Model-based polarized sea ice target decomposition method

Abstract

The invention relates to a model-based polarized sea ice target decomposition method. The method comprises the following steps of: respectively obtaining volume scattering power values PV of three volume scattering models of a sea ice target under horizontal, vertical and random pitching geometry; s2, determining a selected volume scattering model TV according to a strategy of taking a maximum value of the volume scattering power value PV, and obtaining a residual coherence matrix TN; s3, obtaining a residual coherence matrix TN (theta R) after deorientation according to a cross polarization component minimization method; s4, judging the actual scattering mechanism of the target according to the relative size relation of elements at the positions (1, 1) and (2, 2) of the residual coherence matrix TN (theta R); and S5, according to an actual scattering mechanism of the target, adaptively selecting a bisection method of a surface scattering component or a two-surface scattering component of the preference target, and obtaining a two-surface scattering power PD and a surface scattering power value PS.

Description

Technical Field

[0001] The invention relates to the field of polarization decomposition of full-polarization synthetic aperture radar data, and in particular to a model-based polarization sea ice target decomposition method. Background Art

[0002] Sea ice covers about 25 million square kilometers of the Earth's surface and is one of the key factors affecting the global water cycle and climate change. Electromagnetic wave polarization is very sensitive to the dielectric properties and geometric structure of sea ice. Spaceborne polarimetric synthetic aperture radar (PolSAR) can penetrate the surface of the ice layer to carry out high-resolution full-polarimetric imaging all day and all weather, and is widely used in fine monitoring of sea ice. However, the sea ice data obtained by PolSAR is often a mathematical matrix, and it is necessary to use polarization decomposition and other technologies to interpret the physical scattering information required for sea ice identification and classification. Model-based polarization decomposition is committed to decomposing the complex backscattering of sea ice into an additive combination of standard scattering components such as surface scattering caused by rough ice surface, two-sided scattering caused by deformed ice, and volume scattering caused by a large number of ellipsoidal sea ice particles (mainly salt water inclusions) in the ice layer. It has a clear physical meaning and has been successfully applied to the interpretation of sea ice scattering. Despite this, the model-based sea ice decomposition method still has the following two problems: first, the current model decomposition for sea ice only considers a specific volume scattering model, which cannot achieve adaptive matching of the orientation mode of the sea ice particle cloud in the ice layer; second, the existing model-based sea ice decomposition method cannot make full use of the polarization scattering information, and there are residual items in the decomposition results.

[0003] In view of the above two problems, the present invention proposes a model-based polarization sea ice target decomposition algorithm. V , thus uniquely determining the selected volume scattering model T among the three volume scattering models of horizontal, vertical and random V , which solves the problem that the model decomposition of sea ice cannot adaptively match the volume scattering model according to the orientation mode of the ice layer particle cloud. The present invention also adaptively selects the dichotomy of the preferred target surface scattering component or the diffraction component according to the actual scattering mechanism of the target, obtains the power values ​​of the surface scattering and diffraction components, and thus realizes the complete utilization of polarization scattering information while ensuring that the decomposition result will not be degraded. Summary of the invention

[0004] The purpose of the present invention is to propose a model-based polarized sea ice target decomposition method, the specific steps are as follows:

[0005] S1. Based on the three volume scattering models of the salt water inclusions in the ice layer of the sea ice target under horizontal, vertical and random pitch geometry and the polarized sea ice data of the fully polarized synthetic aperture radar, the volume scattering power values ​​P of the sea ice target under the three scattering models are obtained respectively. V ;

[0006] S2, according to the volume scattering power value P V The maximum value strategy determines the selected volume scattering model T V , and obtain the residual coherence matrix T N ;

[0007] S3, according to the method of minimizing cross-polarization components, the residual coherence matrix T after de-orientation is obtained N (θ R );

[0008] S4, according to the residual coherence matrix T N (θ R ) to determine the actual scattering mechanism of the target;

[0009] S5. Adaptively select the preferred target surface scattering component or dichotomy component according to the actual scattering mechanism of the target, and obtain the dichotomy power P D And the surface scattered power value P S .

[0010] Furthermore, the coherence matrices of the horizontal, vertical and random volume scattering models in step S1 are respectively and

[0011]

[0012]

[0013] in, and is the coherence matrix of the horizontal volume scattering model Elements of and is the coherence matrix of the vertical volume scattering model Elements of

[0014] The coherence matrix T of the target is:

[0015]

[0016] Among them, T 11 , T 12 , T 13 , T 21 , T 22, T 23 , T 31 , T 32 and T 33 are the elements of the coherence matrix T;

[0017] The volume scattered power value P to be determined V The calculation formula is: V = trace(T V )f V ;

[0018] Where trace(·) is the matrix rank operator, f V is the weight of the volume scattering component in the total backscattering component. The volume scattering power components corresponding to the three volume scattering models are and

[0019] Furthermore, the step S2 comprises:

[0020] According to the calculated and The maximum value is selected as the final volume scattering power value, that is:

[0021]

[0022] exist and The model that can obtain the maximum power is used as the volume scattering model T V The final model:

[0023]

[0024] Residual coherence matrix T N The expression is: T N =Tf V T V .

[0025] Furthermore, the step S3 comprises:

[0026] According to the method of minimizing cross-polarization components, the residual coherence matrix T after deorientation is obtained. N (θ R ), the coherence matrix T after de-orientation N (θ R ) is:

[0027]

[0028] Where R(θ R ) is the rotation matrix, θ R is the orientation angle, and the orientation angle θ is calculated by minimizing the cross-polarization component method.R The calculation formula is as follows:

[0029]

[0030] And the orientation angle

[0031] Furthermore, the step S4 comprises:

[0032] If T N11 (θ R )≥T N22 (θ R ), then it is determined that the scattering mechanism of the remaining targets is dominated by surface scattering;

[0033] If T N11 (θ R )<T N22 (θ R ), then it is determined that the scattering mechanism of the remaining targets is dominated by two-sided scattering.

[0034] Furthermore, the step S5 comprises:

[0035] If the scattering mechanism of the remaining targets is dominated by surface scattering, the dichotomy method that prefers the surface scattering component is selected at this time to solve the surface scattering power P S and the dihedral scattering power value P D They are:

[0036]

[0037] The remaining target scattering mechanism is dominated by two-sided scattering. At this time, the dichotomy of the two-sided scattering component of the preferred target is selected, and the two-sided scattering power P D And the surface scattered power value P S They are:

[0038]

[0039] Furthermore, the f V The expression is:

[0040]

[0041] The superscript -1 represents the matrix inversion, λ max (·) and λ min (·) represent the maximum and minimum eigenvalues ​​of the matrix.

[0042] The present invention has the following beneficial effects:

[0043] The present invention selects the maximum volume scattering power value P V, thus uniquely determining the selected volume scattering model T among the three volume scattering models of horizontal, vertical and random V , which solves the problem that the model decomposition of sea ice cannot adaptively match the volume scattering model according to the orientation pattern of the ice particle cloud.

[0044] The present invention also adaptively selects a dichotomy method that prefers the target surface scattering component or the dichotomy component according to the actual scattering mechanism of the target, obtains the power values ​​of the surface scattering and dichotomy components, thereby achieving complete utilization of polarization scattering information while ensuring that the decomposition result will not be degraded. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 The present invention is a flow chart of a model-based polarized sea ice target decomposition method of the present invention;

[0046] Figure 2 Pauli decomposition RGB composite image of the experimental area;

[0047] Figure 3 This is a map of the different directional distributions of salt water inclusions in the sea ice layer;

[0048] Figure 4 This is a graph showing the results of the model-based polarized sea ice target decomposition algorithm method. DETAILED DESCRIPTION

[0049] The following is a further detailed description of a model-based polarized sea ice target decomposition method proposed in the present invention in combination with the accompanying drawings and specific embodiments. According to the following description, the advantages and features of the present invention will be clearer. It should be noted that the drawings are in a very simplified form and are not in precise proportions, which are only used to conveniently and clearly assist in explaining the purpose of the embodiments of the present invention. In order to make the purposes, features and advantages of the present invention more obvious and easy to understand, please refer to the accompanying drawings. It should be noted that the structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification for people familiar with this technology to understand and read, and are not used to limit the limiting conditions for the implementation of the present invention, so they have no technical substantive significance. Any structural modification, change in proportional relationship or adjustment of size, without affecting the effects that can be produced by the present invention and the purposes that can be achieved, should still fall within the scope of the technical content disclosed by the present invention.

[0050] like Figure 1 The present invention provides a model-based polarized sea ice target decomposition method, comprising the following steps:

[0051] S1. Based on the three volume scattering models of the salt water inclusions in the ice layer of the sea ice target under horizontal, vertical and random pitch geometry and the polarized sea ice data of the fully polarized synthetic aperture radar, the volume scattering power values ​​P of the sea ice target under the three scattering models are obtained respectively.V ;

[0052] S2, according to the volume scattering power value P V The maximum value strategy determines the selected volume scattering model T V , and obtain the residual coherence matrix T N ;

[0053] S3, according to the method of minimizing cross-polarization components, the residual coherence matrix T after de-orientation is obtained N (θ R );

[0054] S4, according to the residual coherence matrix T N (θ R ) to determine the actual scattering mechanism of the target;

[0055] S5. Adaptively select the preferred target surface scattering component or dichotomy component according to the actual scattering mechanism of the target, and obtain the dichotomy power P D And the surface scattered power value P S .

[0056] The following combined embodiments further illustrate the model-based polarization sea ice target decomposition method proposed in the present invention. The fully polarized synthetic aperture radar data used in this embodiment was acquired by Gaofen-3 in the northeast of Prydz Bay, Antarctica in a fully polarized mode on April 29, 2019, with a resolution of 8 meters. The central geographic coordinates of this data are 73°6' east longitude and 68°1' south latitude. The area is adjacent to China's Antarctic Zhongshan Station (76°22'14.28" east longitude and 69°22'24.76" south latitude). The Prydz Bay waters are located in the Indian Ocean. The regional sea ice is affected by sea surface temperature, wind and updrafts, and has the characteristics of alternating freezing periods (March to September) and melting periods (October to February of the following year). Due to its location in high-latitude sea areas, sea ice changes in Prydz Bay play an important role in studying global climate change. The fully polarized data acquired by Gaofen-3 are Level-1A single-view complex products, Figure 2 Pauli decomposition RGB composite image of the experimental area after multi-view and Refined Lee filtering (5×5) to suppress speckle noise using ESA's PolSARpro software. Combined with the National Ice Center (NIC, http: / / ice.aari.aq / ) obtained the Antarctic sea ice condition information on May 2, 2019. The sea ice in the study area is mainly primary ice, which contains many ellipsoidal brine inclusions. For primary ice, surface scattering caused by the rough surface and the snow-ice interface is dominant, while the secondary volume scattering is mainly contributed by the brine inclusions in the ice layer.

[0057] In step S1, the volume scattering power value P is calculated based on the three volume scattering models of the brine inclusions in the sea ice target under horizontal, vertical and random pitch geometry and the polarized sea ice data of the fully polarized synthetic aperture radar. V .like Figure 3 As shown, the brine inclusions in the ice layer present three different pitch geometries: horizontal, vertical and random. Figure 3 The coherence matrices of the three volume scattering models for different pitch geometries are shown and The simplified expression is:

[0058]

[0059] in, and is the coherence matrix of the horizontal volume scattering model Elements of and is the coherence matrix of the vertical volume scattering model Elements of

[0060] According to the polarization data of the fully polarized synthetic aperture radar obtained by the GF-3 satellite, the coherence matrix T of the target is obtained as follows:

[0061]

[0062] Among them, T 11 , T 12 , T 13 , T 21 , T 22 , T 23 , T 31 , T 32 and T 33 are the elements of the coherence matrix T;

[0063] The volume scattered power value P to be determined V The calculation formula is: V = trace(T V )f V ;

[0064] Where trace(·) is the matrix rank operator, f V is the weight of the volume scattering component in the total backscattering component:

[0065]

[0066] The superscript -1 represents the matrix inversion, λ max (·) and λ min(·) represent the maximum and minimum eigenvalues ​​of the matrix. From the above formula, the volume scattering power components corresponding to the three volume scattering models can be obtained as follows: and

[0067] In step S2, according to the volume scattered power value P in step S1, V The maximum value strategy determines the selected volume scattering model T V , and obtain the residual coherence matrix T N . According to the calculated and The maximum value is selected as the final volume scattering power value, that is:

[0068]

[0069] exist and The model that can obtain the maximum power is used as the volume scattering model T V The final model:

[0070]

[0071] Residual coherence matrix T N The expression is: T N =Tf V T V .

[0072] In step S3, according to the residual coherence matrix T obtained in step S2 N According to the method of minimizing cross-polarization components, the residual coherence matrix T after deorientation is obtained N (θ R ), the coherence matrix T after de-orientation N (θ R ) is:

[0073]

[0074] Where R(θ R ) is the rotation matrix, θ R is the orientation angle, and the orientation angle θ is calculated by minimizing the cross-polarization component method. R The calculation formula is as follows:

[0075]

[0076] And the orientation angle

[0077] In step S4, according to the remaining coherence matrix T in step S3, N (θR ) to determine the actual scattering mechanism of the target.

[0078] If T N11 (θ R )≥T N22 (θ R ), then it is determined that the scattering mechanism of the remaining targets is dominated by surface scattering;

[0079] If T N11 (θ R )<T N22 (θ R ), then it is determined that the scattering mechanism of the remaining targets is dominated by two-sided scattering.

[0080] In step S5, according to the actual scattering mechanism of the target in step S4, the dichotomy method of favoring the target surface scattering component or the dihedral scattering component is adaptively selected to obtain the dihedral scattering power P D And the surface scattered power value P S If the scattering mechanism of the remaining targets is dominated by surface scattering, the dichotomy method that prefers the surface scattering component is selected at this time to solve the surface scattering power P S and the dihedral scattering power value P D They are:

[0081]

[0082] The remaining target scattering mechanism is dominated by two-sided scattering. At this time, the dichotomy of the two-sided scattering component of the preferred target is selected, and the two-sided scattering power P D And the surface scattered power value P S They are:

[0083]

[0084] In this embodiment, the decomposition results of the surface scattering power, the dihedral scattering power and the volume scattering power obtained by the model-based polarization sea ice target decomposition algorithm are as follows: Figure 4 As shown by Figure 4It can be seen intuitively that the dominant scattering mechanism of sea ice in the study area is surface scattering, followed by volume scattering, and there is very little dihedral scattering. In order to quantitatively analyze the scattering mechanism of sea ice in the experimental area, the average value of the ratio of each scattering mechanism component to the total power in the entire study area in the statistical decomposition results is: the surface scattering component accounts for the highest proportion of 75.73%; the volume scattering component is the secondary scattering mechanism, accounting for 21.42%; the dihedral scattering component only accounts for 2.85% of the total power. Therefore, it can be concluded that the main scattering mechanism of sea ice in the experimental area is surface scattering. This is because radar backscattering is easily affected by the roughness of the sea ice surface. The surface roughness of floating ice and broken ice areas is high, so the surface scattering of regional sea ice is dominant. The secondary scattering mechanism in the decomposition results is volume scattering caused by salt water inclusions in the ice layer, indicating that the first ice layer contains a large number of salt water inclusions, which is consistent with the physical properties of the first ice. Since the surface of the first ice is usually relatively flat and there is no large-scale dihedral structure (such as snow hills and ice ridges on the surface of multi-year ice), the dihedral scattering component accounts for the least in the results. In order to verify that the model-based polarization sea ice target decomposition algorithm of the present invention can ensure that the decomposition result is a complete three-component decomposition, the number of pixel points whose surface / dihedral scattering power in the algorithm of this paper is 0 is further counted on the experimental data. There is no zero power value in the decomposition result of the present invention, and the three scattering mechanisms of the sea ice target can be completely separated, so that the backscattering mechanism of the regional sea ice can be more fully understood while realizing the complete utilization of polarization scattering information.

[0085] In addition, another great advantage of the model-based polarized sea ice target decomposition algorithm of the present invention is that the algorithm can adaptively match the volume scattering model according to the orientation mode of the ice particle cloud. In the experimental area, the percentage of pixels using horizontal, vertical and random volume scattering models was statistically analyzed. The percentage of pixels using random models and horizontal models was 19.20% and 10.03% respectively, and the percentage of pixels using the vertical volume scattering model was 70.77%. The statistical results show that the orientation of the salt water inclusions in the sea ice not only presents a dominant vertical orientation, but also has a secondary random orientation and a non-negligible horizontal orientation. Therefore, the present invention uses a volume scattering model under horizontal, vertical and random pitch geometric relationships to describe the scattering mechanism of sea ice, rather than a single vertically oriented volume scattering model. The decomposition results can help to carry out unsupervised classification of sea ice types based on polarization decomposition.

[0086] Although the content of the present invention has been described in detail through the above preferred embodiments, it should be appreciated that the above description should not be considered as a limitation of the present invention. After reading the above content, it will be apparent to those skilled in the art that various modifications and substitutions of the present invention will occur. Therefore, the protection scope of the present invention should be limited by the appended claims.

Claims

1. A model-based polarized sea ice target decomposition method, characterized in that: include: S1. Based on the three volume scattering models of the salt water inclusions in the ice layer of the sea ice target under horizontal, vertical and random pitch geometry and the polarized sea ice data of the fully polarized synthetic aperture radar, the volume scattering power values ​​P of the sea ice target under the three scattering models are obtained respectively. V ; S2, according to the volume scattering power value P V The maximum value strategy determines the selected volume scattering model T V , and obtain the residual coherence matrix T N ; S3, according to the method of minimizing cross-polarization components, the residual coherence matrix T after de-orientation is obtained N (θ R ); S4, according to the residual coherence matrix T N (θ R ) to determine the actual scattering mechanism of the target; S5. Adaptively select the preferred target surface scattering component or dichotomy component according to the actual scattering mechanism of the target, and obtain the dichotomy power P D And the surface scattered power value P S .

2. The model-based polarized sea ice target decomposition method according to claim 1, characterized in that: The coherence matrices of the horizontal, vertical and random volume scattering models in step S1 are respectively and in, and is the coherence matrix of the horizontal volume scattering model Elements of and is the coherence matrix of the vertical volume scattering model Elements of The coherence matrix T of the target is: Among them, T 11 、T 12 、T 13 、T 21 、T 22 、T 23 、T 31 、T 32 and T 33 are the elements of the coherence matrix T; The volume scattered power value P to be determined V The calculation formula is: V = trace(T V )f V ; Where trace(·) is the matrix rank operator, f V is the weight of the volume scattering component in the total backscattering component. The volume scattering power components corresponding to the three volume scattering models are and 3. The model-based polarized sea ice target decomposition method according to claim 2, characterized in that: The step S2 comprises: According to the calculated and The maximum value is selected as the final volume scattering power value, that is: exist and The model that can obtain the maximum power is used as the volume scattering model T V The final model: Residual coherence matrix T N The expression is: T N =Tf V T V .

4. The model-based polarized sea ice target decomposition method according to claim 3, characterized in that: The step S3 comprises: According to the method of minimizing cross-polarization components, the residual coherence matrix T after deorientation is obtained. N (θ R ), the coherence matrix T after de-orientation N (θ R ) is: Where R(θ R ) is the rotation matrix, θ R is the orientation angle, and the orientation angle θ is calculated by minimizing the cross-polarization component method. R The calculation formula is as follows: And the orientation angle 5. The model-based polarized sea ice target decomposition method according to claim 4, characterized in that: The step S4 comprises: If T N11 (θ R )≥T N22 (θ R ), then it is determined that the scattering mechanism of the remaining targets is dominated by surface scattering; If T N11 (θ R )<T N22 (θ R ), then it is determined that the scattering mechanism of the remaining targets is dominated by two-sided scattering.

6. The model-based polarized sea ice target decomposition method according to claim 5, characterized in that: The step S5 comprises: If the scattering mechanism of the remaining targets is dominated by surface scattering, the dichotomy method that prefers the surface scattering component is selected at this time to solve the surface scattering power P S and the dihedral scattering power value P D They are: The remaining target scattering mechanism is dominated by two-sided scattering. At this time, the dichotomy of the two-sided scattering component of the preferred target is selected, and the two-sided scattering power P D And the surface scattered power value P S They are:

7. The model-based polarized sea ice target decomposition method according to claim 2, characterized in that: The f V The expression is: The superscript -1 represents the matrix inversion, λ max (·) and λ min (·) represent the maximum and minimum eigenvalues ​​of the matrix.

Images