CN117826148A - Method and system for identifying coherent point - Google Patents

Abstract

The present invention discloses a method and system for identifying coherent points, belonging to the field of radio measurement technology. The method comprises: obtaining image data of synthetic aperture radar; judging whether the target point of the image data satisfies the following conditions: the amplitude dispersion index of the time series is less than a first threshold, the autocorrelation coefficient is greater than a second threshold, and the average coherence coefficient is greater than a third threshold; if so, the target point is a coherent point. Using multiple detection indicators to identify coherent points can reduce the influence of temporal and spatial incoherence, eliminate non-permanent scatterer targets with serious incoherence, improve the density of coherent points and monitoring accuracy, and improve the accuracy of sedimentation monitoring.

Description

A method and system for identifying coherent points

Technical Field

The invention relates to the technical field of radio measurement, and in particular to a method and system for identifying coherent points.

Background technique

Coherent points, also called Persistent Scatterers (PS), are point targets (single pixels or a group of pixels) that are radiometrically stable in time. Such point targets have the characteristics of strong reflection (high backscattering) and high coherence during the observation period. Radiometrically stable targets are urban infrastructure (such as buildings, bridges, dams, greenhouses, metal buildings, etc.) or natural objects (such as exposed rocks, etc.).

As the operating mileage of subways increases, the scale of newly built subways decreases. Gradually, the subway monitoring market has shifted from monitoring during the construction period to monitoring during the subway operation and maintenance period. However, the deformation characteristics of subway infrastructure (elevated sections and roadbed sections) during the operation and maintenance period are relatively special. It is difficult to identify and extract coherent point targets of subway infrastructure within the urban built-up area only by statistically analyzing the amplitude information of the image. Traditional PS-InSAR coherent point identification usually adopts a single permanent scatterer identification method, such as the coherence coefficient identification method, but the error rate of a single permanent scatterer identification method is high; and coherent point identification is a prerequisite for settlement monitoring. Therefore, based on the traditional PS-InSAR coherent point identification method, some coherent points are difficult to identify, and the accuracy of settlement monitoring along the rail transit line needs to be improved urgently.

Summary of the invention

In view of the above technical problems existing in the prior art, the present invention provides a method and system for identifying coherent points, which adopts a multi-detection index approach to improve the monitoring accuracy of coherent points.

The invention discloses a method for coherent point identification, comprising the following steps: obtaining image data of a synthetic aperture radar; judging whether a target point of the image data satisfies the following conditions: an amplitude dispersion index of a time series is less than a first threshold, an autocorrelation coefficient is greater than a second threshold, and an average coherence coefficient is greater than a third threshold; if so, the target point is a coherent point.

Preferably, the amplitude dispersion index is expressed as:

D＝σ/μ (1)

Where D represents the amplitude dispersion index of the target point time series, σ is the standard deviation of the target point amplitude time series, and μ is the mean of the target point amplitude time series.

Preferably, the average coherence coefficient is expressed as:

γ _k represents the coherence coefficient of the x-th pixel in the k-th interference pair, M(i,j) and S(i,j) are the master and slave images constituting the k-th interference pair, * represents the conjugate multiplication, m and n are the horizontal and vertical sizes of the sliding window, respectively. is the average coherence coefficient of the target point, and N is the logarithm of the adjacent interference pairs of the coherent points.

Preferably, the coherence coefficient is used to eliminate non-permanent scatterer targets with severe decorrelation.

Preferably, the image data includes a time series after registration processing;

The amplitude dispersion index is used to select permanent scatterers whose variation in the time series is less than a first threshold value, so as to reduce the influence of temporal decoherence;

The autocorrelation coefficient is used to reduce the effect of spatial incoherence.

Preferably, the method for obtaining the autocorrelation coefficient comprises:

Preprocess and register the SAR images in sequence;

Perform Fourier transform on the SAR images after image registration to obtain their representation in the frequency domain;

The amplitude signal time series is extracted from the frequency domain representation, and the autocorrelation coefficient of the amplitude signal time series is calculated.

Preferably, the autocorrelation coefficient is calculated as follows:

Y(t,s)=E( _Xt - _μt )( _Xs - _μs ) (5)

Among them, ρ(t,s) represents the autocorrelation coefficient of the amplitude signal time series, t and s represent the moments of the time series respectively, _Xt represents the phase at moment t, _μt represents the mean of the phase at moment t, D() represents the variance, E() represents the mathematical expectation, and Y(t,s) represents the autocovariance function.

Preferably, the target point is used to detect ground subsidence, and the method for detecting ground subsidence includes:

After removing the phase offset and atmospheric effect in the interference pattern of the target point, the deformation rate of the target point is calculated;

The ground settlement is calculated from the vertical component of the deformation rate.

The present invention also provides a system for implementing the above method, comprising an acquisition module and an identification module, wherein the acquisition module is used to obtain image data of a synthetic aperture radar; the identification module is used to determine whether a target point of the image data satisfies the following conditions: an amplitude dispersion index of a time series is less than a first threshold, an autocorrelation coefficient is greater than a second threshold, and a coherence coefficient is greater than a third threshold; if so, the target point is a coherent point.

Preferably, the system further comprises a settlement analysis module, which is used to calculate the deformation rate of the target point after removing the phase offset and atmospheric effect in the interference diagram of the target point; and calculate the ground settlement through the vertical component of the deformation rate.

Compared with the prior art, the beneficial effects of the present invention are as follows: the use of multiple detection indicators to identify coherent points can reduce the impact of temporal incoherence and spatial incoherence, eliminate non-permanent scatterer targets with severe incoherence, increase the density of coherent points and monitoring accuracy, and improve the accuracy of sedimentation monitoring.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a flow chart of a method for coherent point identification according to the present invention;

FIG. 2 is a system logic block diagram of the present invention.

Detailed ways

In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

The present invention is further described in detail below in conjunction with the accompanying drawings:

A method for identifying a coherent point, as shown in FIG1 , comprises the following steps:

Step 101: Obtain image data of synthetic aperture radar, which is also called SAR image. The image data includes a time series after registration processing, and the registration processing is a prior art and will not be described in detail in the present invention.

Step 102: Determine whether the target point of the image data meets the first condition: the amplitude dispersion index of the time series is less than the first threshold, the autocorrelation coefficient is greater than the second threshold, and the average coherence coefficient is greater than the third threshold;

If satisfied, step 103 is executed: the target point is a coherent point.

If not satisfied, the target point is not a coherent point.

The use of multiple detection indicators can improve the density of coherent points and monitoring accuracy, and improve the accuracy of settlement monitoring. Among them, the amplitude dispersion index threshold method and the autocorrelation coefficient threshold method can improve the density of coherent points, and the coherence coefficient threshold method can further improve the monitoring accuracy.

The amplitude dispersion index is used to select permanent scatterers whose time series changes are less than the first threshold T1, and select target points with small time series changes to reduce the impact of temporal decoherence. Permanent scatterers with high credibility are selected based on the statistical relationship between amplitude dispersion and phase deviation under large data volume SAR images.

The amplitude dispersion index can be expressed as:

D＝σ/μ (1)

Among them, D represents the amplitude dispersion index of the target point time series, σ is the standard deviation of the target point amplitude time series, and μ is the mean of the target point amplitude time series. That is, for the time series SAR data after registration processing, the variance and mean of the same target point on the time series multi-images are calculated, and the ratio of the two is used as the measure of the permanent scatterer identification method, and the target point less than the first threshold is selected as a permanent scatterer, that is, D<T1.

The autocorrelation coefficient is used to reduce the impact of spatial incoherence, also known as point target detection method. Through the analysis of the principle of permanent scatterers, it can be seen that the corner reflector effect has high brightness. The target point whose echo signal remains constant over time and whose phase characteristics are also stable can be identified as a candidate permanent scatterer. The time series autocorrelation coefficient between the amplitude signals can be obtained by processing the SAR image, and a certain average spectral coherence can be set to screen out high-quality point targets. The specific steps are as follows:

Step 301: Preprocess the SAR image (such as radiation correction, atmospheric correction and terrain correction, etc.), and perform image registration on the time series SAR image to obtain a time series interferometric pair. Preprocessing can improve the quality and comparability of the image; image registration can ensure that they are in the same geographic coordinates, which is a prerequisite for calculating spectral correlation.

Step 302: Perform Fourier transform on the registered SAR images to obtain their representation in the frequency domain for further spectrum analysis and processing. SAR images record the scattered signals of radar waves on the ground objects, and the amplitude and phase information of these signals actually contain the frequency domain characteristics of the ground objects.

Step 303: Extract the amplitude signal time series from the frequency domain representation and calculate the autocorrelation coefficient of the amplitude signal time series. The autocorrelation coefficient is used to calculate the temporal autocorrelation between the amplitude signals and measure the correlation between the amplitude information (backscattered signal) of the ground objects at different SAR image time phases. The autocorrelation coefficient reflects the spectral correlation. This correlation coefficient calculation method has a wide range of applications in signal processing, data analysis and other fields.

The autocorrelation coefficient is calculated as:

Y(t,s)=E( _Xt - _μt )( _Xs - _μs ) (5)

Among them, ρ(t,s) represents the autocorrelation coefficient of the amplitude signal time series, t and s represent different moments of the time series, _Xt represents the phase at time t, _μt represents the mean of the phase at time t, D() represents the variance, E() represents the mathematical expectation, and Y(t,s) represents the autocovariance function. The amplitude signal time series is expressed as: { _Xt , t∈T}.

The average coherence coefficient is used to eliminate non-permanent scatterer targets with serious decorrelation. This ensures that high-density, high-quality permanent scatterers are finally obtained. After statistical analysis, it can be considered that points with high spatial coherence on SAR images correspond to high coherence areas such as buildings on the ground. Therefore, the average coherence coefficient can be further calculated to select point targets with higher coherence coefficients. Coherence is the most intuitive standard for measuring interferometric phase noise. The coherence coefficient method estimates the average coherence coefficient based on the values of the adjacent pixels around the target pixel.

The average coherence coefficient is expressed as:

γ _k represents the coherence coefficient of the target point in the kth interference pair, M(i,j) and S(i,j) are the master and slave images constituting the kth interference pair, * represents the conjugate multiplication, m and n are the horizontal and vertical sizes of the sliding window, respectively. is the average coherence coefficient of the target point, and N is the logarithm of the adjacent interference pairs of the coherent points. The coherence coefficients of N interference pairs are expressed as: (γ ₁ ,γ ₂ ,…,γ _N )

At surface objects such as vegetation or water bodies, different scattering mechanisms may cause large changes in the phase of radar waves, resulting in decoherence. Multiple scattering sources such as leaves and branches of vegetation cause multiple scattering in different directions, resulting in inconsistent phases of radar waves; secondly, the growth state and structural changes of vegetation will also cause phase changes of radar waves. The main reason for water body decoherence is due to factors such as multiple reflections on the surface. There are phase differences between radar waves from different directions and positions, resulting in reduced coherence in the water area. The average coherence coefficient can quickly identify and eliminate these decoherent data.

The target point is used to detect ground subsidence, and the method for detecting ground subsidence includes:

Step 201: After removing the phase offset and atmospheric effect in the interference graph of the target point, an interference graph is obtained, and the deformation rate of the target point is calculated through the interference graph. Removing the phase offset and atmospheric effect in the interference graph of the target point is a prior art and will not be described in detail in this application.

Step 202: Calculate the ground settlement by the vertical component of the deformation rate.

By using the time series amplitude discrete index threshold method and point target detection method to select permanent scatterer candidate points that are less affected by temporal and spatial incoherence, the density of permanent scatterers in a single monitoring method is greatly improved. On this basis, the coherence coefficient threshold method is used to optimize the recognition of coherent point targets, effectively eliminate the misidentified coherent points, and improve the reliability of the traditional coherent point recognition method. Finally, the terrain phase on these target points is separated to monitor ground subsidence and achieve high-quality monitoring during the subway operation and maintenance period. Points that meet the first threshold, the second threshold, and the third threshold are selected as coherent points (PS points). This multiple information selection method fully considers the various characteristics of coherent points.

A coherence coefficient of 0.8 or more indicates high coherence, and an amplitude dispersion index of 0.15 or less indicates relatively stable amplitude information. The spectral autocorrelation coefficient of man-made structures such as buildings is generally above 0.6. Therefore, in a specific embodiment, the first threshold value T1 is 0.1, the second threshold value T2 is 0.8, and the third threshold value T3 is 0.9, but is not limited thereto.

The method of the present invention can identify radiostable points in time. Level point monitoring data and other field data can also provide references for the identification of coherent points. These point targets have the characteristics of strong reflection (high backscattering) and high coherence during the observation period. Once these points are determined as stable PS candidate points, the displacement model is used to remove the phase offset and atmospheric effects from the interferogram after flattening to obtain the final deformation rate of each pixel.

The present invention also provides a system for implementing the above method, as shown in Figure 2, comprising: an acquisition module 1 and an identification module 2, wherein the acquisition module 1 is used to obtain image data of a synthetic aperture radar; the identification module 2 is used to determine whether a target point of the image data satisfies the following conditions: an amplitude dispersion index of a time series is less than a first threshold, an autocorrelation coefficient is greater than a second threshold, and a coherence coefficient is greater than a third threshold; if so, the target point is a coherent point.

The system further comprises a settlement analysis module 3, which is used to calculate the deformation rate of the target point after removing the phase offset and atmospheric effect in the interference diagram of the target point; and calculate the ground settlement through the vertical component of the deformation rate.

The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims (10)

1. A method of coherent point identification, comprising the steps of:

obtaining image data of a synthetic aperture radar;

judging whether the target point of the image data meets the following conditions: the amplitude dispersion index of the time series is smaller than a first threshold, the autocorrelation coefficient is larger than a second threshold, and the average coherence coefficient is larger than a third threshold;

if yes, the target point is a coherent point.

2. The method of claim 1, wherein the amplitude dispersion index is expressed as:

D＝σ/μ (1)

wherein D represents an amplitude discrete index of the target point time series, sigma is a standard deviation of the target point amplitude time series, and mu is a mean value of the target point amplitude time series.

3. The method of claim 1, wherein the average coherence coefficient is expressed as:

γ _k expressed as the coherence coefficient of the target point under the kth interference pair, M (i, j) and S (i, j) are the master-slave images constituting the kth interference pair, respectively, expressed as performing conjugate multiplication, M and n are the sizes of the sliding window in the transverse and longitudinal directions, respectively,n is the average coherence coefficient of the target point, and is expressed as the logarithm of the adjacent interference pair of the coherence point.

4. A method according to claim 3, wherein the coherence coefficient is used to reject non-permanent scatterer targets that are severely out of correlation.

5. The method of claim 1, wherein the image data comprises a time series after registration processing;

the amplitude dispersion index is used for selecting a permanent scatterer with a change in time sequence smaller than a first threshold value so as to reduce the influence of time incoherence;

the autocorrelation coefficients are used to reduce the effects of spatial incoherence.

6. The method of claim 1, wherein the method of obtaining the autocorrelation coefficients comprises:

preprocessing image data and registering images in sequence;

performing Fourier transform on the image data after image registration to obtain the representation of the image data in a frequency domain;

an amplitude signal time series is extracted from the frequency domain representation and an autocorrelation coefficient of the amplitude signal time series is calculated.

7. The method of claim 6, wherein the autocorrelation coefficients are calculated by:

Y(t，s)＝E(X _t -μ _t )(X _s -μ _s ) (5)

wherein ρ (t, s) is expressed as an autocorrelation coefficient of the time series of amplitude signals, t and s respectively represent the moments of the time series, X _t Phase denoted as time phase at time t, μ _t Expressed as the mean of the time phases at time t, D () represents the taking of the variance, E () represents the mathematical expectation, and Y (t, s) represents the autocovariance function.

8. The method of claim 1, wherein the target point is for detecting ground subsidence, the method of detecting ground subsidence comprising:

after removing phase shift and atmospheric effect in the target point interferogram, calculating the deformation rate of the target point;

the ground subsidence is calculated from the vertical component of the deformation rate.

9. A system for coherence point identification for implementing the method of any of claims 1-8, the system comprising: the acquisition module and the identification module are used for acquiring the data of the data,

the acquisition module is used for acquiring image data of the synthetic aperture radar;

the identification module is used for judging whether the target point of the image data meets the following conditions: the amplitude dispersion index of the time series is smaller than a first threshold, the autocorrelation coefficient is larger than a second threshold, and the coherence coefficient is larger than a third threshold; if yes, the target point is a coherent point.

10. The system of claim 9, further comprising a dip analysis module for calculating a deformation rate of the target point after removing phase shifts and atmospheric effects in the target point interferogram; the ground subsidence is calculated from the vertical component of the deformation rate.