CN103353595B - Meter wave radar height measurement method based on array interpolation compression perception - Google Patents

Abstract

The invention discloses a height measurement method based on an array interpolation compression perception. The height measurement method mainly aims at solving a low elevation height measurement problem under a multipath environment, and especially under low signal to noise ratio and less snapshot environments. The method comprises the following steps of extracting a target signal from a radar echo; acquiring a spatial-domain sparse signal through cancellation and signal reconstruction; using a wave beam formation method to obtain a rough measurement target angle; according to the rough measurement angle, acquiring the spatial domain and dividing the spatial domain; using the array interpolation to acquire a virtual array; according to a matrix transformation relation, acquiring an interpolation transformation matrix and carrying out prewhitening processing on the interpolation transformation matrix; using a whitening interpolation transformation matrix and an observation matrix to acquire an observation signal; using a whitening interpolation transformation matrix and observation signal iteration operation to acquire a target signal estimation value; extracting a target angle from the target signal estimation value so as to acquire a target height. By using the method of the invention, sampling points of the target signal and computation intensity are obviously reduced; sidelobes of a signal power spectrum and a space spectrum are effectively reduced; the method can be used in target tracking.

Description

Based on the meter wave radar height measurement method of array interpolation compression perception

Technical field

The invention belongs to Radar Signal Processing Technology field, particularly compressed sensing and meter wave radar height measurement method, can be used for target following.

Background technology

Domestic and international radar circle generally believes, metre wave radar has anti-stealth capability.Metre wave radar due to its wavelength longer, wave beam is wide, and particularly when measuring low angle target, wave beam beats ground, ground return multipath phenomenon that is strong, target is serious, causes altitude measurement in VHF radar precision low, even complete failure.In radar return signal except the radio wave refration effect that the unevenness of lower atmosphere layer causes, the multi-path interference effect that the mirror-reflection of ground, sea generation in addition and diffuse scattering cause.Multi-path interference has a huge impact the low measurement of elevation precision of radar, and direct wave and multipath reflection ripple signal have strong correlation; The angle of target direct wave incident angle and multipath reflection ripple incident angle is very little, is usually located within a beam angle; Beam split can cause the electrical level flash of Received signal strength, and signal to noise ratio (S/N ratio) fluctuation is larger.During the low elevation angle, topographic relief is very large on the impact of measurement result, especially on the land of the larger sea of sea condition or complex area, reflection clutter in (sea) face, ground is comparatively strong, and echo signal is often submerged in clutter, and the non-stationary and spike of clutter can cause false-alarm probability to increase rapidly.Therefore be difficult to survey in a multi-path environment high, therefore the high problem of the survey of metre wave radar is the difficult problem that radar circle solves not yet very well always.

A high difficult problem is surveyed for solving metric wave preferably, the Major Technology taked mainly contains: 1. increase sky linear content and particularly increase the aperture of antenna in height dimension, to reduce the beam angle of antenna in vertical dimension, improve angular resolution, for the higher elevation angle, make wave beam " not beat ground " and complete elevation carrection; 2. suitably increase the antenna height of antenna, reduce wave beam and upwarp, be beneficial to detecting low-altitude objective.But for low target, " multipath " problem cannot be avoided.

At present, the high method of survey for metre wave radar mainly contains following three classes:

(1) multifrequency beam split altimetry.This method utilizes multiple frequency of operation time-division to work, and its theory is feasible, but requires that the bandwidth of operation of multiple frequency is wider, system complex, does not also have this utility system at present.

(2) based on the meter wave radar height measurement method of beam split.This method utilizes different antennae to divide the phase relation of lobe, determines that the elevation angle, target place is interval, carries out to received signal extracting normalization error signal than width process, finally obtains the height of target according to normalization error signal and elevation angle section scale-checking.It is no more than 1m in the mean square deviation of surface irregularity, and signal to noise ratio (S/N ratio) reaches 16dB, and altimetry precision can reach 1% of distance.The paper " meter wave radar height measurement method based on beam split " that Chen Baixiao etc. delivered in " electronic letters, vol " in June, 2007.This is a kind of low Elevation high method only needing the metre wave radar of 3 antennas in vertical dimension.The method is suitable only for smooth position, and requires higher to the flatness in position, and altimetry precision also can only reach 1% of distance, is difficult to meet the higher actual operation requirements of some precision.

(3) array superresolution processing surveys high method.This method the super resolution technology in Array Signal Processing is applied to differentiate direct-path signal and multipath signal.Comprise proper subspace algorithm and maximum likelihood algorithm.Wherein:

Proper subspace class algorithm, must in the face of the coherence problems of the direct wave caused by multipath transmisstion and multipath signal when being applied to low Elevation height.But when signal source is completely relevant, the order of data covariance matrix will be 1, the existence of coherent source makes signal subspace and noise subspace interpenetrate, cause the steering vector of some coherent source and noise subspace not exclusively orthogonal, this can make the hydraulic performance decline of a lot of classical proper subspace class algorithm, even complete failure.

Maximum likelihood class algorithm idea is simple, superior performance, good performance is all had under high s/n ratio and low signal-to-noise ratio, but it is a nonlinear multidimensional optimization problem that likelihood function solves, need to carry out multi-dimensional grid search, calculated amount exponentially increases along with the increase of target number, and implementation procedure is more complicated.Such as, the paper " the Beam Domain ML meter wave radar height measurement method of Array interpolation " that the people such as the paper " based on the pretreated metre wave radar of difference low elevation angle Processing Algorithm " that the people such as Zhao Guanghui deliver in February, 2009 at " electronics and information journal " and Hu Tiejun deliver in August, 2009 at " electric wave science journal ", and the paper " research of metre wave radar maximum likelihood super-resolution height-finding technique " that the people such as Yang Xueya delivers in September, 2011 in " radar science and technology ".

In said method, method 1 is difficult to realize; Method 2 is only applicable to smooth position, and low precision, cannot practical requirement; Method 3 operand is large, requires that sample number is many, in a multi-path environment hydraulic performance decline, even loses efficacy.Therefore, in the low Elevation high problem process of process, the high method of existing various survey is effective poor, no longer applicable.

Summary of the invention

The object of the invention is to the deficiency for above-mentioned prior art, propose a kind of meter wave radar height measurement method based on array interpolation compression perception, to reduce operand further, the angle measurement accuracy of direction of arrival DOA in raising low signal-to-noise ratio situation.

For achieving the above object, technical thought of the present invention is:

By carrying out Array interpolation to M array element, obtaining array element is P, P > > M, virtual array, thus improve the dimension of array measurement signal, then compression sampling is carried out to the measurement signal of virtual array, obtain target direction of arrival finally by sparse reconstruct.

Specific implementation step comprises as follows:

(1) from radar return, echo signal is extracted, obtain the array manifold matrix V of true array, and clutter cancellation and interference cancellation process are carried out to this echo signal, obtain offseting rear echo signal x and spatial domain sparse signal S, the relation offseting rear echo signal x and spatial domain sparse signal S is as follows:

x=ψS+n，

Wherein, ψ represents super complete redundant dictionary, and its length is that c, n represent white Gaussian noise;

(2) use digital beam froming method DBF to carry out elevation angle bigness scale to the echo signal x after offseting, obtain bigness scale angle [alpha], and then obtain spatial domain, place, echo signal elevation angle Ο;

(3) described spatial domain Ο is divided into P part, P > > M, M represents array number, obtains spatial domain matrix Θ:

Θ=[α _l,α _l+Δα,…,α _r],

Wherein, represent the left margin of Θ, represent the right margin of Θ, represent half-power beam width, Δ α is step-length, Δ α=0.1 °;

(4) Array interpolation process is carried out to true array, obtain the array manifold matrix W of virtual array _i; According to the array manifold matrix W of virtual array _iwith the array manifold matrix W of true array, obtain interpolation transformation matrix B;

(5) pre-whitening processing is carried out to interpolation transformation matrix B, obtain albefaction interpolation transformation matrix T _i;

(6) the echo signal x after offseting is projected to albefaction interpolation transformation matrix T _i, obtain the measurement signal z of virtual array;

(7) tie up observing matrix φ with F × P and compression sampling is carried out to measurement signal z, F < < P, obtain the observation signal y that F × 1 is tieed up;

(8) according to observation signal y and albefaction interpolation transformation matrix T _i, utilize greedy orthogonal matching pursuit method of following the trail of in class algorithm, through type iteration, chooses a locally optimal solution Step wise approximation spatial domain sparse signal S, obtains the estimated value of spatial domain sparse signal S

$\hat{S}=[{\hat{s}}_1,{\hat{s}}_2,\&CenterDot;\&CenterDot;\&CenterDot;,{\hat{s}}_i{,\&CenterDot;\&CenterDot;\&CenterDot;,\hat{s}}_c],$

Wherein, || || ₁represent and ask vectorial 1-norm, s.t represents constraint condition, || || ₂represent and ask vectorial 2-norm, ψ represents super complete redundant dictionary, and c represents the length of super complete redundant dictionary ψ, i=1,2 ..., c, β represent that noise criteria is poor;

(9) objective definition angular range 6=[θ ₁, θ ₂..., θ _i..., θ _c], according to estimated value element and the one-to-one relationship of element of θ, namely with θ _ione_to_one corresponding, obtains target angle measurement result θ _d, d ∈ i:

Wherein, d represents estimated value in non-vanishing element s _dsubscript;

(10) according to target angle measurement result θ _dtarget range R with known, obtains object height by triangular transformation:

H=Rsin(θ _d)。

The present invention compared with prior art tool has the following advantages:

1) the present invention is owing to adopting Array interpolation process to echo signal, reduce the secondary lobe of power spectrum signal and spatial spectrum, effectively improve the performance of meter wave radar height measurement method, for under multi-path environment, especially the low Elevation high problem under low signal-to-noise ratio and the less environment of fast umber of beats, provides a kind of effective solution.

2) the present invention due to adopt observing matrix compression sampling process is carried out to measurement signal, not only reduce operand, improve estimated accuracy, and can when sample number is less the echo signal estimated result of gained more excellent compared with additive method.

Simulation result shows, the present invention can be directly used in the Mutual coupling of coherent signal, and has higher angular resolution.

Accompanying drawing explanation

Advantage and the effect of the inventive method is further illustrated below in conjunction with accompanying drawing.

Fig. 1 is realization flow figure of the present invention;

Fig. 2 is that the present invention and existing two kinds survey the performance curve comparison diagram of high method when signal to noise ratio (S/N ratio) changes;

Fig. 3 is the present invention and existing two kinds of results contrast figure surveying high method and estimate angle on target;

Fig. 4 is that the present invention and a kind of existing method are to the angle error results contrast figure of measured data process.

Embodiment

Content of the present invention and effect is described in detail below in conjunction with accompanying drawing.

With reference to Fig. 1, the present invention includes following steps:

Step 1: extract echo signal from radar return, obtains the array manifold matrix W of true array.

Array radar used is an even linear array vertically placed, and this even linear array is made up of M array element, and array element is spaced apart d.

Suppose have K far field narrow band signal to incide this even linear array, M>K, signal incident angle is α _i, i=1,2 ..., K, then the echo signal that array receives in t is:

X(t)=Vs(t)+n(t)，

Wherein, X is that the array element that M × 1 is tieed up receives data, and n is the white noise that M × 1 is tieed up, and meets zero-mean, variance is σ ²multiple Gaussian distribution, each array element output noise statistical iteration; S=[s ₁, s ₂..., s _i..., s _k] ^tfor the signal phasor that K × 1 is tieed up; W is that M × K ties up array manifold matrix:

W=[v(α ₁),v(α ₂),…，v(α _i),…，v(α _K)]，

Wherein, be the steering vector of i-th echo signal, subscript T represents transposition, and λ represents radar signal wavelength.

Step 2: carry out clutter cancellation and interference cancellation process to echo signal X (t), obtains offseting rear echo signal x; Adopt space lattice division methods to re-construct and offset rear echo signal x, obtain spatial domain sparse signal S.

Owing to belonging to radar signal conventional processing to the clutter cancellation of echo signal X (t) and interference cancellation process, and unnecessarily contact with main contents of the present invention, therefore be not described.

Openness in order to show the spatial domain offseting rear echo signal x, adopting space lattice to divide process to offseting rear echo signal x, being divided into { α by-180 °, space ~ 180 ° ₁, α ₂..., α _u..., α _u, α _ube u angular interval, u=1,2 ..., U, U > > K;

Suppose each α _uall with an echo signal s _ucorresponding, the so just spatial domain sparse signal of structure U × 1 dimension: S=[s ₁, s ₂..., s _u..., s _u] ^t, will offset rear echo signal x and project to S, then in S, only have the element of the K of a physical presence echo signal position non-vanishing, the element of other U-K position is zero, obtains spatial domain sparse signal S:

S=(x-n)ψ ^-1，

Wherein, subscript T represents transposition, and ψ is super complete redundant dictionary; The target information that x with S comprises is consistent, unlike, x is the expression of echo signal at Element space, and S is the expression of echo signal in spatial domain.

From above formula, offset rear echo signal x and also can write:

x=ΨS+n。

Step 3: the echo signal x after using digital beam froming method DBF to offset carries out angle bigness scale, obtains bigness scale angle [alpha], and then obtains spatial domain, place, echo signal elevation angle Ο.

3a) utilize steering vector v (ξ)=[1, e ^{-j2 π sin (ξ)}..., e ^{-j2 π (M-1) sin (ξ)}] ^t, summation is weighted to the echo signal x after offseting, obtains bigness scale angle [alpha]:

$\mathrm{\&alpha;}=\arg \underset{\mathrm{\&xi;}}{\max }\left(\frac{1}{L}\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{\left|v^H\left(\mathrm{\&xi;}\right)x\left(t_l\right)\right|}^2\right),$

Wherein, arg max represents the parameter found and have maximum cost function, and ξ represents target search angular range ,-180 °≤ξ≤180 °, and L represents fast umber of beats, and M represents element number of array, x (t _l) represent the echo signal after offseting, t _lrepresent the sampling time, 1≤l≤L, subscript T represents transposition, and subscript H represents conjugate transpose;

3b) utilize half-power beam width obtain the spatial domain Ο at angle on target place:

Wherein, λ represents radar signal wavelength, and d represents array element distance.

Step 4: described spatial domain Ο is divided into P part, obtains spatial domain matrix Θ, P > > M, and M represents array number:

Θ=[α _l,α _l+Δα，…，α _r]，

Wherein, represent the left margin of Θ, represent the right margin of Θ, represent half-power beam width, Δ α is step-length, Δ α=0.1 °.

Step 5: carry out Array interpolation process to true array, obtains the array manifold matrix W of virtual array _i.

Carrying out Array interpolation process to true array, is add Virtual array between the array element of true array, and to expand the dimension of the array manifold matrix W of true array, the M × P obtaining virtual array ties up array manifold matrix W _i:

W _I=[v _I(α _l)，v _l(α _l+Δα)，…，v _I(α _j)，…，v _I(α _r)]，

Wherein, represent the steering vector of a virtual matrix jth echo signal, M represents element number of array, and subscript T represents transposition, α _j∈ Θ, Θ=[α _l, α _l+ Δ α ..., α _r], Δ α is step-length, Δ α=0.1 °.

Step 6: according to the array manifold matrix W of virtual array _iwith the array manifold matrix W of true array, obtain interpolation transformation matrix B, point following two kinds of situations calculate:

When not considering noise, according to the array manifold matrix W of virtual array _iand the fixed relationship between the array manifold matrix W of true array and interpolation transformation matrix B: B ^hw=W _i, and the steering vector of true array with the steering vector v of virtual array _i(α _j) and interpolation transformation matrix B between fixed relationship:

draw interpolation transformation matrix B:

B=W _IW ^H(WW ^H) ^-1，

Wherein, represent the steering vector of true array,

represent the steering vector of virtual matrix, subscript H represents conjugate transpose, represent the incident angle offseting front echo signal, α _j∈ Θ, Θ=[α _l, α _l+ Δ α ..., α _r], Δ α is step-length, Δ α=0.1 °;

When considering noise, it is the array manifold matrix W according to virtual array _iand the fixed relationship between the array manifold matrix W of true array and interpolation transformation matrix B: B ^h(W+N)=W _i+ N _i, and the steering vector of true array with the steering vector v of virtual array _i(α _j) and interpolation transformation matrix B between fixed relationship:

draw interpolation transformation matrix B:

$B={\mathrm{\&sigma;}}_s^2{W}_I{W}^H{({\mathrm{\&sigma;}}_s^{2}W{W}^H+{\mathrm{\&sigma;}}_n^{2}I)}^{-1},$

Wherein, N represents the noise matrix of true array, N _irepresent the noise matrix of virtual matrix, n represents the noise vector of N, n _irepresent N _inoise vector, for signal power, for noise power, I is unit matrix.

Step 7: carry out pre-whitening processing to interpolation transformation matrix B, obtains albefaction interpolation transformation matrix T _i.

7a) to the autocorrelation matrix R of interpolation transformation matrix B _bcarry out Eigenvalues Decomposition:

R _B=B(B ^HB) ^-1B ^H=QΣQ ^H,

Wherein, Q represents orthogonal matrix, and Q=B, Σ represent diagonal matrix, Σ=(B ^hb) ^-1, subscript H represents conjugate transpose;

7b) according to orthogonal matrix Q and diagonal matrix Σ, obtain albefaction interpolation transformation matrix T by prewhitening formula _i:

T _I=Σ ^1/2Q ^H=(B ^HB) ^-1/2B ^H。

Step 8: the echo signal x after offseting is projected to albefaction interpolation transformation matrix T _i, measurement signal: z=T is tieed up in P × 1 obtaining virtual array _ix=T _iψ S+T _in, wherein, ψ represents super complete redundant dictionary, and n represents white noise, and S represents spatial domain sparse signal.

Step 9: tie up observing matrix φ with F × P and carry out compression sampling to measurement signal z, F < < P, namely reduces the dimension of measurement signal z, obtains the observation signal y that F × 1 is tieed up:

y=φz=φT _Iψs+φT _In。

Step 10: according to observation signal y and albefaction interpolation transformation matrix T _i, utilize greedy orthogonal matching pursuit method of following the trail of in class algorithm, through type iteration, chooses a locally optimal solution Step wise approximation spatial domain sparse signal S, obtains the estimated value of spatial domain sparse signal S

$\hat{S}=[{\hat{s}}_1,{\hat{s}}_2,\&CenterDot;\&CenterDot;\&CenterDot;,{\hat{s}}_i{,\&CenterDot;\&CenterDot;\&CenterDot;,\hat{s}}_c],$

Wherein, || || ₁represent and ask vectorial 1-norm, s.t represents constraint condition, || || ₂represent and ask vectorial 2-norm, ψ represents super complete redundant dictionary, and c represents the length of super complete redundant dictionary ψ, i=1,2 ..., c, β represent that noise criteria is poor.

Step 11: objective definition angular range, theta=[θ ₁, θ ₂..., θ _i..., θ _c], according to estimated value element and the one-to-one relationship of element of θ, namely with θ _ione_to_one corresponding, obtains target angle measurement result θ _d, d ∈ i:

Wherein, d represents estimated value in non-vanishing element s _dsubscript.

Step 12: according to target angle measurement result θ _dtarget range R with known, obtains object height by triangular transformation:

H=Rsin(θ _d)。

Advantage of the present invention and effect are further illustrated by following computer sim-ulation and measured data result:

1. simulated conditions

In simulation process, for equidistantly structuring the formation of vertically arranged 20 horizonally-polarized arraies unit's composition, radar frame height 20m, ground reflection coefficent is-0.95, carrier frequency is 300MHz, only considers the mirror-reflection on ground, interpolation 9 Virtual arrays between every two array elements, total array number of the interpolation battle array obtained is 191, and observing matrix dimension is 20.

2. emulate content

Emulation one: select single static target, be 200km in the distance of target and reference antenna, target angle of going directly is 2 °, multipath reflection angle is-2.01 °, array element signal to noise ratio (S/N ratio) is changed to 30dB from-10dB, fast umber of beats is under the condition of 10, carries out angle measurement accuracy emulation respectively with existing front-rear space smooth multiple signal classification method, alternating projection maximum likelihood method and the present invention to low elevation angle target.Simulation result as shown in Figure 2.Wherein:

Transverse axis represents signal to noise ratio (S/N ratio) change from-10 decibels to 20 decibels, and the longitudinal axis represents angle error;

SS-MUSIC represents the angle error of front-rear space smooth multiple signal classification method when signal to noise ratio (S/N ratio) changes according to transverse axis,

APML represents the angle error of alternating projection maximum likelihood method when signal to noise ratio (S/N ratio) changes according to transverse axis,

IA-CS represents the angle error of the present invention when signal to noise ratio (S/N ratio) changes according to transverse axis.

As can be drawn from Figure 2, for the angle measurement of low elevation angle target, existing front-rear space smooth multiple signal classification method, alternating projection maximum likelihood method angle error are bigger than normal, and angle error of the present invention is minimum.

Emulation two: select single target, at object height 12000m, radial direction flies to 650km from 50km, array element distance half-wavelength, signal to noise ratio (S/N ratio) 10dB, fast umber of beats 10, under the condition that Monte Carlo experiment number of times is 100 times, emulates the impact of the different elevation angle on algorithm estimated accuracy respectively with existing front-rear space smooth multiple signal classification method, alternating projection maximum likelihood method and the present invention.Simulation result as shown in Figure 3.Wherein:

Fig. 3 (a) is the elevation angle of existing front-rear space smooth multiple signal class methods when the distance in target and position changes according to transverse axis;

Fig. 3 (b) is the elevation angle of existing alternating projection maximum likelihood method when the distance in target and position changes according to transverse axis;

Fig. 3 (c) is the elevation angle of the present invention when the distance in target and position changes according to transverse axis.

Transverse axis in Fig. 3 represents that the distance in target and position changes from 0 km to 650 km, and the longitudinal axis represents the elevation angle.

As can be drawn from Figure 3, for the angle measurement of low elevation angle target, existing front-rear space smooth multiple signal classification method, alternating projection maximum likelihood method angle error are bigger than normal, and angle error of the present invention is minimum.

3. to the angle measurement result of certain surveillance radar measured data

Carry out angle measurement process by the present invention and existing front-rear space smooth multiple signal classification method to this surveillance radar measured data, angle error as shown in Figure 4.Wherein:

Transverse axis represents the distance in target and position, the angle error when longitudinal axis represents that distance changes with transverse axis;

SS-MUSIC represents the angle error of front-rear space smooth multiple signal classification method;

IA-CS represents angle error of the present invention.

As can be drawn from Figure 4, existing front-rear space smooth multiple signal classification method angle error is bigger than normal, and angle error of the present invention is less than normal.

Claims (4)

1., based on the high method of survey of array interpolation compression perception, comprise the following steps:

(1) from radar return, echo signal is extracted, obtain the array manifold matrix W of true array, and clutter cancellation and interference cancellation process are carried out to this echo signal, obtain offseting rear echo signal x and spatial domain sparse signal S, the relation offseting rear echo signal x and spatial domain sparse signal S is as follows:

x＝ψS+n，

Wherein, ψ represents super complete redundant dictionary, and its length is that c, n represent white Gaussian noise;

(2) use digital beam froming method DBF to carry out elevation angle bigness scale to the echo signal x after offseting, obtain bigness scale angle [alpha], and then obtain spatial domain, place, echo signal elevation angle Ο;

(3) described spatial domain Ο is divided into P part, P > > M, M represents array number, obtains spatial domain matrix Θ:

Θ＝[α _l,α _l+Δα,…,α _r],

Wherein, represent the left margin of Θ, represent the right margin of Θ, represent half-power beam width, Δ α is step-length, Δ α=0.1 °;

(4) Array interpolation process is carried out to true array, obtain the array manifold matrix W of virtual array _i; According to the array manifold matrix W of virtual array _iwith the array manifold matrix W of true array, obtain interpolation transformation matrix B;

(5) pre-whitening processing is carried out to interpolation transformation matrix B, obtain albefaction interpolation transformation matrix T _i;

(6) the echo signal x after offseting is projected to albefaction interpolation transformation matrix T _i, obtain the measurement signal z of virtual array;

(7) tie up observing matrix φ with F × P and compression sampling is carried out to measurement signal z, F < < P, obtain the observation signal y that F × 1 is tieed up;

(8) according to observation signal y and albefaction interpolation transformation matrix T _i, utilize greedy orthogonal matching pursuit method of following the trail of in class algorithm, through type ≤ β ²iteration, chooses a locally optimal solution Step wise approximation spatial domain sparse signal S, obtains the estimated value of spatial domain sparse signal S

$\hat{S}=[{\hat{s}}_1,{\hat{s}}_2,\&CenterDot;\&CenterDot;\&CenterDot;,{\hat{s}}_i,\&CenterDot;\&CenterDot;\&CenterDot;,{\hat{s}}_c],$

Wherein, || || ₁represent and ask vectorial 1-norm, s.t represents constraint condition, || || ₂represent and ask vectorial 2-norm, i=1,2 ..., c, β represent that noise criteria is poor;

(9) objective definition angular range, theta=[θ ₁, θ ₂..., θ _i..., θ _c], according to estimated value element and the one-to-one relationship of element of θ, namely with θ _ione_to_one corresponding, obtains target angle measurement result θ _d, d ∈ i:

Wherein, d represents estimated value in non-vanishing element subscript;

(10) according to target angle measurement result θ _dtarget range R with known, obtains object height by triangular transformation:

H＝Rsin(θ _d)。

2. the high method of the survey of array interpolation compression perception according to claim 1, digital beam froming method DBF is used in wherein said step (2) to carry out elevation angle bigness scale to the echo signal x after offseting, obtain bigness scale angle [alpha], and then obtain spatial domain, place, echo signal elevation angle Ο, carry out as follows:

2a) utilize steering vector summation is weighted to x, obtains bigness scale angle [alpha]:

$\mathrm{\&alpha;}=\arg \underset{\mathrm{\&xi;}}{\max }\left(\frac{1}{L}\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{\left|v^H\left(\mathrm{\&xi;}\right)x\left(t_l\right)\right|}^2\right),$

Wherein, arg max represents the parameter found and have maximum cost function, and ξ represents target search angular range ,-180 °≤ξ≤180 °, and L represents fast umber of beats, x (t _l) represent the echo signal after offseting, t _lrepresent the sampling time, 1≤l≤L, subscript T represents transposition, and subscript H represents conjugate transpose;

2b) utilize half-power beam width obtain the spatial domain Ο at angle on target place:

Wherein, λ represents radar signal wavelength, and d represents array element distance.

3. the high method of the survey of array interpolation compression perception according to claim 1, wherein described in step (4), Array interpolation process is carried out to true array, add Virtual array between the array element of true array, to expand the dimension of the array manifold matrix W of true array, the M × P obtaining virtual array ties up array manifold matrix W _i:

W _I＝[v _I(α _l),v _I(α _l+Δα),…,v _I(α _j),…,v _I(α _r)]，

Wherein, represent the steering vector of a virtual matrix jth echo signal, α _j∈ Θ.

4. the high method of the survey of array interpolation compression perception according to claim 1, the array manifold matrix W according to virtual array wherein described in step (4) _iwith the array manifold matrix W of true array, obtain interpolation transformation matrix B, point following two kinds of situations calculate:

When not considering noise, according to the array manifold matrix W of virtual array _iand the fixed relationship between the array manifold matrix W of true array and interpolation transformation matrix B: B ^hw=W _i, and the steering vector of true array with a virtual array jth echo signal steering vector v _i(α _j) and interpolation transformation matrix B between fixed relationship: draw interpolation transformation matrix B:

B＝W _IW ^H(WWH) ^-1

Wherein, represent the steering vector of true array,

represent a virtual matrix jth echo signal steering vector, represent the incident angle offseting front echo signal, α _j∈ Θ;

When considering noise, it is the array manifold matrix W according to virtual array _iand the fixed relationship between the array manifold matrix W of true array and interpolation transformation matrix B: B ^h(W+N)=W _i+ N _i, and the steering vector of true array with a virtual array jth echo signal steering vector v _i(α _j) and interpolation transformation matrix B between fixed relationship: draw interpolation transformation matrix B:

$B={\mathrm{\&sigma;}}_s^2{W}_I{W}^H{({\mathrm{\&sigma;}}_s^2{\mathrm{WW}}^H+{\mathrm{\&sigma;}}_n^{2}I)}^{-1},$

Wherein, N represents the noise matrix of true array, N _irepresent the noise matrix of virtual matrix, n represents the noise vector of N, n _irepresent N _inoise vector, for signal power, for noise power, I is unit matrix.