CN113221445B - Method and system for estimating soil salinity by using joint characteristics of remote sensing images - Google Patents
Method and system for estimating soil salinity by using joint characteristics of remote sensing images Download PDFInfo
- Publication number
- CN113221445B CN113221445B CN202110430543.7A CN202110430543A CN113221445B CN 113221445 B CN113221445 B CN 113221445B CN 202110430543 A CN202110430543 A CN 202110430543A CN 113221445 B CN113221445 B CN 113221445B
- Authority
- CN
- China
- Prior art keywords
- soil salinity
- model
- features
- optimal
- resolution
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/194—Terrestrial scenes using hyperspectral data, i.e. more or other wavelengths than RGB
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Mathematical Analysis (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Computational Mathematics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Algebra (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Operations Research (AREA)
- Probability & Statistics with Applications (AREA)
- Remote Sensing (AREA)
- Multimedia (AREA)
- Bioinformatics & Computational Biology (AREA)
- Databases & Information Systems (AREA)
- Astronomy & Astrophysics (AREA)
- Evolutionary Biology (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Image Processing (AREA)
Abstract
The invention provides a method and a system for estimating soil salinity by using the joint characteristics of remote sensing images; firstly, preprocessing a high-resolution second image, obtaining spectral characteristics and index characteristics by using a wave band and wave band combination calculation method, and obtaining typical texture characteristics by using a gray level co-occurrence matrix algorithm; then, screening spectral features, exponential features and texture features by using a Spearman correlation coefficient and a minimum absolute contraction selection operator to form combined features oriented to high-resolution soil salinity remote sensing inversion, and respectively constructing a model by using a partial least square regression method, a Support Vector Machine (SVM), a BP neural network and a random forest modeling method; and finally, determining important parameters of the soil salinity estimation model according to the change of the input characteristic window scale and the characteristic quantity. And evaluating the estimation performance of the model by adopting three indexes of absolute coefficient, root mean square error and residual prediction deviation, and preferably selecting the optimal soil salinity estimation model to estimate the soil salinity.
Description
Technical Field
The invention relates to the field of soil salinity estimation, in particular to a method and a system for estimating soil salinity content by using the joint characteristics of high-resolution remote sensing images.
Background
The method has the advantages that the salt content of the soil in the yellow river delta can be rapidly and accurately evaluated, and the method has important significance for fully knowing the salinization condition of the soil and effectively repairing the soil environment. The remote sensing technology has the advantages of rapidness, no damage, economy and high efficiency, and is widely used for soil salinity estimation and salinization monitoring. However, the low spatial resolution of the medium-resolution remote sensing image generally causes the problem of mixed pixels, and is usually only suitable for salt estimation of bare earth surfaces or regions with low vegetation density, which limits the accuracy of multispectral analysis to a certain extent. Therefore, it is necessary to develop a method and a system for estimating the salinity content of soil by using high-resolution remote sensing images.
Currently, many methods for estimating the salt content of soil by using high-resolution remote sensing images have appeared. For example, "Pereira O J R, melfi A J, montes C R. Image Fusion of Sentinal-2 and CBERS-4 samples for mapping soil coverers in the Wetlands of Pantianal J. International Journal of Image & Data Fusion, 2017, 8 (1-4): 148-172" and "Wang Z, zhang X L, zhang F, et al. High spatial resolution images, which are an important branch of remote sensing development, apparently fail to be used alone for soil salinity estimation due to their lack of spectral information.
Disclosure of Invention
The invention aims to improve the precision of high-resolution remote sensing inversion of soil salinity and provides a method and a system for estimating the soil salinity content by using the joint characteristics of high-resolution remote sensing images.
In order to achieve the purpose, the invention adopts the following technical scheme:
in a first aspect, the invention provides a method for estimating soil salinity content by using combined characteristics of high-resolution remote sensing images, which comprises the following steps:
step A: preprocessing the high-resolution second image, obtaining spectral characteristics and index characteristics by using a wave band and wave band combination calculation method, and obtaining typical texture characteristics by using a gray level co-occurrence matrix algorithm;
and B, step B: screening spectral characteristics, exponential characteristics and texture characteristics by using a Spearman correlation coefficient (r) and a minimum absolute shrinkage selection operator (LASSO) to form a high-resolution soil salinity remote sensing inversion-oriented combined characteristic, and constructing a soil salinity estimation model by respectively using a Partial Least Squares Regression (PLSR), a Support Vector Machine (SVM), a BP neural network (BPNN) and a Random Forest (RF) modeling method;
step C: determining important parameters of a soil salinity estimation model according to the change of the input characteristic window scale and the characteristic quantity; using absolute coefficients (R) 2 ) The estimation performance of the model is evaluated by three indexes, namely Root Mean Square Error (RMSE) and Residual Prediction Deviation (RPD), and an optimal model is optimized (wherein the optimal model needs to meet the following conditions: r 2 Maximum value, RMSE value minimum, RPD value greater than 2 and maximum), soil salinity estimation was performed.
Preferably, in the step a, the high-resolution second image is preprocessed, spectral features and index features are obtained by using a method of band and band combination calculation, and typical texture features are obtained by using a gray level co-occurrence matrix algorithm: and carrying out radiometric calibration and atmospheric correction on the high-resolution second image to obtain the earth surface reflectivity. Spectral characteristics are obtained by using blue (B), green (G), red (R) and Near Infrared (NIR) wave bands of images, exponential characteristics are obtained by using combination calculation among the blue (B), green (G), red (R) and Near Infrared (NIR) wave bands of images, and typical texture characteristics are obtained by using a gray level co-occurrence matrix algorithm.
Further preferably, in the step B, spectral features, exponential features and texture features are screened by using Spearman correlation coefficient (r) and Least Absolute Shrinkage Selection Operator (LASSO) to form a combined feature oriented to high-resolution soil salinity remote sensing inversion: and calculating a Spearman correlation coefficient (r) and a minimum absolute shrinkage selection operator (LASSO) between the spectral characteristics, the exponential characteristics and the texture characteristics and the soil salinity, and sorting the importance of the characteristics by using the minimum absolute shrinkage selection operator (LASSO) to screen out the combined characteristics facing the high-resolution soil salinity remote sensing inversion.
Preferably, the step C extracts the feature that the minimum absolute shrinkage selection operator is greater than a set value under different feature window scales; sorting the importance of the features by using the minimum absolute shrinkage selection operator value, selecting the feature quantity with a set step length according to the importance from high to low in sequence, and respectively performing partial least square regression, a support vector machine and BP nervesCombining a network with a random forest modeling method to construct a soil salinity estimation model; and determining the optimal window size and the optimal characteristic parameter of the model. Using absolute coefficients (R) 2 ) Evaluating the estimation performance of the model by three indexes, namely Root Mean Square Error (RMSE) and Residual Prediction Deviation (RPD), and determining an optimal estimation model (wherein the optimal model needs to meet the following conditions: r 2 Maximum value, RMSE value minimum, RPD value greater than 2 and maximum) were evaluated for soil salinity.
In a second aspect, the present invention further provides a system for estimating soil salinity content by using the combined features of high-resolution remote sensing images, including:
the image preprocessing and feature calculating module comprises: the system is configured to preprocess a high-resolution second image, obtain spectral characteristics and index characteristics by using a wave band and wave band combination calculation method, and calculate typical texture characteristics by using a gray level co-occurrence matrix algorithm;
a feature screening and modeling module: the method comprises the steps that spectral characteristics, exponential characteristics and texture characteristics are screened by using a Spearman correlation coefficient (r) and a minimum absolute shrinkage selection operator (LASSO), combined characteristics facing high-resolution soil salinity remote sensing inversion are formed, and a soil salinity estimation model is constructed by using Partial Least Squares Regression (PLSR), a Support Vector Machine (SVM), a BP neural network (BPNN) and a Random Forest (RF) modeling method respectively;
a model performance evaluation module: determining important parameters of the soil salinity estimation model according to the change of the input characteristic window scale and the characteristic quantity; using absolute coefficients (R) 2 ) Evaluating model estimation performance by three indexes of Root Mean Square Error (RMSE) and Residual Prediction Deviation (RPD), and preferably selecting optimal model (R) 2 Maximum value, RMSE value minimum, RPD value greater than 2 and maximum), soil salinity estimation was performed.
Preferably, in the image preprocessing and feature calculating module, radiometric calibration and atmospheric correction are performed on the high-resolution second image to obtain a surface reflectance, spectral features are obtained by using blue (B), green (G), red (R) and Near Infrared (NIR) bands of the image, exponential features are obtained by using combination calculation among the blue (B), green (G), red (R) and Near Infrared (NIR) bands of the image, and typical texture features are obtained by using a gray level co-occurrence matrix algorithm.
Preferably, in the feature screening and modeling module, a Spearman correlation coefficient (r) between the spectral features, the exponential features, the textural features and the soil salinity and a minimum absolute shrinkage selection operator are calculated, importance ranking is carried out on the features by using the minimum absolute shrinkage selection operator, and combined features facing high-resolution soil salinity remote sensing inversion are screened out.
Preferably, in the model performance evaluation module, the model (R) is estimated according to the optimal feature number modeled at different window scales and the optimal soil salinity 2 Maximum value, minimum RMSE value, RPD value greater than 2 and maximum) accuracy, determining the optimal window size and optimal feature number for soil salinity estimation.
The invention has the beneficial effects that:
firstly, preprocessing a high-resolution second image, obtaining spectral characteristics and index characteristics by using a wave band and wave band combination calculation method, and obtaining typical texture characteristics by using a gray level co-occurrence matrix algorithm; then, spectrum characteristics, exponential characteristics and texture characteristics are screened by using a Spearman correlation coefficient (r) and a minimum absolute shrinkage selection operator (LASSO) to form a combined characteristic facing high-resolution soil salinity remote sensing inversion, and a model is constructed by respectively using a Partial Least Squares Regression (PLSR), a Support Vector Machine (SVM), a BP neural network (BPNN) and a Random Forest (RF) modeling method; and finally, determining important parameters of the soil salinity estimation model according to the change of the input characteristic window scale and the characteristic quantity. Using absolute coefficients (R) 2 ) Evaluating the estimation performance of the model by three indexes of Root Mean Square Error (RMSE) and residual prediction bias (RPD), and preferably selecting the optimal soil salinity estimation model (R) 2 Maximum value, RMSE value minimum, RPD value greater than 2 and maximum) were evaluated for soil salinity.
Drawings
FIG. 1 is a schematic diagram of a method and system for estimating the salinity content of soil using the combined features of high-resolution remote sensing images;
FIG. 2 is a chart of the Spearman correlation coefficient (r) between various characteristics and soil salt content;
FIG. 3 is a combined characteristic diagram for high-resolution soil salinity remote sensing inversion;
4 (a) -4 (f) are feature maps with minimum absolute shrinkage selection operators (LASSO) greater than 0.1 for different feature window scales;
FIG. 5 is a plot of salinity estimation model accuracy and optimal feature number for different feature window scales;
Detailed Description
It is to be understood that the following detailed description is exemplary and is intended to provide further explanation of the invention as claimed. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the invention. As used herein, the singular forms "a", "an", and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof;
according to the research, the salinity index and the vegetation index can be used as indirect indexes for estimating the salinity content of the soil. The high-resolution remote sensing expands the applicable areas of various spectral indexes, so that the high-resolution remote sensing is less limited by the conditions of earth surface vegetation coverage and the like. In addition, the gray level co-occurrence matrix is used as a typical method for extracting the texture characteristics, and the physicochemical properties of the soil can also be expressed. Therefore, the method combines the spectral characteristics, the index characteristics and the texture characteristics, and estimates the soil salinity through the high-resolution remote sensing image is a feasible way. In view of this, the present embodiment provides a method and a system for estimating the salinity content of soil by using the joint features of high-resolution remote sensing images.
The invention is further described with reference to the following figures and examples.
In order to clearly understand the technical features, purposes and effects of the present invention, the remote sensing image spectrum data with high resolution No. two is taken as an example, and the specific implementation of the present invention is described with reference to the accompanying drawings.
The technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.
The method for estimating the salinity content of the soil by using the combined characteristics of the high-resolution remote sensing images comprises the following steps:
step A: preprocessing the high-resolution second image, obtaining spectral characteristics and index characteristics by using a wave band and wave band combination calculation method, and obtaining typical texture characteristics by using a gray level co-occurrence matrix algorithm;
and B: screening spectral characteristics, exponential characteristics and texture characteristics by using a Spearman correlation coefficient (r) and a minimum absolute shrinkage selection operator (LASSO) to form a high-resolution soil salinity remote sensing inversion-oriented combined characteristic, and constructing a soil salinity estimation model by respectively using a Partial Least Squares Regression (PLSR), a Support Vector Machine (SVM), a BP neural network (BPNN) and a Random Forest (RF) modeling method;
and C: determining important parameters of a soil salinity estimation model according to the change of the input characteristic window scale and the characteristic quantity; using absolute coefficients (R) 2 ) The method comprises the following steps of evaluating model estimation performance by three indexes of Root Mean Square Error (RMSE) and Residual Prediction Deviation (RPD), and preferably selecting an optimal model (wherein the optimal model needs to meet the following conditions: r 2 Maximum value, RMSE value minimum, RPD value greater than 2 and maximum), soil salinity estimation was performed.
The high-resolution second-order image is preprocessed, spectral features and index features are obtained by a method of band and band combination calculation, and typical texture features are obtained by a gray level co-occurrence matrix algorithm; the method comprises the following steps:
a: preprocessing the high-resolution second image, obtaining spectral characteristics and index characteristics by using a wave band and wave band combination calculation method, and obtaining typical texture characteristics by using a gray level co-occurrence matrix algorithm.
a: carrying out radiometric calibration and atmospheric correction on the high-resolution second image to obtain the earth surface reflectivity;
b: obtaining spectral characteristics by using image blue (B), green (G), red (R) and Near Infrared (NIR) wave bands;
c: calculating vegetation index, salinity index, brightness index and intensity index by using the combination of blue (B), green (G), red (R) and Near Infrared (NIR) bands of the image;
the calculation formula of the vegetation index is as follows:
NDVI =
(1)
SAVI = 
(2)
EVI = 
(3)
wherein NDVI is a normalized vegetation index, SAVI is a soil adjusted vegetation index L =0.5, and EVI is an enhanced vegetation index salinity.
The formula for the calculation of the index is as follows:
SI1 = 
(4)
SI2 = 
(5)
SI3 = 
(6)
SI4 = 
(7)
wherein SI1 is a salinity index of 1, SI2 is a salinity index of 2, SI3 is a salinity index of 3, and SI4 is a salinity index of 4.
The formula for calculating the luminance index is as follows:
BRI = 
(8)
the calculation formula of the intensity index is as follows:
Int1 = 
(9)
Int2 = 
(10)
wherein Int1 is intensity index 1 and Int2 is intensity index 2.
d: obtaining typical texture features by utilizing a gray level co-occurrence matrix algorithm;
the texture index is calculated as follows:
CON =
(11)
ENT =
(12)
ASM =
(13)
COR =
(17)
wherein the content of the first and second substances,N g is the number of gray levels, g: (i,j) Is in a gray scale matrix ofi,j) The gray value of the picture element is,μis the average value of the gray matrix, delta 2 Is the gray matrix variance.
The spectral characteristics, the exponential characteristics and the texture characteristics are screened by utilizing a Spearman correlation coefficient (r) and a minimum absolute shrinkage selection operator (LASSO), as shown in a figure 2 and a figure 3, the method is based on the Spearman correlation coefficient (r) and the minimum absolute shrinkage selection operator (LASSO) between each characteristic and the soil salinity content, the importance degree of the characteristics is sorted by utilizing the minimum absolute shrinkage selection operator (LASSO), and the combined characteristics facing the high-resolution soil salinity remote sensing inversion are screened out; the method comprises the following steps:
b: spectral characteristics, exponential characteristics and texture characteristics are screened by using a Spearman correlation coefficient (r) and a minimum absolute shrinkage selection operator (LASSO) to form a high-resolution soil salinity remote sensing inversion-oriented combined characteristic, and a soil salinity estimation model is constructed by respectively using Partial Least Squares Regression (PLSR), support Vector Machine (SVM), BP neural network (BPNN) and Random Forest (RF) modeling methods.
a: calculating Spearman correlation coefficients (r) between each feature and soil salinity content;
FIG. 2 is a graphical representation of the Spearman correlation coefficient (r) between various features and soil salt content.
b: calculating a minimum absolute shrinkage selection operator (LASSO), sorting the importance of the features by using the minimum absolute shrinkage selection operator (LASSO), and screening out combined features for high-resolution soil salinity remote sensing inversion;
FIG. 3 is a schematic diagram of combined features screened out by using a minimum absolute shrinkage selection operator (LASSO) and forming the remote sensing inversion oriented to high-resolution soil salinity.
c: and (3) respectively combining the combined characteristics with Partial Least Squares Regression (PLSR), support Vector Machine (SVM), BP neural network (BPNN) and Random Forest (RF) modeling methods to construct a soil salinity estimation model.
According to the method, important parameters of a soil salinity estimation model are determined according to the change of input characteristic window scales and characteristic quantities; using absolute coefficients (R) 2 ) Evaluating the estimation performance of the model by three indexes of Root Mean Square Error (RMSE) and residual prediction bias (RPD), and preferably selecting the optimal soil salinity estimation model (R) 2 Maximum value, RMSE value minimum, RPD value greater than 2 and maximum) were evaluated for soil salinity.
Evaluating the performance of the model; the method comprises the following steps:
in order to more clearly understand the technical features, purposes and effects of the present invention, the feature window scales in this patent are classified into 3 × 3,5 × 5,7 × 7,9 × 9, 11 × 11 and 13 × 13.
Fig. 4 (a) -4 (f) feature maps for which the minimum absolute shrinkage selection operator (LASSO) is greater than 0.1 at the 3 × 3,5 × 5,7 × 7,9 × 9, 11 × 11, and 13 × 13 feature window scales, respectively.
C: and determining important parameters of the soil salinity estimation model according to the change of the input characteristic window scale and the characteristic quantity. Using absolute coefficients (R) 2 ) Evaluating the estimation performance of a model by three indexes including Root Mean Square Error (RMSE) and residual prediction bias (RPD), and preferably selecting an optimal soil salinity estimation model (wherein the optimal model needs to meet the following conditions: r 2 Maximum value, RMSE value minimum, RPD value greater than 2 and maximum) were evaluated for soil salinity.
a: extracting the features of which the minimum absolute shrinkage selection operator (LASSO) is greater than 0.1 under different feature window scales;
b: sorting the importance of the features by using a minimum absolute shrinkage selection operator (LASSO) value, sequentially selecting the quantity of the features by 1 step length according to the importance from high to low, and respectively combining with Partial Least Squares Regression (PLSR), support Vector Machine (SVM), BP neural network (BPNN) and Random Forest (RF) modeling methods to construct a soil salinity estimation model;
FIG. 5 is a graph of salinity estimation model accuracy and optimal feature number curves at different feature window scales.
c: and determining the optimal window size and the optimal characteristic parameter of the model. Using absolute coefficients (R) 2 ) Evaluating the model estimation performance by three indexes of Root Mean Square Error (RMSE) and Residual Prediction Deviation (RPD), and determining the optimal estimation model (R) 2 Maximum value, RMSE value minimum, RPD value greater than 2 and maximum) were evaluated for soil salinity.
Example 2
Based on the same inventive concept, the embodiment also provides a system for estimating soil salinity by using the joint characteristics of the high-resolution remote sensing images, and as the problem solving principle of the system is similar to that of the method for estimating soil salinity by using the joint characteristics of the high-resolution remote sensing images, the implementation of the system can be realized according to the specific steps of the method, and repeated parts are not repeated.
The system for estimating the salt content of the soil by using the joint characteristics of the high-resolution remote sensing image, provided by the embodiment, includes:
the image preprocessing and feature calculating module comprises: the system is configured to preprocess a high-resolution second image, obtain spectral characteristics and index characteristics by using a wave band and wave band combination calculation method, and calculate typical texture characteristics by using a gray level co-occurrence matrix algorithm;
a feature screening and modeling module: the method comprises the steps of screening spectral features, exponential features and textural features by using a Spearman correlation coefficient and a minimum absolute contraction selection operator to form a combined feature oriented to high-resolution soil salinity remote sensing inversion, and constructing a soil salinity estimation model by using a partial least squares regression method, a Support Vector Machine (SVM), a BP neural network method and a random forest modeling method respectively;
a model performance evaluation module: and determining important parameters of the soil salinity estimation model according to the change of the input characteristic window scale and the characteristic quantity. Using absolute coefficients (R) 2 ) Evaluating the estimation performance of the model by three indexes of Root Mean Square Error (RMSE) and residual prediction bias (RPD), and preferably selecting the optimal soil salinity estimation model (R) 2 Maximum value, RMSE valueMinimum, RPD greater than 2 and maximum) were evaluated for soil salinity.
It will be understood by those skilled in the art that all or part of the processes of the above embodiments may be implemented by a computer program, which may be stored in a computer readable storage medium and executed by a computer, and the processes of the above embodiments of the method may be included. The storage medium may be a magnetic disk, an optical disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), or the like.
The technical features of the embodiments described above may be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the embodiments described above are not described, but should be considered as being within the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
Although the embodiments of the present invention have been described with reference to the accompanying drawings, it is not intended to limit the scope of the invention, and it should be understood by those skilled in the art that various modifications and variations can be made without inventive changes in the technical solutions of the present invention.
Claims (5)
1. The method for estimating the soil salinity content by using the combined characteristics of the high-resolution remote sensing images is characterized by comprising the following steps of:
step A: preprocessing the high-resolution second image, obtaining spectral characteristics and index characteristics by using a wave band and wave band combination calculation method, and obtaining typical texture characteristics by using a gray level co-occurrence matrix algorithm;
and B: screening spectral features, exponential features and textural features by using a Spearman correlation coefficient and a minimum absolute contraction selection operator to form combined features oriented to high-resolution soil salinity remote sensing inversion, and constructing a soil salinity estimation model by respectively using a partial least squares regression method, a Support Vector Machine (SVM), a BP neural network and a random forest modeling method;
and C: determining important parameters of a soil salinity estimation model according to the change of the input characteristic window scale and the characteristic quantity; and evaluating the estimation performance of the model by adopting three indexes of absolute coefficient, root mean square error and residual prediction deviation, preferably selecting an optimal model, and estimating the content of the soil salt, wherein the conditions of the optimal model are as follows: the absolute coefficient value is maximum, the root mean square error value is minimum, and the residual prediction deviation value is greater than 2 and is maximum;
calculating spectral characteristics, index characteristics, spearman correlation coefficients between texture characteristics and soil salinity and a minimum absolute shrinkage selection operator in the step B, and sorting the importance of the characteristics by using the minimum absolute shrinkage selection operator to screen out combined characteristics facing high-resolution soil salinity remote sensing inversion;
in the step C, determining an optimal model, an optimal window scale and an optimal characteristic number of the soil salinity estimation model according to the optimal characteristic number of the modeling under different window scales and the change of model precision;
c, extracting the features of which the minimum absolute shrinkage selection operator is larger than a set value under different feature window scales; sorting the importance of the features by using the minimum absolute shrinkage selection operator value, selecting feature quantity according to the importance from high to low in sequence by a set step length, and respectively combining with a partial least square regression method, a support vector machine method, a BP neural network method and a random forest modeling method to construct a soil salinity estimation model; and determining the optimal window size and the optimal characteristic parameter of the model.
2. The method as claimed in claim 1, wherein in the step A, radiometric calibration and atmospheric correction are performed on the high-resolution second image to obtain the earth surface reflectivity; obtaining spectral characteristics by using blue, green, red and near infrared bands of the image; calculating by using the combination of blue, green, red and near infrared bands of the image to obtain a vegetation index, a salinity index, a brightness index and an intensity index; and obtaining typical texture features by utilizing a gray level co-occurrence matrix algorithm.
3. The method of claim 1, wherein the soil salinity estimation model is constructed using combined features respectively combined with partial least squares regression, support vector machines, BP neural networks, and random forest modeling methods.
4. Utilize the joint characteristic of high resolution remote sensing image to carry out soil salinity content estimation's system, characterized by includes:
the image preprocessing and feature calculating module comprises: the system is configured to preprocess a high-resolution second image, obtain spectral characteristics and index characteristics by using a wave band and wave band combination calculation method, and calculate typical texture characteristics by using a gray level co-occurrence matrix algorithm;
a feature screening and modeling module: the method is configured to screen spectral features, exponential features and texture features by using a Spearman correlation coefficient and a minimum absolute contraction selection operator to form a combined feature oriented to high-resolution soil salinity remote sensing inversion, and a soil salinity estimation model is constructed by respectively using a partial least squares regression method, a Support Vector Machine (SVM), a BP neural network method and a random forest modeling method;
a model performance evaluation module: determining important parameters of the soil salinity estimation model according to the change of the input characteristic window scale and the characteristic quantity; evaluating the estimation performance of the model by adopting three indexes of absolute coefficient, root mean square error and residual prediction deviation, and preferably selecting the optimal model to estimate the content of the soil salinity, wherein the conditions of the optimal model are as follows: the absolute coefficient value is maximum, the root mean square error value is minimum, and the residual prediction deviation value is greater than 2 and is maximum;
in the characteristic screening and modeling module, spectral characteristics, spearman correlation coefficients between exponential characteristics and textural characteristics and the salt content of the soil and a minimum absolute shrinkage selection operator are calculated, importance ranking is carried out on the characteristics by using the minimum absolute shrinkage selection operator, and combined characteristics facing high-resolution soil salt remote sensing inversion are screened out;
in the model performance evaluation module, an optimal model for soil salinity estimation, an optimal window scale and an optimal feature number are determined according to the optimal feature number of modeling under different window scales and the change of model precision;
in the model performance evaluation module, extracting the characteristic that the minimum absolute shrinkage selection operator is larger than a set value under different characteristic window scales; sorting the importance of the features by using the minimum absolute shrinkage selection operator value, selecting feature quantity according to the importance from high to low in sequence by a set step length, and respectively combining with a partial least square regression method, a support vector machine method, a BP neural network method and a random forest modeling method to construct a soil salinity estimation model; and determining the optimal window size and the optimal characteristic parameter of the model.
5. The system of claim 4, wherein the image preprocessing and feature calculation module performs radiometric calibration and atmospheric correction on the top-scoring second-order image to obtain surface reflectance, obtains spectral features using blue, green, red and near infrared bands of the image, obtains exponential features using a combination calculation between blue, green, red and near infrared bands of the image, and obtains typical texture features using a gray level co-occurrence matrix algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110430543.7A CN113221445B (en) | 2021-04-21 | 2021-04-21 | Method and system for estimating soil salinity by using joint characteristics of remote sensing images |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110430543.7A CN113221445B (en) | 2021-04-21 | 2021-04-21 | Method and system for estimating soil salinity by using joint characteristics of remote sensing images |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113221445A CN113221445A (en) | 2021-08-06 |
| CN113221445B true CN113221445B (en) | 2023-01-17 |
Family
ID=77088258
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110430543.7A Active CN113221445B (en) | 2021-04-21 | 2021-04-21 | Method and system for estimating soil salinity by using joint characteristics of remote sensing images |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113221445B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113610062B (en) * | 2021-10-08 | 2021-12-10 | 天津市勘察设计院集团有限公司 | Soil sample characteristic extraction method and system and information data processing terminal |
| CN114062439B (en) * | 2021-11-18 | 2022-08-23 | 浙江大学 | Method for jointly estimating salinity of soil profile by using time series remote sensing images |
| CN114076738B (en) * | 2021-11-22 | 2023-04-18 | 浙江大学 | Method for estimating soil salinity of residual straw farmland by using remote sensing construction index |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109344550A (en) * | 2018-11-26 | 2019-02-15 | 国智恒北斗科技集团股份有限公司 | A kind of forest reserves inversion method and system based on domestic high score satellite remote sensing date |
| CN111429980A (en) * | 2020-04-14 | 2020-07-17 | 北京迈高材云科技有限公司 | Automatic acquisition method for material crystal structure characteristics |
| CN111476170A (en) * | 2020-04-09 | 2020-07-31 | 首都师范大学 | Remote sensing image semantic segmentation method combining deep learning and random forest |
| CN111582194A (en) * | 2020-05-12 | 2020-08-25 | 吉林大学 | Multi-temporal high-resolution remote sensing image building extraction method based on multi-feature LSTM network |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105608293B (en) * | 2016-01-28 | 2019-10-11 | 武汉大学 | Merge Biomass retrieval method and system on the woodland of spectrum and textural characteristics |
| CN110046415A (en) * | 2019-04-08 | 2019-07-23 | 中国科学院南京地理与湖泊研究所 | A kind of soil organic matter content remote sensing dynamic playback method of space-time fining |
-
2021
- 2021-04-21 CN CN202110430543.7A patent/CN113221445B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109344550A (en) * | 2018-11-26 | 2019-02-15 | 国智恒北斗科技集团股份有限公司 | A kind of forest reserves inversion method and system based on domestic high score satellite remote sensing date |
| CN111476170A (en) * | 2020-04-09 | 2020-07-31 | 首都师范大学 | Remote sensing image semantic segmentation method combining deep learning and random forest |
| CN111429980A (en) * | 2020-04-14 | 2020-07-17 | 北京迈高材云科技有限公司 | Automatic acquisition method for material crystal structure characteristics |
| CN111582194A (en) * | 2020-05-12 | 2020-08-25 | 吉林大学 | Multi-temporal high-resolution remote sensing image building extraction method based on multi-feature LSTM network |
Non-Patent Citations (2)
| Title |
|---|
| 分数阶微分在土壤盐渍化遥感监测中的应用研究;张东;《中国博士学位论文全文数据库 农业科技辑》;20170915;第二章、第五章-第六章 * |
| 双参数MRI影像组学模型对前列腺良性增生结节和前列腺癌的鉴别价值;邢朋毅 等;《实用医学杂志》;20210228;第38卷(第2期);第107-111页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113221445A (en) | 2021-08-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113221445B (en) | Method and system for estimating soil salinity by using joint characteristics of remote sensing images | |
| US11812684B2 (en) | Computer-implemented methods, computer readable medium and systems for a precision agriculture platform that detects healthy conditions | |
| Brosofske et al. | A review of methods for mapping and prediction of inventory attributes for operational forest management | |
| US20180330245A1 (en) | Computer-implemented methods, computer readable medium and systems for generating an orchard data model for a precision agriculture platform | |
| Peng et al. | Modeling soil organic carbon at regional scale by combining multi-spectral images with laboratory spectra | |
| US10748081B2 (en) | Computer-implemented methods, computer readable medium and systems for a precision agriculture platform that identifies generic anomalies in crops | |
| US7221798B2 (en) | Method for selecting representative endmember components from spectral data | |
| US10395355B2 (en) | Computer-implemented methods, computer readable medium and systems for a precision agriculture platform | |
| US10410334B2 (en) | Computer-implemented methods, computer readable medium and systems for a precision agriculture platform with a satellite data model | |
| Chen et al. | Soil salinity inversion based on differentiated fusion of satellite image and ground spectra | |
| US10943173B2 (en) | Computer-implemented methods, computer readable medium and systems for generating a satellite data model for a precision agriculture platform | |
| CN115271217A (en) | Wheat yield prediction method based on multi-source remote sensing data of unmanned aerial vehicle | |
| CN115015258B (en) | Crop growth vigor and soil moisture association determination method and related device | |
| CN114199800B (en) | Rice sheath blight disease identification method, system, equipment and medium | |
| CN117115669B (en) | Object-level ground object sample self-adaptive generation method and system with double-condition quality constraint | |
| Ashaari et al. | Comparison of model accuracy in tree canopy density estimation using single band, vegetation indices and forest canopy density (FCD) based on Landsat-8 imagery (case study: Peat Swamp Forest in Riau Province) | |
| CN117035066A (en) | Ground surface temperature downscaling method coupling geographic weighting and random forest | |
| CN110070513B (en) | Radiation correction method and system for remote sensing image | |
| CN117075138A (en) | Remote sensing measurement and calculation method, system and medium for canopy height of 30-meter forest in area | |
| Di Michele et al. | Physically based statistical integration of TRMM microwave measurements for precipitation profiling | |
| Dobbs et al. | Using structure-from-motion to estimate cover crop biomass and characterize canopy structure | |
| CN112329856B (en) | Wheat nitrogen accumulation amount prediction method based on spectrum and RGB-D fusion image | |
| Hamzeh et al. | Retrieval of Sugarcane Leaf Area Index From Prisma Hyperspectral Data | |
| Yahia et al. | A new fully constrained least squares-based fusion approach of optical, thermal, and SAR remote sensing data for soil moisture content estimation | |
| Liang et al. | Retrieval of aerosol optical depth (AOD) from the Landsat8 Oli observations over Beijing |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |