the problems of low prediction precision, failure and the like can occur.
The transfer learning brings motivation for solving the problem, the transfer learning does not require that training and test data are subject to the same distribution, and the transfer learning is an important research problem in machine learning, focuses on adapting the previous learning experience to new learning, and improves the new learning efficiency. The spectrum transformation is realized based on the transfer learning algorithm, the data distribution difference of the spectrums of the in-situ soil and the existing laboratory dry soil is reduced, the influence of moisture, impurities and the like on the spectrum data detection is reduced, and the accuracy of in-situ soil nutrient prediction is improved. And by integrating the portable soil collector, the soil collection operation is convenient, the drill bit is used for collecting soil samples to obtain spectral data of the soil samples, and the soil nutrient content is detected. The operation is simple and convenient, the working efficiency is improved, and the popularization and the use are easy.
The invention provides a soil nutrient spectrum detection method based on transfer learning, which comprises the following steps:
obtaining soil to be detected;
pretreating soil to be detected, screening out impurities in the soil to be detected through a screen, and heating to remove water in the soil to be detected;
acquiring soil spectrum data of the preprocessed soil to be detected through a spectrum detection device;
the method comprises the steps of carrying out spectrum transformation on spectrum data of soil to be detected based on a Transfer Component Analysis (TCA) algorithm in Transfer learning in combination with spectrum data of dry soil in a laboratory, obtaining new waveband data through linear transformation on multispectral remote sensing images, and achieving the purposes of reducing data redundancy, compressing data quantity, retaining main information and enhancing useful information;
importing the spectral data of the soil to be detected after spectral transformation into a regression model of the soil nutrient and the spectral reflectivity based on a TCA algorithm for real-time detection of the soil nutrient content, which specifically comprises the following steps:
constructing a source domain and a target domain of soil spectral data, wherein the constructed source domain is laboratory dry soil spectral data, and the constructed target domain is in-situ soil spectral data;
inputting spectral data of a source domain and a target domain
Figure BDA0003188971880000051
And
Figure BDA0003188971880000052
to the regression model, the model is converted into a model,
Figure BDA0003188971880000053
and
Figure BDA0003188971880000054
is a spectral feature vector, ns、ntRespectively obtaining the quantity of source domain samples and the quantity of target domain samples to obtain a parameter L and a central matrix H;
x is then measured by the Maximum Mean Difference (MMD) in the high dimensional regenerative nuclear Hilbert space (RKHS)sAnd XtThe distance formula is as follows:
Figure BDA0003188971880000055
h is RKS, φ ∈ H (3)
The distance of the edge distribution is minimized by mapping phi, but phi is generally highly nonlinear and difficult to seek, so the minimization problem of the distance is converted into a kernel learning problem, and kernel skill k (namely k (x) is utilizedixj)＝φ(xi)Tφ(xj) Equation (3) may be rewritten as a trace of the kernel matrix:
Dist(Xs,Xt)＝trace(KL) (4)
in the formula
Figure BDA0003188971880000061
Is (n)s+nt)×(ns+nt) Composite kernel matrix of dimension, KsAnd KtKernel matrices defined in the source domain and the target domain for the kernel trick k, respectively;
calculating parameters L and a central matrix H, wherein:
Figure BDA0003188971880000062
wherein I is a full 1 matrix;
Figure BDA0003188971880000063
wherein the content of the first and second substances,
Figure BDA0003188971880000064
is a full 1 matrix;
obtaining a kernel matrix K of a source domain and a target domain by utilizing kernel function mapping;
to simplify the calculation, the results are constructed by a dimension reduction method, using (n)s+nt) In x m dimensions
Figure BDA0003188971880000065
The matrix transforms the corresponding eigenvectors to m dimensions, and the obtained kernel matrix is as follows:
Figure BDA0003188971880000066
wherein the content of the first and second substances,
Figure BDA0003188971880000067
is (n)s+nt) X m dimensional matrix, m < ns+nt. Substituting equation (5) into equation (4) yields:
Dist(Xs,Xt)＝trace((KWWTK)L)＝tr(WTKLKW) (6)
minimizing the formula (6), and the obtained W represents the spectrum matrix of the source domain and the target domain after TCA dimensionality reduction, and then the method is converted into the solution:
Figure BDA0003188971880000068
s.t.WTKHKW＝Im (7)
wherein
Figure BDA0003188971880000069
Is a central matrix, and the central matrix is a central matrix,
Figure BDA00031889718800000610
is an all 1 matrix.
Where μ is a balance parameter, the regularization term tr (W)TW) is added to control the complexity of W, adding a constraint WTKHKW＝ImIn order to avoid trivial solutions (i.e., W ═ 0), and finally, by mathematical derivation, the solution for W is (KLK + μ I)-1Outputting the eigenvectors corresponding to the first m eigenvalues of the KHK to obtain data after TCA transfer of the source domain and the target domain, reducing data distribution difference of the source domain and the target domain, and establishing a regression model by using a traditional regression algorithm;
according to the parameter L, the central matrix H and the kernel matrix K, solving (KLK + mu I)-1Obtaining data after TCA spectrum transformation of a source domain and a target domain by using eigenvectors corresponding to the first m eigenvalues of KHK, wherein mu is a balance parameter;
and establishing a soil nutrient and spectral reflectivity regression model for the source domain data and the target domain data after the spectral transformation by using a traditional regression modeling algorithm, and detecting the soil nutrient.
The construction of the regression model comprises the following steps:
constructing a source domain and a target domain of soil spectral data;
calculating data after TCA spectrum transformation of a source domain and a target domain;
and establishing a soil nutrient and spectral reflectivity regression model by using a regression modeling algorithm.
Example 1
If the soil nutrient content of the farmland can be analyzed rapidly and nondestructively before the farmland is cultivated, farmers can be helped to find whether the farmland has the condition of lacking elements or pesticide pollution in time, and various nutrient conditions of the farmland can be adjusted according to the analysis result, so that the aims of planting crops and fertilizing with pertinence are achieved, and the farmers are helped to reduce loss, increase production and income.
Referring to fig. 1-3, a detection system of a migration learning-based soil nutrient spectrum detection method includes:
the soil sampling device comprises a soil sampling drill bit 1, wherein the end part of the soil sampling drill bit 1 is connected with a pipe fitting 2, a soil sample is conveyed into a conveying pipe fitting 2 in a proper condition, a screen filter 3 and a soil measuring box 4 are sequentially arranged in the pipe fitting 2 and used for screening the volume of the soil sample, the small-particle soil sample enters the conveying soil measuring box 4 after being screened, a USB heating sheet is arranged on the outer side surface of the soil measuring box 4 and electrically connected with a built-in power supply and used for heating, drying and conveying the soil sample in the soil measuring box 4 to facilitate detection, and a cross shaft handle 5 is arranged at the end part of the pipe fitting 2 to facilitate downward force application so that the soil sampling drill bit 1 can acquire the soil sample; and a terminal mounting seat is arranged outside the cross shaft handle 5 and used for fixing a terminal.
The soil testing device 6 is fixed in the pipe fitting 2, the USB interface 7 connected with the soil testing device 6 is arranged on the surface of the pipe fitting 2, the pipe fitting is connected with a fixed terminal (such as a mobile phone) through the USB interface 7, or is connected with a power supply through the USB interface 7, the mobile phone is used as a main control module, soil nutrient spectrum detection system software based on transfer learning is installed, regression models of different types of soil nutrients and spectrum reflectivity based on a transfer learning algorithm TCA are stored in the software, spectrum reflectivity data of the current in-situ soil to be tested after being subjected to TCA spectrum transformation are obtained in real time, then the regression models of different types of soil nutrients and spectrum reflectivity based on the transfer learning algorithm TCA in the system installed on the mobile phone are called, real-time prediction of the current soil nutrients to be tested is achieved, and the final result is displayed on a screen of the mobile phone. The soil testing device is characterized in that a spectrum detection module is arranged in the soil testing device 6, a power supply, a built-in light source and a micro spectrometer chip are arranged in the spectrum detection module, the built-in light source and the micro spectrometer chip are electrically connected with the power supply, the spectrum detection module is used for performing spectrum detection on a soil sample and performing spectrum transformation on spectrum data, the built-in light source and the micro spectrometer chip are both connected with a reflection probe 9 through an optical fiber 8, the reflection probe 9 is connected with a glass slide on the surface of the soil testing box 4 and is used for collecting the spectrum data of the soil sample, and a migration component analysis (TCA) algorithm in migration learning is integrated in the micro spectrometer chip to perform spectrum transformation, so that the data distribution difference between an in-situ soil spectrum and an existing laboratory dry soil spectrum is reduced.
The surface of the pipe fitting 2 is provided with a replacing port, a screw rod 10 is arranged in the replacing port, the screw rod 10 is in threaded connection with a spiral ring 11 installed outside a gasket 12, and a screen 13 is filled inside the gasket 12. The screen cloth of different apertures of convenient replacement comes from the definition soil aperture that sieves, also conveniently surveys soil box 4 in soil discharge.
The above disclosure is only for a few specific embodiments of the present invention, however, the present invention is not limited to the above embodiments, and any variations that can be made by those skilled in the art are intended to fall within the scope of the present invention.

Claims (6)

1. A soil nutrient spectrum detection method based on transfer learning is characterized by comprising the following steps:
obtaining soil to be detected;
acquiring soil spectrum data of soil to be detected;
performing spectrum transformation on the spectrum data of the soil to be detected based on a migration component analysis algorithm in migration learning and combining with the spectrum data of the dry soil in a laboratory to obtain new band data;
importing the spectral data of the soil to be detected after spectral transformation into a regression model of soil nutrient and spectral reflectivity based on a TCA algorithm to perform real-time detection on the content of the soil nutrient;
the construction of the regression model comprises the following steps:
constructing a source domain and a target domain of soil spectral data;
calculating data after TCA spectrum transformation of a source domain and a target domain;
and establishing a soil nutrient and spectral reflectivity regression model by using a regression modeling algorithm.
2. The soil nutrient spectrum detection method based on transfer learning of claim 1, wherein the real-time detection of the content of the soil nutrients is performed by a regression model of the soil nutrients and the spectrum reflectivity based on the TCA algorithm, and specifically comprises the following steps:
constructing a source domain and a target domain of soil spectral data;
inputting spectral data of a source domain and a target domain
Figure FDA0003188971870000011
And
Figure FDA0003188971870000012
obtaining a parameter L and a central matrix H in the regression model;
obtaining a kernel matrix K of a source domain and a target domain by utilizing kernel function mapping;
according to the parameter L, the central matrix H and the kernel matrix K, solving (KLK + mu I)-1Obtaining data after TCA spectrum transformation of a source domain and a target domain by using eigenvectors corresponding to the first m eigenvalues of KHK, wherein mu is a balance parameter;
and establishing a soil nutrient and spectral reflectivity regression model for the source domain data and the target domain data after the spectral transformation by using a regression modeling algorithm, and using the regression model for detecting the soil nutrient content.
3. The migration learning-based soil nutrient spectrum detection method as claimed in claim 2, wherein the parameter L is calculated by the formula:
Figure FDA0003188971870000021
wherein I is an all 1 matrix.
4. The migration learning-based soil nutrient spectrum detection method as claimed in claim 2, wherein the calculation formula of the central matrix H is as follows:
Figure FDA0003188971870000022
wherein the content of the first and second substances,
Figure FDA0003188971870000023
is an all 1 matrix.
5. The migration learning-based soil nutrient spectral detection method as claimed in claim 2, wherein the source domain is laboratory dry soil spectral data, and the target domain is in-situ soil spectral data.
6. The migration learning based soil nutrient spectrum detection method as claimed in claim 1, further comprising:
the method comprises the following steps of pretreating soil to be detected, wherein the pretreatment comprises the following steps: impurities in the soil to be detected are removed through screening by a screening device, and water in the soil to be detected is removed through heating of the USB heating sheet.
CN202110871418.XA 2021-07-30 2021-07-30 Soil nutrient spectrum detection method based on transfer learning Pending CN113607669A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110871418.XA CN113607669A (en) 2021-07-30 2021-07-30 Soil nutrient spectrum detection method based on transfer learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110871418.XA CN113607669A (en) 2021-07-30 2021-07-30 Soil nutrient spectrum detection method based on transfer learning

Publications (1)

Publication Number Publication Date
CN113607669A true CN113607669A (en) 2021-11-05

Family

ID=78306242

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110871418.XA Pending CN113607669A (en) 2021-07-30 2021-07-30 Soil nutrient spectrum detection method based on transfer learning

Country Status (1)

Country Link
CN (1) CN113607669A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114088661A (en) * 2021-11-18 2022-02-25 云南省烟草农业科学研究院 Online prediction method for chemical components in tobacco leaf curing process based on transfer learning and near infrared spectrum

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103712923A (en) * 2013-12-23 2014-04-09 浙江大学 Method for eliminating moisture influence factor in field in-situ soil measurement spectrums
CN105300897A (en) * 2015-11-14 2016-02-03 华东交通大学 Portable soil nutrient fast-detection device for field
CN105469109A (en) * 2015-11-19 2016-04-06 中国地质大学（武汉） Transfer learning method based on class centroid alignment and for remote sensing image classification
CN212432969U (en) * 2020-04-23 2021-01-29 吉林省林业科学研究院 Portable soil nutrient detection device
CN112748232A (en) * 2020-12-22 2021-05-04 中国科学院合肥物质科学研究院 Real-time detection system for in-situ heavy metal distribution of soil profile

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103712923A (en) * 2013-12-23 2014-04-09 浙江大学 Method for eliminating moisture influence factor in field in-situ soil measurement spectrums
CN105300897A (en) * 2015-11-14 2016-02-03 华东交通大学 Portable soil nutrient fast-detection device for field
CN105469109A (en) * 2015-11-19 2016-04-06 中国地质大学（武汉） Transfer learning method based on class centroid alignment and for remote sensing image classification
CN212432969U (en) * 2020-04-23 2021-01-29 吉林省林业科学研究院 Portable soil nutrient detection device
CN112748232A (en) * 2020-12-22 2021-05-04 中国科学院合肥物质科学研究院 Real-time detection system for in-situ heavy metal distribution of soil profile

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
XIAOLI LI等: "Boosting the generalization ability of Vis-NIR-spectroscopy-based regression models through dimension reduction and transfer learning", 《COMPUTERS AND ELECTRONICS IN AGRICULTURE 》 *
ZHENGJUN QIU等: "Transfer learning method for plastic pollution evaluation in soil using NIR sensor", 《SCIENCE OF THE TOTAL ENVIRONMENT》 *
郑文瑞等: "土壤速效磷近红外迁移学习预测方法研究", 《分析测试学报》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114088661A (en) * 2021-11-18 2022-02-25 云南省烟草农业科学研究院 Online prediction method for chemical components in tobacco leaf curing process based on transfer learning and near infrared spectrum
CN114088661B (en) * 2021-11-18 2024-03-29 云南省烟草农业科学研究院 Tobacco leaf baking process chemical composition online prediction method based on transfer learning and near infrared spectrum

Similar Documents

Publication Publication Date Title
Virlet et al. Field Scanalyzer: An automated robotic field phenotyping platform for detailed crop monitoring
CN104931470B (en) A kind of pesticide residue detection device and detection method based on fluorescent high spectral technology
CN1837787A (en) Non-destructive precise determination method for biophysical parameters of cotton
Zeng et al. Quantitative visualization of photosynthetic pigments in tea leaves based on Raman spectroscopy and calibration model transfer
CN101210875A (en) Damage-free measurement method for soil nutrient content based on near infrared spectra technology
CN101210876A (en) Rice nutrient information measuring method based on visible/near infrared multispectral imaging
CN101915738A (en) Method and device for rapidly detecting nutritional information of tea tree based on hyperspectral imaging technique
CN101021472A (en) Portable multiway crop leaf nitrogen nutrient index nondestructive monitoring device
CN111751376B (en) Rice nitrogen nutrition estimation method based on canopy image feature derivation
CN107607486B (en) Soil total nitrogen detection method and device
CN103278503B (en) Multi-sensor technology-based grape water stress diagnosis method and system therefor
CN102564962A (en) System and method for detecting nutritional components in fruit trees
CN105527246A (en) Portable near infrared spectrum detection system
CN108732137B (en) Model and method for estimating plant species diversity based on hyperspectral remote sensing data
Li et al. Assessing a soil-removed semi-empirical model for estimating leaf chlorophyll content
CN115235997A (en) Soil texture inversion method based on satellite hyperspectral image
CN108613943B (en) Near-infrared single-grain crop component detection method based on spectrum morphology transfer
Xu et al. Monitoring ratio of carbon to nitrogen (C/N) in wheat and barley leaves by using spectral slope features with branch-and-bound algorithm
Wang et al. Rapid detection of chlorophyll content and distribution in citrus orchards based on low-altitude remote sensing and bio-sensors
CN114202675A (en) Crop growth parameter determination method and system based on multispectral remote sensing image of unmanned aerial vehicle
CN101059427A (en) Method for quickly non-destructive measurement for nitrogen content of tea using multiple spectrum imaging technology
Song et al. Development of crop chlorophyll detector based on a type of interference filter optical sensor
CN113607669A (en) Soil nutrient spectrum detection method based on transfer learning
CN109001143B (en) Intermediate infrared spectroscopy for sensitively predicting quality characteristics of ephedra
CN205506679U (en) Portable near infrared spectroscopy detecting system

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: 20211105