CN-116804755-B - Polarization SAR three-component target decomposition method based on model decomposition

CN116804755BCN 116804755 BCN116804755 BCN 116804755BCN-116804755-B

Abstract

The invention provides a polarized SAR three-component target decomposition method based on model decomposition, which can solve the problems of volume scattering overestimation and negative energy of scattered components in the polarized SAR target decomposition and realize the model full-parameter optimization solution of the polarized SAR target decomposition. According to the method, second-order statistics of polarized SAR are analyzed, the maximum eigenvalue or residual matrix trace of the residual matrix after the minimization based on model decomposition is taken as an objective function, an optimization problem is established under the constraint condition that each scattered component power is non-negative and the residual matrix is semi-positive, the optimization problem is a convex optimization problem, and a global optimal solution of all parameters can be obtained by solving the model. The method has the outstanding advantages that the method is not limited to polarized SAR target decomposition under the scattering symmetry condition, can rapidly acquire the full-parameter optimal solution of the polarized SAR target decomposition based on the multi-basic scattering component, and provides guarantee for high-precision polarized SAR target decomposition.

Inventors

WANG TINGTING
JIANG PENGHUI
DING ZHIQUAN
QIN TIANQI
XIANG HONGLI
YAN YISHENG

Assignees

四川航天燎原科技有限公司

Dates

Publication Date: 20260512
Application Date: 20230511

Claims (10)

1. A polarized SAR three-component target decomposition method based on model decomposition is characterized by comprising the following steps: under a single-station backscattering system, acquiring polarization scattering matrix information and solving a polarization coherence matrix According to imaging scene information, determining a bulk scattering model, a secondary scattering model and a surface scattering model T V 、T D 、T S , and according to a depolarization coherent matrix T, primarily calculating a solving result of a polarized SAR target decomposition model, wherein the solving result comprises energy P V of the bulk scattering model, energy P D of the secondary scattering model and energy P S of the surface scattering model; if the solving result is not abnormal, outputting the solving result; If the solving result is abnormal, solving a polarized SAR target residual decomposition model through a cost function and a constraint condition, and recalculating the energy of the bulk scattering model, the energy of the secondary scattering model and the energy of the surface scattering model, wherein the residual decomposition model meets the condition that T=P V T V +P D T D +P S T S +T R ,T R is a residual matrix after decomposition, The cost function is used for shrinking the characteristic root of T R , and the constraint condition comprises at least one of P V ≥0,P D ≥0,P S ≥0,T R is more than or equal to 0.
2. The method according to claim 1, wherein the method further comprises: And outputting an imaging analysis result according to the energy of the bulk scattering model, the energy of the secondary scattering model and the energy of the surface scattering model, wherein the imaging analysis result is used for identifying grasslands, buildings and oceans.
3. The method of claim 1, wherein the solution results in anomalies, comprising at least one of: A negative value exists in the energy P S of the secondary scattering and/or the energy P D of the surface scattering component; duty cycle of bulk scatter component in coherence matrix Greater than the weight γ, λ 1 、λ 2 、λ 3 is 3 eigenvectors of the polarization coherence matrix T.
4. The method of claim 3, wherein the value of γ is 0.1 to 0.9.
5. The method of claim 1, wherein the cost function satisfies: Lambda (T R ) is the characteristic root of T R .
6. The method of claim 1, wherein the cost function satisfies: trace (T R ) is used to indicate the sum of the feature roots of the residual matrix T R .
7. The method of claim 1, wherein the energy P V ＝4*T 33 of the preliminary computed volume scattering model; The energy P D of the preliminary calculated secondary scattering model and the energy P S of the surface scattering model satisfy: If T 11 -2T 33 >T 22 -T 33 , then a=0, Simultaneously, the energy of secondary scattering and the energy of surface scattering components are calculated preliminarily as respectively P S ＝x 11 +|T 12 | 2 /x 11 ,P D ＝x 22 -|T 12 | 2 /x 11; If T 11 -2T 33 <T 22 -T 33 , then beta=0, Simultaneously, the energy of secondary scattering and the energy of surface scattering components are calculated preliminarily as respectively P S ＝x 11 -|T 12 | 2 /x 22 ,P D ＝x 22 +|T 12 | 2 /x 22 ; Wherein x 11 ＝T 11 -2T 33 ,x 22 ＝T 22 -T 33 .
8. The method of claim 1, wherein the polarization coherence matrix T satisfies T=UCU -1 , C is the polarization covariance matrix.
9. The method of claim 1, wherein the initially calculated polarized SAR target decomposition model satisfies T = P V T V +P D T D +P S T S .
10. An electronic device for performing the method of any one of claims 1 to 9.

Description

Polarization SAR three-component target decomposition method based on model decomposition Technical Field The application relates to the technical field of polarized SAR target classification recognition design, in particular to a polarized SAR three-component target decomposition method based on model decomposition. Background The synthetic aperture radar (SYNTHETIC APERTURE RADAR, SAR) is used as a high-resolution microwave imaging radar, can realize two-dimensional high-resolution imaging of a target, and has the characteristics of all-weather, all-weather and the like. The polarized SAR (PolSAR) can comprehensively record the target scattering characteristic information by utilizing the vector characteristic of electromagnetic waves on the basis of SAR. Compared with the common single-polarization SAR, the polarization SAR system not only can utilize the image power information, but also can quantitatively reflect the target characteristic difference according to the relative phase information among channels. As an important method for extracting information from the PolSAR image, the polarization target decomposition technology is widely applied to the fields of parameter inversion, target classification, polarization calibration and the like, and is an important link of PolSAR application. Although the traditional target decomposition method based on eigenvalue decomposition has clear mathematical background, the method does not directly establish the corresponding relation between eigenvalue eigenvectors and different scattering mechanisms, so the power and other parameters of the scattering mechanisms are difficult to determine. The method for decomposing the polarized target based on the basic scattering model is characterized in that the scattering mechanism of a scatterer is decomposed into basic scattering components such as bulk scattering, secondary scattering and surface scattering, and the scattering mechanism of the target is interpreted by analyzing physical parameters such as energy of each basic scattering component. The decomposition method based on the basic scattering model has definite physical significance and has been widely used. However, in the aspect of model parameter solution, a step-by-step solution method is mostly adopted for polarized target decomposition based on a basic scattering model, and the method prioritizes the bulk scattering component, so that the contribution of the bulk scattering is often easy to overestimate, and the residual secondary scattering and the surface scattering mechanism can be caused to generate negative power conditions contrary to the physical principle. Meanwhile, the step-by-step solving method does not solve the problem of full-parameter power solving of basic scatterers such as bulk scattering and secondary scattering. These all result in the advantages of the PolSAR not being well developed. Disclosure of Invention The application provides an optimal polarization SAR three-component target decomposition method based on model decomposition. The method considers the correlation between the magnitude of the eigenvalue and the energy of the corresponding scattering component, proposes to analyze the eigenvalue of the residual matrix after the target decomposition, and constrains the energy of the residual matrix by minimizing the maximum eigenvalue of the residual matrix or minimizing the trace of the residual matrix. And constructing an optimized mathematical model of semi-positive definite programming (SDP) by restraining the power of each scattering component to carry out full-parameter optimization solution on the PolSAR decomposition, so as to realize the full decomposition of the target. The method takes non-negative basic scattering power as constraint conditions, can enable the decomposition result not to have negative power, and provides technical support for PolSAR accurate target decomposition. In a first aspect, a polarized SAR three-component target decomposition method based on model decomposition is provided, which is characterized by comprising: under a single-station backscattering system, acquiring polarization scattering matrix information and solving a polarization coherence matrix According to imaging scene information, determining a bulk scattering model, a secondary scattering model and a surface scattering model T V、TD、TS, and according to a depolarization coherent matrix T, primarily calculating a solving result of a polarized SAR target decomposition model, wherein the solving result comprises energy P V of the bulk scattering model, energy P D of the secondary scattering model and energy P S of the surface scattering model; if the solving result is not abnormal, outputting the solving result; If the solving result is abnormal, solving a polarized SAR target residual decomposition model through a cost function and a constraint condition, and recalculating the energy of the bulk scattering model, the energy of the sec