CN112950492B - Full-polarization SAR image denoising method based on self-adaptive anisotropic diffusion - Google Patents
Full-polarization SAR image denoising method based on self-adaptive anisotropic diffusion Download PDFInfo
- Publication number
- CN112950492B CN112950492B CN202110117113.XA CN202110117113A CN112950492B CN 112950492 B CN112950492 B CN 112950492B CN 202110117113 A CN202110117113 A CN 202110117113A CN 112950492 B CN112950492 B CN 112950492B
- Authority
- CN
- China
- Prior art keywords
- scattering
- fully
- adaptive
- gradient
- self
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000009792 diffusion process Methods 0.000 title claims abstract description 31
- 230000010287 polarization Effects 0.000 claims abstract description 91
- 238000001914 filtration Methods 0.000 claims abstract description 79
- 239000011159 matrix material Substances 0.000 claims abstract description 34
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 12
- 230000007246 mechanism Effects 0.000 claims abstract description 11
- 239000003550 marker Substances 0.000 claims abstract description 8
- 230000003044 adaptive effect Effects 0.000 claims description 63
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000003064 k means clustering Methods 0.000 claims description 4
- 238000006243 chemical reaction Methods 0.000 claims description 2
- 238000010276 construction Methods 0.000 claims description 2
- 230000006978 adaptation Effects 0.000 claims 1
- 230000002401 inhibitory effect Effects 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 21
- 230000001629 suppression Effects 0.000 description 12
- 230000001427 coherent effect Effects 0.000 description 7
- 230000005855 radiation Effects 0.000 description 7
- 239000013535 sea water Substances 0.000 description 7
- 238000011158 quantitative evaluation Methods 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 238000005388 cross polarization Methods 0.000 description 5
- 238000011156 evaluation Methods 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 4
- 230000014759 maintenance of location Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000000007 visual effect Effects 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 239000000126 substance Substances 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 150000001875 compounds Chemical class 0.000 description 2
- 230000021615 conjugation Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000002592 echocardiography Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000035515 penetration Effects 0.000 description 2
- 230000000704 physical effect Effects 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
- 238000004458 analytical method Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 229910052500 inorganic mineral Inorganic materials 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000011707 mineral Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000003305 oil spill Substances 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 238000004451 qualitative analysis Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 239000002689 soil Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
- G06T2207/10044—Radar image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20004—Adaptive image processing
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a self-adaptive anisotropic diffusion-based full-polarization SAR image denoising method, which comprises the following steps: s1, acquiring the fully polarized SAR data, and extracting a covariance matrix of the fully polarized SAR data; s2, extracting polarization scattering entropy based on the covariance matrix; s3, selecting a gradient self-adaptive filtering window based on polarization scattering entropy; s4, performing Freeman-Durden decomposition on the covariance matrix to obtain the Bragg scattering, secondary scattering and volume scattering power of each pixel in the fully-polarized SAR data; s5, extracting the ground object scattering label graph based on the Bragg scattering, secondary scattering and volume scattering power of each pixel; s6, denoising the fully polarized SAR image by adopting a self-adaptive anisotropic diffusion method of a scattering mechanism based on a gradient self-adaptive filtering window and a ground feature scattering marker map. The invention can fully retain polarization information in the original full-polarization SAR image while inhibiting image noise.
Description
Technical Field
The invention relates to the technical field of polarimetric SAR image denoising, in particular to a polarimetric SAR image denoising method based on self-adaptive anisotropic diffusion.
Background
Polar Synthetic Aperture Radar (polar Synthetic Aperture Radar, polar sar) measures a ground object target by using different polarization transceiving combinations of antennas, can acquire more abundant information compared with the Synthetic Aperture Radar, and has important research value and wide application prospect. However, due to the inherent defect of the imaging mechanism, the PolSAR image inevitably introduces speckle noise, which not only causes the visual quality of the image to be degraded, but also causes difficulty in subsequent classification, identification and other processing. Speckle suppression is therefore an essential step in polarimetric SAR image processing.
Synthetic Aperture Radar (SAR) was born in the 60's of the 20 th century, is an active earth observation system, and overcomes the limitation of spatial resolution due to the size of the antenna in the conventional Radar system. The SAR uses a small antenna to move at a constant speed along a long linear array and radiate signals, and then echoes received at different positions are subjected to coherent processing, so that a radar image is obtained. At present, SAR has become one of the most rapidly and effectively developed sensors in microwave remote sensing, and as an active system, the SAR has all-weather and all-time working capability, can observe the ground in different microwave frequency bands and different polarization states, and can even obtain the subsurface information through ground tables and vegetation.
Polar Synthetic Aperture Radar (polar Synthetic Aperture Radar, polar sar) is a new type of Synthetic Aperture Radar that can measure the polarization characteristics of a radiation signal. The PolSAR can measure the amplitude of the ground object echo and record the phase difference of the echoes under different polarization state combinations by alternately transmitting and simultaneously receiving the electromagnetic wave pulses with orthogonal polarization states, and can be used for analyzing the ground object echo information in any polarization state. Due to the unique superiority of the PolSAR in obtaining the target complete polarization information, the application field and the value of the SAR are greatly expanded. Therefore, PolSAR has been widely used in agriculture (crop resolution, growth state estimation, etc.), forests (biological estimation, fire disaster relief, etc.), geology (structure observation, mineral resource detection, etc.), hydrology (soil temperature and humidity estimation, inversion, etc.), oceanography (wave characteristic estimation, oil spill detection, etc.), search and rescue (crash aircraft, ship location, etc.), military (military target identification, etc.). The research aiming at PolSAR is also generally concerned and paid attention to, and has become one of the hot spots of the international radar system and the technical development.
Speckle (Speckle) is a deterministic interference phenomenon inherent to SAR (including PolSAR) image data due to the system imaging mechanism. The coherent speckles appear on the SAR image as violent random fluctuation of pixel values, and the intensity value of each pixel point does not directly reflect the radar reflection coefficient of a ground target any more, so that the visual quality of the image is poor, the target identification by utilizing the SAR image becomes very complicated, and the false alarm probability of the target identification are increased. Therefore, denoising of the polarized SAR image becomes an essential important step in interpretation of the polarized SAR image. In the denoising process of the polarized SAR image, the method has very important significance for maintaining the polarization information of the image besides inhibiting speckle noise.
The development of the polarized SAR speckle suppression algorithm goes through three stages of single-polarized SAR image denoising, multi-polarized SAR image denoising and full-polarized SAR image denoising. At present, the research for the suppression of the coherent speckle of the fully polarimetric SAR at home and abroad is mainly carried out on the suppression of the coherent speckle in three directions, namely a space domain, a frequency domain and a polarization domain, and the methods have the following advantages and disadvantages:
a spatial domain based approach. The method based on the spatial domain mainly calculates the similarity between the neighborhood pixels and the target pixels according to the statistical characteristics of data, so as to obtain the denoised estimated image. The method has the advantages of reducing the resolution loss caused by multiple situations, and has the disadvantages that the filtered image is fuzzy, a lot of detail information is lost, and the filtering effect depends on the selection of the size of the filtering window.
Frequency domain based methods. The method based on the frequency domain mainly utilizes the wavelet transform theory and the Bayes method to restrain the speckle. The method has the advantages that a proper wavelet base can be selected according to the characteristics of data, and the method has the defects that high-frequency information is difficult to distinguish, and polarization information is lost in the filtered image, so that a subsequent image classification and segmentation task is influenced to a certain extent.
A polarimetric domain based approach. The polarization domain-based method mainly utilizes the optimal combination of scattering matrixes to calculate the linear correlation among all polarization channels, and finally obtains the denoised image. The method has the advantages of being capable of processing multi-polarization and multi-time phase data, and has the disadvantages of introducing mutual crosstalk among polarization channels and being difficult to maintain polarization information of the data.
With the continuous development of the technology and the change of the polarization mode from single polarization to multi-polarization, more new coherent spot suppression methods are proposed, such as: a speckle suppression method based on a scattering model, a speckle suppression method based on a partial differential equation, and the like. These methods achieve good results in terms of speckle suppression, but also have some significant disadvantages: while speckle noise is inhibited, polarization information specific to the fully-polarized SAR data cannot be well reserved, the SAR polarization information is important information for ground object target identification, loss of the polarization information causes data information loss, and inconvenience is brought to subsequent tasks such as target detection and classification.
Therefore, it is important to provide a full-polarization SAR image denoising method based on adaptive anisotropic diffusion to maintain polarization information of the SAR image while suppressing noise.
Disclosure of Invention
The invention aims to provide a full-polarization SAR image denoising method based on self-adaptive anisotropic diffusion, which aims to solve the technical problems in the prior art and can fully retain polarization information in an original full-polarization SAR image while inhibiting image noise.
In order to achieve the purpose, the invention provides the following scheme: the invention provides a self-adaptive anisotropic diffusion-based full-polarization SAR image denoising method, which comprises the following steps:
s1, acquiring the fully polarized SAR data, and extracting a covariance matrix of the fully polarized SAR data;
s2, extracting polarization scattering entropy based on the covariance matrix of the fully polarized SAR data;
s3, selecting a gradient adaptive filtering window based on the polarization scattering entropy;
s4, performing Freeman-Durden decomposition on the covariance matrix of the fully-polarized SAR data to obtain the Bragg scattering, secondary scattering and volume scattering power of each pixel in the fully-polarized SAR data;
s5, extracting a ground object scattering label graph based on the power of Bragg scattering, secondary scattering and volume scattering of each pixel in the fully polarized SAR data;
s6, denoising the fully polarized SAR image by adopting a self-adaptive anisotropic diffusion method of a scattering mechanism based on a gradient self-adaptive filtering window and a ground feature scattering marker map.
Preferably, the step S3 specifically includes:
calculating the vertical gradient and the horizontal gradient of the polarization scattering entropy, and calculating the initial direction and the size of the self-adaptive filtering window based on the vertical gradient and the horizontal gradient of the polarization scattering entropy;
constructing a horizontal filtering window based on the initial size of the self-adaptive filtering window, rotating the horizontal filtering window based on the initial direction of the self-adaptive filtering window, calculating the coordinates of each vertex of the self-adaptive filtering window, and calculating the coordinates of each vertex of the self-adaptive filtering window;
traversing the search window, acquiring overlapped pixels in the search window and the adaptive filtering window, calculating the vertical gradient algebraic sum and the horizontal gradient algebraic sum of all the overlapped pixels, and updating the direction of the adaptive filtering window by utilizing the vertical gradient algebraic sum and the horizontal gradient algebraic sum;
and calculating the sum of the absolute values of the vertical gradients and the sum of the absolute values of the horizontal gradients of all overlapped pixels in the search window and the adaptive filtering window, and updating the size of the adaptive filtering window by using the sum of the absolute values of the vertical gradients and the sum of the absolute values of the horizontal gradients until the size of the adaptive filtering window meets a preset threshold.
Preferably, the initial direction α of the adaptive filter window0(i, j) is represented by formula 7:
α0(i,j)＝arctan(Iy(i,j)/Ix(i,j))……………………7
wherein, Ix(i,j)、Iy(i, j) are the vertical gradient and the horizontal gradient of the polarization scattering entropy at the pixel (i, j), respectively, and i, j are the x-axis and y-axis coordinates of the pixel, respectively.
Preferably, the vertex coordinates of the search window are as follows:
in the formula, Xmin、XmaxRespectively the minimum value and the maximum value of the abscissa of the top point of the search window; y ismin、YmaxRespectively the minimum value and the maximum value of the vertical coordinate of the top point of the search window; x is the number of1、x2、x3、x4Respectively are the horizontal coordinates of four vertexes of the self-adaptive filtering window; y is1、y2、y3、y4Respectively are the vertical coordinates of four vertexes of the self-adaptive filtering window; ceil () represents rounded up and floor () represents rounded down.
Preferably, the step S5 specifically includes:
s5.1, performing initial clustering on each pixel in the fully-polarized SAR data based on the power of Bragg scattering, secondary scattering and volume scattering of each pixel in the fully-polarized SAR data;
s5.2, based on the initial clustering result, carrying out merged clustering on all pixels in the fully-polarized SAR data by adopting a Wishart distance;
s5.3, based on the merged clustering result, performing iterative classification on each pixel in the fully-polarized SAR data by adopting a Wishart distance;
and S5.4, based on the iterative classification result, adopting a K-means clustering algorithm to redistribute each pixel in the fully-polarized SAR data to generate a ground object scattering label graph.
Preferably, in the step S5.1, all pixels in the fully-polarized SAR data are initially clustered according to the power of bragg scattering, secondary scattering, and volume scattering.
Preferably, in the step S5.3, Wishart distances between each pixel in the fully-polarized SAR data and a plurality of clustering centers in the merged clustering result are respectively calculated, and each pixel in the fully-polarized SAR data is reclassified according to the distance calculation result.
Preferably, the step S6 specifically includes:
s6.1, based on the gradient adaptive filtering window and the ground feature scattering marker map, eliminating neighborhood pixels in the gradient adaptive filtering window, wherein the scattering types of the neighborhood pixels are different from those of the central pixel;
s6.2, calculating a gradient operator based on the residual pixels in the gradient adaptive filtering window;
s6.3, calculating a self-adaptive gradient threshold value based on a gradient operator;
s6.4, constructing a self-adaptive diffusion model based on the self-adaptive gradient threshold;
and S6.5, performing iterative computation based on the adaptive diffusion model to complete denoising of the fully-polarized SAR image.
The invention discloses the following technical effects:
the invention provides a self-adaptive anisotropic diffusion-based fully-polarized SAR image denoising method, which comprises the steps of utilizing a gradient to calculate a filtering self-adaptive filtering window, and simultaneously utilizing a target decomposition theory to obtain different scattering mechanism types so as to obtain a ground feature scattering labeled graph.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
FIG. 1 is a flow chart of a full-polarization SAR image denoising method based on adaptive anisotropic diffusion according to the present invention;
FIG. 2 is a schematic diagram of a full-polarization SAR image denoising method based on adaptive anisotropic diffusion in the embodiment of the present invention;
FIG. 3 is a diagram illustrating a polarization scattering entropy extraction result according to an embodiment of the present invention;
fig. 4 is a schematic diagram of a Freeman-Durden decomposition result of a covariance matrix of fully-polarized SAR data in the embodiment of the present invention; wherein, FIG. 4(a), FIG. 4(b), FIG. 4(c) are Ps、Pd、PvA decomposition result schematic diagram of the components;
FIG. 5 is a schematic view of a feature scattering mark in an embodiment of the present invention;
FIG. 6 is a schematic diagram of a denoising result of a fully polarized SAR image in an embodiment of the present invention; FIG. 6(a) and FIG. 6(b) are schematic diagrams of denoising results of an HH channel and a VV channel, respectively;
FIG. 7 is a comparison graph of denoising results of different denoising methods in the embodiment of the present invention; FIG. 7(a) is an original image, and FIGS. 7(b), 7(c), 7(d), 7(e), 7(f), 7(g) and 7(h) are schematic diagrams of denoising results of the method, SARD, Re-Lee, Boxcar, Gauss, IDAN and Mean-shift, respectively;
FIG. 8 is a polarization characteristic diagram of an original image and data denoised by the method of the present invention under different plan states in the embodiment of the present invention; fig. 8(a) and 8(b) are polarization characteristic diagrams of an original image co-polarization channel (oil film region) and a denoised image co-polarization channel (oil film region), respectively; fig. 8(c) and fig. 8(d) are polarization characteristic diagrams of an original image cross-polarization channel (oil film region) and a denoised image cross-polarization channel (oil film region), respectively; fig. 8(e) and 8(f) are polarization characteristic diagrams of an original image co-polarization channel (seawater region) and a denoised image co-polarization channel (seawater region), respectively; fig. 8(e) and 8(f) are polarization characteristic diagrams of an original image co-polarization channel (seawater region) and a denoised image co-polarization channel (seawater region), respectively; fig. 8(g) and 8(h) are polarization characteristic diagrams of an original image cross polarization channel (seawater region) and a denoised image cross polarization channel (seawater region), respectively.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
The terms of art referred to in the present invention are explained as follows:
full polarization SAR: the synthetic aperture radar sensor comprises four polarization modes of HH, HV, VH and VV, and has the characteristics of all-weather penetration and fog penetration.
Speckle noise: the Synthetic Aperture Radar (SAR) results from a polarized SAR imaging mechanism, caused by the coherent superposition of a large number of scattering unit reflections. The coherent speckles cause signal intensity variations between adjacent pixels, and visually appear as granular noise.
Anisotropy: the term "chemical or physical property" refers to a property in which all or part of a chemical or physical property of a substance changes with a change in direction and shows a difference in direction. The representation on the image is centered on a point, and the characteristics of the image are different in all directions of the point.
Referring to fig. 1-2, the present embodiment provides a full-polarization SAR image denoising method based on adaptive anisotropic diffusion, including the following steps:
s1, acquiring the fully polarized SAR data, and extracting a covariance matrix of the fully polarized SAR data;
the extraction of the covariance matrix of the fully polarimetric SAR data comprises a polarization scattering matrix S and a polarization covariance matrix C3The extraction of (1).
The polarized scattering matrix S can completely describe the electromagnetic scattering characteristics of the SAR target, and the definition is shown as formula (1):
wherein S isHVRepresenting horizontally polarized transmission, vertically polarized reception; sVHRepresenting vertical polarization transmission, horizontal polarization reception; sHHRepresenting horizontally polarized transmission and reception; sVVRepresenting vertically polarized transmission and reception. The diagonal elements of the polarized scattering matrix represent the incident field and scattered field relationship for the same polarization, referred to as the "co-polarized" term. The off-diagonal elements represent the incident angle and fringe field relationships for the orthogonal polarization mode, referred to as the "cross-polarization" term.
When the reciprocity theorem is satisfied, there is SHV＝SVHDecomposing the scattering matrix S by using a Lexicograter basis to obtain a polarization scattering vector
The definition is shown as formula (2):
in the formula
The norm of the target vector can be guaranteed to be equal to the total scattering power (Span), i.e., the 'conservation of total power' is satisfied.
When S isHV＝SVHThen, the target satisfies the reciprocity theorem, at which time the polarization scattering vector
Can be simplified into a three-dimensional polarization scattering vector, which is defined as shown in formula (3):
polarization covariance matrix C3Scattering vector by polarization
And polarization scattering vector
Conjugate transpose vector of
Formation of outer product of C3The definition is shown in formula (4):
wherein, T represents the conjugation transpose and conjugation, respectively, and <. indicates the overall average value.
S2, extracting polarization scattering entropy based on the covariance matrix of the fully polarized SAR data;
by polarizing covariance matrix C for 3X 3 Hermite3Diagonalization is carried out, and a polarization covariance matrix C is obtained through calculation3The eigenvalues and eigenvectors of (a) are as shown in equation (5):
wherein Σ is C3Is a 3 x 3 non-negative real diagonal matrix, and the main diagonal element of sigma is C3Characteristic value λ of1,λ2,λ3Having a characteristic value of λ1≥λ2≥λ3≥0，UCIs C3Of the eigenvector matrix, UC＝[e1e2e3]Is a 3 x 3 special unitary matrix SU (3) where e1,e2,e3Are respectively C3The feature vector of (2).
The polarization scattering entropy H is defined as shown in equation (6):
in the formula (I), the compound is shown in the specification,
the polarization scattering entropy describes the random scattering degree of the distributed scatterer, and is represented as single scattering when the value of the polarization scattering entropy is lower, and is represented as random scattering when the value of the polarization scattering entropy is higher. For example, when the radar beam irradiates the sea surface to be cleaned, the surface bragg scattering mainly occurs, and the scattering mechanism is relatively single, so the value of H is small; when the radar wave beam irradiates the seawater covered by the oil film, surface Bragg scattering is generated, and simultaneously, mirror surface scattering exists, so that the scattering process is relatively complex, the scattering randomness is strong, and the value of H is large. The extracted polarization scattering entropy is shown in fig. 3.
S3, selecting a gradient adaptive filtering window based on the polarization scattering entropy;
the invention provides a gradient self-adaptive window selection method, which reflects image detail change by utilizing gradient change of polarization scattering entropy, and enables a homogeneous region window with smaller gradient to be larger, a heterogeneous region window with larger gradient to be smaller and an edge region window to be narrower according to the direction and the size of a gradient self-adaptive selection filter window.
The specific method for selecting the gradient adaptive filtering window comprises the following steps:
s3.1, calculating the vertical gradient I of polarization scattering entropyx(I, j), horizontal gradient Iy(i, j), wherein i, j are x-axis and y-axis coordinates of the pixel respectively.
S3.2 vertical gradient I based on polarization scattering entropyx(I, j), horizontal gradient Iy(i, j), calculating an adaptive filter window QchooseThe initial direction and size of the device.
Let a be the maximum square window size, α0(i, j) is the initial direction of the adaptive filter window, as shown in equation (7), w0(i, j) is the initial width of the adaptive filtering window, as shown in equation (8), l0(i, j) is the initial length of the adaptive filtering window, as shown in equation (9).
α0(i,j)＝arctan(Iy(i,j)/Ix(i,j))……………………(7)
w0(i,j)＝a/(|Ix(i,j)|+1)………………………(8)
l0(i,j)＝a/(|Iy(i,j)|+1)………………………(9)
Wherein, Ix(i,j)、Iy(i, j) are the vertical gradient and the horizontal gradient of the polarization scattering entropy at the pixel (i, j), respectively.
S3.3, construction of Long loWidth woThe horizontal filtering window of (1) rotating the horizontal filtering window based on the initial direction of the adaptive filtering window and calculating the adaptive filtering window QchooseCoordinates of each vertex; given a single pixel centered QchooseInitial width w ofoLength l, length loAnd direction alphaoThen the conversion formula is as shown in formula (10):
in the formula, alphaoIs the initial direction of the adaptive filtering window; x is the number of1、x2、x3、x4Respectively, the abscissa, y, of the four vertices of the adaptive filter window1、y2、y3、y4Respectively, the ordinate of the four vertexes of the adaptive filtering window.
S3.4, based on self-adaptive filtering window QchooseCalculates the search window Q of each vertex coordinatesearchIs defined as shown in formula (11):
in the formula, ceil () represents rounding up, floor () represents rounding down.
S3.5, traversing the search window QsearchAcquiring overlapped pixels in a search window and an adaptive filtering window, and calculating the vertical gradient algebraic sum I of all the overlapped pixelsys(I, j) algebraic sum with horizontal gradient Ixs(i, j), which is defined as shown in formula (12).
In the formula, N represents the number of pixels overlapping in the search window and the adaptive filter window.
S3.6 algebraic sum I using vertical gradientys(I, j), horizontal gradient algebraic sum Ixs(i, j) updating the direction θ of the adaptive filter windowmDefined as shown in formula (13):
θm(i,j)＝arctan(Iys(i,j)/Ixs(i,j))…………………(13)。
s3.7, calculating the sum I of the absolute values of the vertical gradients of all overlapped pixels in the search window and the adaptive filtering windowyc(I, j), sum of absolute values of horizontal gradients Ixc(i, j), which is defined as shown in formula (14):
s3.8, utilizing sum I of absolute values of vertical gradientsyc(I, j), sum of absolute values of horizontal gradients Ixc(i, j) updating the size of the adaptive filtering window, and setting the updated selection window width as wm(i, j) of length lm(i, j), then it is defined as shown in formula (15):
in the formula, wm(i,j)、lmAnd (i, j) are the width and the length of the updated adaptive filtering window respectively.
S3.9, when the size of the self-adaptive filtering window meets the formula (16), stopping updating, wherein the width of the self-adaptive filtering window is wm+1(i, j) of length lm+1(i, j) in the direction of θm+1：
In the formula, T1And T2Is a preset threshold.
S4, performing Freeman-Durden decomposition on the covariance matrix of the fully-polarized SAR data to obtain the Bragg scattering, secondary scattering and volume scattering power of each pixel in the fully-polarized SAR data;
the Freeman-Durden decomposition is a technique that uses a three-component scattering mechanism model for polarimetric SAR observation, which does not require the use of any terrestrial measurement data. Under the condition of satisfying the reciprocity theorem, the covariance matrix of the target fully-polarized SAR data is represented again by using Freeman-Durden decomposition, and the definition is shown as formula (17):
in the formula (f)v、fd、fsAre all coefficients, and fv、fd、fsThe multiplied three matrixes respectively represent covariance matrixes of volume scattering, secondary scattering and Bragg scattering, alpha and beta are respectively a secondary scattering parameter and a Bragg scattering parameter, and the values of the alpha and the beta are represented by
Positive and negative of (2) are determined.
In Freeman-Durden decomposition, fv、fd、fsThe coefficient satisfies the equation relation shown in equation (18)Is described. Wherein when
When α is-1; when in use
When β is 1.
According to the principle of conservation of total polarization power, the final expression of the Freeman-Durden decomposition is obtained, the definition of which is shown in the formula (19), and the result of the Freeman-Durden decomposition is shown in FIG. 4.
In the formula, Ps、Pd、PvThe respective powers of bragg scattering, secondary scattering, and volume scattering are shown.
S5, extracting a ground object scattering label graph based on the power of Bragg scattering, secondary scattering and volume scattering of each pixel in the fully polarized SAR data;
in the polarized SAR image, different surface features have different scattering characteristics, and different surface feature scattering characteristics can be used for describing differences of different types of surface features or the same type of surface features in the aspect of polarized scattering, so the denoising of the polarized SAR image should be performed based on the same scattering mechanism.
The method for extracting the ground object scattering marker map specifically comprises the following steps:
s5.1, performing initial clustering on each pixel in the fully-polarized SAR data based on the power of Bragg scattering, secondary scattering and volume scattering of each pixel in the fully-polarized SAR data;
in different scattering types, all pixels in the fully-polarized SAR data are initially clustered according to the power of Bragg scattering, secondary scattering and volume scattering to obtain 30 initial categories.
S5.2, based on the initial clustering result, carrying out merged clustering on all pixels in the fully-polarized SAR data by adopting a Wishart distance;
the distance metric equation is shown in equation (20):
in the formula, VpAnd VqClass mean values, D, representing the covariance matrices of class p and class q in the initial clustering result, respectivelypqRepresenting the distance between the pth and qth classes in the initial clustering result, if D between two classes in the initial clustering resultpqIf the shortest, the two types are merged; in this embodiment, all initial clusters of different scattering types will eventually be merged into 9 classes.
S5.3, based on the merged clustering result, performing iterative classification on each pixel in the fully-polarized SAR data by adopting a Wishart distance;
respectively calculating the Wishart distance between each pixel in the fully-polarized SAR data and 9 clustering centers in the combined clustering result, and reclassifying each pixel in the fully-polarized SAR data according to the distance calculation result, wherein in order to make the classification result more convergent, the embodiment utilizes an iterative Wishart classifier to perform iterative calculation for 5-10 times. The Wishart distance metric equation is shown in equation (21):
wherein C represents a polarization covariance matrix of the pixel, VfRepresenting the f-th class center in the merged clustering result.
And S5.4, based on the iterative classification result, adopting a K-means clustering algorithm to redistribute each pixel in the fully-polarized SAR data to generate a ground object scattering label graph.
The K-means clustering algorithm is a clustering analysis algorithm for iterative solution, and specifically comprises the following steps:
firstly, based on an iterative classification result, randomly selecting a plurality of pixels as an initial K-means clustering center; in this embodiment, 3 pixels are randomly selected.
Then, the sum of absolute differences between each pixel in the fully-polarized SAR data and each K-means clustering center is calculated, and each pixel in the fully-polarized SAR data is allocated to the K-means clustering center closest to the pixel based on the sum of absolute differences.
The K-means cluster center and the pixels assigned to the K-means cluster center represent a cluster. Each pixel is allocated, the K-means cluster center is recalculated according to the existing pixels in the cluster, the process is repeated until no object is reallocated to a different cluster, and finally, a labeled surface feature map, namely a surface feature scattering label map, is generated, as shown in fig. 5. The K-means distance metric equation is shown in equation (22):
wherein, P is the dimension of data, where P is 1; x is the number ofjIs a pixel; c. CjIs the cluster center.
S6, denoising the fully polarized SAR image by adopting a self-adaptive anisotropic diffusion method of a scattering mechanism based on a gradient self-adaptive filtering window and a ground feature scattering marker map. The method specifically comprises the following steps:
s6.1, based on the gradient adaptive filtering window and the ground feature scattering marker map, eliminating neighborhood pixels in the gradient adaptive filtering window, wherein the scattering types of the neighborhood pixels are different from those of the central pixel;
s6.2, calculating a gradient operator based on the residual pixels in the gradient adaptive filtering window
As shown in equation (23):
in which B is ∈ [1, B ]]And b is an integer; i (x)b,yb) Representing a neighborhoodPixel, I (x, y) represents the center pixel; b is the number of neighborhood pixels with the same scattering type as the central pixel in the gradient adaptive filtering window; η is the distance between the neighborhood pixel and the center pixel.
S6.3, calculating an adaptive gradient threshold k based on a gradient operator, wherein the adaptive gradient threshold k is shown as a formula (24):
in the formula (I), the compound is shown in the specification,
is the average of the gradients within the gradient adaptive filter window.
S6.4, constructing an adaptive diffusion model based on the adaptive gradient threshold, wherein the adaptive diffusion model is shown as a formula (25):
in the formula, g () is an adaptive diffusion function, and s represents an adaptive diffusion function variable.
S6.5, performing iterative computation based on the adaptive diffusion model to complete denoising of the fully-polarized SAR image, as shown in formula (26):
where m is the number of iterations, Δ t is the time interval, λ (I)0-Im(x, y)) is a fidelity term, and λ is a fidelity coefficient; through the fidelity term, the denoised full-polarization SAR image can be close to the original image as much as possible, and polarization information in the image is effectively kept. The denoising result of the fully polarized SAR image is shown in fig. 6.
In order to further verify the effectiveness and reliability of the self-adaptive anisotropic diffusion-based full-polarization SAR image denoising method, in the embodiment, the method of the invention is used to compare with the denoising results of the polarization SAR image obtained by 6 classical denoising methods, such as SARD, Re-Lee, Boxcar, Gauss, IDAN, and Mean-shift, and the denoising results are shown in FIG. 7. The speckle suppression method comprises the steps of conducting quantitative and qualitative evaluation on speckle suppression results by using 6 denoising evaluation indexes of polarization characteristic diagram, equivalent visual number ENL, radiation resolution, mean value, variance and covariance matrix difference, wherein the 6 denoising evaluation indexes are used for conducting quantitative and qualitative evaluation on speckle suppression results, the polarization characteristic diagram is shown in figure 8 to be a qualitative analysis result, the equivalent visual number is shown in table 1 to be a quantitative evaluation result, the radiation resolution is shown in table 2 to be a quantitative evaluation result, the image mean value is shown in table 3 to be a quantitative evaluation result, the image variance is shown in table 4 to be a quantitative evaluation result, and the covariance matrix difference is shown in table 5 to be a quantitative evaluation result. The polarization characteristic diagram is an echo power diagram of the transmitting-receiving antenna in different polarization states, is an advantageous tool for representing the target polarization characteristic, and the polarization characteristic retentivity of the polarization filtering algorithm is visually evaluated by generally adopting a method for comparing the consistency degrees of the polarization characteristic diagrams before and after polarization filtering, wherein the closer the polarization characteristic diagrams are, the stronger the polarization characteristic retentivity of the algorithm is, and the more reliable the algorithm is; the equivalent vision ENL can measure the smoothness degree of the filtered image, and if the ENL of a certain homogeneous region is larger, the smoother the region is, and the better the filtering performance is; radiation resolution, which represents the ability to distinguish backscatter coefficients of a polarized SAR target, is a measure of the ability of a polarized SAR system to distinguish between adjacently distributed targets. The quality of denoising directly influences polarized SAR image interpretation and quantitative application, and the size of the radiation resolution is determined by the amount of speckle noise to be eliminated, so that the good polarized SAR denoising method can improve the radiation resolution; the image mean value is the average intensity of the whole image and reflects the average gray level of the image, namely the average backscattering coefficient of the target contained in the image; the image variance represents the degree of deviation of all points in the image area from the mean value, and reflects the nonuniformity of the image; the mean and variance of an image are indexes reflecting the overall characteristics of the image, so the mean of the image should be maintained as much as possible while the variance of the image is reduced. The covariance matrix difference Dod can measure the fidelity of the covariance matrix of the image before and after filtering, the polarization information retention performance of the polarization SAR denoising method can be quantitatively evaluated by comparing the covariance matrix difference of the image before and after filtering, if the Dod is smaller, the change of the covariance matrix is smaller, and the polarization information retention degree is higher.
TABLE 1
TABLE 2
TABLE 3
TABLE 4
TABLE 5
First, from qualitative perspective evaluation, it can be seen from fig. 7 that the polarization characteristic diagrams of different polarization states extracted from the filtering result of the method of the present invention are almost not different from the polarization characteristic diagram extracted from the original image, which indicates that the polarization information retention capability of the method of the present invention is strong. In addition, from quantitative evaluation, it is seen from tables 1 and 2 that the equivalent vision and the radiation resolution of the method of the present invention are both significantly better than those of other methods, from table 3, it can be seen that the mean value of the filtered image of the method of the present invention is consistent with that of the original image, and from table 4, it is seen that the variance of the filtering results of Re-Lee and IDAN methods is smaller, but from table 3, the difference between the mean value of the filtered image of the two methods and the original image is larger, so it can be seen from the combination of tables 3 and 4 that the filtered image of the method of the present invention is better than those of other methods. As can be seen from Table 5, the covariance matrix of the method of the present invention has the least difference, so the polarization information retention of the method of the present invention is better than that of other methods. This shows that the polarization information of the polarized SAR data can be more effectively maintained based on the adaptive anisotropic diffusion method compared with other filtering methods. Therefore, the invention is superior to other filtering methods in terms of speckle suppression and polarization information preservation.
The above-described embodiments are merely illustrative of the preferred embodiments of the present invention, and do not limit the scope of the present invention, and various modifications and improvements of the technical solutions of the present invention can be made by those skilled in the art without departing from the spirit of the present invention, and the technical solutions of the present invention are within the scope of the present invention defined by the claims.
Claims (5)
1. The full-polarization SAR image denoising method based on the self-adaptive anisotropic diffusion is characterized by comprising the following steps:
s1, acquiring the fully polarized SAR data, and extracting a covariance matrix of the fully polarized SAR data;
s2, extracting polarization scattering entropy based on the covariance matrix of the fully polarized SAR data;
s3, selecting a gradient adaptive filtering window based on the polarization scattering entropy;
s4, performing Freeman-Durden decomposition on the covariance matrix of the fully-polarized SAR data to obtain the Bragg scattering, secondary scattering and volume scattering power of each pixel in the fully-polarized SAR data;
s5, extracting a ground object scattering label graph based on the power of Bragg scattering, secondary scattering and volume scattering of each pixel in the fully polarized SAR data;
s6, denoising the fully polarized SAR image by adopting a self-adaptive anisotropic diffusion method of a scattering mechanism based on a gradient self-adaptive filtering window and a ground feature scattering marker map;
the step S3 specifically includes:
calculating the vertical gradient and the horizontal gradient of the polarization scattering entropy, and calculating the initial direction and the size of the self-adaptive filtering window based on the vertical gradient and the horizontal gradient of the polarization scattering entropy;
construction of Long loWidth woThe conversion formula is shown as follows:
in the formula, alphaoIs the initial direction of the adaptive filtering window; x is the number of1、x2、x3、x4Respectively, the abscissa, y, of the four vertices of the adaptive filter window1、y2、y3、y4Respectively are the vertical coordinates of four vertexes of the self-adaptive filtering window;
constructing a horizontal filtering window based on the initial size of the self-adaptive filtering window, rotating the horizontal filtering window based on the initial direction of the self-adaptive filtering window, calculating the coordinates of each vertex of the self-adaptive filtering window, and calculating the coordinates of each vertex of the self-adaptive filtering window;
traversing the search window, acquiring overlapped pixels in the search window and the adaptive filtering window, calculating the vertical gradient algebraic sum and the horizontal gradient algebraic sum of all the overlapped pixels, and updating the direction of the adaptive filtering window by utilizing the vertical gradient algebraic sum and the horizontal gradient algebraic sum;
calculating the sum of absolute values of vertical gradients and the sum of absolute values of horizontal gradients of all overlapped pixels in the search window and the adaptive filtering window, and updating the size of the adaptive filtering window by using the sum of absolute values of vertical gradients and the sum of absolute values of horizontal gradients until the size of the adaptive filtering window meets a preset threshold;
updating the size of the adaptive filtering window, and setting the width of the updated selection window as wm(i, j) of length lm(i, j), which is defined by the following formula:
in the formula, wm(i,j)、lm(i, j) are updated adaptations, respectivelyThe width and length of the filtering window;
initial direction alpha of the adaptive filter window0(i, j) is represented by formula 7:
α0(i,j)＝arctan(Iy(i,j)/Ix(i,j))……………………7
wherein, Ix(i,j)、Iy(i, j) are respectively a vertical gradient and a horizontal gradient of polarization scattering entropy at the pixel (i, j), and i, j are respectively x-axis coordinates and y-axis coordinates of the pixel;
the vertex coordinates of the search window are shown as follows:
in the formula, Xmin、XmaxRespectively the minimum value and the maximum value of the abscissa of the top point of the search window; y ismin、YmaxRespectively the minimum value and the maximum value of the vertical coordinate of the top point of the search window; x is the number of1、x2、x3、x4Respectively are the horizontal coordinates of four vertexes of the self-adaptive filtering window; y is1、y2、y3、y4Respectively are the vertical coordinates of four vertexes of the self-adaptive filtering window; ceil () represents rounded up and floor () represents rounded down.
2. The method for denoising a fully polarimetric SAR image based on adaptive anisotropic diffusion according to claim 1, wherein the step S5 specifically includes:
s5.1, performing initial clustering on each pixel in the fully-polarized SAR data based on the power of Bragg scattering, secondary scattering and volume scattering of each pixel in the fully-polarized SAR data;
s5.2, based on the initial clustering result, carrying out merged clustering on all pixels in the fully-polarized SAR data by adopting a Wishart distance;
s5.3, based on the merged clustering result, performing iterative classification on each pixel in the fully-polarized SAR data by adopting a Wishart distance;
and S5.4, based on the iterative classification result, adopting a K-means clustering algorithm to redistribute each pixel in the fully-polarized SAR data to generate a ground object scattering label graph.
3. The method for denoising a fully-polarized SAR image based on adaptive anisotropic diffusion according to claim 2, wherein in the step S5.1, all pixels in the fully-polarized SAR data are initially clustered according to the power of Bragg scattering, secondary scattering and volume scattering.
4. The self-adaptive anisotropic diffusion-based denoising method for the fully-polarized SAR image as claimed in claim 2, wherein in step S5.3, Wishart distances between each pixel in the fully-polarized SAR data and a plurality of clustering centers in the merged clustering result are respectively calculated, and each pixel in the fully-polarized SAR data is reclassified according to the distance calculation result.
5. The method for denoising a fully polarimetric SAR image based on adaptive anisotropic diffusion according to claim 1, wherein the step S6 specifically includes:
s6.1, based on the gradient adaptive filtering window and the ground feature scattering marker map, eliminating neighborhood pixels in the gradient adaptive filtering window, wherein the scattering types of the neighborhood pixels are different from those of the central pixel;
s6.2, calculating a gradient operator based on the residual pixels in the gradient adaptive filtering window;
s6.3, calculating a self-adaptive gradient threshold value based on a gradient operator;
s6.4, constructing a self-adaptive diffusion model based on the self-adaptive gradient threshold;
and S6.5, performing iterative computation based on the adaptive diffusion model to complete denoising of the fully-polarized SAR image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110117113.XA CN112950492B (en) | 2021-01-28 | 2021-01-28 | Full-polarization SAR image denoising method based on self-adaptive anisotropic diffusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110117113.XA CN112950492B (en) | 2021-01-28 | 2021-01-28 | Full-polarization SAR image denoising method based on self-adaptive anisotropic diffusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112950492A CN112950492A (en) | 2021-06-11 |
| CN112950492B true CN112950492B (en) | 2022-04-29 |
Family
ID=76238497
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110117113.XA Expired - Fee Related CN112950492B (en) | 2021-01-28 | 2021-01-28 | Full-polarization SAR image denoising method based on self-adaptive anisotropic diffusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112950492B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116091345B (en) * | 2022-12-23 | 2023-10-03 | 重庆大学 | Anisotropic diffusion medical image denoising method, system and storage medium based on local entropy and fidelity terms |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102096921A (en) * | 2011-01-10 | 2011-06-15 | 西安电子科技大学 | SAR (Synthetic Aperture Radar) image change detection method based on neighborhood logarithm specific value and anisotropic diffusion |
| CN103778613A (en) * | 2014-02-21 | 2014-05-07 | 武汉大学 | Polarization SAR image filtering method for window self-adaptation |
| CN104318246A (en) * | 2014-10-20 | 2015-01-28 | 西安电子科技大学 | Depth self-adaption ridgelet network based polarimetric SAR (Synthetic Aperture Radar) image classification |
| CN104951789A (en) * | 2015-07-15 | 2015-09-30 | 电子科技大学 | Quick landslide extraction method based on fully polarimetric SAR (synthetic aperture radar) images |
| CN105160353A (en) * | 2015-08-18 | 2015-12-16 | 西安电子科技大学 | Polarimetric SAR data ground object classification method based on multiple feature sets |
| CN106778884A (en) * | 2016-12-23 | 2017-05-31 | 辽宁工程技术大学 | The Classification of Polarimetric SAR Image method of plane is decomposed based on scattering entropy and three-component |
| CN107229933A (en) * | 2017-05-11 | 2017-10-03 | 西安电子科技大学 | The freeman/ Eigenvalues Decomposition methods of adaptive volume scattering model |
| CN107644408A (en) * | 2017-09-27 | 2018-01-30 | 安徽大学 | Full polarization radar image denoising method based on anisotropic diffusion |
| CN108387898A (en) * | 2018-02-27 | 2018-08-10 | 河海大学 | The polarization characteristic parameter extracting method chosen based on homologous pints |
| CN108876746A (en) * | 2018-06-28 | 2018-11-23 | 广东工业大学 | A kind of SAR image method for reducing speckle and system based on anisotropic diffusion equation |
| CN110363105A (en) * | 2019-06-25 | 2019-10-22 | 电子科技大学 | A kind of full polarimetric SAR speckle suppression method |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103824084B (en) * | 2014-03-12 | 2016-11-02 | 西安电子科技大学 | Classification of Polarimetric SAR Image method based on SDIT and SVM |
| CN109740475B (en) * | 2018-12-25 | 2020-04-21 | 杭州世平信息科技有限公司 | Ground scene classification method for remote sensing image |
-
2021
- 2021-01-28 CN CN202110117113.XA patent/CN112950492B/en not_active Expired - Fee Related
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102096921A (en) * | 2011-01-10 | 2011-06-15 | 西安电子科技大学 | SAR (Synthetic Aperture Radar) image change detection method based on neighborhood logarithm specific value and anisotropic diffusion |
| CN103778613A (en) * | 2014-02-21 | 2014-05-07 | 武汉大学 | Polarization SAR image filtering method for window self-adaptation |
| CN104318246A (en) * | 2014-10-20 | 2015-01-28 | 西安电子科技大学 | Depth self-adaption ridgelet network based polarimetric SAR (Synthetic Aperture Radar) image classification |
| CN104951789A (en) * | 2015-07-15 | 2015-09-30 | 电子科技大学 | Quick landslide extraction method based on fully polarimetric SAR (synthetic aperture radar) images |
| CN105160353A (en) * | 2015-08-18 | 2015-12-16 | 西安电子科技大学 | Polarimetric SAR data ground object classification method based on multiple feature sets |
| CN106778884A (en) * | 2016-12-23 | 2017-05-31 | 辽宁工程技术大学 | The Classification of Polarimetric SAR Image method of plane is decomposed based on scattering entropy and three-component |
| CN107229933A (en) * | 2017-05-11 | 2017-10-03 | 西安电子科技大学 | The freeman/ Eigenvalues Decomposition methods of adaptive volume scattering model |
| CN107644408A (en) * | 2017-09-27 | 2018-01-30 | 安徽大学 | Full polarization radar image denoising method based on anisotropic diffusion |
| CN108387898A (en) * | 2018-02-27 | 2018-08-10 | 河海大学 | The polarization characteristic parameter extracting method chosen based on homologous pints |
| CN108876746A (en) * | 2018-06-28 | 2018-11-23 | 广东工业大学 | A kind of SAR image method for reducing speckle and system based on anisotropic diffusion equation |
| CN110363105A (en) * | 2019-06-25 | 2019-10-22 | 电子科技大学 | A kind of full polarimetric SAR speckle suppression method |
Non-Patent Citations (5)
| Title |
|---|
| 一种自适应的混合Freeman /Eigenvalue 极化分解模型;何连，秦其明，任华忠;《国土资源遥感》;20170630;第29卷(第2期);全文 * |
| 基于GA-WNN 的极化SAR 海洋溢油检测方法研究;陈伟民, 丁亚雄, 宋冬梅, 王 斌等;《海洋科学》;20181231;第42卷(第1期);全文 * |
| 基于场景异质性的极化SAR相干斑滤波方法;行晓黎;《中国博士学位论文全文数据库信息科技辑》;20180715;第2018年卷(第7期);全文 * |
| 基于改进扩散系数的各向异性扩散相干斑抑制算法;李艳华;《中国优秀硕士学位论文数据库信息科技辑》;20160315;第2016年卷(第3期);全文 * |
| 基于新极化特征参数的SAR 海洋溢油检测;任慧敏，宋冬梅等;《遥感技术与应用》;20200831;第35卷(第4期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112950492A (en) | 2021-06-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Huang et al. | Under-foliage object imaging using SAR tomography and polarimetric spectral estimators | |
| Frey et al. | Analyzing tomographic SAR data of a forest with respect to frequency, polarization, and focusing technique | |
| Ertin et al. | GOTCHA experience report: Three-dimensional SAR imaging with complete circular apertures | |
| CN112558066B (en) | Dual polarized SAR image system | |
| Cloude et al. | Target detection beneath foliage using polarimetric synthetic aperture radar interferometry | |
| CN101369019A (en) | Polarization interference synthetic aperture radar three-dimensional imaging method based on polarization data amalgamation | |
| Chen et al. | Mitigation of azimuth ambiguities in spaceborne stripmap SAR images using selective restoration | |
| Shahrezaei et al. | Fractal analysis and texture classification of high-frequency multiplicative noise in SAR sea-ice images based on a transform-domain image decomposition method | |
| Rambour et al. | Introducing spatial regularization in SAR tomography reconstruction | |
| Stojanovic et al. | Joint space aspect reconstruction of wide-angle SAR exploiting sparsity | |
| Iqbal et al. | SAR image despeckling by selective 3D filtering of multiple compressive reconstructed images | |
| Nielsen et al. | Direction-of-arrival estimation for radar ice sounding surface clutter suppression | |
| CN112950492B (en) | Full-polarization SAR image denoising method based on self-adaptive anisotropic diffusion | |
| Volosyuk et al. | Phenomenological description of coherent radar images based on the concepts of the measure of set and stochastic integral | |
| Omati et al. | The comparative study of three nonparametric methods of SAR tomography for building reconstruction | |
| Stagliano et al. | Ship detection from SAR images based on CFAR and wavelet transform | |
| Demirci et al. | An investigation of the performances of polarimetric target decompositions using GB-SAR imaging | |
| Brigui et al. | New SAR target imaging algorithm based on oblique projection for clutter reduction | |
| Dos Santos et al. | Polarimetric responses and scattering mechanisms of tropical forest in the Brazilian Amazon | |
| Karachristos et al. | A Review on PolSAR Decompositions for Feature Extraction | |
| Azzaz et al. | Classification of radar echoes using fractal geometry | |
| Aghababaee et al. | Full 3D DEM generation in urban area by improving estimation from SAR tomography | |
| Wecklich et al. | Production of a global forest/non-forest map utilizing TanDEM-X interferometric SAR data | |
| Akbari | Multitemporal analysis of multipolarization synthetic aperture radar images for robust surface change detection | |
| Prakash et al. | Emerging techniques of polarimetric interferometric synthetic aperture radar for scattering-based characterization |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20220429 |