CN113126051B

CN113126051B - Airborne multichannel SAR interference effective baseline estimation method and device

Info

Publication number: CN113126051B
Application number: CN202110245580.0A
Authority: CN
Inventors: 傅东宁; 廖桂生; 黄岩; 许京伟; 李婕
Original assignee: Xidian University
Current assignee: Xidian University
Priority date: 2021-03-05
Filing date: 2021-03-05
Publication date: 2023-09-19
Anticipated expiration: 2041-03-05
Also published as: CN113126051A

Abstract

The invention provides an airborne multichannel SAR interference effective baseline estimation method and device, which are used for correcting phase deviation of phase interferograms of adjacent channels caused by track-crossing motion by a two-dimensional method and compensating the phase deviation of the track-crossing motion; and then, detecting a moving target of the corrected phase interference diagram to obtain a distance direction resolution unit and an azimuth direction resolution unit of the moving target in the corrected interference phase diagram, and further performing pulse compression and mapping by using a corrected average median filtering method to obtain an accurate focusing image, so that the estimation precision of the effective baseline of a distance compression domain and the equivalent speed of the carrier relative to the moving target along the track is improved, and the effective baseline and the equivalent speed along the track with higher precision are obtained. And estimating the high-precision target inclined plane distance velocity according to the higher-precision effective baseline and the track equivalent velocity, and finally repositioning the moving target in the interference phase map according to the high-precision target inclined plane distance velocity. Therefore, the invention can improve the accuracy of the positioning of the moving target, thereby improving the imaging effect of the focused image in the radar system.

Description

Airborne multichannel SAR interference effective baseline estimation method and device

Technical Field

The invention belongs to the technical field of radar signal processing, and particularly relates to an airborne multichannel SAR interference effective baseline estimation method and device.

Background

The synthetic aperture radar ground moving target detection (Synthetic Aperture Radar Ground Moving Target Indication, SAR-GMTI) technique is capable of effectively imaging and locating moving targets in synthetic aperture radar (Synthetic Aperture Radar, SAR) images. However, the performance of SAR-GMTI techniques depends on the motion parameters of moving targets, as range-oriented velocities of the targets can cause range-migration and azimuth-offset in SAR images, while azimuth-oriented velocities can obscure focused targets. Thus, accurate estimation of motion parameters is critical for imaging and positioning of moving objects.

Typically, single channel SAR systems generate multiple images with different doppler channel fields of view or different ground moving object detection and parameter estimation. For a single-channel SAR system, a search algorithm with larger calculation amount such as Radon transformation, fractional Fourier transformation or polynomial Fourier transformation is generally required to estimate the motion parameters of a target, and in order to reduce the calculation complexity and obtain the robust performance, the multi-channel SAR system has become an important tool for realizing the ground motion target detection function due to higher system freedom degree and high resolution. Meanwhile, for a multichannel SAR system based on Along-track interference (Along-Track Interferometric, ATI), the motion parameters of the target can be obtained through the interference phase, so that the calculated amount can be greatly reduced, and the parameter estimation precision can be improved. However, due to the possible yaw conditions of the on-board system, the phase center of the multi-channel SAR system may deviate from the intended motion profile. In this case, the interference phase depends on the effective baseline length, which is defined as the projection of the physical baseline onto the platform movement trajectory. Thus, in a multi-channel SAR system, accurate estimation of the effective baseline is critical for focused imaging and repositioning of moving objects.

In practical applications, it is difficult to estimate the effective baseline directly from the received echo signals. Over the last decades, researchers have made a great deal of effort on how to accurately estimate the effective baseline. The conventional median filtering method is generally adopted to compensate the motion of the radial component of the radar, and the interference of clutter and noise signals to the interference phase of the target echo signals causes the oscillation of an interference phase estimation curve, so that the effect is poor under the condition of low signal-to-noise ratio (Signal to Clutter and Noise Ratio, SCNR), which seriously affects the performance of effective baseline and motion parameter estimation. Meanwhile, in practical situations, the carrier usually generates motion errors in two dimensions of along-track (along track) and across-track (yaw) at the same time, so that a relatively complex baseline deviation is caused, and the accurate estimation of the traditional effective baseline is an estimation method based on the condition that only ideal vertical baseline deviation is considered to be generated. This further reduces the performance of the conventional method.

Therefore, how to accurately estimate the effective baseline and complete the focused imaging and repositioning of the moving target under the condition of actual yaw of the carrier, especially under the condition of low SCNR, becomes a key problem for realizing the detection of the moving target when the airborne multichannel SAR system works in the mode of Along-track interference synthetic aperture (Along-Track InterferometricSynthetic Aperture Radar, ATI-SAR).

Disclosure of Invention

In order to solve the problems in the prior art, the invention provides an airborne multichannel SAR interference effective baseline estimation method and device. The technical problems to be solved by the invention are realized by the following technical scheme:

in a first aspect, the invention provides a method for estimating an effective baseline of airborne multichannel SAR interference, which comprises the following steps:

acquiring a coarse focusing image mapped by echo signals of adjacent channels from a radar system;

the coarse focusing image is an image which is mapped by carrying out azimuth pulse compression on the data of the echo signal subjected to the distance pulse compression by an original azimuth matched filter;

determining an interference phase map of adjacent channels based on the focused image;

determining a main frequency band phase interference pattern of the adjacent channel by using a band-pass filter aiming at the interference phase pattern of the adjacent channel;

the phase offset of the interference phase map of the main frequency band caused by the track-crossing motion error is calibrated by adopting a two-dimensional correction method, and a corrected interference phase map is obtained;

detecting a moving target in the corrected interference phase diagram to obtain a distance direction resolution unit and an azimuth direction resolution unit of the moving target in the corrected interference phase diagram;

Based on the distance direction resolution unit and the azimuth direction resolution unit where the moving target is located, a first interference phase of the distance direction resolution unit where the moving target is located, a first transformation rate of the first interference phase, a second interference phase of the distance direction resolution unit where the strong clutter is located and a second change rate of the second interference phase are obtained through calculation;

estimating the equivalent speed and the effective baseline of the carrier relative moving target along the track based on the first change rate and the second change rate until the equivalent speed and the effective baseline along the track are accurately estimated;

modifying the original azimuth matched filter by using the equivalent speed along the track to obtain a modified azimuth matched filter;

carrying out azimuth pulse compression and mapping on the data of the echo signal subjected to the distance pulse compression by using a corrected azimuth matched filter to obtain a precise focusing image;

extracting a moving target interference phase from the precisely focused image;

estimating the inclined plane distance velocity of the moving target by using the target interference phase, the precisely estimated effective baseline and the precisely estimated equivalent velocity along the track to obtain the precisely estimated inclined plane distance velocity;

According to the precisely estimated inclined plane distance-to-speed, calculating to obtain the azimuth resolution unit offset of the moving target;

and repositioning the moving target in the precisely focused image according to the azimuth resolution unit offset.

In a second aspect, the present invention provides an airborne multichannel SAR interference effective baseline estimation device, including:

the acquisition module is used for acquiring a coarse focusing image mapped by echo signals of adjacent channels from the radar system;

the coarse focusing image is an image which is obtained by compressing and mapping the echo signal data by the original azimuth matching filter after compressing the distance pulse;

a first determining module, configured to determine an interference phase map of adjacent channels based on the focused image;

the second determining module is used for determining a main frequency band phase interference pattern of the adjacent channel by using a band-pass filter aiming at the interference phase pattern of the adjacent channel;

the correction module is used for correcting the phase offset of the interference phase map caused by the track crossing motion error by adopting a two-dimensional correction method to obtain a corrected interference phase map;

the detection module is used for detecting the moving target in the corrected interference phase diagram to obtain a distance direction resolution unit and an azimuth direction resolution unit where the moving target is located in the corrected interference phase diagram;

The first calculation module is used for calculating and obtaining a first interference phase of the distance resolution unit where the moving target is located, a first transformation rate of the first interference phase, a second interference phase of the distance resolution unit where the strong clutter is located and a second change rate of the second interference phase based on the distance resolution unit where the moving target is located and the azimuth resolution unit;

the estimating module is used for estimating the equivalent speed and the effective baseline of the relative moving target of the carrier along the track based on the first change rate and the second change rate until the equivalent speed and the effective baseline along the track are accurately estimated;

the modification module is used for modifying the original azimuth matched filter by using the equivalent speed along the track to obtain a modified azimuth matched filter;

the compression module is used for carrying out azimuth pulse compression and mapping on the data of the echo signals subjected to the distance pulse compression by using the corrected azimuth matched filter to obtain a precise focusing image;

the extraction module is used for extracting the interference phase of the moving target from the precisely focused image;

the speed correction module is used for estimating the inclined plane distance velocity of the moving target by using the target interference phase, the precisely estimated effective baseline and the precisely estimated equivalent speed along the flight path to obtain the precisely estimated inclined plane distance velocity;

The second calculation module is used for calculating and obtaining the azimuth resolution unit offset of the moving target according to the precisely estimated inclined plane distance-to-speed;

and the positioning module is used for repositioning the moving target in the precisely focused image according to the azimuth resolution unit offset.

Drawings

Fig. 1 is a schematic flow chart of an airborne multichannel SAR interference effective baseline estimation method provided by an embodiment of the present invention;

FIG. 2 is a schematic diagram of the spatial geometry of a multi-channel SAR system in yaw situation provided by the present disclosure;

FIG. 3 is a graph showing the effect of comparing the true interference phase of a signal containing target and clutter with that of a signal containing only the target;

FIG. 4 is a graph showing the effect of comparing the true interference phase of a target-containing & clutter signal with the true interference phase of a target-only signal provided by the present invention;

FIG. 5 is a plot of interference phase and fit after passing through an average median filter (L=1000, l=200) for various SCNR provided by the present invention;

FIG. 6 is a graph comparing average median filtering with median filtering applied to measured SAR data in accordance with the present subject matter;

FIG. 7 is a root mean square error comparison plot of the effective baseline estimate of the method and median filter provided by the present invention;

FIG. 8 is a coarse focus dual channel image provided by the present invention;

FIG. 9 is a diagram of an interference phase provided by the present invention;

FIG. 10 is a graph of SAR-GMTI detection after cross track motion error correction provided by the present invention;

FIG. 11 is a distance-only pulse compressed image of an illuminated scene provided by the present invention;

FIG. 12 is an interference phase obtained after distance pulse compression provided by the present invention;

FIG. 13 is a precisely focused moving target image provided by the present invention;

FIG. 14 is an effect diagram of the repositioning of a moving object provided by the present invention;

fig. 15 is a block diagram of an airborne multichannel SAR interference effective baseline estimation device provided by the present invention.

Detailed Description

The present invention will be described in further detail with reference to specific examples, but embodiments of the present invention are not limited thereto.

Example 1

As shown in fig. 1, the method for estimating the airborne multichannel SAR interference effective baseline provided by the invention comprises the following steps:

s1, acquiring a coarse focusing image mapped by echo signals of adjacent channels from a radar system;

the coarse focusing image is an image which is obtained by compressing the echo signal data by the original azimuth matching filter through distance pulse and then mapping the azimuth pulse;

referring to fig. 2, subgraph a-subgraph b of fig. 2 shows a schematic diagram of the operation of the airborne multi-channel radar in the dual-channel SAR mode under front side view conditions. The radar transmits a signal and is then received by two independent array elements. At a synthetic aperture time T _a In the radar, the radar transmits a linear frequency modulation signal, and the transmitted kth radio frequency pulse signal can be expressed as:

In sub-figure a the physical baseline length is the distance between the phase centers of the two physical aperture antennas (solid line elements) and the effective baseline length is the projection of the physical baseline length onto the motion trajectory, i.e. the distance between the two projected elements (dashed line elements). Baseline shifts occur when the vehicle is yawing. At time T _a In the method, the baseband echo pulse of the moving target P of the kth transmitting pulse received by the channel n and the baseband echo pulse expression of the ground clutter scattering bin i are respectively as follows

The slant-distance expressions given by the Taylor expansion and according to the space geometrical relation are respectively that

Wherein Δyz _n Is the track-crossing (yaw) error of channel n on the target pitch, expressed as

Δyz _n ≈Δy·sinφ+Δz·cosφ

The present invention uses a range-doppler algorithm for the focused imaging process. Due to track-crossing (track) motion error deltayz _n Far less than the pitch of the vehicle to the target and scene, therefore Δyz _n The effect of the term on the distance-to-pulse compression can be ignored. In the imaging process, a first-order wedge stone transformation method is adopted for compensating the range migration. Thus, the distance-wise pulse compressed signal of the moving object P and the echo of the clutter scattering bin i can be expressed as:

Symbol A _rp And A _rc The equivalent gain coefficients of the target echo and clutter scattering surface element echo signals of each channel after distance compression are respectively represented, and the characteristics of all channels are assumed to be the same. Setting R _0,n ＝R ₀ +Δyz _n ，R _c0,n ＝R _c0 +Δyz _n The equivalent nearest slope distances of the moving object P and the clutter scattering bin with respect to the channel n are shown, respectively. The distance pulse compression signal of the moving target echo is processed by adopting fast Fourier transform, and the envelope movement and the small amplitude change of the window function are ignored, so that the frequency spectrum function of the distance pulse compression signal can be obtained:

for the target and clutter scattering bin in the same range-resolution unit there is R _c0 ＝R ₀ For acquiring a coarse focus image, it is assumed that the target moves much slower than the carrier, i.e. v _rx ≈V _a . The matched filter for performing azimuth pulse compression processing on the data of the moving target echo subjected to distance pulse compression can be designed as follows:

after azimuth pulse compression processing is completed on the data of the moving target echo subjected to distance pulse compression, a coarse focusing imaging result of the moving target can be obtained:

wherein:

A _ap the convolution operation is represented by the gain of the target signal after the azimuth compression for each channel (assuming that all the channel gains are the same). The same matched filter is adopted, the clutter scattering surface element echo signals of all channels can be subjected to approximate accurate focusing imaging processing, and the expression is that

A _ac The gain of the echo of the clutter is scattered for the clutter after the azimuth pulse compression. The above result is approximately accurate for a relatively stationary clutter scattering bin. However, the moving object is blurred in the coarse focus imaging result, and the position of the present image is inaccurate or even erroneous. Thus, the track-crossing motion error coarsely focuses the imaging junction of the moving target which is inherently blurred and inaccurate or wrong in positionThe effect of the fruit is not significant. However, when the airborne multichannel SAR system works in an ATI-SAR mode, a relatively accurate effective baseline parameter needs to be obtained, so that the movement speed of the target can be estimated relatively accurately, and the fine focusing imaging and repositioning processing of the target are finally completed, so that a relatively clear and relatively accurate image of the target position is obtained. At this time, the track-crossing motion error can seriously affect the estimation accuracy of the effective baseline, so that the estimation accuracy of the target motion parameter is reduced, even the estimation error is caused, and the performance of the radar on the fine focusing imaging and repositioning processing of the target is seriously affected. Thus, for an on-board multi-channel SAR system, accurate estimation of the effective baseline is critical to the fine focus imaging and repositioning process of the moving target.

Since the invention is based on the working principle of the airborne multichannel SAR system, the proposed parameters of the invention are introduced in the process of explanation, and therefore, the meaning of the parameters appearing in the invention is explained first, see table 1.

Table 1 parameter definition table

S2, determining an interference phase diagram of an adjacent channel based on the focusing image;

s3, determining a main frequency band phase interference pattern of the adjacent channel by using a band-pass filter aiming at the interference phase pattern of the adjacent channel;

s4, calibrating phase offset of an interference phase map caused by the track crossing motion error by adopting a two-dimensional correction method to obtain a corrected interference phase map;

s5, detecting a moving target in the corrected interference phase diagram to obtain a distance direction resolution unit and an azimuth direction resolution unit of the moving target in the corrected interference phase diagram;

s6, calculating to obtain a first interference phase of the distance-direction resolution unit where the moving target is located, a first transformation rate of the first interference phase, a second interference phase of the distance-direction resolution unit where the strong clutter is located and a second change rate of the second interference phase based on the distance-direction resolution unit where the moving target is located and the azimuth-direction resolution unit;

as an optional manner of the present invention, the step of S6 includes:

Step a: based on a distance direction resolution unit and an azimuth direction resolution unit where the moving target is located, calculating a first interference phase of the distance direction resolution unit where the moving target is located by using a preset first interference phase formula, and calculating a first conversion rate of the first interference phase by using a preset first rate calculation formula;

step b: based on a distance direction resolution unit and an azimuth direction resolution unit where the moving target is located, calculating a second interference phase of the distance direction resolution unit where the moving target is located by using a preset second interference phase formula, and calculating a second change rate of the second interference phase by using a preset second rate calculation formula;

after calibrating the interference phase offset caused by the track crossing (yaw), the remaining track following (motion along the track) error can be accounted for in the estimation problem of the effective baseline, i.e., d _rx ＝d _n +Δd _n . The first interference phase of the data of the distance resolution unit where the moving object P is located is expressed as

Wherein, the first interference phase formula is:

the first rate calculation formula is:

the second interference phase formula is:

the second rate calculation formula is:

S7, estimating the equivalent speed and the effective baseline of the carrier relative to the moving target along the track based on the first change rate and the second change rate, and obtaining the accurately estimated equivalent speed and the effective baseline along the track;

wherein the equivalent velocity v along the track _rx And an effective baseline d _rx Pre-estimated values of v are respectively v _e And d _e ：

Arc top symbol on letter headRepresenting an estimate, the above estimate is only correct under ideal conditions. However, in actual SAR data, the interference phase of the target may be affected by contamination of surrounding clutter and noise, which may cause the actual interference phase curve to be far from the target phase curve. Also, noise can cause the interference phase to oscillate around the ideal phase curve. Noise-inducedErrors are produced in estimating the interferometric phase slope and as the signal-to-noise ratio deteriorates, the estimated interferometric phase profile may suffer from phase wrapping problems with greater noise power. To solve this problem, the conventional median filtering method used in the previous studies can improve the performance under the influence of noise. However, conventional median filters do not work well under conditions where practical radars tend to work with low signal-to-noise ratios. Thus, clutter pollution and noise effects need to be processed simultaneously before the effective baseline is estimated.

In practice the power of the target is typically slightly greater than the power of the clutter. Clutter pollution and noise effects can be addressed by taking the desired operation.

S8, modifying the original azimuth matched filter by using the equivalent speed along the track to obtain a modified azimuth matched filter;

the modified azimuth matched filter is as follows:

s9, carrying out azimuth pulse compression and mapping on the data of the echo signals subjected to the distance pulse compression by using the corrected azimuth matched filter to obtain a precise focusing image;

wherein, the precisely focused image can be expressed as:

s10, extracting a moving target interference phase from the precisely focused image;

wherein the target interference phase may be expressed as:

s11, estimating the inclined plane distance velocity of the moving target by using the target interference phase, the precisely estimated effective baseline and the precisely estimated equivalent velocity along the track to obtain the precisely estimated inclined plane distance velocity;

the correction of the inclined plane distance velocity can obtain an accurately estimated inclined plane distance velocity, and the accurately estimated inclined plane distance velocity is as follows:

s12, calculating to obtain the azimuth resolution unit offset of the moving target according to the precisely estimated inclined plane distance to speed;

The offset of the azimuth resolution unit is as follows:

S＝round(R _p0 v _ry /ρ _a v _e )。

s13, repositioning the moving target in the precisely focused image according to the azimuth resolution unit offset.

Example two

As an alternative embodiment of the present invention, the step of determining the primary band phase interferograms of the adjacent channels using band pass filters for the interferograms of the adjacent channels:

step a: performing fast Fourier transform on the two-dimensional data of the interference phase map of the adjacent channel in the azimuth direction so as to enable the azimuth direction represented by the two-dimensional data to be converted into a Doppler domain, and obtaining interference data of the Doppler domain;

step b: carrying out spectrum peak value search on interference data of a Doppler domain, and determining a peak value interval of the interference phase map;

step c: extracting primary frequency band data in the peak interval by using a band-pass filter;

step d: and carrying out inverse fast Fourier transform on the main frequency band data in the azimuth direction to obtain a main frequency band interference phase diagram of the adjacent channel.

Example III

As an optional embodiment of the present invention, the step of calibrating the phase offset of the interference phase map of the main frequency band caused by the cross-track motion error by using a two-dimensional correction method, and obtaining the corrected interference phase map includes:

step a: grouping the two-dimensional interference data of the main frequency band interference phase diagram according to the equal azimuth units to obtain a grouped result;

Wherein, two-dimensional interference data includes: the last group of data comprises an excessive number of azimuth resolution units and corresponding distance resolution units, and the number of the azimuth resolution units and the corresponding distance resolution units contained in the grouping data except the last group is the same;

step b: sequencing interference phases of all corresponding azimuth resolution units aiming at each distance resolution unit;

step c: selecting a first number of interference phases from each group of ordered interference phases in a mode of selecting from the middle to the two sides, and calculating an average value;

step d: taking the average value as an estimated value of the phase offset of the group of distance-to-resolution units;

step e: and compensating the interference phase corresponding to each distance resolution unit for the estimated value of the corresponding phase offset, and obtaining the interference phase diagram after correction.

It is assumed here that the yaw angle varies slowly over a slow time, approximately not exceeding a constant N ₁ And a plurality of azimuth resolution units. Assume that the magnitude of the interference phase diagram to be calibrated is N _s ×N _e ，N _s Is the number of pixels in the distance direction, N _e Is the number of pixels in azimuth, divides the main band interference phase map into A group. Each group has a size of N, except that the last group includes the remaining lateral distance pixels _s ×N ₁ ，/>Is a downward rounding operation; for each range direction resolution unit, N is the total of all the corresponding azimuth direction resolution units ₁ Sequencing the interference phases; from N ₁ Taking k median phases from the phases, and taking an average value of the median phases as an estimated value of the phase offset of each group of the distance resolution units, wherein the magnitude of k does not change the estimated value of the phase significantly; compensating each group of distance to the resolution unit for the corresponding estimated value of the phase offset, and exceeding the phase number value (-pi, pi)]Is folded back into the main value interval.

Example IV

As an optional embodiment of the present invention, the step of detecting the moving object in the corrected interference phase diagram to obtain the distance direction resolution unit and the azimuth direction resolution unit where the moving object is located in the corrected interference phase diagram includes:

step a: detecting a moving target in the corrected interference phase diagram by using a preset interference detector to obtain a parameter matrix;

each parameter of the parameter matrix represents a distance direction resolution unit and a parameter corresponding to the azimuth direction resolution unit where the moving target is located;

Step b: and determining parameters exceeding a preset threshold value in the parameter matrix as a distance direction resolution unit and a direction resolution unit where the moving target is located.

For an on-board multi-channel SAR system, the performance of the ground moving target detection method depends on the coherence between adjacent channels. However, position errors of the phase center, amplitude-phase errors, and image co-registration errors may reduce coherence and degrade performance. Moving object detection is performed here using ATI detection techniques. The maximum non-ambiguous ATI speed for the phase detector is:

the test statistic of the phase detector can be expressed as

Phi is the interference phase function, eta _Φ Representing a threshold for detecting moving objects, which is determined by the required false alarm rate. If the probability density function can be accurately represented, the threshold can be calculated with the probability of a false alarm. Otherwise, monte Carlo simulation is used to estimate the threshold and detect moving objects. Therefore, the minimum detection speed is

If only the interference phase or amplitude is considered, the false alarm probability will be high for an airborne radar system. The invention adopts the amplitude and phase (Magnitude and Phase, M & P) combined detection method to detect the moving target, can greatly reduce the number of false targets, and the M & P combined detector, namely the preset interference detector, has the output expression of

ζ(r,k)＝ξ _l,k (1-cosφ _r,k )

Wherein, xi _l,k And phi _l,k Representing the real amplitude and phase values of the interference data at the (l, k) position on the corresponding interference phase map, r=1, 2, …, N _l And k=1, 2, …, K being the distance resolution element and the azimuth resolution element numbers, N _l And K is the maximum number of distance-and azimuth-resolved units of the interferogram, respectively.

Example five

As an optional embodiment, the step of estimating the equivalent speed along the track and the effective baseline of the relative moving target of the carrier based on the first change rate and the second change rate, waiting until the estimated equivalent speed along the track and the effective baseline are accurate includes:

step a: for the first interference phase and the second interference phase, respectively acquiring a first number of first phases of the first interference phase and a first number of second phases of the second interference phase each time by using a sliding window;

wherein the median filter can be used to slide in the first and second interference phases respectively in a sliding window manner, and l first phases and l second phases are obtained each time until the sliding is completed;

where l is the length of the median filter.

Step b: sequencing the first phase and the second phase respectively to obtain a first phase vector of a plurality of groups of moving targets and a second phase vector of a plurality of groups of strong clutter;

Step c: determining a first useful phase in the first phase vector and a second useful phase in the second phase vector;

as an alternative embodiment, step c comprises:

step c1: selecting a second number of phases from the first phase vectors of each group and averaging the second number of phases from the second phase vectors of each group and averaging the second number of phases from the first phase vectors of each group by means of the middle position to two sides;

the second number is a preset value, and can be changed according to actual conditions.

Step c2: and carrying out expected operation on the average value obtained by the first phase vector of each group to obtain a first useful phase of each group, and carrying out expected operation on the average value obtained by the first phase vector of each two groups to obtain a second useful phase of each group.

Step d: drawing an interference phase curve of the moving object by using the first useful phase, and drawing an interference phase curve of the strong hybrid wave by using the second useful phase;

step e: determining a first slope of an interference phase curve of a moving object and a second slope of an interference phase curve of a strong hybrid wave;

step f: correcting the first rate of change using the first slope and correcting the second rate of change using a second slope until an accurately estimated along-track equivalent velocity and effective baseline.

Wherein a least squares estimation method may be used to estimate a first slope of the interference phase curve of the moving object and a second slope of the interference phase curve of the strong clutter.

Taking the first interference phase as an example, assume that the moving object is located at a distance from the first interference phase of the resolution unitThe length is L, the median filter length is L, L phases can be obtained in each sliding window operation, the obtained phases are ordered from small to large once, and a total of L-l+1 groups of first phase vectors are obtained. The obtained first phase vectors of each group are respectively subjected to average median filtering, namely, each group selects a first quantity of a median phases (around the median value) to form a median vector phi _si And performing a desired operation to obtain a first useful phase m of the set of first phase vectors _i ：

m _i ＝mean{φ _si },i＝1,...,L-l+1，length(φ _si )＝a

The magnitude of a is determined by an abnormal value generated by noise, all first useful phases are sequentially combined into a vector, and a curve is drawn, wherein the curve serves as an interference phase curve of a moving object. Finally, the least square method is applied to obtain an estimated value k of the inclined slope of the interference phase curve, and the interference phase curve can be expressed as a linear function form:

therefore, the slope of the interference phase curve of the moving object is the first slope, and the calculation and acquisition process of the second slope is the same as that of the second slope, and the description is omitted here.

The quantitative analysis and final effects of the present invention can be further illustrated by the following theoretical analysis and experimental simulation fruit figures.

Referring to fig. 3 and 4, the effect of clutter pollution and noise on effective baseline estimation performance is analyzed.

The ideal interference phase of the moving object shown in the interference phase expression formula of the distance-to-resolution cell data where the moving object P is located is not practically available. The echo signals received by the radar contain clutter, targets and noise, which means that clutter pollution and noise influence exist in a resolution unit where the extracted target signals are located, echoes of clutter scattering surface elements of different landforms can seriously pollute the interference phase, and an interference phase estimation curve of the targets can be influenced. Assume that the motion target echo signal function vector after distance pulse compression received by two channels of the radar is

To simplify the analysis, the common part outside after the extraction phase is represented by β. The interference image after the target echo pulse is overlapped with the echo of one clutter scattering bin is:

where ε represents the texture random variable of the clutter and is omitted hereinafter, i.e. the focus of attention is on the uniform region. The interference phase expectations of the range-wise compressed signal can be expressed as:

Wherein, the liquid crystal display device comprises a liquid crystal display device,is the average power of the clutter and ρ is the coherence coefficient between the channels. From the interference phase expectation expression of the distance-to-compression signal, it can be seen that the cross-correlation term of the target and clutter is eliminated after the expectation is taken. Thus, the desire for the interference phase is expressed by the following equation:

wherein, the liquid crystal display device comprises a liquid crystal display device,if the signal-to-noise ratio (Signal to Clutter Ratio, SCR) goes towards zero, the interference phase is expected to be close to +.>When the SCR is changed to infinity, the interference phase is expected to approach +.>In fact, the +.>And->Not numericallyOften, the difference function between them is defined here:

wherein R is ₀ Representing the closest skew of the target and clutter scattering bins to the radar because they have R in the same range-resolved unit ₀ ＝R _c0 . The above can then be rewritten as

Because ofVery small, the above formula can be written approximately: />

Wherein:

it is obvious that the process is not limited to,is a slope of about 10 ^-3 A linear function of the order rad/s with respect to slow time, whereas α is a cosine function which varies slowly around zero, to a much smaller extent than +.>And->

TABLE 2 actual SAR System operating parameters

Parameter type	(symbol)	Numerical value
			Carrier frequency	f _c	8 850MHz
Bandwidth of a communication device	B	40MHz
			Sampling rate	f _s	60MHz
Pulse repetition frequency	PRF	1kHz
			Minimum pitch	R _o	7.7475km
Platform speed	V _a	115m/s
			Physical baseline	d	0.28m

Here again, simulations are used to prove the correctness of clutter pollution analysis. The track-following speed of the moving object is 2m/s, and other parameters are shown in Table 2. The invention compares the two conditions based on the interference phase in the interference image expression after the superposition of the target echo pulse and the clutter scattering bin echo, one is the interference phase of the target signal only, and the other is the actual interference phase of the target clutter adding signal. The pair of interferometric phase curves at different SCR values is shown as subgraphs a-d in fig. 3, regardless of noise conditions. It can be seen that clutter contamination causes the actual interference phase to deviate from that of the target signal alone, and the degree of deviation becomes progressively greater as the SCR decreases. At the same time, when the clutter approaches but is not equal to the target power, the actual interference phase curve oscillates dramatically (like a saw tooth). This is mainly due to the cross terms in the interference image expression after superposition of the target echo pulse and one clutter scattering bin echo. When SCR deteriorates, the self term of the clutter (relative to the cross phase) dominates.

The slope of the oscillation curves in sub-graph b and sub-graph d in fig. 3 is difficult to confirm. The interference phase of the range-wise pulse compressed signal is expected to effectively cancel the effects of the cross terms. Thus, a 0.25 second smoothing window is used here to simulate the expectations of the actual interference phase (target plus clutter scene). As shown in fig. 4, sub-graphs a-d, the actual interference phase (dashed line) is expected to be nearly linear. This indicates that the desired effect of the crossover term on the interference phase is eliminated. Obviously, as SCR improves, the two curves get closer and closer. In addition, the slopes of the two curves are almost identical, differing by about 10 ^-3 The magnitude of rad/s. For an effective baseline estimate, the error in slope estimation after the expectation is made is acceptable.

Fig. 5 and 6: performance analysis based on different SCNR conditions of real data.

The average median filtering combines the traditional median filtering with the expected operation, and reduces the fluctuation of the interference phase caused by surrounding clutter and noise. In order to demonstrate the effectiveness of the mean median filtering method proposed by the present invention, a target signal is added here on the real SAR data of the distance to the compressed domain. The system parameters are shown in Table 2. The averaged median filtered interference phase curve is shown in figure 5. Solid points are interference phase points and bold lines are curves of average median filtered phase. As SCNR increases, outliers may decrease significantly. Since the average median filter has good robustness, the estimated phase curves are substantially the same by varying the SCNR. Thus, the proposed average median filter is robust to different SCNRs.

A comparison of a conventional median filter with an average median filter is shown in fig. 6. In this example, a target of 1.19dB with scnr=1.19 dB is added to the same actual SAR data. All interference phases are deleted, but the median filtering and the average median filtering of the two curves are preserved. It can be seen that at low SCNR conditions, the curve of the average median filter is smoother than the curve of the median filter. So that the slope can be estimated more accurately using the least squares method.

Fig. 7 and 8: theoretical performance analysis of mean median filtering

The experiment gives the analysis result of the method of the invention based on the theoretical performance of the Cramer-Rao Bound (CRB) and uncertainty in measurement. From the equation effective baseline expression, it can be seen that the accuracy of the effective baseline estimate can be determined from the slope estimation error of the interferometric phase curve and the carrier velocity error:

wherein: Δk _c Is the estimated error of the slope of the interferometric phase curve, deltaV _a Is an error in the speed of the vehicle. Assuming that the interferometric phase curve slope and carrier velocity have independent error contributions, the standard deviation of the effective baseline can be calculated as

Wherein:and->The standard deviation of the interference phase estimate and the carrier velocity, respectively. The uncertainty of the carrier speed is mainly caused by the motion error of the carrier itself; due to coefficient terms [ -k _c λR _c0 /(4πV _2a )] ² Far less than [ lambda R ] _c0 /(4πV _a )] ² The estimation error of the slope of the interferometric phase curve will therefore be the main error source.

The standard deviation of the slope estimate can be expressed in terms of uncertainty of the interference phase:

wherein: sigma (sigma) _Φ Represents uncertainty of the interference phase and:

wherein the number of effective interference phases is L-L. The uncertainty of the interference phase can be estimated by measuring:

wherein phi is _i Represents the ith measurement of interference phase, Φ _i Representing the slope estimate of the ith interference phase. For an average filter, Φ _i ＝m _i . The CRB of the simultaneous interference phase can be calculated from the following standard formula:

where ρ represents the correlation coefficient between channels, and |ρ|= (1+1/CNR) ^-1 。

Wherein: CNR stands for the hetero-noise ratio. The above equation is based on the assumption that the CNR of each channel is the same. Fig. 7 shows 1000 monte carlo simulations of root mean square error as a function of noise-to-noise ratio for an effective baseline estimate. The radar parameters are shown in table 2. The method with median filter presented in this simulation is compared. Clutter and noise are both based on complex gaussian distributions. In this case, only the estimation error of the slope, which is the main source of the error, may be considered. As shown in fig. 8, the proposed mean median filtering approach approximates CRB with more accurate performance than conventional median filtering.

Performance analysis based on measured data

The effectiveness of the method is checked by processing data obtained by a group of real X-band airborne multichannel radars working in an ATI-SAR mode, the working parameters of the system are shown in table 2, some vehicles move along a highway, the positions of all vehicles are calibrated in advance, and the accuracy of the inclined plane distance-to-speed estimation is verified by using the highway scene data.

The dual channel coarse focus image processed by the range-doppler imaging algorithm is shown in fig. 9, while the original interference phase is shown in sub-graph a in fig. 9. It can be seen that there is a significant difference between the different azimuthal units and that the effects of the cross-track phase offset on the apparent noise speckle in the original interferogram should be eliminated before the effective baseline and motion parameter estimation is performed. As illustrated in step 1, the azimuthal cells may be divided into several groups by a bandpass filtering process based on the approximate interferometric phase offset. Here select N ₁ =750, and even N by trial discovery ₁ At=1 000, both values can still obtain similar performance. Selecting an average of k intermediate phases to represent the effective phase offset for each group eliminates the phase offset caused by the track-crossing motion. In this example, the parameter k is set to 10, and the k value has little influence on the phase shift correction. In Table 3, the phase shifts corresponding to the different k values are given, and the visible phase shift changes only slightly And (5) melting. The results after using the two-dimensional calibration method are shown in sub-graph b of fig. 9, and it can be seen that the noise caused by the phase shift is mostly filtered out, leaving a cleaner interference phase. .

TABLE 3 phase shift at the value of k failure

k	10	20	30	40
					Phase shift (rad)	-0.0069	-0.0048	-0.0058	-0.0067

FIG. 10 compares the correction of the track-crossing motion error with the phase detection a, the amplitude detection b and M&ATI detection result of P joint detection method c, wherein false alarm probability is set to 10 ^-6 . Since the scene of the region of interest in this scene data is relatively uniform, the scene is detected in amplitude, phase, and M&The probability density function of the P joint detector may use the above formula. Here in M&The probability density function of the P joint detector is exemplified:

p _ζ (ζ)＝(v ₀ /πζ) ^1/2 exp(-v ₀ ζ),ζ,v ₀ ＞0

v ₀ ＝2mρ/(1-ρ ² ) M represents the number of observations and ρ represents the complex correlation coefficient of the two channels. By solving the following function, M can be obtained&Threshold of P joint detector:

as shown in fig. 10, only the phase or amplitude detection method is adopted, and although a moving object is detected, some false objects (false alarms) are also detected at the same time. In contrast, most moving objects can be better detected with an M & P joint detector, and only when the vehicle is moving very slowly, one object is missed, which is already below the minimum detectable speed at the detector. Here, two moving objects of interest are selected for illustration. In sub-graphs a-c, the range resolution unit where the two targets are located can be approximately determined by detection using an M & P joint detector.

The result of only performing range pulse pressure on all scenes is shown in fig. 11, and the amplifying region containing one strong clutter scattering bin and two moving targets is shown in the sub-graph b-sub-graph d, wherein the range migration has been corrected, thus obtaining the effect that the echo power of clutter and targets is slightly stronger than that of the background, and improving the parameter estimation accuracy.

Clutter points for interferometry and distance compressed signals of moving objects are extracted. The distance-oriented pulse compression signal of the clutter scattering bin and the moving object for interference processing is extracted. Since the cross-track correction can be considered constant over 1000 azimuth resolution units, the effective baseline can also be considered a constant during this azimuth time. The interference phase and estimated curve extracted by applying the mean median filtering method proposed by the present invention are given in fig. 12, sub-plot a-sub-plot c. It can be seen from each sub-graph that the curve formed by the discrete points has an approximately constant slope, but is closer to a diagonal line, but that some outliers present can affect the slope estimation. By adopting the average median filtering method provided by the invention, the abnormal value caused by clutter pollution and noise influence can be eliminated. The phase vector length is designed to be l=1000 and the filter length is designed to be l=250, and in each average median filter, a=20 median phase points are selected for expectation, so that an estimated curve can be smooth enough to be convenient for least square fitting. Finally, the estimated effective baseline length is 0.24m, and estimated values of the track following speeds of the two moving targets are 1.5 m/s and 1.9 m/s respectively.

The track following speeds of the two targets are relatively close, and the test proves that the two targets can be accurately focused when the track following speed is 1.7m/s in fig. 13, and the 1.7m/s can be regarded as an approximate true value. If the baseline error caused by the yaw of the carrier is not corrected, the baseline length is 0.28m, and the obtained estimated values of the inclined plane distance and the velocity of the two targets are respectively 1.05 m/s and 1.59 m/s; however, after the method provided by the invention is applied to effective baseline estimation, the inclined plane distance-to-speed estimated values of the two targets are 1.2 m/s and 1.84 m/s respectively, and the accuracy is obviously improved. The azimuth resolution element offset caused by the bevel distance to speed can be given by:

S＝round(R _p0 v _ry /ρ _a v _e )

wherein ρ is _a Indicating the azimuthal resolution. Using the conventional method, the azimuth resolved unit offset amounts of the two targets are calculated as 659 th and 1 st units by the above formula of the azimuth resolved unit offset amounts. By applying the effective baseline estimation method provided by the invention, the offset of the resolution units of the two targets are 764 units and 1160 units respectively.

In fig. 14, comparing the proposed method with the conventional method, it is verified that the inclined plane distance-to-velocity estimation has a more accurate. As shown in fig. 14, sub-graph a, in the case of using the conventional method of not estimating the effective baseline and performing the subsequent correction, two targets are positioned in the middle of the bidirectional highway, which is obviously erroneous. And the two targets are accurately positioned to one side of the bidirectional road after being processed by the method provided by the invention, and the effectiveness of the method is verified.

As shown in fig. 15, the device for estimating the airborne multichannel SAR interference effective baseline provided by the invention comprises:

an acquiring module 151, configured to acquire a coarse focusing image mapped by echo signals of adjacent channels from a radar system;

a first determining module 152, configured to determine an interference phase map of adjacent channels based on the focused image;

a second determining module 153, configured to determine, for the interference phase map of the adjacent channel, a main band phase interference map of the adjacent channel using a band-pass filter;

the correction module 154 is configured to calibrate, in the main band interval, a phase offset of an interference phase map caused by a track-crossing motion error by using a two-dimensional correction method, and obtain a corrected interference phase map;

the detection module 155 is configured to detect a moving object in the corrected interference phase diagram, so as to obtain a distance direction resolution unit and an azimuth direction resolution unit where the moving object is located in the corrected interference phase diagram;

a first calculating module 156, configured to calculate, based on the distance-direction resolution unit and the azimuth-direction resolution unit where the moving target is located, a first interference phase of the distance-direction resolution unit where the moving target is located, a first transformation rate of the first interference phase, a second interference phase of the distance-direction resolution unit where the strong clutter is located, and a second change rate of the second interference phase;

A estimating module 157, configured to estimate a track equivalent speed and an effective baseline of the relative moving target of the carrier based on the first change rate and the second change rate, and wait until the track equivalent speed and the effective baseline are accurately estimated;

a modifying module 158, configured to modify the original azimuth matched filter by using a track equivalent speed of the carrier relative to a target, so as to obtain a modified azimuth matched filter;

the compression module 159 is configured to perform azimuth pulse compression and mapping on the data of the echo signal subjected to the distance pulse compression by using the corrected azimuth matched filter, so as to obtain a precise focusing image;

an extracting module 160, configured to extract a moving target interference phase from the precisely focused image;

a speed correction module 161, configured to estimate a bevel distance velocity to the moving object using the target interference phase, the precisely estimated effective baseline, and the precisely estimated along-track equivalent speed, and obtain a precisely estimated bevel distance velocity;

the second calculating module 162 is configured to calculate, according to the precisely estimated inclined plane distance velocity, an azimuth resolution unit offset of the moving target;

A positioning module 163, configured to reposition the moving object in the precisely focused image according to the azimuth resolution unit offset.

The foregoing is a further detailed description of the invention in connection with the preferred embodiments, and it is not intended that the invention be limited to the specific embodiments described. It will be apparent to those skilled in the art that several simple deductions or substitutions may be made without departing from the spirit of the invention, and these should be considered to be within the scope of the invention.

Claims

1. An airborne multichannel SAR interference effective baseline estimation method is characterized by comprising the following steps:

extracting a moving target interference phase from the precisely focused image;

2. The method of claim 1, wherein the step of determining the dominant band phase interferogram of the adjacent channel using a band pass filter for the interferogram of the adjacent channel:

performing fast Fourier transform on the two-dimensional data of the interference phase map of the adjacent channel in the azimuth direction so as to enable the azimuth direction represented by the two-dimensional data to be converted into a Doppler domain, and obtaining interference data of the Doppler domain;

carrying out spectrum peak value search on interference data of a Doppler domain, and determining a peak value interval of the interference phase map;

extracting primary frequency band data in the peak interval by using a band-pass filter;

and carrying out inverse fast Fourier transform on the main frequency band data in the azimuth direction to obtain a main frequency band interference phase diagram of the adjacent channel.

3. The method of claim 1, wherein the step of calibrating the phase offset of the primary band interference phase map caused by the cross-track motion error using a two-dimensional correction method, and obtaining the corrected interference phase map comprises:

grouping the two-dimensional interference data of the main frequency band interference phase diagram according to the equal azimuth units to obtain a grouped result;

wherein the two-dimensional interference data includes: the last group of data comprises an excessive number of azimuth resolution units and corresponding distance resolution units, and the number of the azimuth resolution units and the corresponding distance resolution units contained in the grouping data except the last group is the same;

sequencing interference phases of all corresponding azimuth resolution units aiming at each distance resolution unit;

selecting a first number of interference phases from each group of ordered interference phases in a mode of selecting from the middle to the two sides, and calculating an average value;

taking the average value as an estimated value of the phase offset of the distance resolution unit;

and compensating the interference phase corresponding to each distance resolution unit for the estimated value of the corresponding phase offset, and obtaining the interference phase diagram after correction.

4. The method according to claim 1, wherein the step of performing moving object detection in the corrected interference phase map to obtain a distance-direction resolution unit and an azimuth-direction resolution unit where the moving object is located in the corrected interference phase map includes:

detecting a moving target in the corrected interference phase diagram by using a preset interference detector to obtain a parameter matrix;

and determining parameters exceeding a preset threshold value in the parameter matrix as a distance direction resolution unit and a direction resolution unit where the moving target is located.

5. The method according to claim 1, wherein the step of calculating, based on the range-wise resolution unit and the azimuth-wise resolution unit where the moving object is located, a first interference phase of the range-wise resolution unit where the moving object is located, a first transformation rate of the first interference phase, a second interference phase of the range-wise resolution unit where the strong clutter is located, and a second change rate of the second interference phase includes:

based on a distance direction resolution unit and an azimuth direction resolution unit where the moving target is located, calculating a first interference phase of the distance direction resolution unit where the moving target is located by using a preset first interference phase formula, and calculating a first conversion rate of the first interference phase by using a preset first rate calculation formula;

And calculating a second interference phase of the distance-direction resolution unit where the moving object is located by using a preset second interference phase formula based on the distance-direction resolution unit and the azimuth-direction resolution unit where the moving object is located, and calculating a second change rate of the second interference phase by using a preset second rate calculation formula.

6. The method of claim 1, wherein the step of estimating the vehicle relative moving target along the track equivalent speed and the effective baseline based on the first rate of change and the second rate of change, waiting until the along track equivalent speed and the effective baseline are accurately estimated comprises:

for the first interference phase and the second interference phase, respectively acquiring a first number of first phases of the first interference phase and a first number of second phases of the second interference phase each time by using a sliding window;

sequencing the first phase and the second phase respectively to obtain a first phase vector of a plurality of groups of moving targets and a second phase vector of a plurality of groups of strong clutter;

determining a first useful phase in the first phase vector and a second useful phase in the second phase vector;

Drawing an interference phase curve of the moving object by using the first useful phase, and drawing an interference phase curve of the strong hybrid wave by using the second useful phase;

determining a first slope of an interference phase curve of a moving object and a second slope of an interference phase curve of a strong hybrid wave;

correcting the first rate of change using the first slope and correcting the second rate of change using the second slope until the along-track equivalent velocity and effective baseline are accurately estimated.

7. The method of claim 6, wherein the step of acquiring a first number of first phases of the first interference phase and a first number of second phases of the second interference phase each time using a sliding window for the first interference phase and the second interference phase, respectively, comprises:

sliding in the first interference phase and the second interference phase respectively in a sliding window mode by using a median filter, and obtaining l first phases and l second phases each time until the sliding is completed;

where l is the length of the median filter.

8. The method of claim 6, wherein the steps of determining a first useful phase in the first phase vector and determining a second useful phase in the second phase vector comprise:

Selecting a second number of phases from the first phase vectors of each group and averaging the second number of phases from the second phase vectors of each group and averaging the second number of phases from the first phase vectors of each group by means of the middle position to two sides;

and carrying out expected operation on the average value obtained by the first phase vector of each group to obtain a first useful phase of each group, and carrying out expected operation on the average value obtained by the first phase vector of each two groups to obtain a second useful phase of each group.

9. The method of claim 6, wherein the step of determining a first slope of the interference phase curve of the moving object and a second slope of the interference phase curve of the strong clutter comprises:

a least square estimation method is used for estimating a first slope of an interference phase curve of a moving object and a second slope of an interference phase curve of a strong hybrid wave.

10. An airborne multichannel SAR interference effective baseline estimation device, comprising: