CN110516698A - Complete polarization image polarization decomposing method, apparatus, electronic equipment and storage medium - Google Patents

Complete polarization image polarization decomposing method, apparatus, electronic equipment and storage medium Download PDF

Info

Publication number
CN110516698A
CN110516698A CN201910637291.8A CN201910637291A CN110516698A CN 110516698 A CN110516698 A CN 110516698A CN 201910637291 A CN201910637291 A CN 201910637291A CN 110516698 A CN110516698 A CN 110516698A
Authority
CN
China
Prior art keywords
scattering
object pixel
value
angle
pixel
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
Application number
CN201910637291.8A
Other languages
Chinese (zh)
Other versions
CN110516698B (en
Inventor
王宇
禹卫东
刘秀清
王春乐
吕继宇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Institute of Electronics of CAS
Original Assignee
Institute of Electronics of CAS
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Institute of Electronics of CAS filed Critical Institute of Electronics of CAS
Priority to CN201910637291.8A priority Critical patent/CN110516698B/en
Publication of CN110516698A publication Critical patent/CN110516698A/en
Application granted granted Critical
Publication of CN110516698B publication Critical patent/CN110516698B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2134—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on separation criteria, e.g. independent component analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Pure & Applied Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Evolutionary Biology (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computational Mathematics (AREA)
  • Algebra (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Databases & Information Systems (AREA)
  • Investigating Or Analysing Materials By Optical Means (AREA)
  • Eye Examination Apparatus (AREA)
  • Image Processing (AREA)

Abstract

The embodiment of the invention discloses a kind of methods of complete polarization image polarization decomposing, comprising: n of each pixel independent hiding characteristic value in p object pixel in acquisition complete polarization image；Principal component analysis is carried out to n of each pixel in the p object pixel independent hiding characteristic value, obtains independent hiding the corresponding n weighted value of characteristic value with described n；Determine m weighted value in the n weighted value, wherein the m weighted value is all larger than n-m weighted value among the n weight in addition to the m weighted value, m < n；Obtain the corresponding m independent hiding characteristic value of the m weighted value；Hiding characteristic value independent for m described in the p object pixel is all larger than q object pixel of the first preset threshold, and the volume scattering power of the q object pixel, q < p are determined using the first volume scattering model.The embodiment of the invention also discloses device, electronic equipment and the computer storage mediums of a kind of polarization classification of complete polarization image.

Description

Complete polarization image polarization decomposing method, apparatus, electronic equipment and storage medium
Technical field
The present invention relates to the goal decomposition technology in polarimetric synthetic aperture radar field more particularly to a kind of complete polarization image poles Change decomposition method, device, electronic equipment and computer storage medium.
Background technique
With high resolution synthetic aperture radar (Synthetic Aperture Radar, SAR) image and polarization measurement The maturation of technology, polarization SAR carry important role in civil and military field.Polarization target decomposition technology is as polarization One of important branch of SAR, to more fully understand that target scattering mechanism provides useful information.Decomposition technique is broadly divided into two A part: decomposition based on model and based on the decomposition of characteristic value-feature vector.Wherein, the decomposition based on model and physics dissipate It penetrates that mechanism is directly related, can be realized by the way that coherence matrix to be decomposed into the combination of each scattering ingredient to target scattering mechanism Effectively description.Nearest decades, many decomposition methods for improving volume scattering model, which are suggested, overcomes volume scattering energy mistake Estimation and negative scattering energy problem, but additional calculating is so that method low efficiency.Due to the diversity of natural scene, method Validity is still an open problem.And in existing goal decomposition method, there is estimation to volume scattering energy and asked Topic.
Summary of the invention
In view of this, an embodiment of the present invention is intended to provide a kind of complete polarization image polarization decomposing method, apparatus, electronic equipment And computer storage medium, at least partly solve the above problems.
In order to achieve the above objectives, the technical scheme of the present invention is realized as follows:
A kind of method that the embodiment of the present invention proposes complete polarization image polarization decomposing, which comprises
N of each pixel independent hiding characteristic value in p object pixel in acquisition complete polarization image；
Principal component analysis is carried out to n of each pixel in the p object pixel independent hiding characteristic value, is obtained and institute State the n independent hiding corresponding n weighted value of characteristic value；
Determine m weighted value in the n weighted value, wherein the m weighted value be all larger than the n weight it In n-m weighted value in addition to the m weighted value, m < n；
Obtain the corresponding m independent hiding characteristic value of the m weighted value；
Hiding characteristic value independent for m described in the p object pixel is all larger than q target of the first preset threshold Pixel determines the volume scattering power of the q object pixel, q < p using the first volume scattering model.
In above scheme, the method also includes:
For p-q object pixel in the p object pixel in addition to the q object pixel, using the second body Scattering model determines the volume scattering power of the p-q object pixel.
In above scheme, n of each pixel independent hiding feature in p object pixel in the acquisition complete polarization image Value, comprising:
Obtain the ensemble average value < [T] of the coherence matrix of the complete polarization image >,
Wherein, k is Pauli (Pauli) base,S is the complete polarization The collision matrix of image,Wherein, SHHFor Vertical Launch, the back scattering of vertical reception, SHVIt is vertical Transmitting, horizontal received back scattering, SVHFor the back scattering of horizontal emission, vertical reception, SVVIt is connect for horizontal emission, level The back scattering of receipts；
It obtains after carrying out orientation angles compensation to<[T]>through the compensated coherence matrix of the angle of orientation<[T ']>,
Wherein,θ indicates orientation angles, [R (θ)] is tenth of the twelve Earthly Branches spin matrix；
Described n independent hiding characteristic value, comprising:
Wherein, A indicates that the amplitude of rotation coherence matrix, B indicate the amplitude center of rotation coherence matrix, θ0Indicate initial phase Place value, ω indicate angular frequency；Re [] indicates that real part, Im [] indicate imaginary part, and Angle { } is for obtaining element in multiple axis Phase value；A_T′ijIndicate T 'ijAmplitude, B_T 'ijIndicate T 'ijAmplitude center, θ0_T′ijIndicate T 'ijInitial phase value.
In above scheme, the first volume scattering modelExpression formula are as follows:
Wherein, ψ indicates spiral angle.
In above scheme, the second volume scattering modelExpression formula are as follows:
Wherein, τ be calculating process amount, τ=< | SHH|2>/<|SVV|2>。
In above scheme, the method also includes:
It is obtained after the compensated coherence matrix of the angle of orientation<[T ']>carries out spiral angle compensation through the angle of orientation and spiral shell to described The compensated coherence matrix of swing angle<[T "]>,
Wherein,[R (ψ)] is the tenth of the twelve Earthly Branches Transformation matrix.
In above scheme, the method also includes:
For the q object pixel, line scattering ingredient, spiral are being subtracted in the coherence matrix of the q object pixel After scattering ingredient and volume scattering ingredient, according to the first decision threshold C1Determine leading scattering mechanism；
C1=T '11-T′22+fc/2-2Re{γ}|fw|-2m1fv, wherein m1For calculating process amount, m1=(cos (4 θ)- 15)/60, fcIt is right for spiral scattering component | SVV|2Contribution, fvIt is right for volume scattering component | SVV|2Contribution, fwFor line scattering point Amount pair | SVV| 2Contribution, γ be calculating process amount, γ=SHH/SVV；
Work as C1When > 0, surface scattering ingredient is leading scattering mechanism；Work as C1When≤0, dihedral angle scattering is leading scattering machine System.
In above scheme, the method also includes:
For the p-q object pixel, the p-q object pixel coherence matrix subtract line scattering ingredient, After spiral scatters ingredient and volume scattering ingredient, leading scattering mechanism is determined according to the second decision threshold；
C2=T '11-T′22+fc/2-2Re{γ}|fw|-2m2fv, wherein m2For calculating process amount,
Work as C2When > 0, surface scattering ingredient is leading scattering mechanism；Work as C2When≤0, dihedral angle scattering is leading scattering machine System.
The embodiment of the present invention provides a kind of device of complete polarization image polarization decomposing, and described device includes:
Image processing unit, the n independence for obtaining each pixel in p object pixel in complete polarization image hide spy Value indicative；
Main control unit carries out principal component for the independent hiding characteristic value of n to each pixel in the p object pixel Analysis obtains n weighted value corresponding with the hiding characteristic value of n independence；Determine m in the n weighted value Weighted value, wherein the m weighted value is all larger than n-m weight among the n weight in addition to the m weighted value Value, m < n；Obtain the corresponding m independent hiding characteristic value of the m weighted value；
Polarization decomposing unit is all larger than first for hiding characteristic value independent for m described in the p object pixel Q object pixel of preset threshold determines the volume scattering power of the q object pixel, q < p using the first volume scattering model.
In above scheme, the polarization decomposing unit is also used to: for removing the q target in the p object pixel P-q object pixel other than pixel determines the volume scattering power of the p-q object pixel using the second volume scattering model.
In above scheme, described image processing unit is specifically used for: obtaining the collection of the coherence matrix of the complete polarization image It closes average value<[T]>,
Wherein, k is Pauli base,S is the complete polarization image Collision matrix,Wherein, SHHFor Vertical Launch, the back scattering of vertical reception, SHVFor Vertical Launch, water Put down received back scattering, SVHFor the back scattering of horizontal emission, vertical reception, SVVIt is received backward for horizontal emission, level Scattering；
It obtains after carrying out orientation angles compensation to<[T]>through the compensated coherence matrix of the angle of orientation<[T ']>,
Wherein,θ indicates orientation angles, [R (θ)] is tenth of the twelve Earthly Branches spin matrix；
Described n independent hiding characteristic value, comprising:
Wherein, A indicates that the amplitude of rotation coherence matrix, B indicate the amplitude center of rotation coherence matrix, θ0Indicate initial phase Place value, ω indicate angular frequency；Re [] indicates that real part, Im [] indicate imaginary part, and Angle { } is for obtaining element in multiple axis Phase value；A_T′ijIndicate T 'ijAmplitude, B_T 'ijIndicate T 'ijAmplitude center, θ0_T′ijIndicate T 'ijInitial phase value.
In above scheme, described image processing unit is also used to: to it is described through the compensated coherence matrix of the angle of orientation < It obtains after [T ']>progress spiral angle compensation through the angle of orientation and the compensated coherence matrix of helical angle<[T "]>,
Wherein,[R (ψ)] is the tenth of the twelve Earthly Branches Transformation matrix.
In above scheme, the polarization decomposing unit is also used to: for the q object pixel, in the q target The coherence matrix of pixel is after subtracting line scattering ingredient, spiral scattering ingredient and volume scattering ingredient, according to the first decision threshold C1 Determine leading scattering mechanism；
C1=T '11-T′22+fc/2-2Re{γ}|fw|-2m1fv, wherein m1For calculating process amount, m1=(cos (4 θ)- 15)/60, fcIt is right for spiral scattering component | SVV|2Contribution, fvIt is right for volume scattering component | SVV|2Contribution, fwFor line scattering point Amount pair | SVV|2Contribution, γ be calculating process amount, γ=SHH/SVV；
Work as C1When > 0, surface scattering ingredient is leading scattering mechanism；Work as C1When≤0, dihedral angle scattering is leading scattering machine System.
In above scheme, the polarization decomposing unit is also used to: for the p-q object pixel, at the p-q The coherence matrix of object pixel is after subtracting line scattering ingredient, spiral scattering ingredient and volume scattering ingredient, according to the second decision threshold Value C2Determine leading scattering mechanism；
C2=T '11-T′22+fc/2-2Re{γ}|fw|-2m2fv, wherein m2For calculating process amount,
Work as C2When > 0, surface scattering ingredient is leading scattering mechanism；Work as C2When≤0, dihedral angle scattering is leading scattering machine System.
The embodiment of the present invention provides a kind of electronic equipment, comprising: transceiver, memory, processor and is stored in described deposit The computer program executed on reservoir and by the processor；
The processor is connect with the transceiver and the memory respectively, for by executing the computer journey Sequence, the method for realizing any of the above-described complete polarization image polarization decomposing.
The embodiment of the present invention provides a kind of computer storage medium, and the computer storage medium is stored with computer journey Sequence；After the computer program is performed, the method that can be realized any of the above-described complete polarization image polarization decomposing.
The method of complete polarization image polarization decomposing provided in an embodiment of the present invention, by obtaining p mesh in complete polarization image N independent hiding characteristic value of each pixel in pixel is marked, it is independent hiding to n of each pixel in the p object pixel Characteristic value carries out principal component analysis, obtains n weighted value corresponding with the hiding characteristic value of n independence；Determine the n M weighted value in a weighted value, wherein the m weighted value is all larger than among the n weight except the m weighted value N-m weighted value in addition, m < n；Obtain the corresponding m independent hiding characteristic value of the m weighted value；For the p mesh M independent hiding characteristic value described in mark pixel is all larger than q object pixel of the first preset threshold, using the first volume scattering mould Type determines the volume scattering power of the q object pixel, q < p；It is pre- by being all larger than first to described m independent hiding characteristic value If q object pixel of threshold value determines the volume scattering power of the q object pixel using the first volume scattering model, reduce volume Outer volume scattering power calculating process, and reduce in the prior art to complete polarization image all pixels using identical volume scattering mould Type calculates the problem of volume scattering energy caused by volume scattering power crosses estimation.
Detailed description of the invention
Attached drawing generally shows each embodiment discussed herein by way of example and not limitation.
Fig. 1 is a kind of method flow schematic diagram of complete polarization image polarization decomposing of the embodiment of the present invention；
Fig. 2 is a kind of method flow schematic diagram of complete polarization image polarization decomposing of the embodiment of the present invention；
Fig. 3 is a kind of flow diagram of multielement polarization image decomposition method of the embodiment of the present invention；
Fig. 4 is a kind of apparatus structure schematic diagram of complete polarization image polarization decomposing of the embodiment of the present invention；
Fig. 5 is a kind of multielement polarization image decomposer structural schematic diagram of the embodiment of the present invention；
Fig. 6 is the structural schematic diagram of electronic equipment provided in an embodiment of the present invention.
Specific embodiment
The characteristics of in order to more fully hereinafter understand the embodiment of the present invention and technology contents, with reference to the accompanying drawing to this hair The realization of bright embodiment is described in detail, appended attached drawing purposes of discussion only for reference, is not used to limit the embodiment of the present invention.
Fig. 1 is a kind of method flow schematic diagram of complete polarization image polarization decomposing of the embodiment of the present invention, as shown in Figure 1, this The method of the complete polarization image polarization decomposing of inventive embodiments the following steps are included:
Step 101, the hiding characteristic value of n independence of each pixel in p object pixel in complete polarization image is obtained.
Specifically, the pretreatment operations such as exquisite residue (Lee) filtering are carried out to complete polarization image data, it is relevant obtains removal Complete polarization image data after spot noise；Independence is calculated using the complete polarization image data after the removal coherent speckle noise Hide characteristic value.
In some embodiments, the n independence for obtaining each pixel in p object pixel in complete polarization image is hidden Characteristic value, comprising:
The ensemble average value<[T]>of the coherence matrix of the complete polarization image is obtained,
Wherein, k is Pauli (Pauli) base,S is the complete polarization The collision matrix of image,Wherein, SHHFor Vertical Launch, the back scattering of vertical reception, SHVIt is vertical Transmitting, horizontal received back scattering, SVHFor the back scattering of horizontal emission, vertical reception, SVVIt is connect for horizontal emission, level The back scattering of receipts；
It obtains after carrying out orientation angles compensation to<[T]>through the compensated coherence matrix of the angle of orientation<[T ']>,
Wherein,θ indicates orientation angles, [R (θ)] it is tenth of the twelve Earthly Branches spin matrix；
Described n independent hiding characteristic value, comprising:
Wherein, A indicates that the amplitude of rotation coherence matrix, B indicate the amplitude center of rotation coherence matrix, θ0Indicate initial phase Place value, ω indicate angular frequency；Re [] indicates that real part, Im [] indicate imaginary part, and Angle { } is for obtaining element in multiple axis Phase value；A_T′ijIndicate T 'ijAmplitude, B_T 'ijIndicate T 'ijAmplitude center, θ0_T′ijIndicate T 'ijInitial phase value.
Wherein, it includes abundant information relevant to polarization matrix rotation effect that the independence, which hides feature,.In essence, The feature of these parameters is directly related with the scattering phenomenon in rotation, and can potentially reflect the characteristic of scatterer.To spy The sign biggish construction zone of fuzzy region, the especially angle of orientation has preferable characterization to act on.
Step 102, principal component analysis is carried out to n independent hiding characteristic value of each pixel in the p object pixel, Obtain n weighted value corresponding with the hiding characteristic value of n independence.
In embodiments of the present invention, principal component analysis is a kind of statistical method.By orthogonal transformation by one group there may be The variable of correlation is converted to one group of linearly incoherent variable, this group of variable after conversion is principal component.Pass through principal component point Analysis, weighted value of the available each group variable in principal component.
Step 103, m weighted value in the n weighted value is determined, wherein the m weighted value is all larger than the n N-m weighted value among a weight in addition to the m weighted value, m < n.
By selecting the maximum m weighted value of weighted value in the n weighted value, the spy to scatterer can be effectively excluded Property the lower independence of correlation hide characteristic value.
Step 104, the corresponding m independent hiding characteristic value of the m weighted value is obtained.
By selecting m higher to the characteristic correlation of scatterer independent hiding characteristic value, reduce to scattering bulk properties The calculation amount of judgement.
Step 105, hiding characteristic value independent for m described in the p object pixel is all larger than the first preset threshold Q object pixel, the volume scattering power of the q object pixel, q < p are determined using the first volume scattering model.
In some embodiments, the first volume scattering modelExpression formula are as follows:
Wherein, ψ indicates spiral angle.
Specifically, the first volume scattering modelIncluding through orientation angle compensation and phase angle is compensated inclines Oblique dihedral angle scattering model
In some embodiments, m independent hiding characteristic value described in the p object pixel is all larger than the first default threshold Q object pixel of value is artificial atural object region, is usedDetermine the volume scattering power of the q object pixel, It can be effectively reduced the problem of volume scattering energy to construction zone crosses estimation.
Fig. 2 is a kind of method flow schematic diagram of complete polarization image polarization decomposing of the embodiment of the present invention, as shown in Fig. 2, this The method of the complete polarization image polarization decomposing of inventive embodiments the following steps are included:
Step 201, the hiding characteristic value of n independence of each pixel in p object pixel in complete polarization image is obtained.
In some embodiments, the n independence for obtaining each pixel in p object pixel in complete polarization image is hidden Characteristic value, comprising:
The ensemble average value<[T]>of the coherence matrix of the complete polarization image is obtained,
Wherein, k is Pauli base,S is the complete polarization image Collision matrix,Wherein, SHHFor Vertical Launch, the back scattering of vertical reception, SHVFor Vertical Launch, water Put down received back scattering, SVHFor the back scattering of horizontal emission, vertical reception, SVVIt is received backward for horizontal emission, level Scattering；
It obtains after carrying out orientation angles compensation to<[T]>through the compensated coherence matrix of the angle of orientation<[T ']>,
Wherein,θ indicates orientation angles, [R (θ)] is tenth of the twelve Earthly Branches spin matrix；
Described n independent hiding characteristic value, comprising:
Wherein, A indicates that the amplitude of rotation coherence matrix, B indicate the amplitude center of rotation coherence matrix, θ0Indicate initial phase Place value, ω indicate angular frequency；Re [] indicates that real part, Im [] indicate imaginary part, and Angle { } is for obtaining element in multiple axis Phase value；A_T′ijIndicate T 'ijAmplitude, B_T 'ijIndicate T 'ijAmplitude center, θ0_T′ijIndicate T 'ijInitial phase value.
Step 202, principal component analysis is carried out to n independent hiding characteristic value of each pixel in the p object pixel, Obtain n weighted value corresponding with the hiding characteristic value of n independence.
Principal component analysis is a kind of statistical method.By orthogonal transformation, there may be the conversions of the variable of correlation by one group For one group of linear incoherent variable, this group of variable after conversion is principal component.By principal component analysis, available each group becomes Measure the weighted value in principal component.
Step 203, m weighted value in the n weighted value is determined, wherein the m weighted value is all larger than the n N-m weighted value among a weight in addition to the m weighted value, m < n.
By selecting the maximum m weighted value of weighted value in the n weighted value, the spy to scatterer can be effectively excluded Property the lower independence of correlation hide characteristic value.
Step 204, the corresponding m independent hiding characteristic value of the m weighted value is obtained.
By selecting m higher to the characteristic correlation of scatterer independent hiding characteristic value, reduce to scattering bulk properties The calculation amount of judgement.
Step 205, hiding characteristic value independent for m described in the p object pixel is all larger than the first preset threshold Q object pixel, the volume scattering power of the q object pixel, q < p are determined using the first volume scattering model.
In some embodiments, the first volume scattering modelExpression formula are as follows:
Wherein, ψ indicates spiral angle.
Specifically, the first volume scattering modelIncluding through orientation angle compensation and phase angle is compensated inclines Oblique dihedral angle scattering model
In some embodiments, m independent hiding characteristic value described in the p object pixel is all larger than the first default threshold Q object pixel of value is artificial atural object region, is usedDetermine the volume scattering power of the q object pixel, It can be effectively reduced the problem of volume scattering energy to construction zone crosses estimation.
Step 206, it for p-q object pixel in the p object pixel in addition to the q object pixel, adopts The volume scattering power of the p-q object pixel is determined with the second volume scattering model.
In some embodiments, the second volume scattering modelExpression formula are as follows:
Wherein, τ be calculating process amount, τ=< | SHH|2>/<|SVV|2>, ψ indicates spiral angle.
Specifically, the second volume scattering modelIncluding compensated general through orientation angle compensation and phase angle Volume scattering model
In some embodiments, p-q object pixel in the p object pixel in addition to the q object pixel For natural feature on a map region, useIt determines the volume scattering power of the p-q object pixel, can correctly characterize naturally The volume scattering energy of object area.
In some embodiments, it is desirable to carry out spiral angle compensation through the compensated coherence matrix of the angle of orientation<[T ']>to described After obtain through the angle of orientation and the compensated coherence matrix of helical angle<[T "]>,
Wherein,[R (ψ)] is the tenth of the twelve Earthly Branches Transformation matrix.
Step 207, it for the q object pixel, is scattering into the coherence matrix of the q object pixel subtracting line Divide, after spiral scattering ingredient and volume scattering ingredient, according to the first decision threshold C1Determine leading scattering mechanism；
C1=T '11-T′22+fc/2-2Re{γ}|fw|-2m1fv, wherein m1For calculating process amount, m1=(cos (4 θ)- 15)/60, fcIt is right for spiral scattering component | SVV|2Contribution, fvIt is right for volume scattering component | SVV|2Contribution, fwFor line scattering point Amount pair | SVV|2Contribution, γ be calculating process amount, γ=SHH/SVV；
Specifically, work as C1When > 0, surface scattering ingredient is leading scattering mechanism；Work as C1When≤0, dihedral angle scattering is leading Scattering mechanism.
In some embodiments, the q object pixel is construction zone, according to the first decision threshold C1Determine Leading scattering mechanism, it is possible to reduce the crossing for volume scattering of construction zone is estimated.
Step 208, it for the p-q object pixel, dissipates in the coherence matrix of the p-q object pixel subtracting line After penetrating ingredient, spiral scattering ingredient and volume scattering ingredient, according to the second decision threshold C2Determine leading scattering mechanism；
C2=T '11-T′22+fc/2-2Re{γ}|fw|-2m2fv, wherein m2For calculating process amount,
Specifically, work as C2When > 0, surface scattering ingredient is leading scattering mechanism；Work as C2When≤0, dihedral angle scattering is leading Scattering mechanism.
Fig. 3 is a kind of flow diagram of multielement polarization image decomposition method of the embodiment of the present invention, as shown in figure 3, this Method of the inventive embodiments based on the above complete polarization image polarization decomposing, a kind of multielement polarization picture breakdown of exemplary offer Method the following steps are included:
Step 301: the complete polarization image data obtained using polarimetric synthetic aperture radar is calculated independent hiding feature and taken Value.
Specifically, the pretreatment operations such as exquisiteness Lee filtering are carried out to polarization image data, after obtaining removal coherent speckle noise Image data.
The collision matrix of complete polarization image data is
Pauli base are as follows:
According to the relationship of Pauli base and coherence matrix, obtain the ensemble average value of the coherence matrix of the polarization image < [T] > are as follows:
Through the compensated coherence matrix of orientation angles<[T ']>are as follows:
Wherein, [R (θ)] is tenth of the twelve Earthly Branches spin matrix
θ indicates orientation angles,
It is existing characteristic parameter that the independence, which hides feature, specifically expression formula are as follows:
Wherein, A indicates that the amplitude of rotation coherence matrix, B indicate the amplitude center of rotation coherence matrix, θ0Indicate initial phase Place value, ω indicate angular frequency；Re [] indicates that real part, Im [] indicate imaginary part, and Angle { } is for obtaining element in multiple axis Phase value；A_T′ijIndicate T 'ijAmplitude, B_T 'ijIndicate T 'ijAmplitude center, θ0_T′ijIndicate T 'ijInitial phase value.
It includes abundant information relevant to polarization matrix rotation effect that the independence, which hides feature,.These are joined in essence Several features is directly related with the scattering phenomenon in rotational domain, and may potentially reflect the characteristic of scatterer.To character modules Paste region especially has preferable characterization to act on the radar line of sight direction biggish construction zone of angle.
Step 302: using the hiding feature carrying out principal component analysis, choose to have man-made land object area and preferably characterize Three hiding characteristic values carry out artificial atural object area with histogram thresholding method and extract, and obtain two points of extraction results.
Specifically, feature is hidden to all independences and carries out principal component analysis, each independence is calculated and hides spy The shared weight of sign, and according to gained weight select three optimal hiding characteristic values as Urban Areas building extract according to According to.
The Urban Areas building extraction is realized by histogram thresholding method, is exactly to select specifically Know that the partial pixel point of atural object as training sample, the hiding feature value of training sample point is counted in the form of histogram, is led to It crosses and chooses threshold value appropriate, complete the extraction of man-made land object area.
Described two points extract result and refer to: when value is greater than the threshold value chosen, the logical value of pixel is set to 1, the picture Vegetarian refreshments belongs to man-made land object area；When value is less than the threshold value chosen, the logical value of pixel is set to 0, and pixel belongs to nature Ground object area.After the logical value of all pixels determines, two points of extraction results are obtained.
Step 303: selecting corresponding improved volume scattering model using two points of extraction results.
Specifically, two points of extractions result can be used for two classification of the progress of the type described in region: when extraction end value When being 1, pixel belongs to man-made land object area, volume scattering model selection the first volume scattering model；When extracting end value is 0, Pixel belongs to natural feature on a map region, volume scattering model selection the second volume scattering model.
In the present invention, the coherence matrix decomposed for multielement needs to carry out angle of orientation compensating operation and spiral shell before decomposition Swing angle compensating operation.
Through orientation angles compensation and the compensated coherence matrix of spiral angle are as follows:
Wherein,
In above formula, ψ indicates that spiral angle, [R (ψ)] are unitary transformation matrix.
After the completion of the phase angle compensation and spiral angle compensating operation, element T23Real part and T13Imaginary part can become At 0, unknown element becomes seven from original nine, reduces the number for needing the unknown quantity solved, meanwhile, it is preferably minimum T is changed33The value of element.
The heretofore described improved volume scattering model for being characterized to different types of ground objects includes through being orientated Angle compensation and the compensated first volume scattering model of phase angle and the second volume scattering model.
The initializer of the volume scattering model is as follows:
Wherein, τ=< | SHH|2>/<|SVV|2>, θ indicates angle of orientation angle value above-mentioned,WithRespectively Indicate the first volume scattering model and the second volume scattering model.
When pixel belongs to man-made land object area, volume scattering model selection is compensated through orientation angles compensation and phase angle The first volume scattering model afterwards, the expression of model are as follows:
When pixel belongs to natural feature on a map region, volume scattering model selection is compensated through orientation angles compensation and phase angle The second volume scattering model afterwards, the expression of model are as follows:
Wherein, θ and ψ respectively indicate orientation angles and spiral angle.
Step 304: obtaining multielement decomposition result using the selected scattering model and image data.
Specifically, volume scattering power is calculated using the selected scattering model, and is calculated and is remained with multielement decomposition method Under scattering component performance number.
The multielement decomposition method is compared with existing multielement decomposition method, the difference is that for determining two Face angle/surface scattering accounts for the value of the threshold value of main body.Specifically, coherence matrix is scattering into subtracting line scattering ingredient, spiral Point and volume scattering ingredient after, for determining that dihedral angle/surface scattering accounts for the threshold value value of main body are as follows:
C=T '11-T′22+fc/2-2Re{γ}|fw|-2mfv
Wherein, fc, fvAnd fwIt is right to respectively represent spiral scattering component, volume scattering component and line scattering component | SVV|2Tribute It offers, γ=SHH/SVV.If selecting the first volume scattering model, m=(cos (4 θ) -15)/60, conversely, if selecting the second volume scattering mould Type, then
Fig. 4 is a kind of apparatus structure schematic diagram of complete polarization image polarization decomposing of the embodiment of the present invention, as shown in figure 4, this The device of the complete polarization image polarization decomposing of inventive embodiments includes: image processing unit 401, main control unit 402 and polarization point Solve unit 403, in which:
Image processing unit 401, the n independence for obtaining each pixel in p object pixel in complete polarization image are hidden Hide characteristic value.
In some embodiments, described image processing unit 401, is specifically used for:
The ensemble average value<[T]>of the coherence matrix of the complete polarization image is obtained,
Wherein, k is Pauli base,S is the complete polarization image Collision matrix,Wherein, SHHFor Vertical Launch, the back scattering of vertical reception, SHVFor Vertical Launch, water Put down received back scattering, SVHFor the back scattering of horizontal emission, vertical reception, SVVIt is received backward for horizontal emission, level Scattering；
It obtains after carrying out orientation angles compensation to<[T]>through the compensated coherence matrix of the angle of orientation<[T ']>,
Wherein,θ indicates orientation angles, [R (θ)] it is tenth of the twelve Earthly Branches spin matrix；
Described n independent hiding characteristic value, comprising:
Wherein, A indicates that the amplitude of rotation coherence matrix, B indicate the amplitude center of rotation coherence matrix, θ0Indicate initial phase Place value, ω indicate angular frequency；Re [] indicates that real part, Im [] indicate imaginary part, and Angle { } is for obtaining element in multiple axis Phase value；A_T′ijIndicate T 'ijAmplitude, B_T 'ijIndicate T 'ijAmplitude center, θ0_T′ijIndicate T 'ijInitial phase value.
Main control unit 402 is led for the independent hiding characteristic value of n to each pixel in the p object pixel Constituent analysis obtains n weighted value corresponding with the hiding characteristic value of n independence；It determines in the n weighted value M weighted value, wherein the m weighted value is all larger than among the n weight n-m power in addition to the m weighted value Weight values, m < n；Obtain the corresponding m independent hiding characteristic value of the m weighted value.
Principal component analysis is a kind of statistical method.By orthogonal transformation, there may be the conversions of the variable of correlation by one group For one group of linear incoherent variable, this group of variable after conversion is principal component.By principal component analysis, available each group becomes Measure the weighted value in principal component.
By selecting the maximum m weighted value of weighted value in the n weighted value, the spy to scatterer can be effectively excluded Property the lower independence of correlation hide characteristic value.
By selecting m higher to the characteristic correlation of scatterer independent hiding characteristic value, reduce to scattering bulk properties The calculation amount of judgement.
Polarization decomposing unit 403 is all larger than for hiding characteristic value independent for m described in the p object pixel Q object pixel of the first preset threshold determines the volume scattering power of the q object pixel, q using the first volume scattering model <p。
In some embodiments, the first volume scattering modelExpression formula are as follows:
Wherein, ψ indicates spiral angle.
Specifically, the first volume scattering modelIncluding through orientation angle compensation and phase angle is compensated inclines Oblique dihedral angle scattering model
In some embodiments, m independent hiding characteristic value described in the p object pixel is all larger than the first default threshold Q object pixel of value is construction zone, is usedDetermine the volume scattering power of the q object pixel, energy The problem of volume scattering energy to construction zone crosses estimation is enough effectively reduced.
In some embodiments, the polarization decomposing unit 403 is also used to for removing the q in the p object pixel P-q object pixel other than a object pixel determines the volume scattering of the p-q object pixel using the second volume scattering model Power.
In some embodiments, the second volume scattering modelExpression formula are as follows:
Wherein, τ be calculating process amount, τ=< | SHH|2>/<|SVV|2>, ψ indicates spiral angle.
Specifically, the second volume scattering modelIncluding compensated general through orientation angle compensation and phase angle Volume scattering model
In some embodiments, p-q object pixel in the p object pixel in addition to the q object pixel For natural feature on a map region, useIt determines the volume scattering power of the p-q object pixel, can correctly characterize naturally The volume scattering energy of object area.
In some embodiments, described image processing unit 401, is also used to: to described compensated relevant through the angle of orientation Matrix<[T ']>obtains after carrying out spiral angle compensation through the angle of orientation and the compensated coherence matrix of helical angle<[T "]>,
Wherein,[R (ψ)] is the tenth of the twelve Earthly Branches Transformation matrix.
In some embodiments, the polarization decomposing unit 403, is also used to: for the q object pixel, in the q The coherence matrix of a object pixel determines after subtracting line scattering ingredient, spiral scattering ingredient and volume scattering ingredient according to first Threshold value C1Determine leading scattering mechanism；
C1=T '11-T′22+fc/2-2Re{γ}|fw|-2m1fv, wherein m1For calculating process amount, m1=(cos (4 θ)- 15)/60, fcIt is right for spiral scattering component | SVV|2Contribution, fvIt is right for volume scattering component | SVV|2Contribution, fwFor line scattering point Amount pair | SVV|2Contribution, γ be calculating process amount, γ=SHH/SVV；
Specifically, work as C1When > 0, surface scattering ingredient is leading scattering mechanism；Work as C1When≤0, dihedral angle scattering is leading Scattering mechanism.
In some embodiments, the q object pixel is construction zone, according to the first decision threshold C1Determine Leading scattering mechanism, it is possible to reduce estimation is crossed to the volume scattering energy of construction zone.
In some embodiments, the polarization decomposing unit 403, is also used to: for the p-q object pixel, in institute The coherence matrix of p-q object pixel is stated after subtracting line scattering ingredient, spiral scattering ingredient and volume scattering ingredient, according to second Decision threshold determines leading scattering mechanism；
C2=T '11-T′22+fc/2-2Re{γ}|fw|-2m2fv, wherein m2For calculating process amount,
Specifically, work as C2When > 0, surface scattering ingredient is leading scattering mechanism；Work as C2When≤0, dihedral angle scattering is leading Scattering mechanism.
In some embodiments, the p-q object pixel is natural feature on a map region, according to the second decision threshold C2 Determine leading scattering mechanism, can the scattering mechanism to natural feature on a map region correctly characterized.
Fig. 5 is a kind of multielement polarization image decomposer structural schematic diagram of the embodiment of the present invention, as shown in figure 5, this hair The multielement polarization image decomposer of bright embodiment includes: parameter calculating module 501, optimization extraction module 502, selecting module 503 and decomposing module 504, in which:
Parameter calculating module 501 calculates only for carrying out the pretreatment operations such as exquisiteness Lee filtering to polarization image data Vertical hiding feature value, is sent to optimization extraction module for the value that all independences being calculated hide feature.
Optimize extraction module 502, the independence for being sent using parameter calculating module 501 is hidden feature and carries out data Dimensionality reduction, the feature that optimization is extracted for man-made land object area and then obtains the extraction of man-made land object area as a result, and will obtain Two points of extraction results of man-made land object area be sent to selecting module.
Selecting module 503, two points of extraction results selection pair of the artificial atural object for being sent using optimization extraction module 502 Selected volume scattering model is sent to decomposing module 504 by the improved volume scattering model answered.
Decomposing module 504, the selected volume scattering model and image data for being sent using selecting module 503 obtain more Element decomposition result.
The parameter calculating module 501 is specifically used for carrying out polarization image data the pretreatments behaviour such as exquisiteness Lee filtering Make, the image data after obtaining removal coherent speckle noise；It is calculated using the image data after the removal coherent speckle noise Independent hiding characteristic value.
It is the characteristic parameter proposed, specific expression formula that the independence, which hides feature, are as follows:
Wherein, it includes abundant information relevant to polarization matrix rotation effect that the independence, which hides feature,.In essence The feature of these parameters is directly related with the scattering phenomenon in rotational domain, and may potentially reflect the characteristic of scatterer.It is right The feature Fuzzy region biggish construction zone of the especially angle of orientation has preferable characterization to act on.
The optimization extraction module 502 carries out principal component analysis, meter specifically for hiding feature to all independences Calculation obtains each independence and hides weight shared by feature, and selects three optimal hiding characteristic values as cities and towns according to gained weight The foundation that local building is extracted.
The Urban Areas building extraction is realized by histogram thresholding method, is exactly to select specifically Know that the partial pixel point of atural object is chosen suitable as training sample by the hiding feature value of statistics with histogram training sample point When threshold value, complete man-made land object area extraction.
Two points of extraction results refer to, if the value of the hiding feature of pixel is greater than the threshold value chosen, pixel Point logical value is set to 1, which belongs to man-made land object area；When value is less than the threshold value chosen, the logical value of pixel is set It is 0, pixel belongs to natural feature on a map region.After all pixels point logical value determines, two points of extraction results are obtained.
The selecting module 503 is specifically used for the type described in region and carries out two classification: when extracting end value is 1, as Vegetarian refreshments belongs to man-made land object area, first volume scattering model of the volume scattering model selection after angle compensation；When extraction end value When being 0, pixel belongs to natural feature on a map region, second volume scattering model of the volume scattering model selection after angle compensation.
In 503 execution of selecting module movement, the coherence matrix decomposed for multielement is orientated before decomposition Angle compensation operation and helical angle compensating operation.
After the completion of the angle compensation operation, element T23Real part and T13Imaginary part will become 0, unknown element is from original Nine become seven, reduce the number of unknown quantity for needing to solve, meanwhile, preferably minimize T33The value of element.
The heretofore described improved volume scattering model for being characterized to different types of ground objects includes through being orientated Angle compensation and the compensated first volume scattering model of phase angle and the second volume scattering model.
The initializer of the volume scattering model is as follows:
Orientation angles compensation process is written as:<[T ']>=[R (θ)]<[T]>[R (θ)]
Spiral angle compensation process is written as:<[T "]>=[R (ψ)]<[T ']>[R (ψ)]
Wherein,
When pixel belongs to natural feature on a map region, volume scattering model selection is compensated through orientation angles compensation and phase angle The second volume scattering model afterwards, the expression of model are as follows:
When pixel belongs to man-made land object area, volume scattering model selection is compensated through orientation angles compensation and phase angle The first volume scattering model afterwards, the expression of model are as follows:
Wherein, τ=< | SHH|2>/<|SVV|2>, θ and ψ indicate that rotation angle, specific expression formula have been presented above.
The decomposing module 504, specifically for being calculated using the selected scattering model, and with multielement decomposition method The performance number of remaining scattering component.
The multielement decomposition method is compared with existing multielement decomposition method, the difference is that for determining two Face angle/surface scattering accounts for the value of the threshold value of main body.Specifically, coherence matrix is scattering into subtracting line scattering ingredient, spiral Point and volume scattering ingredient after, for determining that dihedral angle/surface scattering accounts for the threshold value value of main body are as follows:
C=T '11-T′22+fc/2-2Re{γ}|fw|-2mfv
Wherein, if pixel to be asked belongs to natural feature on a map region, the improved second volume scattering model of volume scattering model selection,If pixel to be asked belongs to man-made land object area, volume scattering model selection improved first Volume scattering model, m=(cos (4 θ) -15)/60.Decision threshold used in method can be adapted for different types of ground objects.
In order to realize the embodiment of the present invention complete polarization image polarization decomposing method, the embodiment of the invention provides one kind The electronic equipment of structural schematic diagram as shown in Figure 6, as shown in fig. 6, the electronic equipment 610 of the embodiment of the present invention includes: processor 61 and the memory 62 for storing the computer program that can be run on a processor, wherein
The processor 61 is for executing present invention any complete polarization figure when running the computer program As polarization decomposing method the step of.
Certainly, when practical application, as shown in fig. 6, the electronic equipment can also include at least one communication interface 63.Electronics Various components in equipment are coupled by bus system 64.It is understood that bus system 64 for realizing these components it Between connection communication.Bus system 64 further includes that power bus, control bus and status signal are total in addition to including data/address bus Line.But for the sake of clear explanation, various buses are all designated as bus system 64 in Fig. 6.
Wherein, communication interface 63, for being interacted with other equipment.
Specifically, the application clothes that the processor 61 can be applied by communication interface 63 to the correspondence called side Business device sends operating result inquiry request, obtains the operating result that the called side that the application server is sent applies.
It will be understood by those skilled in the art that memory 62 can be volatile memory or nonvolatile memory, It may include both volatile and non-volatile memories.Wherein, nonvolatile memory can be read-only memory (ROM, Read Only Memory), programmable read only memory (PROM, Programmable Read-Only Memory), erasable compile Journey read-only memory (EPROM, Erasable Programmable Read-Only Memory), electrically erasable are read-only Memory (EEPROM, Electrically Erasable Programmable Read-Only Memory), magnetic random are deposited Access to memory (FRAM, ferromagnetic random access memory), flash memory (Flash Memory), magnetic Memory surface, CD or CD-ROM (CD-ROM, Compact Disc Read-Only Memory)；Magnetic surface storage It can be magnetic disk storage or magnetic tape storage.Volatile memory can be random access memory (RAM, Random Access Memory), it is used as External Cache.By exemplary but be not restricted explanation, the RAM of many forms can With, such as static random access memory (SRAM, Static Random Access Memory), synchronous static random-access Memory (SSRAM, Synchronous Static Random Access Memory), dynamic random access memory (DRAM, Dynamic Random Access Memory), Synchronous Dynamic Random Access Memory (SDRAM, Synchronous Dynamic Random Access Memory), double data speed synchronous dynamic RAM (DDRSDRAM, Double Data Rate Synchronous Dynamic Random Access Memory), enhanced synchronous dynamic random Access memory (ESDRAM, Enhanced Synchronous Dynamic Random Access Memory), synchronized links Dynamic random access memory (SLDRAM, SyncLink Dynamic Random Access Memory), direct rambus Random access memory (DRRAM, Direct Rambus Random Access Memory).Description of the embodiment of the present invention is deposited Reservoir 62 is intended to include but is not limited to the memory of these and any other suitable type.
In the embodiment of the present invention, a kind of computer readable storage medium is additionally provided, is mentioned for storing in above-described embodiment The calculation procedure of confession, to complete step described in the method for aforementioned complete polarization image polarization decomposing.Computer readable storage medium can To be the storage such as FRAM, ROM, PROM, EPROM, EEPROM, Flash Memory, magnetic surface storage, CD or CD-ROM Device；Be also possible to include one of above-mentioned memory or any combination various equipment, as mobile phone, computer, intelligent appliance, Server etc..
Disclosed method in several embodiments of the method provided herein, in the absence of conflict can be any group It closes, obtains new embodiment of the method.
Disclosed feature in several method or apparatus embodiments provided by the present invention, in the absence of conflict can be with Any combination obtains new embodiment of the method or Installation practice.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any Those familiar with the art in the technical scope disclosed by the present invention, can easily think of the change or the replacement, and should all contain Lid is within protection scope of the present invention.Therefore, protection scope of the present invention should be based on the protection scope of the described claims.

Claims (16)

1. a kind of method of complete polarization image polarization decomposing, which is characterized in that the described method includes:
N of each pixel independent hiding characteristic value in p object pixel in acquisition complete polarization image；
Principal component analysis is carried out to n independent hiding characteristic value of each pixel in the p object pixel, is obtained and the n A independence hides the corresponding n weighted value of characteristic value；
Determine m weighted value in the n weighted value, wherein the m weighted value is all larger than among the n weight and removes N-m weighted value other than the m weighted value, m < n；
Obtain the corresponding m independent hiding characteristic value of the m weighted value；
Hiding characteristic value independent for m described in the p object pixel is all larger than q target picture of the first preset threshold Element determines the volume scattering power of the q object pixel, q < p using the first volume scattering model.
2. the method according to claim 1, wherein the method also includes:
For p-q object pixel in the p object pixel in addition to the q object pixel, using the second volume scattering Model determines the volume scattering power of the p-q object pixel.
3. according to the method described in claim 2, it is characterized in that, every in p object pixel in the acquisition complete polarization image The independent hiding characteristic value of n of a pixel, comprising:
The ensemble average value<[T]>of the coherence matrix of the complete polarization image is obtained,
Wherein, k is Pauli Pauli base,S is the complete polarization image Collision matrix,Wherein, SHHFor Vertical Launch, the back scattering of vertical reception, SHVFor Vertical Launch, water Put down received back scattering, SVHFor the back scattering of horizontal emission, vertical reception, SVVIt is received backward for horizontal emission, level Scattering；
It obtains after carrying out orientation angles compensation to<[T]>through the compensated coherence matrix of the angle of orientation<[T ']>,
Wherein,θ indicates orientation angles, [R (θ)] For tenth of the twelve Earthly Branches spin matrix；
Described n independent hiding characteristic value, comprising:
Wherein, A indicates that the amplitude of rotation coherence matrix, B indicate the amplitude center of rotation coherence matrix, θ0Indicate initial phase value, ω indicates angular frequency；Re [] indicates that real part, Im [] indicate imaginary part, and Angle { } is used to obtain the phase of element in multiple axis Value；A_T′ijIndicate T 'ijAmplitude, B_T 'ijIndicate T 'ijAmplitude center, θ0_T′ijIndicate T 'ijInitial phase value.
4. according to the method described in claim 3, it is characterized in that, the first volume scattering modelExpression formula Are as follows:
Wherein, ψ indicates spiral angle.
5. according to the method described in claim 4, it is characterized in that, the second volume scattering modelExpression formula are as follows:
Wherein, τ be calculating process amount, τ=< | SHH|2>/<|SVV|2>。
6. according to the method described in claim 5, it is characterized in that, the method also includes:
It is obtained after the compensated coherence matrix of the angle of orientation<[T ']>carries out spiral angle compensation through the angle of orientation and helical angle to described Compensated coherence matrix<[T "]>,
Wherein,[R (ψ)] is unitary transformation Matrix.
7. according to the method described in claim 6, it is characterized in that, the method also includes:
For the q object pixel, line scattering ingredient, spiral scattering are being subtracted in the coherence matrix of the q object pixel After ingredient and volume scattering ingredient, according to the first decision threshold C1Determine leading scattering mechanism；
C1=T '11-T′22+fc/2-2Re{γ}|fw|-2m1fv, wherein m1For calculating process amount, m1=(cos (4 θ) -15)/60, fcIt is right for spiral scattering component | SVV|2Contribution, fvIt is right for volume scattering component | SVV|2Contribution, fwIt is right for line scattering component | SVV |2Contribution, γ be calculating process amount, γ=SHH/SVV；
Work as C1When > 0, surface scattering ingredient is leading scattering mechanism；Work as C1When≤0, dihedral angle scattering is leading scattering mechanism.
8. the method according to the description of claim 7 is characterized in that the method also includes:
For the p-q object pixel, line scattering ingredient, spiral are being subtracted in the coherence matrix of the p-q object pixel After scattering ingredient and volume scattering ingredient, according to the second decision threshold C2Determine leading scattering mechanism；
C2=T '11-T′22+fc/2-2Re{γ}|fw|-2m2fv, wherein m2For calculating process amount,
Work as C2When > 0, surface scattering ingredient is leading scattering mechanism；Work as C2When≤0, dihedral angle scattering is leading scattering mechanism.
9. a kind of device of complete polarization image polarization decomposing, which is characterized in that described device includes:
Image processing unit, for obtaining the hiding feature of n independence of each pixel in p object pixel in complete polarization image Value；
Main control unit carries out principal component point for the independent hiding characteristic value of n to each pixel in the p object pixel Analysis obtains n weighted value corresponding with the hiding characteristic value of n independence；Determine m power in the n weighted value Weight values, wherein the m weighted value is all larger than n-m weighted value among the n weight in addition to the m weighted value, m<n；Obtain the corresponding m independent hiding characteristic value of the m weighted value；
It is default to be all larger than first for hiding characteristic value independent for m described in the p object pixel for polarization decomposing unit Q object pixel of threshold value determines the volume scattering power of the q object pixel, q < p using the first volume scattering model.
10. device according to claim 9, which is characterized in that the polarization decomposing unit is also used to: for the p P-q object pixel in object pixel in addition to the q object pixel determines the p-q using the second volume scattering model The volume scattering power of a object pixel.
11. device according to claim 10, which is characterized in that described image processing unit is specifically used for: described in acquisition The ensemble average value<[T]>of the coherence matrix of complete polarization image,
Wherein, k is Pauli Pauli base,S is the complete polarization image Collision matrix,Wherein, SHHFor Vertical Launch, the back scattering of vertical reception, SHVFor Vertical Launch, water Put down received back scattering, SVHFor the back scattering of horizontal emission, vertical reception, SVVIt is received backward for horizontal emission, level Scattering；
It obtains after carrying out orientation angles compensation to<[T]>through the compensated coherence matrix of the angle of orientation<[T ']>,
Wherein,θ indicates orientation angles, [R (θ)] For tenth of the twelve Earthly Branches spin matrix；
Described n independent hiding characteristic value, comprising:
Wherein, A indicates that the amplitude of rotation coherence matrix, B indicate the amplitude center of rotation coherence matrix, θ0Indicate initial phase value, ω indicates angular frequency；Re [] indicates that real part, Im [] indicate imaginary part, and Angle { } is used to obtain the phase of element in multiple axis Value；A_T′ijIndicate T 'ijAmplitude, B_T 'ijIndicate T 'ijAmplitude center, θ0_T′ijIndicate T 'ijInitial phase value.
12. device according to claim 11, which is characterized in that described image processing unit is also used to: to described through taking It obtains after carrying out spiral angle compensation to the coherence matrix<[T ']>after angle compensation through the compensated relevant square of the angle of orientation and helical angle Battle array < [T "] >,
Wherein,[R (ψ)] is unitary transformation Matrix.
13. device according to claim 12, which is characterized in that the polarization decomposing unit is also used to: for the q A object pixel, the q object pixel coherence matrix subtract line scattering ingredient, spiral scattering ingredient and volume scattering at After point, according to the first decision threshold C1Determine leading scattering mechanism；
C1=T '11-T′22+fc/2-2Re{γ}|fw|-2m1fv, wherein m1For calculating process amount, m1=(cos (4 θ) -15)/60, fcIt is right for spiral scattering component | SVV|2Contribution, fvIt is right for volume scattering component | SVV|2Contribution, fwIt is right for line scattering component | SVV |2Contribution, γ be calculating process amount, γ=SHH/SVV；
Work as C1When > 0, surface scattering ingredient is leading scattering mechanism；Work as C1When≤0, dihedral angle scattering is leading scattering mechanism.
14. device according to claim 13, which is characterized in that the polarization decomposing unit is also used to: for the p- Q object pixel, the p-q object pixel coherence matrix subtract line scattering ingredient, spiral scattering ingredient and body dissipate After penetrating ingredient, leading scattering mechanism is determined according to the second decision threshold；
C2=T '11-T′22+fc/2-2Re{γ}|fw|-2m2fv, wherein m2For calculating process amount,
Work as C2When > 0, surface scattering ingredient is leading scattering mechanism；Work as C2When≤0, dihedral angle scattering is leading scattering mechanism.
15. a kind of electronic equipment, comprising: transceiver, memory, processor and be stored on the memory and by the processing The computer program that device executes；
The processor is connect with the transceiver and the memory respectively, executes the computer program for passing through, real The method that any one of existing claim 1 to 8 provides.
16. a kind of computer storage medium, the computer storage medium is stored with computer program；The computer program quilt After execution, the method that any one of claim 1 to 8 provides can be realized.
CN201910637291.8A 2019-07-15 2019-07-15 Polarization decomposition method and device for full polarization image, electronic equipment and storage medium Active CN110516698B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910637291.8A CN110516698B (en) 2019-07-15 2019-07-15 Polarization decomposition method and device for full polarization image, electronic equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910637291.8A CN110516698B (en) 2019-07-15 2019-07-15 Polarization decomposition method and device for full polarization image, electronic equipment and storage medium

Publications (2)

Publication Number Publication Date
CN110516698A true CN110516698A (en) 2019-11-29
CN110516698B CN110516698B (en) 2022-03-11

Family

ID=68623631

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910637291.8A Active CN110516698B (en) 2019-07-15 2019-07-15 Polarization decomposition method and device for full polarization image, electronic equipment and storage medium

Country Status (1)

Country Link
CN (1) CN110516698B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116797845A (en) * 2023-07-05 2023-09-22 中国科学院空天信息创新研究院 Unsupervised reduced polarization classification method based on scattering mechanism
CN117390354A (en) * 2023-12-12 2024-01-12 江西师范大学 GoaC compensation-based polarization target decomposition method

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102999762A (en) * 2012-10-25 2013-03-27 西安电子科技大学 Method for classifying polarimetric SAR (synthetic aperture radar) images on the basis of Freeman decomposition and spectral clustering
CN104463805A (en) * 2014-12-16 2015-03-25 西安电子科技大学 Polarimetric SAR coherent speckle noise suppression method based on homogeneity saliency and direction selection
CN104459656A (en) * 2014-12-31 2015-03-25 中国科学院空间科学与应用研究中心 Target orientation angle compensation method for fully polarimetric synthetic aperture radar

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102999762A (en) * 2012-10-25 2013-03-27 西安电子科技大学 Method for classifying polarimetric SAR (synthetic aperture radar) images on the basis of Freeman decomposition and spectral clustering
CN104463805A (en) * 2014-12-16 2015-03-25 西安电子科技大学 Polarimetric SAR coherent speckle noise suppression method based on homogeneity saliency and direction selection
CN104459656A (en) * 2014-12-31 2015-03-25 中国科学院空间科学与应用研究中心 Target orientation angle compensation method for fully polarimetric synthetic aperture radar

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
CHEN SIWEI 等: "Uniform polarimetric matrix rotation theory", 《 2013 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM-IGARSS》 *
XIANG DELIANG: "Multiple‐component polarimetric decomposition with new volume scattering models for PolSAR urban areas", 《IET RADAR,SONAR&ＮAVIGATION》 *
张腊梅 等: "极化SAR图像目标分解方法的研究进展", 《电子与信息学报》 *
朱岱寅 等: "高分辨率极化合成孔径雷达成像研究进展", 《数据采集与处理》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116797845A (en) * 2023-07-05 2023-09-22 中国科学院空天信息创新研究院 Unsupervised reduced polarization classification method based on scattering mechanism
CN116797845B (en) * 2023-07-05 2024-01-26 中国科学院空天信息创新研究院 Unsupervised reduced polarization classification method based on scattering mechanism
CN117390354A (en) * 2023-12-12 2024-01-12 江西师范大学 GoaC compensation-based polarization target decomposition method
CN117390354B (en) * 2023-12-12 2024-04-26 江西师范大学 Polarization target decomposition method based on GoaC compensation

Also Published As

Publication number Publication date
CN110516698B (en) 2022-03-11

Similar Documents

Publication Publication Date Title
CN109840588A (en) Neural network model training method, device, computer equipment and storage medium
CN110244271A (en) Radar emitter sorting recognition methods and device based on multiple simultaneous compressed transform
CN103425998B (en) SAR target identification method under circumstance of occlusion
CN109712153A (en) A kind of remote sensing images city superpixel segmentation method
CN110516698A (en) Complete polarization image polarization decomposing method, apparatus, electronic equipment and storage medium
AU2021101142A4 (en) Method and System for Constructing Digital Rock
CN111275686B (en) Method and device for generating medical image data for artificial neural network training
CN110516728B (en) Polarized SAR terrain classification method based on denoising convolutional neural network
CN112446357B (en) SAR automatic target recognition method based on capsule network
CN108897042A (en) Content of organic matter earthquake prediction method and device
Wei et al. Synthesis of large realistic iris databases using patch-based sampling
CN107301641A (en) A kind of detection method and device of Remote Sensing Imagery Change
CN109613617B (en) Underground water detection method and system based on magnetic resonance response signal parameter extraction
CN112381845B (en) Rock core image generation method, model training method and device
CN110069969A (en) A kind of certification fingerprint identification method based on pseudorandom integration
CN107909595A (en) Extra large land clutter Scene Segmentation based on amplitude Yu energy compaction measure product
CN107944352A (en) A kind of Vehicle License Plate Recognition System and method based on picture retrieval method
CN105842665A (en) Spectrum weighting-based SAR image sidelobe suppression method
Zhao et al. WAEGAN: A GANs-based data augmentation method for GPR data
WO2009088524A1 (en) System for and method of enhancing images using fractals
CN115201245A (en) Rock core analysis method and device based on nuclear magnetic resonance simulation, equipment and medium
CN115131548A (en) SAR image ship target detection method fusing frequency domain significance
CN107610130A (en) Extra large land clutter Scene Segmentation based on the ratio between amplitude and phase linearity
CN113960551A (en) Clutter image generation method and target detection method for SAR image
CN106600535A (en) Fullhole logging image generation method

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