CN115372970A - Remote sensing extraction method for crops SAR in mountainous and hilly areas - Google Patents
Remote sensing extraction method for crops SAR in mountainous and hilly areas Download PDFInfo
- Publication number
- CN115372970A CN115372970A CN202210996592.1A CN202210996592A CN115372970A CN 115372970 A CN115372970 A CN 115372970A CN 202210996592 A CN202210996592 A CN 202210996592A CN 115372970 A CN115372970 A CN 115372970A
- Authority
- CN
- China
- Prior art keywords
- time sequence
- scattering
- remote sensing
- radar
- component
- 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.)
- Granted
Links
- 238000000605 extraction Methods 0.000 title claims description 42
- 230000010287 polarization Effects 0.000 claims abstract description 68
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 42
- 238000000034 method Methods 0.000 claims abstract description 36
- 239000011159 matrix material Substances 0.000 claims abstract description 24
- 238000012216 screening Methods 0.000 claims abstract description 9
- 238000007781 pre-processing Methods 0.000 claims abstract description 8
- 238000011160 research Methods 0.000 claims description 27
- 238000005070 sampling Methods 0.000 claims description 15
- 238000007637 random forest analysis Methods 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 9
- 238000012937 correction Methods 0.000 claims description 8
- 230000015572 biosynthetic process Effects 0.000 claims description 5
- 150000001875 compounds Chemical class 0.000 claims description 5
- 238000003786 synthesis reaction Methods 0.000 claims description 5
- 230000001427 coherent effect Effects 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 claims description 3
- 244000038559 crop plants Species 0.000 claims description 2
- 230000001186 cumulative effect Effects 0.000 claims description 2
- 238000011156 evaluation Methods 0.000 claims description 2
- 230000003287 optical effect Effects 0.000 abstract description 4
- 238000000691 measurement method Methods 0.000 abstract description 3
- 241000196324 Embryophyta Species 0.000 description 11
- 238000005516 engineering process Methods 0.000 description 11
- 230000017260 vegetative to reproductive phase transition of meristem Effects 0.000 description 8
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 8
- 235000006008 Brassica napus var napus Nutrition 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 6
- 240000000385 Brassica napus var. napus Species 0.000 description 5
- 230000028161 membrane depolarization Effects 0.000 description 5
- 239000002689 soil Substances 0.000 description 5
- 239000013598 vector Substances 0.000 description 5
- 238000003384 imaging method Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 230000009977 dual effect Effects 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000000149 penetrating effect Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- 238000012935 Averaging Methods 0.000 description 1
- 240000002791 Brassica napus Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000002592 echocardiography Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 210000000056 organ Anatomy 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 230000007903 penetration ability Effects 0.000 description 1
- 230000000243 photosynthetic effect Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/88—Radar or analogous systems specially adapted for specific applications
- G01S13/89—Radar or analogous systems specially adapted for specific applications for mapping or imaging
- G01S13/90—Radar or analogous systems specially adapted for specific applications for mapping or imaging using synthetic aperture techniques, e.g. synthetic aperture radar [SAR] techniques
- G01S13/9021—SAR image post-processing techniques
Landscapes
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Radar, Positioning & Navigation (AREA)
- Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Physics & Mathematics (AREA)
- Image Processing (AREA)
Abstract
A method for extracting crops SAR in mountainous and hilly areas in a remote sensing mode comprises the following steps: s1, preprocessing dual-polarization SAR time sequence data and extracting a time sequence scattering matrix; s2, carrying out three-component compact polarization decomposition on the time sequence scattering matrix to obtain a secondary scattering component, a volume scattering component and a surface scattering component, and constructing a three-component compact polarization decomposition radar vegetation index; s3, decomposing the radar vegetation index based on three-component compact polarization, and constructing a radar vegetation index standard time sequence curve; s4, screening the extracted time sequence combination of the optimal crop planting area; and S5, comparing the similarity between the time sequence curve of the pixels to be classified and the standard time sequence curve pixel by using the time sequence combination extracted from the optimal crop planting area according to the radar vegetation index time sequence standard curve, and judging the pixels to be judged as the pixel classes with the maximum similarity. The invention solves the problems that the terrain of mountainous and hilly areas is complex, the mountainous and hilly areas are interfered by meteorological conditions, the optical remote sensing data is difficult to obtain, and the work efficiency of the crop area ground measurement method is low.
Description
Technical Field
The invention relates to the technical field of crop remote sensing, in particular to a method for extracting SAR remote sensing of crops in mountainous and hilly areas.
Background
In mountainous and hilly areas in south China, the traditional crop area ground measurement method is low in working efficiency, cannot obtain a real-time and dynamic crop planting area monitoring result, and consumes a large amount of manpower and material resources. The remote sensing technology can monitor the crop area in real time, dynamically and in a large range, and becomes an important data source and a main technical means for monitoring the crop planting area in mountainous and hilly areas in recent years. Mountainous and hilly areas in south China are not only complex in terrain, but also often interfered by weather conditions such as cloud, rain and fog, and stable and effective optical remote sensing data acquisition is difficult. With the development of Radar remote sensing technology, the C-band Synthetic Aperture Radar (SAR) is applied to the fields of agricultural resource investigation, land resource utilization, crop type identification and the like in mountainous and hilly areas by virtue of the penetrating capability of the SAR all day long, all weather and without being influenced by weather conditions such as cloud, rain, fog and the like, and the sensitivity to crops and soil structures and characteristics. Sentinel No. 1 (Sentinel-1) SAR data is the only SAR data source available for free download. The Sentinel-1SAR satellite consists of two polar orbit satellites in the same orbital plane, and successfully transmits in 4-month and 3-month 2014 and 25-month 2016 4-month. The satellite-mounted C-band synthetic aperture radar has 4 imaging modes, wherein an IW (Interferometric Wide weather) mode acquires 250 kilometer-long data at a spatial resolution (single view) of 5 meters multiplied by 20 meters (distance direction multiplied by azimuth direction), and has the advantages of high spatial resolution and large coverage range. The Sentinel-1 radar data covering the main crop producing area in south China is revisited once every 12 days, and data guarantee can be provided for remote sensing extraction of the crop planting area in mountain areas and hilly areas in south China.
The scattering of microwaves by the crop itself and the interaction between the crop and radar microwaves are complex processes. With the rapid development of microwave imaging technology, when a large amount of SAR data is obtained, how to extract target crop scattering characteristics from the SAR data and identify crops becomes one of the key problems in the research of crop microwave remote sensing application. The polarization decomposition technology can separate polarization characteristics caused by different scattering mechanisms of various ground objects, and the main scattering characteristics of the target ground object can be effectively extracted based on the polarization decomposition technology so as to extract the target ground object. The more representative polarization decomposition models are mainly divided into two types, one is coherent object polarization decomposition based on scattering matrix, such as Pauli decomposition, krogger decomposition and Cameron decomposition, and the other is incoherent object polarization decomposition based on scattering matrix second-order moment, such as Freeman decomposition, H/A/alpha decomposition and Van Zyl decomposition. Most of the ground objects in the nature are distributed target scatterers, belong to incoherent targets, and the incoherent target polarization decomposition method is adopted to analyze and describe the natural ground objects so as to better accord with the microwave scattering characteristics of the natural ground objects. The polarimetric decomposition technology is mainly researched and applied to the fully polarimetric SAR data, and the Sentinel-1SAR data only has two wave bands of VV and VH. In recent years, a new compact polarimetric decomposition technology brings new ideas and technical supports for polarimetric decomposition of dual-polarized Sentinel-1SAR data and ground object polarimetric feature extraction. The compact polarization SAR is a dual-polarization SAR system essentially, and the compact polarization decomposition technology can be applied to dual-polarization SAR data. At present, 3 compact polarization SAR working modes are proposed, namely a pi/4 mode for transmitting a 45-degree linear polarization wave and receiving H and V linear polarization waves; a Dual Circular Polarization (DCP) mode for transmitting left-hand or right-hand circularly polarized waves and receiving left-hand and right-hand circularly polarized waves; a Hybrid Polarization (HP) mode, which transmits left-hand or right-hand circularly polarized waves, receives H and V linearly polarized waves, is also called a Circular polarization Transmit and Linear Receive (CTLR) mode. Compared with the traditional linear dual-polarization SAR, the compact polarization SAR can store the phase of an echo signal, and the signal combination mode is more flexible, so that richer scattering information can be obtained, and a result similar to the fully-polarized SAR data is obtained in many applications. In addition, the Radar Vegetation Index (RVI) is a comprehensive index for describing ground feature scattering characteristics after enhancement operation of each wave band of the multi-polarization Radar, and can reduce interference factors such as Radar system noise, microwave incident angle and plant individual difference to a certain extent to describe vegetation microwave scattering characteristics. The currently commonly used radar vegetation indexes are mainly: calculating a radar vegetation index Freeman _ RVI of volume scattering, dihedral angle scattering and surface scattering based on Freeman polarization decomposition; and calculating radar vegetation indexes Van _ RVI of characteristic values based on H/A/alpha polarization decomposition and calculating radar vegetation indexes Kim _ RVI based on backscattering strength. However, the above 3 radar vegetation indexes need to be constructed based on a full polarization radar data polarization decomposition technology, and are difficult to apply to dual polarization radar data. The penetration ability of the microwave enables radar imaging to be affected by various factors such as crop canopies, soil roughness, soil moisture and the like, and the missing wave band information of the dual-polarization radar brings great difficulty for dual-polarization radar data interpretation and radar vegetation index construction, so that a new radar vegetation index suitable for dual-polarization radar data needs to be constructed based on a compact polarization decomposition technology.
The crops have specific and regular phenological and plant characteristics which change with the phenological. The single-time phase SAR image is influenced by interference factors such as radar system noise, microwave incident angles, plant individual differences and the like, the same ground object represents different microwave scattering characteristics or different ground objects represent the same microwave scattering characteristics, and the remote sensing classification and crop extraction precision based on the SAR image are greatly reduced by the phenomenon. The time series remote sensing data is used for identifying and extracting the crop planting area, the crop growth rule and the phenological characteristics are fully utilized, and the remote sensing classification and crop extraction precision and accuracy can be effectively improved. In recent years, researchers at home and abroad have conducted more researches in remote sensing classification and crop extraction application based on multi-temporal optical vegetation indexes. The algorithm is mainly used for detecting the similarity between time sequences, obtains an optimal curved path by adjusting the corresponding relation between elements of an input time sequence and a reference time sequence, and measures the similarity degree of the time sequences by using the value on the path. The similarity between the input time sequence and the reference time sequence is measured by calculating the distance between the two sequences, so that the distance measure of the time sequence plays a crucial role in the DTW algorithm. The two most commonly used distance measurement indexes are the Euclidean distance and the dynamic time warping distance respectively, but the Euclidean distance can only process time sequences with equal length, time axis warping is not allowed, and the dynamic time warping distance overcomes the defect of the Euclidean distance and becomes the preferred distance measurement index for processing time sequence data.
Disclosure of Invention
Aiming at the problems in the background technology, in order to obtain the regional crop planting distribution and the regional crop planting area in mountainous and hilly areas in southern China, the compact three-component polarization decomposition radar vegetation index which can be used for dual-polarization SAR data is constructed; screening the extracted time sequence combination of the optimal crop planting area by using a random forest algorithm and combining ground sample point data; and performing remote sensing extraction based on dual-polarization SAR data on the crop planting distribution in the mountainous and hilly areas by using a dynamic time-warping waveform similarity classification method.
The invention provides a remote sensing extraction method for crops SAR in mountainous and hilly areas, which comprises the following steps:
s1, preprocessing dual-polarization SAR time sequence data and extracting a time sequence scattering matrix;
s2, carrying out three-component compact polarization decomposition on the time sequence scattering matrix to obtain a secondary scattering component, a volume scattering component and a surface scattering component, and constructing a three-component compact polarization decomposition radar vegetation index;
s3, decomposing the radar vegetation index based on the three-component compact polarization, and constructing a radar vegetation index standard time sequence curve by combining ground sampling points;
s4, screening the time sequence combination extracted from the optimal crop planting area by combining the ground sample point data;
and S5, referring to the radar vegetation index time sequence standard curve, using the time sequence combination extracted from the optimal crop planting area, comparing the similarity between the time sequence curve of the pixel to be classified and the time sequence curve of the standard pixel one by one, and judging the pixel to be judged as the pixel category with the maximum similarity.
The invention has the beneficial effects that: the method solves the problems that the mountainous area and hilly area in south China have complex terrain, are often interfered by meteorological conditions such as cloud, rain, fog and the like, are difficult to acquire optical remote sensing data, and the work efficiency of the crop area ground measurement method is low.
Drawings
In order that the invention may be more readily understood, it will be described in more detail with reference to specific embodiments thereof that are illustrated in the accompanying drawings. These drawings depict only typical embodiments of the invention and are not therefore to be considered to limit the scope of the invention.
FIG. 1 is a technical scheme of an embodiment of the process of the present invention.
FIG. 2 is a study area and GF-2 optimal vegetation identification false color composite image of the study area demonstrating the method of the present invention.
FIG. 3 is a graph of the phenological period of winter rape in the study area shown in FIG. 2.
Fig. 4 shows the time series curves (100 points each) of a typical feature.
Figure 5 shows the remote sensing classification of the study area.
Detailed Description
Embodiments of the present invention will be described below with reference to the accompanying drawings so that those skilled in the art can better understand the present invention and can carry out the present invention, but the illustrated embodiments are not intended to limit the present invention, and technical features in the following embodiments and embodiments can be combined with each other without conflict, wherein like parts are denoted by like reference numerals.
In order to obtain the regional crop planting distribution and the regional crop planting area in hilly areas in southern China, the method comprises the steps of firstly carrying out compact three-component polarization decomposition on Sentinel-1 data and analyzing the microwave scattering time sequence characteristics of typical ground objects in a research area, and constructing a three-component compact polarization decomposition radar vegetation index by utilizing secondary scattering, body scattering and surface scattering components; then, screening a time sequence data combination extracted from the optimal crop planting area by using a random forest algorithm and combining ground sample point data; and finally, performing remote sensing extraction based on dual-polarization SAR data on the crop planting distribution in mountainous and hilly areas in southern China by using a dynamic time warping waveform similarity classification method.
The method of the present invention is described below with reference to fig. 1.
S1, preprocessing dual-polarization SAR time sequence data and extracting a time sequence scattering matrix. The pretreatment comprises the following steps: orbit correction, radiometric calibration, band synthesis, terrain correction, etc.
And S2, carrying out three-component compact polarization decomposition on the preprocessed time sequence scattering matrix to obtain scattering components (secondary scattering, volume scattering and surface scattering components), and constructing a three-component compact polarization decomposition radar vegetation index.
And S3, decomposing the radar vegetation index based on the three-component compact polarization of typical ground objects in the research area, and constructing a radar vegetation index standard time sequence curve by combining ground sampling points.
And S4, screening the extracted time sequence combination of the optimal crop planting area by using a random forest algorithm and combining the ground sampling point data.
And S5, referring to a typical ground object radar vegetation index time sequence standard curve of the research area, using the time sequence combination extracted from the optimal crop planting area, comparing the similarity between the time sequence curve of the pixel to be classified and the standard time sequence curve pixel by pixel based on a dynamic time warping waveform similarity classification method, and judging the pixel to be judged as the pixel class with the maximum similarity. And combining other types of ground objects to obtain the final crop planting distribution result.
The process of the present invention is described in more detail below.
In step S2, the construction process of compact three-component polarization decomposition and dual-polarization radar vegetation index is as follows.
The VV and VH dual-polarized SAR data scattering matrix S is as follows:
the compact polarization data can be represented by a target vector and can also be obtained by transforming a polarization basis through a complete polarization scattering matrix:
wherein, [1,e jδ ] T Is a set of polarization bases, and different values are taken for delta, so that target vectors of different polarization states can be traversed.
Stokes vector expression in compact polarization mode uses S = [ S ] 0 S 1 S 2 S 3 ] T The description is as follows:
in the formula (I), the compound is shown in the specification,
and (·) denotes taking the real and imaginary parts of the complex number, S, respectively 0 Representing the total power of the electromagnetic wave, S 1 Representing the horizontal or vertical linear polarization component power value, S 2 Representing the value of the power of the linear polarization component at an angle of inclination of 45 DEG or 135 DEG, S 3 The power value of the right-hand circularly polarized component.
The polarized wave may be decomposed into a depolarization component and a full polarization component,
in the formula, the polarizability m is the quantity reflecting the depolarization degree of the target echo and can be obtained by Stokes parameters,
compact three-component polarization decomposition decomposes the target scattering process into the sum of volume scattering (V), surface scattering (S) and even-order scattering components (D), and builds a corresponding model for each scattering mechanism.
Bulk scattering is described as multiple scattering by a layer of randomly oriented particles, and the bulk scattering model can be described as:
the surface scattering model can be described as:
in the formula (I), the compound is shown in the specification,
is the phase difference between the H and V polarized channels.
The even order scattering model can be described as:
in the formula (I), the compound is shown in the specification,
R gV ,R tV ,R gH ,R tH is the Fresnel coefficient, subscripts g, t represent the surface and crop plants, respectively, V, H represent the vertical and horizontal polarizations, respectively,
is the phase difference between the H and V polarized channels.
Corresponding, respective sub-matrices J of the compact polarimetric complex interference coherent matrix (ij) Can be decomposed into a weighted sum of the 3 scatter components of bulk scatter, surface scatter and even scatter:
in the formula (f) v ,，f s ，f d Are the weights of the bulk, surface and even-order scatter components, respectively, and are represented as complex numbers consisting of the amplitude F and the phase Φ. J. the design is a square 11 And J 22 Only contains polarization information, so the phase of each scattering component is zero; j. the design is a square 12 And J 21 Including the interference information, the phase of each scattered component can be resolved.
3 scattering models are substituted for the formula (9) to obtain
By representing the 3 kinds of scattering components by using Stokes vectors, the target echo can be decomposed into:
comparing equation (4) and equation (11), since bulk scattering is described as a completely random scattering process, the degree of polarization is zero, corresponding to the depolarization component of the wave; surface scattering and even-order scattering are described as deterministic scattering processes, with a high degree of polarization that can correspond approximately to the fully polarized component of the wave. Thus, the power of the bulk scattering component is approximately equal to the power of the echo depolarization component.
From observed compact polarization data S 1 And S 2 Separately calculating the polarization m of the base line two-terminal received back wave according to the formula (5) 1 And m 2 Because the cross-correlation matrix is obtained by performing interference processing on the two groups of target vectors, the power of the volume scattering component can be obtained by performing geometric averaging on the depolarization components of the two groups of echoes:
substitution of formula (12) for formula (10), J 11 、J 22 、J 12 、J 2 And P v Are all alreadyKnowing that f can be solved by the equation d 、f s α and β.
How to construct the radar vegetation index based on the compact three-component polarization decomposition in step S3 is described in detail below.
The incident wave of the radar enters the vegetation to be scattered for multiple times, the scattering echo of the radar is random scattering wave, and the radar is mainly subjected to volume scattering by the vegetation. Therefore, the ground object with the larger proportion of the bulk scattering to the total scattering is more likely to be vegetation. In order to better distinguish crops in a research region from other typical objects, the radar vegetation index RVI based on compact three-component polarization decomposition is constructed 3-c As shown in equation (12).
In the formula, P s Denotes the surface scattered power, P d Denotes the even-order scattered power, P v Representing the bulk scattering power.
RVI 3-c Is in the range of [0,1]. When the radar wave irradiation area is water or bare land, the bulk scattering power P v Theoretically tending toward zero, RVI 3-c Also tends to zero; when the observation field contains forest land or crops, the energy of radar microwave penetrating through vegetation or crop canopy and ground to generate single scattering and radar microwave incident on ground and reflected to trunk (stem) to generate dihedral angle reflection echo is reduced, and surface scattering P s And even scattering P d Reduced power, bulk scattering power P v It will increase.
And editing, adding attributes and projection conversion are carried out on the ground sampling points, and the ground sampling points are covered on the time sequence SAR remote sensing image to obtain time sequence remote sensing pixel values corresponding to the ground sampling point space. And performing confidence analysis on the pixel value of each moment, and taking the mean value of the data in the 90% confidence interval of each moment as the standard time series curve value of the moment. And connecting the numerical values at all times to construct a typical land object radar vegetation index standard time sequence curve of the research area.
Step S4 is described in detail below: and screening the extracted time sequence combination of the optimal crop planting area by using a random forest algorithm and combining the ground sample point data.
And (5) dynamic time warping.
Suppose two time series M = { M = { (M) 1 ,m 2 ,m 3 ,…,m n And N = { N = } 1 ,n 2 ,n 3 ,…,n n And the number of bits is p and q respectively. To solve the optimal curved path, a distance matrix T of two sequence elements is constructed m×n 。
Wherein d (m) of the distance element i ,n j ) Represents a point m i And n j The matching relationship between the two; all the curved matrix elements satisfying the constraint condition in the matrix T form a matching relationship of the whole time series M and N, that is, a curved path. The constraint conditions of the path are as follows:
1) The start and end positions of the path must be d (m) 1 ,n 1 ) And d (m) p ,n q )。
2) The elements on the curved path must follow the neighboring points downwards, i.e. the positions are guaranteed to be continuous and the distance increases monotonically.
In fact, there are many curved paths that can satisfy the above constraints, and an optimal path needs to be found among many possible paths, so that the sum of the distance elements on the path is minimized.
And (3) finding an optimal time curved path by adopting dynamic programming, wherein the optimal time curved path can be obtained by calculating a cumulative distance matrix:
so that the sum of the curved path distance elements is minimized, i.e.
In the formula, x i The coordinate of the ith point on a certain path; d (x) i ) Is a coordinate x i A corresponding distance element; k is within the range of [ max (m, n), m + n-2]Representing the total number of elements on the path.
The invention adopts a random forest time sequence screening method. The Sentinel-1 radar data covering the hilly areas of the middle and lower reaches of Yangtze river of China is revisited once every 12 days, at least about 10 scenes of images can be obtained in the whole growth period of crops, and if s Jing Shixu remote sensing images are obtained in total, 2 scenes of images can be obtained s -1 sequential combination. The invention utilizes a random forest method to evaluate the importance of the time sequence data, analyzes the importance of each time sequence data and preferably selects the optimal time sequence combination.
The importance evaluation of the time sequence data is carried out by using the random forest, and the contribution degree of each time sequence data on the trees in the random forest is quantified by using a measuring index and sequenced. In the present invention, the degree of contribution of the time series is expressed by a Gini index (Gini index). Suppose that there are j time-series data X in total 1 ,X 2 ,X 3 ,…,X j Random forest has i trees and C categories, and each feature X needs to be calculated j The importance score (VIM) of the kini index of (a).
The calculation formula of the kini index of the ith tree node q is
Wherein p is qc And the proportion of the class c in the node q is shown, namely the probability that the class marks of two samples are inconsistent is randomly extracted from the node q.
Characteristic X j The importance of the node q in the ith tree, i.e., the variation of the Gini index before and after the node q branches
Wherein the content of the first and second substances,
and
respectively representing the kini indexes of two new nodes after branching.
If characteristic X j The node appearing in decision tree i is set Q, then X j The importance of the ith tree is
For the i trees in random forest, the importance scores after normalization are:
step S5 is described in detail below.
And calculating the DTW distance between the time sequence of the pixels to be classified and the standard time sequence of all types of ground objects. If c types of ground objects are shared, the maximum DTW distance between the time sequence of the pixels to be classified and the standard time sequence of the ground objects of all types is recorded as D c-max ＝[d 1-max ,d 2-max ,…,d c-max ]. Traversing the optimal crop planting area, extracting pixels on the time sequence combination data, and calculating the DTW distance D = [ D ] between the time sequence curve of the pixel to be classified and the standard time sequence curve of various ground features pixel by pixel 1 ,d 2 ,…,d c ]If d is i = min (D) and D i <d i-max Then the pixel class is determined as i-class ground object. And if the ground object type meeting the condition does not exist, judging the pixel as unclassified. And combining other types of ground objects to obtain the final crop planting distribution result.
The method of the present invention was tested.
The study areas were as follows: the invention takes rape in hilly areas of Qidong county in south of Hunan of China rape main producing area as a research object, verifies the SAR image crop extraction based on compact three-component polarization decomposition vegetation index and dynamic time regular waveform similarityFeasibility and applicability of the method. The research area is distributed along Hunan river, and the coverage area is 1024km 2 The land form belongs to the hilly area and is located in subtropical monsoon climate areas, the soil type is mainly red soil, and the main crop planting system is winter rape-one season rice-two cropping system in one year. Winter rape in the research area is sown from 10 months every year to harvested in 5 middle months of the next year: the seedling stage is from 11 th to 12 th, and the flowering period, silique period and mature period of the winter rape are sequentially followed from 1 st, 3 rd, 4 th and 5 th last months of the following year. The phenological period of winter rape in the study area is shown in figure 3.
Remote sensing data: remote sensing data adopts Sentine-1SLC (Single Look complete) data covering a research area, the data is in an Interferometric Wide swap (IW) mode, and the polarization mode is VV + VH dual polarization. The 13-scene Sentine-1 remote sensing image is used together, imaging time is sequentially 12-month and 9-day 2020, 12-month and 21-day 2020, 1-month and 2-day 2021, 1-month and 14-day 2021, 1-month and 26-day 2021, 2-month and 7-day 2021, 2-month and 19-day 2021, 3-month and 3-day 2021, 3-month and 15-day 2021, 3-month and 27-day 2021, 4-month and 8-day 2021, 4-month and 20-day 2021, 5-month and 2-day 2021, relative orbit numbers are 11, and the rape growth period in a research region is shown in fig. 3. The invention uses SNAP software to sequentially carry out preprocessing such as orbit correction, radiation calibration, wave band synthesis, terrain correction and the like on downloaded Sentinel-1 data, and resamples the spatial resolution of the data to 20m multiplied by 20m. The invention also collects GF-2 multispectral remote sensing images of 4 scenes rape flowering phase in a research area, and a typical test area and GF-2 optimal vegetation identification false color synthetic graph are shown in figure 2. The vegetation in fig. 2 is shown as red or pink, where red is the woodland and pink is the rape.
Ground sampling points: the data acquisition time of the ground sample points is 3 months and 3 days in 2021, and the data acquisition time corresponds to the most obvious full flowering period of rape plant characteristics. The method is characterized in that 1000 ground sample points are collected, wherein 200 rape, water, bare land, forest land and building sample points are respectively collected, the singular sample points are used for constructing a standard time sequence curve of 5 typical land object radar vegetation indexes in a research area, and the even sample points are used for verifying land object classification and rape extraction accuracy. Editing, adding attributes and projection conversion are carried out on the ground sample points, and the ground sample points are covered on a test Sentinel-1 remote sensing image.
The results of the study are as follows. RVI 3-c Time series data analysis: the Sentnel-1 time series data of a research area from 12 months to 5 months in 2020 to 2021 are preprocessed, wherein the preprocessing comprises preprocessing such as orbit correction, radiometric calibration, wave band synthesis, terrain correction and the like, and a scattering matrix (S matrix) is extracted. Carrying out compact three-component polarization decomposition on the preprocessed Sentinel-1 time sequence data S matrix to obtain even scattering P d Volume scattering P d And surface scattering P s And (4) components. And constructing a three-component compact polarization radar vegetation index by utilizing even-order scattering, volume scattering and surface scattering components. And constructing a three-component compact polarization radar vegetation index standard time sequence curve of typical ground objects in the research area by combining the ground sampling points. The standard time sequence curves of five typical ground objects of water bodies, bare land, rape, woodland and buildings are shown in figure 4.
As shown in FIG. 4, RVIs of 5 typical land objects, water bodies, bare land, forest land and buildings in the research area 3-c The variation in the value in time series is small. RVI of bodies of water and open areas 3-c Small value and RVI of water body 3-c Lower than RVI of bare land 3-c This is due to the fact that both water and open ground appear predominantly as surface scattering on radar images, with a surface scattering component P s Larger, volume scattering component P v Is small; RVI of a building 3-c The value is centered because the building appears mainly as even-order scatter and volume scatter in the radar image, the even-order scatter component P d And the volume scattering component P v The values are all larger and respectively act on RVI 3-c Numerator and denominator of value, making RVI of building 3-c The value is centered; RVI of forest land 3-c The value is high because the woodland is mainly shown as volume scattering on the radar image, and the volume scattering component P v Higher. RVI of oilseed rape 3-c The variation of the values in time sequence is large, the RVI of the rape 3-c The value is gradually increased along with the growth and development of the rape, the plant volume is smaller when the rape is just sown, and the volume scattering component P expressed on the radar image v Smaller, as rape blooms and grows, the volume of the rape is increased, and the volume scattering component P is v And also becomes larger. Especially at early flowering and floweringAt the end of flowering, when the rape flowers or the siliques grow, the plant volume is obviously increased, RVI 3-c The value also increased significantly. In the later stage of rape pod, the photosynthetic organ of rape is gradually changed into rape pod, the leaf begins to decline, the plant volume becomes small, and RVI 3-c The value decreases.
And (3) time sequence screening: due to the influence of the radar sensor, a large amount of coherent noise exists on a radar image, the scene number of the time sequence SAR image participating in classification is increased, the remote sensing classification precision and the rape extraction precision cannot be improved necessarily, but the remote sensing classification speed and efficiency can be reduced certainly. The method combines a random forest method and a Gini index to evaluate the importance of the time sequence data, analyzes the importance of each time sequence data and preferably selects the optimal time sequence combination. In the whole growth period of winter rape in a research area, 13 sceneinel-1 remote sensing images are obtained, and 213-1 (8191) time sequence combinations are provided. The invention uses 500 (singular number) sampling points actually measured on the ground to calculate the standard time sequence curve RVI of the compact polarization radar vegetation index of each ground feature 3-c And the other 500 (even number) sampling points take the overall precision as a precision inspection index to verify the remote sensing classification and the rape extraction precision.
The remote sensing classification and the overall rape extraction precision are used as indexes, 9 time sequence combinations with high classification or extraction precision are respectively selected for analysis, and the combined data list and the rape classification precision are shown as table 1.
TABLE 1 high precision time sequence combined data List
From the contribution of the 13 time series remote sensing data in table 1 to the overall accuracy fluctuation of rape extraction, the 5 time series data which have the largest contribution to the accuracy fluctuation of rape extraction are data of 19 days at 2 months in 2021, 26 days at 1 month in 2021, 9 days at 12 months in 2020, 7 days at 2 months in 2021 and 21 days at 12 months in 2020. The time sequence data which greatly contributes to the overall accuracy fluctuation of rape extraction is mostly remote sensing data of rape seedling stage. On one hand, the plant characteristics of the rape at the seedling stage are not obvious, the distinguishing degree with bare land is low, and the interaction among the images at all the stages has great influence on the remote sensing classification and the rape precision fluctuation; on the other hand, the rape seedling stage corresponds to the winter and early spring of a research area, the typical feature difference of each ground feature of the research area is large, and particularly the feature difference of the ground feature of a water body and a bare land is large. In the multi-factor synthesis, the SAR remote sensing image with better quality obtained in the rape seedling stage has positive effects on the extraction of the rape or the classification of the ground objects. The time sequence combination of high rape extraction precision comprises remote sensing data of 3-15 days in 2021, 4-8 days in 2021 and 5-2 days in 2021, the dates respectively correspond to the full flowering period, the silique period and the mature period of the rape, rape plants in the periods comprise special flowers or siliques of the rape, the plant characteristics are obvious, the distinguishing degree with other land features is high, and the high rape extraction precision can be obtained by participating in remote sensing classification. Most of time sequences of high rape extraction accuracy comprise remote sensing data of 14 days at 1 month and 14 days at 2021, 19 days at 2 months at 2021 and 3 days at 3 months at 2021, the dates respectively correspond to a bolting period, a flowering period and a flowering period of rape, rape plants at the periods have large characteristic difference with typical ground objects such as forest lands, buildings and the like, and high rape extraction accuracy can be obtained by participating in remote sensing classification. Most time series with high remote sensing classification accuracy comprise remote sensing data of 21 days 12 and 21 months in 2020, 2 days 1 and 2 months in 2021, 7 days 2 and 7 months in 2021, 3 and 3 days 3 and 3 months in 2021, 15 days 3 and 3 months and 27 days 3 and 3 months in 2021, the data can be roughly divided into two time periods which respectively correspond to winter and early spring of a research area, and the characteristic difference of each typical ground feature in the research area is large, so the remote sensing classification accuracy is high.
Regional rape extraction and precision verification: in the selection of time sequence combination, remote sensing classification and rape extraction precision (the total precision of remote sensing classification is more than 60 percent, and the total precision of rape extraction is more than 70 percent) are considered, the remote sensing classification and rape extraction of a research area are carried out by using 8-scene time sequence data combination of 12 and 9 days in 2020, 1 and 2 days in 2021, 2 and 2 days in 2021, 7 and 2 months in 2021, 3 and 3 days in 2021, 15 and 15 days in 3 and 27 days in 2021, 4 and 8 days in 2021 and 5 and 2 days in 2021, and the remote sensing classification result and the regional rape extraction precision are verified as shown in fig. 5 and table 2.
TABLE 2 regional rape extraction accuracy verification
As can be seen from the accuracy of fig. 5 and table 2, most of the features are classified accurately, and some of the features are classified incorrectly. The remote sensing classification overall precision and rape F-1 coefficient of the research area are 61.80% and 85.64% respectively. The research results show that the rape planting area extraction method based on the compact three-component polarization decomposition radar vegetation index and the dynamic time-warping waveform similarity classification method has higher precision in the extraction of the rape planting area in the research area, and also show the stability of the compact three-component polarization decomposition radar vegetation index and the feasibility of the dynamic time-warping waveform similarity classification method in the time sequence extraction of the rape planting area.
The embodiments described above are merely preferred specific embodiments of the present invention, and the present specification uses the phrases "in one embodiment," "in another embodiment," "in yet another embodiment," or "in other embodiments," which may each refer to one or more of the same or different embodiments in accordance with the present disclosure. General changes and substitutions by one skilled in the art within the technical scope of the present invention should be included in the scope of the present invention.
Claims (10)
1. A method for extracting crops SAR remote sensing in mountainous and hilly areas is characterized by comprising the following steps:
s1, preprocessing dual-polarization SAR time sequence data and extracting a time sequence scattering matrix;
s2, carrying out three-component compact polarization decomposition on the time sequence scattering matrix to obtain a secondary scattering component, a volume scattering component and a surface scattering component, and constructing a three-component compact polarization decomposition radar vegetation index;
s3, decomposing the radar vegetation index based on the three-component compact polarization, and constructing a radar vegetation index standard time sequence curve by combining ground sampling points;
s4, screening the time sequence combination extracted from the optimal crop planting area by combining the ground sample point data;
and S5, referring to the radar vegetation index time sequence standard curve, using the time sequence combination extracted from the optimal crop planting area, comparing the similarity between the time sequence curve of the pixels to be classified and the standard time sequence curve pixel by pixel, and judging the pixels to be judged as the pixel categories with the maximum similarity.
2. The SAR remote sensing extraction method for crops on mountainous and hilly areas as claimed in claim 1,
in step S2, compact three-component polarization decomposition decomposes the target scattering process into the sum of bulk scattering, surface scattering and even-order scattering components, and builds a corresponding model for each scattering mechanism.
3. The SAR remote sensing extraction method for crops on mountainous and hilly areas as claimed in claim 2, wherein in step S2,
the volume scattering model is:
the surface scattering model is:
in the formula (I), the compound is shown in the specification,
is the phase difference of the H and V polarized channels;
the even order scattering model is:
in the formula (I), the compound is shown in the specification,
R gV ,R tV ,R gH ,R tH is the Fresnel coefficient, subscripts g, t represent the surface and crop plants, respectively, V, H represent the vertical and horizontal polarization, respectively,
is the phase difference of the H and V polarized channels.
4. The SAR remote sensing extraction method for crops on mountainous and hilly areas as claimed in claim 3,
in step S2, the respective sub-matrices J of the coherent matrix are compacted and polarized to form a complex interference matrix (ij) Obtaining a weighted sum of the volume scatter component, the surface scatter component, and the even-order scatter component:
in the formula (f) v ,，f s ，f d The weights of the bulk, surface and even-order scatter components, respectively, F is the amplitude and Φ is the phase.
5. The method for SAR remote sensing extraction of crops on mountainous and hilly areas as claimed in claim 4, wherein in step S2, a three-component compact polarimetric decomposition radar vegetation index is constructed by:
calculating the power of the volume scattering, surface scattering and even-order scattering components;
constructing a three-component compact polarization decomposition radar vegetation index:
in the formula, P s Denotes the surface scattered power, P d Denotes the even-order scattered power, P v Representing the bulk scattering power.
6. The method for SAR remote sensing extraction of crops in mountainous and hilly areas as claimed in claim 1, wherein step S3 comprises:
editing, adding attributes and projection conversion are carried out on the ground sampling points, and the ground sampling points are covered on a time sequence SAR remote sensing image to obtain time sequence remote sensing pixel values corresponding to the ground sampling point space;
performing confidence analysis on the pixel value at each moment, and taking the mean value of the data in the confidence interval set at each moment as the standard time series curve value of the moment;
and connecting the numerical values at all times to construct a typical land object radar vegetation index standard time sequence curve of the research area.
7. The method for SAR remote sensing extraction of crops in mountainous and hilly areas as claimed in claim 1, wherein step S4 comprises:
finding an optimal time curved path by calculating a cumulative distance matrix so that the sum of curved path distance elements is minimum;
and (4) performing time sequence data importance evaluation by using a random forest method, analyzing the importance of each time sequence data and preferably selecting the optimal time sequence combination.
8. The method for SAR remote sensing extraction of crops in mountainous and hilly areas as claimed in claim 1, wherein step S5 comprises:
calculating DTW distances between the sampling point time sequence and all standard time sequences;
traversing the optimal crop planting area, extracting pixels on the time sequence combination data, and calculating the DTW distance d between the time sequence curve of the pixel to be classified and the time sequence curve of various ground features one by one i ；
If the distance d i = is the smallest of all distances and d i <d i-max Then the image element type is judged as i-type ground object, wherein d i-max Indicating the maximum distance DTW between the sample time series and the standard time series of the category.
9. The SAR remote sensing extraction method for crops on mountainous and hilly areas as claimed in claim 1,
in step S1, the preprocessing includes: orbit correction, radiometric calibration, waveband synthesis and terrain correction.
10. The method for SAR remote sensing extraction of crops in mountainous and hilly areas as claimed in claim 1, wherein step S5 further comprises: based on a dynamic time warping waveform similarity classification method, comparing the similarity between a time sequence curve of pixels to be classified and a standard time sequence curve pixel by pixel; and
and combining other types of ground objects to obtain the final crop planting distribution result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210996592.1A CN115372970B (en) | 2022-08-19 | 2022-08-19 | SAR remote sensing extraction method for crops in mountainous and hilly areas |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210996592.1A CN115372970B (en) | 2022-08-19 | 2022-08-19 | SAR remote sensing extraction method for crops in mountainous and hilly areas |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115372970A true CN115372970A (en) | 2022-11-22 |
| CN115372970B CN115372970B (en) | 2024-06-28 |
Family
ID=84065401
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210996592.1A Active CN115372970B (en) | 2022-08-19 | 2022-08-19 | SAR remote sensing extraction method for crops in mountainous and hilly areas |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115372970B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116224271A (en) * | 2023-01-10 | 2023-06-06 | 南京工业大学 | Sea surface weak target detection method based on polarized scattering characteristics |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5469167A (en) * | 1993-10-18 | 1995-11-21 | The United States Of America As Represented By The Secretary Of The Army | Synthetic aperture radar for nonlinear trajectories using range relative doppler processing and invariant mapping |
| JP2005140607A (en) * | 2003-11-06 | 2005-06-02 | National Institute Of Information & Communication Technology | Method and system for processing polarization synthetic aperture radar image |
| CN104360347A (en) * | 2014-11-03 | 2015-02-18 | 北京农业信息技术研究中心 | Method and device for monitoring crop harvesting progress |
| US20170010353A1 (en) * | 2015-07-08 | 2017-01-12 | Conocophillips Company | Terrestrial imaging using multi-polarization synthetic aperture radar |
| CN108509836A (en) * | 2018-01-29 | 2018-09-07 | 中国农业大学 | Crop yield estimation method based on double-polarized synthetic aperture radar and crop model data assimilation |
| CN108761456A (en) * | 2018-05-07 | 2018-11-06 | 中国农业科学院农业资源与农业区划研究所 | A kind of inversion method of leaf area index of crop |
| US20190179009A1 (en) * | 2017-12-08 | 2019-06-13 | International Business Machines Corporation | Crop classification and growth tracking with synthetic aperture radar |
| CN114202535A (en) * | 2021-12-15 | 2022-03-18 | 广东省科学院广州地理研究所 | Crop planting area extraction method and device |
| WO2022057319A1 (en) * | 2020-09-21 | 2022-03-24 | 河南大学 | Garlic crop recognition method based on coupled active/passive remote sensing images on cloud platform |
-
2022
- 2022-08-19 CN CN202210996592.1A patent/CN115372970B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5469167A (en) * | 1993-10-18 | 1995-11-21 | The United States Of America As Represented By The Secretary Of The Army | Synthetic aperture radar for nonlinear trajectories using range relative doppler processing and invariant mapping |
| JP2005140607A (en) * | 2003-11-06 | 2005-06-02 | National Institute Of Information & Communication Technology | Method and system for processing polarization synthetic aperture radar image |
| CN104360347A (en) * | 2014-11-03 | 2015-02-18 | 北京农业信息技术研究中心 | Method and device for monitoring crop harvesting progress |
| US20170010353A1 (en) * | 2015-07-08 | 2017-01-12 | Conocophillips Company | Terrestrial imaging using multi-polarization synthetic aperture radar |
| US20190179009A1 (en) * | 2017-12-08 | 2019-06-13 | International Business Machines Corporation | Crop classification and growth tracking with synthetic aperture radar |
| CN108509836A (en) * | 2018-01-29 | 2018-09-07 | 中国农业大学 | Crop yield estimation method based on double-polarized synthetic aperture radar and crop model data assimilation |
| CN108761456A (en) * | 2018-05-07 | 2018-11-06 | 中国农业科学院农业资源与农业区划研究所 | A kind of inversion method of leaf area index of crop |
| WO2022057319A1 (en) * | 2020-09-21 | 2022-03-24 | 河南大学 | Garlic crop recognition method based on coupled active/passive remote sensing images on cloud platform |
| CN114202535A (en) * | 2021-12-15 | 2022-03-18 | 广东省科学院广州地理研究所 | Crop planting area extraction method and device |
Non-Patent Citations (3)
| Title |
|---|
| 吴尚蓉: "基于三分量分解优化模型的农用地SAR 影像提取方法", 农 业 工 程 学 报, vol. 31, no. 2, 31 January 2015 (2015-01-31), pages 266 - 273 * |
| 孙政;周清波;杨鹏;王迪;: "基于星载极化SAR数据的农作物分类识别进展评述", 中国农业资源与区划, no. 11, 25 November 2019 (2019-11-25) * |
| 宋茜;周清波;吴文斌;胡琼;余强毅;唐华俊;: "农作物遥感识别中的多源数据融合研究进展", 中国农业科学, no. 06, 9 June 2015 (2015-06-09) * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116224271A (en) * | 2023-01-10 | 2023-06-06 | 南京工业大学 | Sea surface weak target detection method based on polarized scattering characteristics |
| CN116224271B (en) * | 2023-01-10 | 2024-06-28 | 南京工业大学 | Sea surface weak target detection method based on polarized scattering characteristics |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115372970B (en) | 2024-06-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Betbeder et al. | Assimilation of LAI and dry biomass data from optical and SAR images into an agro-meteorological model to estimate soybean yield | |
| CN109345555B (en) | Method for identifying rice based on multi-temporal multi-source remote sensing data | |
| Ranson et al. | Boreal forest ecosystem characterization with SIR-C/XSAR | |
| Kellndorfer et al. | Toward consistent regional-to-global-scale vegetation characterization using orbital SAR systems | |
| Costa et al. | Biophysical properties and mapping of aquatic vegetation during the hydrological cycle of the Amazon floodplain using JERS-1 and Radarsat | |
| CN112711989B (en) | Corn straw coverage estimation method based on radar remote sensing and optical remote sensing | |
| CN114387516B (en) | Single-season rice SAR (synthetic aperture radar) identification method for small and medium-sized fields in complex terrain environment | |
| JP2011167163A (en) | Method of generating paddy rice crop yield forecasting model, and method of forecasting crop yield of paddy rice | |
| CN108766203B (en) | Compact polarization rice mapping method and system | |
| Yang et al. | Interpreting RADARSAT-2 quad-polarization SAR signatures from rice paddy based on experiments | |
| CN115372971A (en) | Time sequence SAR image crop extraction method based on CTLR and DTW K-means | |
| CN115372970A (en) | Remote sensing extraction method for crops SAR in mountainous and hilly areas | |
| Shen et al. | High-resolution distribution maps of single-season rice in China from 2017 to 2022 | |
| Otukei et al. | Estimation and mapping of above ground biomass and carbon of Bwindi impenetrable National Park using ALOS PALSAR data | |
| Nzimande et al. | Mapping the spatial distribution of the yellowwood tree (Podocarpus henkelii) in the Weza-Ngele forest using the newly launched Sentinel-2 multispectral imager data | |
| Frison et al. | Using satellite scatterometers to monitor continental surfaces | |
| RU2443977C1 (en) | Method of estimating distribution and reserves of resources and endangered species of plants in large territories | |
| Benabdelouahab et al. | Using SAR data to detect wheat irrigation supply in an irrigated semi-arid area | |
| Hernandez-Figueroa et al. | Sugarcane precision monitoring by drone-borne P/L/C-band DInSAR | |
| Lam-Dao | Rice crop monitoring using new generation synthetic aperture radar (SAR) imagery | |
| CN114397276B (en) | Regional soil humidity monitoring method based on equivalent precipitation estimation method | |
| Haldar et al. | Identification of tall fibre crop (jute) using multi-temporal RADARSAT data in rainfed areas of eastern indogangetic plain | |
| Liang et al. | The Application of Compact Polarization Decomposition in the Construction of a Dual-Polarization Radar Index and the Effect Evaluation of Rape Extraction | |
| CN115937067A (en) | Crop planting area extraction method based on polarization decomposition and ground statistical texture | |
| Dong et al. | Inversion of forest canopy height in south of China by integrating GLAS and MERSI: The case of Jiangxi province in China |
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 |