CN108872989B - An accurate search method for PS-InSAR based on maximum periodogram - Google Patents

Abstract

The invention discloses a PS-InSAR accurate search method based on a maximum periodogram, relates to the field of signal processing, and provides a PS-InSAR accurate search method based on a maximum periodogram aiming at indexes of scene elevation and deformation rate measurement accuracy of a satellite-borne SAR. The method mainly comprises three steps, namely: carrying out phase filtering on the satellite-borne long-time sequence differential interference image, and screening out permanent scatterer pixels by utilizing a dispersion amplitude criterion; step two: roughly estimating elevation and deformation rate values of the permanent scatterers based on a maximum periodogram criterion; step three: and taking the result of the rough estimation of the maximum periodogram as an initial value, and carrying out unconstrained accurate search on the elevation value and the deformation rate value by utilizing a maximum time sequence correlation coefficient-based criterion.

Description

An accurate search method for PS-InSAR based on maximum periodogram

technical field

The invention belongs to the field of signal processing, in particular to a PS-InSAR precise search method based on a maximum periodogram.

Background technique

Synthetic Aperture Radar Interferometry (InSAR) is a promising surface deformation detection technology. It has the advantages of both synthetic aperture radar technology and interferometry technology. It can measure the three-dimensional information and deformation information of ground targets in a large area and with high precision. However, InSAR technology is very sensitive to factors such as atmospheric errors, remote sensing satellite orbit errors, surface conditions, and spatial and temporal irrelevance. Therefore, in 1999, Italian scholars Ferrettit and Rocca proposed the permanent scatterer technology, which not only got rid of the constraints of spatial de-correlation and temporal de-correlation, but also allowed atmospheric errors to be estimated and corrected due to the availability of a large number of SAR data. PS technology greatly enhances the environmental adaptability and accuracy of interferometry. It is a major technological breakthrough in the field of interferometry and has great potential for practical applications.

Synthetic Aperture Radar Differential Interferometry (PS-InSAR) technology based on permanent scatterers is essentially interferometry. Multiple SAR images in a time series use one of them as the main image. After image registration, the interference forms a time series. Interferometric phase; use orbital and external DEM to remove flat ground and terrain effects to form a time series differential interferometric phase. Among them, the interference phase is mainly composed of the phase caused by the elevation, the phase caused by the terrain deformation, and the winding value of the phase caused by the atmospheric delay. Separating each interference phase component of the PS point from the winding phase is the key to accurately estimate the elevation and deformation. .

Currently, accurate search methods for elevation and deformation rate estimation based on PS-InSAR have received extensive attention. Among them, the more common methods are the unconstrained search method based on the maximum time series correlation coefficient and the maximum periodogram method. The unconstrained search method based on the maximum time series correlation coefficient uses the principle of the quasi-Newton method to find the optimal value of the objective function. This method requires a small amount of calculation and has a fast convergence speed, but the stability is insufficient and is greatly affected by the setting of the initial value. Large; the maximum periodogram method is to perform spectrum transformation on complex signals, transform from t-b space to h-v space, and find the value corresponding to the peak of the spectrum as the optimal value, without the need to set the initial value, and the estimation accuracy is high and robust. The performance is better, but the phase noise has a great influence on the estimation results.

With the widening of the application scope of surface deformation monitoring requirements, the application requirements are increasing, and the accuracy requirements are also increasing. It is necessary to propose a new search method to estimate the elevation difference and deformation rate more accurately. The invention combines the maximum period method and the unconstrained search method based on the maximum time series correlation coefficient, takes the rough estimation result of the maximum period diagram as the initial value, and searches the elevation and deformation rate values unconstrained and precisely based on the maximum time series correlation coefficient criterion. Because the unconstrained search method based on the maximum temporal correlation coefficient can suppress noise, the estimation accuracy of elevation and deformation rate can be further improved.

SUMMARY OF THE INVENTION

The main purpose of the present invention is to solve the above problems, and propose a PS-InSAR precise search method based on the maximum periodogram. The present invention can further improve the estimation accuracy of the elevation and deformation rate of the scene.

The invention provides a PS-InSAR precise search method based on the maximum periodogram, which mainly includes the following steps:

Step 1: Perform phase filtering on the long-time series differential interference image, and filter out the permanent scatterer pixels using the dispersion amplitude criterion;

Step 2: Roughly estimate the elevation and deformation rate of the permanent scatterer based on the maximum periodogram criterion;

Step 3: Take the rough estimation result of the maximum periodogram as the initial value, and accurately search for the elevation and deformation rate values without constraints based on the maximum time series correlation coefficient criterion.

The advantages of the PS-InSAR precise search method based on the maximum periodogram of the present invention are:

(1) Practicality. The PS-InSAR precise search method based on the maximum periodogram proposed by the present invention is carried out on the basis of the existing method, and the realization difficulty is reduced.

(2) Effectiveness. The PS-InSAR precise search method based on the maximum periodogram proposed by the present invention can effectively perform high-precision elevation and deformation rate estimation for strong point targets in the scene.

Description of drawings

Fig. 1 is the flow chart of PS-InSAR precise search method based on maximum periodogram.

Fig. 2 is the SAR image and elevation map of the simulated terrain, wherein Fig. 2(a) is the original SAR image of the simulated terrain, and Fig. 2(b) is the elevation map of the simulated terrain.

Figure 3 is the reconstruction results of three different search methods on the simulated terrain elevation, among which, Figure 3(a) is the unconstrained search method based on the maximum time series correlation coefficient, Figure 3(b) is the maximum periodogram method, Figure 3( c) is the exact search method of PS-InSAR based on the maximum periodogram.

Figure 4 shows the reconstruction results of three different search methods for the simulated terrain deformation rate. Figure 4(a) is the unconstrained search method based on the maximum timing correlation coefficient, Figure 4(b) is the maximum periodogram method, and Figure 4 (c) is the exact search method of PS-InSAR based on the maximum periodogram.

Detailed ways

The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

The present invention is a PS-InSAR precise search method based on the maximum periodogram, and the method flowchart is shown in Figure 1, which specifically includes the following steps:

Step 1: perform registration, differential interferometric operation and differential interferometric phase filtering on M long-time series SAR single-view complex images to obtain processed differential interferometric SAR images. Next, the normalized amplitude variance of each observed image is calculated as:

Among them, M is the total number of long-term sequence SAR single-view complex images, and g _m is the m-th differential interferometric SAR image. The amplitude dispersion index D _A has the following relationship with the normalized amplitude variance μ:

D _A = 1/(1-μ)-1 (1.2)

A certain threshold k is set, and the amplitude dispersion index of a pixel is less than the threshold, then the pixel is selected as a permanent scatterer pixel;

Step 2: PS-InSAR solves the estimation problem of the following objective function,

Among them, M is the total number of long-time series SAR single-view complex images, λ is the wavelength of the electromagnetic wave, g _m is the m-th differential interferometric SAR image, b _{⊥ m} is the vertical baseline, r and θ are the PS point and the main satellite The slant moment and side view angle corresponding to the antenna, t _m is the imaging time interval between the secondary image and the main image, h is the elevation value of the PS point, and v is the deformation rate value of the PS point along the radar line of sight. In order to accurately estimate the above timing correlation Elevation h and deformation rate v when the coefficient γ _p (h, v) is the largest. First, use the maximizing periodogram method to solve the periodogram estimation of each long-time series differential interference image:

和形变速率

作为最大化时序相关系数这一目标函数的初步估计值；After Fourier transform, the complex signal is transformed from tb space to hv space, and then the corresponding elevation at the peak of the spectrum is transformed

and deformation rate

As a preliminary estimate of the objective function of maximizing the time series correlation coefficient;

Step 3: Re-establish the objective function,

f _p (h,v)=-|γ _p (h,v)| ² (1.5)

γ _p (h, v) is the time series correlation coefficient, h is the elevation value of the PS point, and v is the deformation rate value of the PS point along the radar line of sight. The maximization problem of the objective function γ _p is transformed into the minimization of f _p problem, using the quasi-Newton method,

Among them, with d as the search direction, H _i+1 as an approximate matrix of the inverse matrix of the Hesse matrix of f _p , and the initial value as the starting point of the search, call fminunc in Matlab to solve the optimal problem to obtain the exact elevation and deformation rate measurements.

Implementation example

In order to illustrate the effectiveness of the present invention, the verification experiments of the following single control point and simulated terrain are carried out. The simulation parameters of the example are shown in Table 1, and Table 2 provides three different methods (including: no Constrained search method, maximum periodogram method, PS-InSAR accurate search method based on maximum periodogram) comparison results of time-consuming and accuracy of estimating the elevation and deformation rate of a single control point, Table 3 gives the above three methods to estimate the simulated terrain elevation Compared with the relative accuracy of the deformation rate, Fig. 2 shows the SAR image of the simulated terrain and the elevation map of the simulated terrain, and Fig. 3 and Fig. 4 show the reconstruction results respectively.

Table 1 Part of the simulation parameters of the implementation example

Using the parameters in Table 1, first randomly set the elevation and deformation rate of a single control point, and use different methods to estimate it. The time-consuming, elevation accuracy, deformation rate accuracy, and average objective function value of a thousand cycles are as follows: As shown in Table 2, it can be seen from the comparison that although the PS-InSAR precise search method based on the maximum periodogram takes a little longer, the estimation accuracy and the average objective function value are the best. The elevation and deformation rate of the terrain are estimated, and further simulation verification is performed. The deformation rate is set to 0. The PS-InSAR precise search method based on the maximum periodogram is also optimal for the relative accuracy of the simulated terrain elevation and deformation rate estimation, and The elevation and deformation rate of the simulated terrain are reconstructed. From the elevation reconstruction results, it can be seen that the reconstruction result of the unconstrained search method based on the maximum time series correlation coefficient is the worst, and no obvious terrain fluctuations can be seen; the maximum periodogram method It is close and optimal to the reconstructed image of the elevation by the PS-InSAR precise search method based on the maximum periodogram; it can be seen from the deformation rate reconstruction results that the reconstruction result of the unconstrained search method based on the maximum time series correlation coefficient is also the best. Only the estimation accuracy of the deformation rate of the terrain bar edge is better; the maximum periodogram method and the PS-InSAR accurate search method based on the maximum periodogram can accurately estimate the deformation rate of the entire simulated terrain; the above simulation results effectively prove the present invention The validity and practicability of the PS-InSAR precise search method based on the maximum periodogram can effectively improve the estimation accuracy of the elevation and deformation rate of the permanent scatterer.

Table 2 Comparison results of time-consuming and accuracy of three different methods for estimating control point elevation and deformation rate

Table 3 Comparison results of three different methods for estimating the relative accuracy of simulated terrain elevation and deformation rate

Claims (2)

1. A PS-InSAR accurate search method based on a maximum periodogram is characterized in that: the method comprises the following steps:

the method comprises the following steps: carrying out phase filtering on the long-time sequence differential interference image, and screening out permanent scatterer pixels by utilizing a dispersion amplitude criterion;

step two: roughly estimating elevation and deformation rate values of the permanent scatterers based on a maximum periodogram criterion; the method specifically comprises the following steps: PS-InSAR solves the estimation problem of the following target functions,

wherein M is the total amplitude of the long-time sequence SAR single-vision complex image, lambda is the wavelength of the electromagnetic wave, g_mFor the mth differential interference SAR image, b_⊥mIs a vertical baseline, r and θ are the skew moment and side view angle of the PS point corresponding to the primary satellite antenna, t_mFor the imaging time interval of the auxiliary image and the main image, h is the elevation value of the PS point, v is the deformation speed value of the PS point along the radar sight line direction, and in order to accurately estimate the time sequence correlation coefficient gamma_p(h, v) the maximum elevation h and deformation rate v, firstly, solving the periodic map estimation of each long time sequence differential interference image by using a maximum periodogram method:

transforming the complex signal from t-b space to h-v space through Fourier transform, and then transforming the corresponding elevation at the peak of the frequency spectrum

And rate of deformation

A preliminary estimate as an objective function of maximizing the time-series correlation coefficient;

step three: taking the result of the rough estimation of the maximum periodogram as an initial value, and based on the maximum time sequence correlation coefficient criterion, performing unconstrained accurate search on elevation and deformation rate values, wherein the method specifically comprises the following steps:

the objective function is re-established and,

f_p(h,v)＝-|γ_p(h,v)|²(1.5)

γ_p(h, v) are time sequence correlation coefficients, h is an elevation value of the PS point, v is a deformation rate value of the PS point along the radar sight line direction, and an objective function gamma is calculated_pThe maximization problem of (1) is converted into_pThe minimization problem of (2) is solved by adopting a quasi-Newton method,

wherein d is the search direction, H_i+1Is f_pThe initial value is used as the starting point of searching, and fminuc in Matlab is called to solve the optimal problem so as to obtain accurate elevation and deformation rate measurement values.

2. The PS-InSAR accurate search method based on the maximum periodogram as claimed in claim 1, wherein: the first step is specifically as follows: carrying out registration, differential interference operation and differential interference phase filtering on the M long-time sequence SAR single-vision complex images to obtain processed differential interference SAR images, and then calculating the normalized amplitude variance mu of each observation image as follows:

wherein M is the total amplitude of the long-time sequence SAR single-vision complex image, g_mFor the m-th differential interference SAR image, the amplitude dispersion index D_AThe following relationship exists with the normalized amplitude variance μ:

D_A＝1/(1-μ)-1 (1.2)

and setting a certain threshold k, and selecting a pixel as a permanent scatterer pixel if the amplitude dispersion index of the pixel is smaller than the threshold.

Images