CN102262223A - Method for decomposing coherent targets based on scattering matrixes - Google Patents

Abstract

The invention discloses a kind of method for decomposing coherent targets based on collision matrix, comprising: obtains the full polarimetric SAR data of target area; Extract the collision matrix of the full polarimetric SAR data

; By the collision matrix

Be decomposed into trihedral angle component collision matrix, 45 degree dihedral angle component collision matrixes of dihedral angle component collision matrix and rotation be concerned with and, formula is as follows: It sets restrictive condition and solves parameter lambda i, φ i, ψ; According to decomposition result composograph. It is of the invention various method for decomposing coherent targets are united based on the method for decomposing coherent targets of collision matrix, versatility it is more preferable, the polarization parameter of extraction can more accurately react the scattering signatures of target (atural object), improve the accuracy of automatic target detection.

Description

A kind of relevant target decomposition method based on scattering matrix

Technical field

The present invention relates to the full polarimetric SAR data processing field, relate in particular to a kind of relevant target decomposition method based on scattering matrix.

Background technology

The polarization characteristic of target, comprise surfaceness, symmetry, directionality etc., main relevant with its shape and structure, so polarization SAR (synthetic aperture radar, being synthetic-aperture radar) view data can provide more terrestrial object information than single-channel SAR data, and the polarization characteristic of target can promote the development of automatic target identification (ATR) algorithm, wherein the extracting method of polarization information just target decompose (TD), the scattering target of complexity is decomposed into the synthetic of some less basic scattering targets.

Target resolution theory (TD) is broadly divided into two classes: relevant target decomposes (CTD) and incoherent target is decomposed (ICTD).Incoherent target decomposition is the decomposition at target covariance matrix, coherence matrix, Mueller matrix or Stokes matrix, mainly comprises Huynen decomposition, Cloude decomposition and Freeman-Durden decomposition etc.Along with the development of polarization and high-resolution measuring technique, resolution element is more little, and the number of the scattering center that it contains is just few more, and application just can accurately be reacted the scattering signatures of target more based on the parameter of the relevant target decomposition method extraction of scattering matrix.Up to the present, the relevant decomposition methods of multiple classics such as Pauli decomposition, SDH decomposition, Cameron decomposition and TSVM decomposition model have been arranged, and different decomposition methods can obtain similar polarization parameter.

When basic scatterer by arbitrarily, but have the transmitting during irradiation of definite polarization state, all can radar signal reflected, and scattering matrix comprises all information that can predict reflected signal.Under the linear polarization base, scattering matrix can be expressed as:

$\tilde{S}=\left(\begin{array}{cc}{s}_{11}& s_{12}\\ s_{21}& s_{22}\end{array}\right),$ s _ij∈C，i，j∈{1，2} (1)

Based on two fundamental characteristics---reciprocity and the symmetry of radar target, Cameron has proposed a kind of relevant target decomposition method:

$\stackrel{\→}{S}=a[\cos {\mathrm{\θ}}_{\mathrm{rec}}(\cos \mathrm{\τ}{\hat{S}}_{\max }+\sin \mathrm{\τ}{\hat{S}}_{\min })+\sin {\mathrm{\θ}}_{\mathrm{rec}}{\hat{S}}_{\mathrm{nr}}]---\left(2\right)$

Wherein

It is scattering matrix

Vector form,

Represent nonreciprocal scattering component scattering matrix,

With

Represent maximum and minimum symmetrical scattering component scattering matrix in the reciprocity scattering component respectively.

Be scattering matrix and reciprocity subspace W _RecBetween angle, the expression scattering matrix is obeyed the degree of reciprocity principle.

The symmetrical degree of expression reciprocity scattering component.

After the decomposition, Cameron with the symmetrical scattering component of maximum with normalized complex vector

Represent:

${\hat{S}}_{\max }={\mathrm{ae}}^{\mathrm{i\ρ}}R\left(\mathrm{\ψ}\right)\stackrel{\→}{\mathrm{\Λ}}\left(z\right),$ a∈IR ⁺，ρ，ψ∈(-π，π] (3)

Wherein

$\stackrel{\→}{\mathrm{\Λ}}\left(z\right)=\frac{1}{\sqrt{1+{\left|z\right|}^2}}\left(\begin{array}{c}1\\ 0\\ 0\\ z\end{array}\right),$ z∈C，|z|≤1

A is the range value of scattering matrix, IR ⁺The expression nonnegative real number, ρ represents excess phase, ψ represents the orientation angle of scatterer, R (ψ) expression rotational transform operator.Multiple parameter z can be used for describing the scattering type of the symmetrical scatterers of being studied, and can be expressed as in the complex plane unit circle a bit.

2002, Touzi proved that maximum symmetrical scattering component can be expressed as form:

${\stackrel{\→}{S}}_{\max }={\exp }^{j{\mathrm{\φ}}_{S_a}}\·\sqrt{{\mathrm{\α}}^2+{\mathrm{\ϵ}}^2}\·[\cos \left(\mathrm{\η}\right)\·{\stackrel{\→}{S}}_a+\sin \left(\mathrm{\η}\right){\exp }^{j({\mathrm{\φ}}_{S_b}-{\mathrm{\φ}}_{S_a})}\·{\stackrel{\→}{S}}_b]---\left(4\right)$

Wherein

Be Pauli basis matrix [S _a] and [S _b] vector form, represent trihedral angle and dihedral angle scatterer respectively. The phase differential of expression trihedral angle-dihedral angle passage under the very high situation of passage coherence, can provide the useful information of target scattering body.Parameter η is positioned at interval [0, pi/2], is illustrated in the Scattering of Vector direction under trihedral angle-dihedral angle base, and the information of the scattering type of relevant maximum symmetrical scattered portion is provided.

In addition, Cameron pointed out once that there was orthogonality in maximum with minimum symmetrical scattering component in the decomposition result, and this has limited the span of minimum symmetrical scattering component greatly, for the maximum symmetrical scattering component of determining

Minimum symmetrical scattering component can be expressed as:

${\hat{S}}_{\min }=e^{j{\mathrm{\φ}}_{\min }}R({\mathrm{\ψ}}_{\max }\±\frac{\mathrm{\π}}{4}){\hat{S}}_b---\left(5\right)$

Following formula shows except that phase factor φ _MinOutside, minimum symmetrical scattering component

Feature can be by the diagonalization parameter ψ of the symmetrical scattering component of maximum _MaxDetermine.

The inventor finds, still do not research and analyse contact between these decomposition models and the difference between the similar polarization parameter at present, and the decomposition method that neither one is unified is used and had certain limitation.

Summary of the invention

The problems referred to above in view of prior art, the purpose of this invention is to provide a kind ofly various relevant target decomposition methods can be united, versatility is better based on the relevant target decomposition method of scattering matrix, make the polarization parameter that extracts to react the scattering signatures of target (atural object) more accurately, improve the accuracy of automatic target identification.

To achieve these goals, the invention provides a kind of relevant target decomposition method, comprising based on scattering matrix:

S11: the full polarimetric SAR data of obtaining the target area;

S12: the scattering matrix that extracts described full polarimetric SAR data Wherein

S ₁₂=S ₂₁

S13: with described scattering matrix

Be decomposed into trihedral angle component scattering matrix, dihedral angle component scattering matrix and rotate the relevant of 45 degree dihedral angle component scattering matrixes and, formula is as follows:

Wherein,

Expression trihedral angle component scattering matrix,

Expression dihedral angle component scattering matrix, λ _i, φ _iThe intensity and the phase factor of back three components, λ are decomposed in expression respectively _i〉=0, i=1,2,3, R (ψ) expression rotational transform operator, ψ represents the orientation angle of target scattering body;

S14: set restrictive condition and find the solution parameter lambda _i, φ _i, ψ;

S15: according to the decomposition result composograph.

As preferably, in the described S14 step, the setting restrictive condition is ψ=0.

As preferably, in the described S14 step, set restrictive condition and be

As preferably, in the described S14 step, set restrictive condition and be

λ wherein _MaxExpression λ ₂Maximal value, λ _MinExpression λ ₃Minimum value,

Expression

Maximal value.

As preferably, in the described S14 step, set restrictive condition and be

As preferably, after the described S15 step, also comprise:

S16: according to composograph identification object region type of ground objects.

The invention has the beneficial effects as follows: with various relevant target decomposition methods unite, versatility is better, the polarization parameter of extraction can react the scattering signatures of target (atural object) more accurately, improves the accuracy of automatic target identification.

Description of drawings

Fig. 1 is the process flow diagram based on the relevant target decomposition method of scattering matrix of the embodiment of the invention one;

Fig. 2 is that the relevant target decomposition method setting restrictive condition based on scattering matrix of the embodiment of the invention one is the composograph (to the complete polarization SAR data of certain target area) of ψ=0;

Fig. 3 is that the relevant target decomposition method setting restrictive condition based on scattering matrix of the embodiment of the invention one is

Composograph (to the complete polarization SAR data of certain target area) with set restrictive condition and be The comparison diagram of composograph (to the complete polarization SAR data of certain target area);

Fig. 4 is that the relevant target decomposition method setting restrictive condition based on scattering matrix of the embodiment of the invention one is

Composograph (to the complete polarization SAR data of certain target area);

Fig. 5 be the embodiment of the invention one based on decompose in the relevant target decomposition method of scattering matrix parameter analyze comparison diagram (be followed successively by from left to right Cameron decompose in scattering type parameter α and TSVM decomposition model scattering type intensity α among the maximum symmetrical scattering component scattering type parameter η, Cloude alpha-beta decomposition model _s);

Fig. 6 be the embodiment of the invention one analyze comparison diagram (left-to-right is followed successively by Cameron and decomposes the orientation angle parameter | ψ | (eliminating fuzzy back), Cloude decomposition model deflection parameter beta/2) based on decomposing parameter in the relevant target decomposition method of scattering matrix;

Fig. 7 be the embodiment of the invention one analyze comparison diagram (Cameron decomposes and TSVM decomposition model helicity parameter) based on decomposing parameter in the relevant target decomposition method of scattering matrix;

Fig. 8 is analyzing comparison diagram (Cameron decomposes parameter η and TSVM decomposition model parameter 2 τ based on decomposing parameter in the relevant target decomposition method of scattering matrix of the embodiment of the invention one _m);

Fig. 9 is the synoptic diagram of finding the solution based on the relevant target decomposition method of scattering matrix of the embodiment of the invention one;

Figure 10 is the process flow diagram based on the relevant target decomposition method of scattering matrix of the embodiment of the invention two.

Embodiment

Describe embodiments of the invention in detail below in conjunction with accompanying drawing.Among the embodiment, subscript m ax represents maximal value, and subscript m in represents minimum value.

S11: the full polarimetric SAR data of obtaining the target area;

S12: the scattering matrix that extracts described full polarimetric SAR data

Wherein

S ₁₂=S ₂₁Step S11 and step S12 are steps well known to those skilled in the art, do not repeat them here.

S13: with described scattering matrix Be decomposed into trihedral angle component scattering matrix, dihedral angle component scattering matrix and rotate the relevant of 45 degree dihedral angle component scattering matrixes and, formula is as follows:

$\tilde{S}={\mathrm{\λ}}_1{e}^{j{\mathrm{\φ}}_1}{\tilde{S}}_a+{\mathrm{\λ}}_2{e}^{j{\mathrm{\φ}}_2}R\left(\mathrm{\ψ}\right){\tilde{S}}_b+{\mathrm{\λ}}_3{e}^{j{\mathrm{\φ}}_3}R(\mathrm{\ψ}\±\frac{\mathrm{\π}}{4}){\tilde{S}}_b---\left(6\right)$

Wherein,

Expression trihedral angle component scattering matrix,

Expression dihedral angle component scattering matrix, λ _i, φ _iThe intensity and the phase factor of back three components, λ are decomposed in expression respectively _i〉=0, i=1,2,3, R (ψ) expression rotational transform operator, ψ represents the orientation angle of target scattering body; This formula can be brought formula (3) into by formula (4), (5) and obtain.The decomposition model of formula (6) has 7 unknown parameter (λ _i, φ _i, ψ), but have only 6 equations (respectively corresponding co-polarization and cross-polarized real part and imaginary part), so there is infinite multiresolution in this decomposition model.Generally speaking, set a restrictive condition in advance, just can obtain one group and separate.

S14: set restrictive condition and find the solution parameter lambda _i, φ _i, ψ;

Situation one:

The setting restrictive condition is ψ=0, and then formula (6) can be reduced to the form that Pauli decomposes:

$\tilde{S}={\mathrm{\λ}}_1{e}^{j{\mathrm{\φ}}_1}{\tilde{S}}_a+{\mathrm{\λ}}_2{e}^{j{\mathrm{\φ}}_2}{\tilde{S}}_b+{\mathrm{\λ}}_3{e}^{j{\mathrm{\φ}}_3}{\tilde{S}}_c---\left(7\right)$

The advantage that Pauli decomposes is that it is very simple, corresponding qualifications is also very simple, yet orientation angle is an important parameter describing the target scattering body, be defined as 0 (being the target level orientation, Horizontally Orientation) simply and under most of situation, can not describe actual conditions exactly.

In the Cloude alpha-beta decomposition model that is decomposed into the basis with Pauli, parameter beta is used for describing the target direction angle, can be described as under the form of formula (6):

$\mathrm{\β}={\tan }^{-1}\left(\frac{{\mathrm{\λ}}_3}{{\mathrm{\λ}}_2}\right)---\left(8\right)$

Prove that in the mode of theory following formula is incorrect for the statement at target direction angle according to the description of front.

Situation two:

The setting restrictive condition is

If λ ₂〉=λ ₃, this decomposition model can be reduced to:

$\tilde{S}={\mathrm{\λ}}_1{e}^{j{\mathrm{\φ}}_1}{\tilde{S}}_a+{\mathrm{\λ}}_2{e}^{j{\mathrm{\φ}}_2}R\left(\mathrm{\ψ}\right){\tilde{S}}_b+{\mathrm{\λ}}_3{e}^{j{\mathrm{\φ}}_2+\frac{\mathrm{\π}}{2}}R(\mathrm{\ψ}\±\frac{\mathrm{\π}}{4}){\tilde{S}}_b---\left(9\right)$

$={\mathrm{\λ}}_1{e}^{j{\mathrm{\φ}}_1}{\tilde{S}}_a+({\mathrm{\λ}}_2-{\mathrm{\λ}}_3)e^{j{\mathrm{\φ}}_2}R\left(\mathrm{\ψ}\right){\tilde{S}}_b+{\mathrm{\λ}}_3{e}^{j{\mathrm{\φ}}_2}R\left(\mathrm{\ψ}\right){\tilde{S}}_H$

Wherein

Expression conveyor screw scattering matrix:

${\tilde{S}}_H={\tilde{S}}_b+e^{j\frac{\mathrm{\π}}{2}}R(\±\frac{\mathrm{\π}}{4}){\tilde{S}}_b=\left[\begin{array}{cc}1& \±j\\ \±j& -1\end{array}\right]---\left(10\right)$

Formula this moment (6) is the equivalent form of value of SDH decomposition just.

Situation three:

The setting restrictive condition is:

$\frac{{\mathrm{\λ}}_{\max }}{{\mathrm{\λ}}_{\min }}=\max \left\{\frac{{\mathrm{\λ}}_2}{{\mathrm{\λ}}_3}\right\}---\left(11\right)$

When following formula satisfies, Also reach maximal value, one, two components of formula this moment (6) are concerned with and are the maximum symmetrical scattered portion of target, and three-component is minimum symmetrical scattered portion, and promptly the decomposition model of formula (6) is that symmetrical scattering matrix carries out the equivalent form of value that Cameron decomposes.

Situation four:

The setting restrictive condition is:

${\mathrm{\φ}}_3={\mathrm{\φ}}_1\±\frac{\mathrm{\π}}{2}---\left(12\right)$

When

Formula (6) is represented with vector form:

$\stackrel{\&RightArrow;}{S}={\mathrm{\&lambda;}}_1{e}^{j{\mathrm{\&phi;}}_1}\&CenterDot;\left[\begin{array}{c}1\\ 0\\ 1\end{array}\right]+{\mathrm{\&lambda;}}_2{e}^{j{\mathrm{\&phi;}}_2}\&CenterDot;\left[\begin{array}{c}\mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">cos</figure-callout>}2\mathrm{\&psi;}\\ \mathrm{<figure-callout\; id="2"\; label="sin"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">sin</figure-callout>}2\mathrm{\&psi;}\\ -\mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">cos</figure-callout>}2\mathrm{\&psi;}\end{array}\right]+{\mathrm{\&lambda;}}_3{e}^{j{\mathrm{\&phi;}}_3}\&CenterDot;\left[\begin{array}{c}-\mathrm{<figure-callout\; id="2"\; label="sin"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">sin</figure-callout>}2\mathrm{\&psi;}\\ \mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">cos</figure-callout>}2\mathrm{\&psi;}\\ \mathrm{<figure-callout\; id="2"\; label="sin"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">sin</figure-callout>}2\mathrm{\&psi;}\end{array}\right]---\left(13\right)$

Be similar to the TSVM method, following formula is mapped to Pauli base S _a, S _b, S _cOn, obtain:

$\left[\begin{array}{c}{\mathrm{\&lambda;}}_1{e}^{j{\mathrm{\&phi;}}_1}\\ \mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">cos</figure-callout>}2\mathrm{\&psi;}{\mathrm{\&lambda;}}_2{e}^{j{\mathrm{\&phi;}}_2}-\mathrm{<figure-callout\; id="2"\; label="sin"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">sin</figure-callout>}2\mathrm{\&psi;}{\mathrm{\&lambda;}}_3{e}^{j{\mathrm{\&phi;}}_3}\\ \mathrm{<figure-callout\; id="2"\; label="sin"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">sin</figure-callout>}2\mathrm{\&psi;}{\mathrm{\&lambda;}}_2{e}^{j{\mathrm{\&phi;}}_2}+\mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">cos</figure-callout>}2\mathrm{\&psi;}{\mathrm{\&lambda;}}_3{e}^{j{\mathrm{\&phi;}}_3}\end{array}\right]=\left[\begin{array}{ccc}1& 0& 0\\ 0& \mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">cos</figure-callout>}2\mathrm{\&psi;}& -\mathrm{<figure-callout\; id="2"\; label="sin"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">sin</figure-callout>}2\mathrm{\&psi;}\\ 0& \mathrm{<figure-callout\; id="2"\; label="sin"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">sin</figure-callout>}2\mathrm{\&psi;}& \mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">cos</figure-callout>}2\mathrm{\&psi;}\end{array}\right]\left[\begin{array}{c}{\mathrm{\&lambda;}}_1{e}^{j{\mathrm{\&phi;}}_1}\\ {\mathrm{\&lambda;}}_2{e}^{j{\mathrm{\&phi;}}_2}\\ {\mathrm{\&lambda;}}_3{e}^{j{\mathrm{\&phi;}}_3}\end{array}\right]---\left(14\right)$

When restrictive condition is

The time, following formula and TSVM decomposition model equivalence.

When

The time, formula (6) being mapped as on the Pauli base:

$\left[\begin{array}{c}{\mathrm{\&lambda;}}_1{e}^{j{\mathrm{\&phi;}}_1}\\ \mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">cos</figure-callout>}2\mathrm{\&psi;}{\mathrm{\&lambda;}}_2{e}^{j{\mathrm{\&phi;}}_2}-\mathrm{<figure-callout\; id="2"\; label="sin"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">sin</figure-callout>}2\mathrm{\&psi;}{\mathrm{\&lambda;}}_3{e}^{j{\mathrm{\&phi;}}_3}\\ \mathrm{<figure-callout\; id="2"\; label="sin"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">sin</figure-callout>}2\mathrm{\&psi;}{\mathrm{\&lambda;}}_2{e}^{j{\mathrm{\&phi;}}_2}+\mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">cos</figure-callout>}2\mathrm{\&psi;}{\mathrm{\&lambda;}}_3{e}^{j{\mathrm{\&phi;}}_3}\end{array}\right]=\left[\begin{array}{ccc}1& 0& 0\\ 0& \mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">cos</figure-callout>}2\mathrm{\&psi;}& -\mathrm{<figure-callout\; id="2"\; label="sin"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">sin</figure-callout>}2\mathrm{\&psi;}\\ 0& \mathrm{<figure-callout\; id="2"\; label="sin"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">sin</figure-callout>}2\mathrm{\&psi;}& \mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">cos</figure-callout>}2\mathrm{\&psi;}\end{array}\right]\left[\begin{array}{c}{\mathrm{\&lambda;}}_1{e}^{j{\mathrm{\&phi;}}_1}\\ {\mathrm{\&lambda;}}_2{e}^{j{\mathrm{\&phi;}}_2}\\ -{\mathrm{\&lambda;}}_3{e}^{j{\mathrm{\&phi;}}_3}\end{array}\right]---\left(15\right)$

When restrictive condition is The time, following formula and TSVM decomposition model equivalence.

What the decomposition model of formula (6) was not determined separates, but we can simplify it under left and right sides rounding polarization base:

$S_{\mathrm{RL}}={\mathrm{\&lambda;}}_1{e}^{j{\mathrm{\&phi;}}_1}{e}^{\mathrm{j\&pi;}}---\left(16\right)$

$S_{\mathrm{RR}}={\mathrm{\&lambda;}}_2{e}^{j{\mathrm{\&phi;}}_2}{\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="HSA00000126241100051.png"\; state="\{\{state\}\}">e</figure-callout>}}^j+{\mathrm{\&lambda;}}_3{e}^{j{\mathrm{\&phi;}}_3}{e}^{j(2\mathrm{\&psi;}+\frac{\mathrm{\&pi;}}{2})}---\left(17\right)$

$S_{\mathrm{LL}}={\mathrm{\&lambda;}}_2{e}^{j{\mathrm{\&phi;}}_2}{e}^{j(-2\mathrm{\&psi;}+\mathrm{\&pi;})}+{\mathrm{\&lambda;}}_3{e}^{j{\mathrm{\&phi;}}_3}{e}^{j(-2\mathrm{\&psi;}+\frac{\mathrm{\&pi;}}{2})}---\left(18\right)$

Be not difficult to find out two parameter lambda of first scattering component from formula (16) ₁And φ ₁Can by about revolve component S _RLDefinite fully, and irrelevant with qualifications, therefore for different decomposition methods, the intensity and the phase factor of trihedral angle (ball) scattering component all are on all four.Three-component is mutually orthogonal in the model (6) in addition, so following formula is set up:

${\left|\right|\tilde{S}\left|\right|}^2={{\mathrm{\&lambda;}}_1}^2+{{\mathrm{\&lambda;}}_2}^2+{{\mathrm{\&lambda;}}_3}^2---\left(19\right)$

Norm wherein || || be by inner product operator () definition:

$\left|\right|\tilde{S}\left|\right|=\sqrt{(\stackrel{\&RightArrow;}{S},\stackrel{\&RightArrow;}{S})}---\left(20\right)$

Again because λ ₁Have nothing to do with qualifications, then λ ₂With λ ₃Quadratic sum be constant.

Multiplied each other in equation (17) and (18), again with λ ₂ ²+ λ ₃ ²Carry out normalization, obtain:

$\frac{S_{\mathrm{RR}}\&CenterDot;S_{\mathrm{LL}}}{{{\mathrm{\&lambda;}}_2}^2+{{\mathrm{\&lambda;}}_3}^2}=\frac{{{\mathrm{\&lambda;}}_2}^2}{{{\mathrm{\&lambda;}}_2}^2+{{\mathrm{\&lambda;}}_3}^2}{e}^{j(2{\mathrm{\&phi;}}_2+\mathrm{\&pi;})}+\frac{{{\mathrm{\&lambda;}}_3}^2}{{{\mathrm{\&lambda;}}_2}^2+{{\mathrm{\&lambda;}}_3}^2}{e}^{j(2{\mathrm{\&phi;}}_3+\mathrm{\&pi;})}---\left(21\right)$

Following formula can be regarded the convex combination of two vectors on the unit circumference as, as shown in figure 10.Stain in the unit circle is represented the left end expression formula position of equation (21), and expression is vectorial respectively for 2 of unit circumference blues

With

The expression that its middle distance black color dots is nearer

The two line segment lengths ratio of being cut apart by black color dots As can be seen from Figure 2, cross the determined phase factor φ of any straight line of black color dots ₂, φ ₃And two the line segment length ratio can carry out finding the solution of decomposition model (6) as restrictive condition.

As previously mentioned, the SDH decomposition is a restrictive condition

The equivalent form of value of following decomposition model (6), in this case, among Fig. 2 on the unit circle two vectorial phase differential be π, promptly the line of two vectors advanced true origin.It should be noted that ratio this moment

Reach maximal value, and this just Cameron decompose restrictive condition to scattering model (6), therefore we may safely draw the conclusion: SDH decomposes and decomposes with Cameron is equivalent equivalence.

(H, V) under the linear orthogonal base, SDH decomposes and can be expressed as:

Wherein θ is an orientation angle,

The expression absolute phase, the three-component basis matrix is expressed as:

Because the basis matrix of following formula does not carry out normalization as the Pauli base, and conveyor screw component scattering matrix can be expressed as dihedral angle and the dihedral angle sum of rotating π/4 that phase differential is a pi/2, the three-component intensity λ that Cameron decomposes _iThere is following relation with SDH three-component weight:

${\mathrm{\&lambda;}}_1=\sqrt{2}{k}_S$

${\mathrm{\&lambda;}}_2=\sqrt{2}(k_D+\frac{1}{2}{k}_H)---\left(24\right)$

$2{\mathrm{\&lambda;}}_3=\sqrt{2}{k}_H$

As can be seen, relevant decomposition model of existing several classics and parameter difference thereof and contact can be obtained by formula (6) setting restrictive condition.Be that formula (6) is a kind of relevant decomposition model of Unified Form, set different restrictive conditions and can obtain different decomposition models, can set restrictive condition according to concrete needs in the practical application, make the polarization parameter that extracts to react the scattering signatures of target (atural object) more accurately, and then improve the accuracy of automatic target identification.

S15: according to the decomposition result composograph.The synthetic image of different decomposition models is different, and the accuracy of the scattering signatures of the polarization parameter reaction target (atural object) that decomposition model obtains is also different, and is also just different with the degree that conforms to of actual conditions.Therefore the quality of decomposition model is directly connected to the quality of composograph.

Below select L-band ESAR full polarimetric SAR data to experimentize, use the decomposition model (setting restrictive condition) of present embodiment to decompose, and decomposition result and extracting parameter are compared analysis, unitarity, the correctness of checking decomposition model.ESAR is the airborne polarization SAR system that is developed by German NASA (DLR), and the experiment area of choosing (target area) is positioned at German Oberpfaffenhofen, data acquisition time on September 30th, 2000,2.8 meters * 1.7 meters of the resolution of experimental data.This area's type of ground objects distributes more, comprises forest, city, airport, and near multiple meadow, airport and parking lot.Setting restrictive condition as shown in Figure 2 is the composograph (being equivalent to the RGB composite diagram that Paul i decomposes, HH+VV (blueness) wherein, HH-VV (redness), 2HV (green)) of ψ=0, and setting restrictive condition shown in Figure 3 is

Composograph (be equivalent to the RGB composite diagram that Cameron decomposes, λ ₂(redness), λ ₃(green), λ ₁(blueness)) and set restrictive condition and be

Composograph (be equivalent to the RGB composite diagram that SDH decomposes, k _d(redness), k _h(green), k _s(blueness)) comparison diagram, setting restrictive condition shown in Figure 4 is

Composograph (be equivalent to the RGB composite diagram that TSVM decomposes, λ ₁(blueness), λ ₂(redness), λ ₃(green)), embody color distortion with gray scale in each composograph.As can be seen:

1) two of the composite diagram of Fig. 2 and Fig. 3 composite diagrams relatively, for the simple relatively surface scattering zone of types of ground objects such as airfield runway and meadow, two decomposition result difference are little, and for zones such as forest, city and parking lots, (many volume scatterings in wood land and trunk and ground rescattering because scattering process complicates, city spriochaeta polyspira scattering and dihedral angle scattering, parking lot dihedral angle scattering is more), define ψ=0 (situation 1) simply and make two kinds of decomposition result that apparent in view difference arranged.

2) two composite diagrams among Fig. 3 compare, and almost 100% satisfies the corresponding relation of formula (24), and promptly the equivalence that Cameron decomposes and SDH decomposes has obtained further checking.

3) two of the composite diagram of the composite diagram of Fig. 4 and Fig. 2, Fig. 3 composite diagram difference are bigger.This be because its restrictive condition with Bigger with other decomposition model difference.The method that Touzi finds the solution the TSVM decomposition model is at first to calculate goal orientation angle and elimination by the Huynen method, and then carry out each invariable rotary CALCULATION OF PARAMETERS, but for relevant target, the deflection difference of calculating in Huynen method and Cloude model, the circular polarisation method (SDH decomposition) is little, and decomposition result that obtains in this case and Cloude decomposition result difference are little.Fig. 4 finds the solution by model shown in Figure 9, each parameter find the solution the order suc as formula (25).

${\mathrm{\&lambda;}}_1,{\mathrm{\&phi;}}_1\&DoubleRightArrow;{\mathrm{\&phi;}}_2\&DoubleRightArrow;{\mathrm{\&phi;}}_3\&DoubleRightArrow;{\mathrm{\&lambda;}}_2,{\mathrm{\&lambda;}}_3\&DoubleRightArrow;\mathrm{\&psi;}---\left(25\right)$

As shown in Figure 5, Cameron decomposition model and Cloude decomposition model difference in the expression of scattering type parameter is little, and this is because alpha parameter is to pass through formula in the Cloude decomposition model

Obtain, and Cameron decomposes parameter η by formula η=tan ^-1(λ _Max/ λ ₁) obtain, whole zone is distributed as the master with natural feature on a map, even the area, cities and towns, the conveyor screw scattering is also less, the minimum symmetrical scattering component λ during Cameron decomposes _MinLess, and two decompose bases and all satisfy orthogonality, makes As can be seen from Figure 5, only rely on the scattering type parameter to distinguish relatively difficulty (as meadow and forest) of different atural objects, can introduce dihedral angle-trihedral angle phase differential parameter

Improve terrain classification.45 ° of dihedral angle component phase differences of trihedral angle component and rotation are pi/2 in the TSVM decomposition model, then relevant and the point that can be expressed as on the symmetrical scattering and spatial complex plane unit circle circumference, the TSVM decomposition model scattering type parameter α of this two component _sBy formula Obtain, and decompose η=tan for Cameron ^-1(λ _Max/ λ ₁), λ ₂＜λ _Max, so parameter alpha among Fig. 5 _sBe significantly less than parameter η, only work as λ _Max＜＜λ ₁, i.e. scattering type parameter η → 0 o'clock, two parameter difference are less.

As shown in Figure 6, though as can be seen from the figure the basic trend that atural object distributes can be reflected in β/2, but for the target scattering body, no matter helicity high or low (being that the conveyor screw component was big or little during SDH decomposed), β/2 and orientation angle parameter | ψ | all there is some difference, because Cameron decomposes with SDH and decomposes equivalence, parameter | ψ | consistent with the orientation angle that the circular polarisation method obtains, promptly β/2 can not be reacted the orientation angle of target accurately.

Be illustrated in figure 7 as, the difference that can clearly find out two parameters from figure causes by the difference of two calculation method of parameters, during Cameron decomposes,

And in the TSVM decomposition, τ _m=tan ^-1(λ ₃/ λ ₁)/2, λ _Min＜λ ₃, τ＜τ then _mIt should be noted that in addition for meadow type shown in Figure 2 and wood land and can not distinguish, similar with Cameron decomposition scattering type parameter η among Fig. 3.Analysis by the front, 45 ° of dihedral angle component phase differences of trihedral angle component and rotation are pi/2 in the TSVM model, this two component relevant and can be expressed as point on the symmetrical scattering and spatial complex plane unit circle circumference then, according to the sign of Touzi for symmetrical scatterers, parameter 2 τ _mCan regard as 45 ° of dihedral angles of expression trihedral angle and rotation relevant and the scattering type of symmetrical scatterers, this moment, trihedral angle-dihedral angle phase differential was a pi/2, parameter alpha _sThe helicity of target then can be described.Cameron decomposes parameter η and TSVM model parameter 2 τ _mComparative effectiveness as shown in Figure 8.

More as can be known,, all there is mutual inner link by parameter though have difference between each decomposition method.

The relevant target decomposition method based on scattering matrix of the embodiment of the invention is united the relevant decomposition model of several classics, the difference of the parameter of each relevant target decomposition method and contact also obtain embodying, therefore can set restrictive condition according to concrete needs in the practical application, make the polarization parameter that extracts to react the scattering signatures of target (atural object) more accurately, and then improve the accuracy of automatic target identification.

The process flow diagram based on the relevant target decomposition method of scattering matrix of inventive embodiments two as shown in figure 10 on the basis of embodiment one, after the described S15 step, also comprises:

S16: according to composograph identification object region type of ground objects.

Since the embodiment of the invention based on the relevant target decomposition method of scattering matrix with various relevant target decomposition methods unite, versatility is better, the polarization parameter that extracts can react the scattering signatures of target (atural object) more accurately, therefore can be according to composograph identification object region type of ground objects, recognition accuracy also correspondingly improves.

Above embodiment is an exemplary embodiment of the present invention only, is not used in restriction the present invention, and protection scope of the present invention is defined by the claims.Those skilled in the art can make various modifications or be equal to replacement the present invention in essence of the present invention and protection domain, this modification or be equal to replacement and also should be considered as dropping in protection scope of the present invention.

Claims (6)

1. the relevant target decomposition method based on scattering matrix is characterized in that, comprising:

S11: the full polarimetric SAR data of obtaining the target area;

S12: the scattering matrix that extracts described full polarimetric SAR data Wherein

S ₁₂=S ₂₁

S13: with described scattering matrix

Be decomposed into trihedral angle component scattering matrix, dihedral angle component scattering matrix and rotate the relevant of 45 degree dihedral angle component scattering matrixes and, formula is as follows:

Wherein,

Expression trihedral angle component scattering matrix, Expression dihedral angle component scattering matrix, λ _i, φ _iThe intensity and the phase factor of back three components, λ are decomposed in expression respectively _i〉=0, i=1,2,3, R (ψ) expression rotational transform operator, ψ represents the orientation angle of target scattering body;

S14: set restrictive condition and find the solution parameter lambda _i, φ _i, ψ;

S15: according to the decomposition result composograph.

2. the relevant target decomposition method based on scattering matrix according to claim 1 is characterized in that in the described S14 step, the setting restrictive condition is ψ=0.

3. the relevant target decomposition method based on scattering matrix according to claim 1 is characterized in that, in the described S14 step, the setting restrictive condition is

4. the relevant target decomposition method based on scattering matrix according to claim 1 is characterized in that, in the described S14 step, the setting restrictive condition is λ wherein _MaxExpression λ ₂Maximal value, λ _MinExpression λ ₃Minimum value,

Expression Maximal value.

5. the relevant target decomposition method based on scattering matrix according to claim 1 is characterized in that, in the described S14 step, the setting restrictive condition is

6. according to the described relevant target decomposition method of one of claim 1 to 5, it is characterized in that, after the described S15 step, also comprise based on scattering matrix:

S16: according to composograph identification object region type of ground objects.

Images