CN103064082B - Microwave imaging method based on direction dimension random power modulation - Google Patents

Abstract

The invention discloses a microwave imaging method based on direction dimension random power modulation. The direction dimension power distribution of radar transmitting signals is modulated randomly, a compressive sensing theory is combined to an imaging radar system, and accordingly imaging time is reduced. Meanwhile, the power modulation information of a target echo and a transmit signal is used as the common input of an imaging algorithm, a super-resolution result is obtained by using the sparse reconstruction capacity of a convex optimization algorithm, and accordingly the quality of images is improved. The microwave imaging method can overcome the defects of long imaging time and low resolution ratio and the like in close distance and wide viewing angle imaging application of a narrow beam scanning radar, and the practical process of the microwave radar in the field of close distance and wide viewing angle imaging is promoted.

Description

Microwave imaging method based on the random power modulation of azimuth dimension

Technical field

The present invention relates to microwave imaging field, be specially a kind of microwave imaging method based on the random power modulation of azimuth dimension.

Background technology

Due to microwave and millimeter wave, can penetrate the nonmetallic substances such as cloud and mist, vegetation, clothing, case and bag, so microwave and millimeter wave radar is not only widely used in the remote imaging field such as military surveillance, geographical remote sensing, also can be used for the closely imaging scenes such as lossless detection.Be limited to applied environment, closely imaging radar is difficult to adopt synthetic aperture technique to improve azimuthal resolution, so such radar often adopts large aperture antenna (array), with narrow beam scan mode, completes imaging.Yet closely the imaging angular field of view of imaging radar is conventionally larger, it is longer that narrow beam scans required imaging time, is difficult to meet the practical application request of current fast detecting or even real time imagery.Except imaging time is long, microwave closely imaging radar to be difficult to promote another practical major reason be that imaging resolution and practical application request still have certain gap, therefore super-resolution is processed for microwave imaging radar closely and is seemed particularly important, it can improve actual imaging resolution under the prerequisite that does not increase system antenna (array) bore, restoring image detail, improves picture quality.Constantly surging along with closely microwave imaging demands such as safety detection, lossless detections, needs a kind of imaging new method that not only can reduce sweep time but also take into account hyperresolution at present badly.

Summary of the invention

The object of the invention is to provide a kind of microwave imaging method based on the random power modulation of azimuth dimension, long to solve microwave radar imaging technique imaging time, the problem of image resolution rate variance.

In order to achieve the above object, the technical solution adopted in the present invention is:

Microwave imaging method based on the random power modulation of azimuth dimension, it is characterized in that: to radar emission signal in the spatial power of the azimuth dimension Stochastic Modulation in addition that distributes, the power transmitting in azimuth dimension is distributed with irradiating order random variation, transmit all not identical in the power distribution of azimuth dimension, each time transmits and do not have regularity or relevance between the distribution of azimuth dimension power at every turn; Using the power modulation information of the radar observation result that repeatedly irradiation obtains and radar emission signal as common input quantity, utilize protruding optimization process algorithm to obtain last image.

The described microwave imaging method based on the random power modulation of azimuth dimension, is characterized in that: utilize reconfigurable antenna technology or phased-array technique, change transmitter pattern function, realize and transmitting in the power modulation of azimuth dimension.

The described microwave imaging method based on the random power modulation of azimuth dimension, is characterized in that: the power transmitting in space distributes and do not have obvious main beam, and its emittance covers whole observation area.

The described microwave imaging method based on the random power modulation of azimuth dimension, is characterized in that: protruding optimization process adopts l ₁norm optimization algorithm, preferably BP algorithm, OMP algorithm.

The described microwave imaging method based on the random power modulation of azimuth dimension, is characterized in that: the information that the observed result that each radar emission signal irradiation obtains has comprised all targets in whole observation area.

The described microwave imaging method based on the random power modulation of azimuth dimension, it is characterized in that: for closely, wide visual angle imaging system, utilize the described random power modulation of the azimuth dimension that transmits, take the protruding optimization process that radar observation result and the Stochastic Modulation content that transmits be common input quantity.

The present invention, by the in addition Stochastic Modulation that distributes of the azimuth dimension power to radar emission signal, is fused to compressive sensing theory in imaging radar system, forms a kind of compression sampling system, thereby reduces imaging time.Meanwhile, using radar observation result and the common input quantity of the power modulation information transmitting as imaging algorithm, utilize the sparse re-configurability of convex optimized algorithm to obtain super-resolution imaging result, thereby improve picture quality.

Accompanying drawing explanation

Fig. 1 is traditional wave beam scanning imagery principle schematic.

Fig. 2 is image-forming principle schematic diagram of the present invention.

Fig. 3 tradition scanning imagery simulation result figure.

Fig. 4 the invention process Case Simulation imaging results figure.

Embodiment

Microwave imaging method based on the random power modulation of azimuth dimension, to radar emission signal in the spatial power of the azimuth dimension Stochastic Modulation in addition that distributes, the power transmitting in azimuth dimension is distributed with irradiating order random variation, transmit all not identical in the power distribution of azimuth dimension, each time transmits and do not have regularity or relevance between the distribution of azimuth dimension power at every turn; Using the power modulation information of the radar observation result that repeatedly irradiation obtains and radar emission signal as common input quantity, utilize protruding optimization process algorithm to obtain last image.

Utilize reconfigurable antenna technology or phased-array technique, change transmitter pattern function, realize and transmitting in the power modulation of azimuth dimension.

The power transmitting in space distributes and does not have obvious main beam, and its emittance covers whole observation area.

Protruding optimization process adopts l ₁norm optimization algorithm, preferably BP algorithm, OMP algorithm.

The information that the observed result that each radar emission signal irradiation obtains has comprised all targets in whole observation area.

For closely, wide visual angle imaging system, utilize the described random power modulation of the azimuth dimension that transmits, take the protruding optimization process that radar observation result and the Stochastic Modulation content that transmits be common input quantity.

Compressed sensing is a kind of information acquisition and signal processing theory framework of novelty, this theory can be briefly described below: the signal x that length is N, it is expressed as x=Ψ Θ on orthogonal basis Ψ, and wherein Ψ is N * N orthogonal dimension basis matrix, and Θ is that ordered series of numbers vector is maintained in N * 1.If the nonzero term number K in coefficient is less than N, claim that the expression of x in Ψ territory is that K item is sparse.Use and the incoherent measurement matrix of Ψ Φ _{m * N}coefficient is converted, can obtain the observed result of M * 1 dimension (M<N)

y=ΦΘ=ΦΨ ^Tx=A ^CSx

A wherein ^cS=Φ Ψ ^t, be called as sensing matrix, if A ^cSsatisfied constraint isometry condition (when Φ is taken as stochastic matrix, A ^cSgenerally can meet RIP condition), and M>=Klog (N/K), can adopt so protruding optimization method (is often l ₁norm optimization, specific algorithm comprises orthogonal matching pursuit OMP, base is followed the trail of BP etc.), with very high probability, from observed result y, recover original signal x, rejuvenation is formulated as

min||Ψ ^Tx|| _l?s.t.?y=ΦΨ ^Tx

An important application of compressed sensing is compression sampling, with the speed far below carrying out Qwest's frequency, completes the sampling of sparse signal, and intactly carries the contained information of original signal.Take this theory as basis, can be by loading time domain or spatial domain Stochastic Modulation (the random sensing matrix in corresponding compressed sensing) at receiving end, structure compression sampling system.Microwave radar is a kind of active imaging system, so its Stochastic Modulation process not only can be placed in during receiving end also can be fused to and transmit, and makes the compression sampling mode of radar more flexible.

The Stochastic Modulation that the present invention distributes by radar emission aspect dimension power, has been fused to compressive sensing theory in imaging system, proposes a kind of new target information compression sampling scheme and super-resolution imaging method.The method and traditional beam scanning are imaged on has very large difference, Fig. 1 and Fig. 2 illustrate respectively both imaging process in basic thought and specific implementation.The narrow beam of tradition scanning imagery radar emission concentration of energy, carries out point by point scanning with narrow beam to observed object region, and scanning result is stitched together and can obtains observation area image.In Fig. 1, also illustrated the mathematics implication that traditional scanning imagery process is corresponding, if the target scattering in observation area for information N * 1 dimensional vector x represent, in scanning process, each irradiation only can obtain the scattered information within the scope of narrow beam, therefore radar directional diagram Jacobian matrix Φ can think a unit matrix, scanning result is the vectorial y identical with x dimension, finally scanning result y is reset to splicing and just can obtain the two dimensional image of N pixel.The different introducings that are the random power modulation of azimuth dimension of the present invention and traditional scanning imagery maximum, as shown in Figure 2, introducing due to Stochastic Modulation, radar is distributed with irradiating order random variation at the power of azimuth dimension, the power of each irradiation distributes all not identical, and irradiation energy no longer concentrates within the scope of narrow beam, but all there is distribution in whole observation area.In radar system, the power of azimuth dimension distributes and is expressed by antenna radiation pattern function, therefore pattern function of the present invention and traditional scanning radar have very large difference: each direction of illumination figure does not have clear and definite main beam, and does not have any regularity or relevance between the pattern function of each irradiation yet.The difference of new method and traditional scanning imagery is not only embodied on pattern function, also show reception & disposal aspect: image the information that not merely relies on reception observed result to provide obtain, but reception observed result and pattern function information co-treatment obtain.Fig. 2 has illustrated the mathematics implication that imaging process of the present invention is corresponding equally, the corresponding sensing matrix A of each radar directional diagram function irradiating ^cSin a row vector, because each time irradiate to adopt different Stochastic Modulation, so A ^cSrow vector between be uncorrelated, guaranteed A ^cSmeet the desired constraint isometry of compressed sensing condition.Because each irradiation energy has covered whole observation area, so each, irradiate the information that the observed result that obtains has comprised all targets in whole observation area, and unlike scanning imagery, only comprise the information of narrow beam range of exposures internal object.According to compressive sensing theory, needing only target information is sparse in the expression of certain transform domain, so (M<<N) measurement result y and the pattern function matrix A much smaller than target information with dimension ^cSfor initial conditions, utilize l ₁norm optimization method just can reconstruct the image of N pixel.The present invention, by the random power modulation in the orientation that transmits, incorporates compressive sensing theory in imaging system design dexterously, and compression sampling that not only can realize target information, significantly reduces imaging time, and can obtain super-resolution imaging effect.

Describe embodiments of the invention below in detail, the example of described embodiment is shown in the drawings.Below by the embodiment being described with reference to the drawings, be exemplary, only for explaining the present invention, and can not be interpreted as limitation of the present invention.

For simplifying the analysis, first consider one-dimensional case, the research conclusion of one dimension is easy to be generalized to two dimension.If the target scattering strength function in observation area is σ (θ) (θ is position angle), in irradiation area, only have K target, target scattering function σ (θ) itself just has very strong sparse property.If radar directional diagram function is s (θ), the echoed signal power receiving is so

$p=\&Integral;{\left|s\left(\mathrm{\&theta;}\right)\right|}^2\mathrm{\&sigma;}\left(\mathrm{\&theta;}\right)\mathrm{d\&theta;}---\left(3\right)$

By σ (θ) and s (θ) discretize, discrete counting as N, formula can be expressed as follows

$p=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left|s\left({\mathrm{\&theta;}}_n\right)\right|}^2\mathrm{\&sigma;}\left({\mathrm{\&theta;}}_n\right)---\left(4\right)$

If irradiate M time, can obtain M time so observed result, establishing the observed result irradiating for the m time is p (m), pattern function is s _m(θ _n), the pattern function that irradiates for the m time of take is the matrix that row vector that m is capable forms a M * N.According to formula, M time irradiation process can be used following matrix representation

$\left[\begin{array}{c}p\left(1\right)\\ p\left(2\right)\\ .\\ .\\ .\\ p\left(M\right)\end{array}\right]=\left[\begin{array}{cccc}{\left|s_1\left({\mathrm{\&theta;}}_1\right)\right|}^2& {\left|s_1\left({\mathrm{\&theta;}}_2\right)\right|}^2& ...& {\left|s_1\left({\mathrm{\&theta;}}_N\right)\right|}^2\\ {\left|s_2\left({\mathrm{\&theta;}}_1\right)\right|}^2& {\left|s_2\left({\mathrm{\&theta;}}_2\right)\right|}^2& ...& {\left|s_2\left({\mathrm{\&theta;}}_N\right)\right|}^2\\ & .& & \\ & .& & \\ & .& & \\ {\left|s_M\left({\mathrm{\&theta;}}_1\right)\right|}^2& {\left|s_M\left({\mathrm{\&theta;}}_2\right)\right|}^2& ...& {\left|s_M\left({\mathrm{\&theta;}}_N\right)\right|}^2\end{array}\right]\left[\begin{array}{c}\mathrm{\&sigma;}\left({\mathrm{\&theta;}}_1\right)\\ \mathrm{\&sigma;}\left({\mathrm{\&theta;}}_2\right)\\ .\\ .\\ .\\ \mathrm{\&sigma;}\left({\mathrm{\&theta;}}_N\right)\end{array}\right]---\left(5\right)$

Contrast formula is known, and this matrix is just corresponding with the compressed sensing process that formula (1) represents, therefore, if pattern function matrix [| s _m(θ _n) | ²] _{m * N}meet the equidistant condition of the desired constraint of compressive sensing theory, and irradiate number of times M>=Klog (N/K), so just can from measurement result few in number, reconstruct by protruding optimization method the super resolution image of target.

Below in conjunction with imaging scene; by accompanying drawing and specific embodiments, principle of the present invention is described in detail; because the imaging scene the present invention relates to is wider; the comparatively classical embodiment of only take is below example; provide the specific embodiment of the present invention; but, should not limit practical application of the present invention and protection domain with this.

Analogue system adopts 16 * 16 aerial array, and array element is omnidirectional's point source antenna, and evenly arranges.Beam scanning imaging regulates array element first phase in phased array mode, and point by point scanning is carried out in imaging observation region, and imaging process needs to irradiate for 1530 times altogether, and Fig. 3 is the imaging results that traditional beam sweeping method obtains.Utilize new method that the present invention puies forward, adopt the mode of randomly changing first phase to realize the Stochastic Modulation that azimuth dimension power distributes, general needs more than 400 irradiation can complete imaging, sometimes even irradiate just energy imaging for more than 100 time, and the resolution of institute of the present invention extracting method is significantly better than beam scanning imaging, at some details places, the imaging effect at (for example trigger place) is better, and Fig. 4 is the imaging results that institute of the present invention extracting method obtains.The formation method that the present invention carries, than traditional scan imaging method, not only can reduce scanning times, shorten imaging time, and imaging resolution is also better than the latter.

Claims (6)

1. the microwave imaging method based on the random power modulation of azimuth dimension, it is characterized in that: to radar emission signal in the spatial power of the azimuth dimension Stochastic Modulation in addition that distributes, the spatial power transmitting in azimuth dimension is distributed with irradiating order random variation, transmit it is all not identical in the spatial power of azimuth dimension, to distribute at every turn, and the transmit spatial power of azimuth dimension of each time does not exist regularity or relevance between distributing; Using the power modulation information of the observed result that repeatedly irradiation obtains and radar emission signal as common input quantity, utilize protruding optimization process algorithm to obtain last image.

2. the microwave imaging method based on the random power modulation of azimuth dimension according to claim 1, is characterized in that: utilize reconfigurable antenna technology or phased-array technique, change transmitter pattern function, realize and transmitting in the power modulation of azimuth dimension.

3. the microwave imaging method based on the random power modulation of azimuth dimension according to claim 1, is characterized in that: the spatial power transmitting in azimuth dimension distributes and do not have obvious main beam, and its emittance covers whole observation area.

4. the microwave imaging method based on the random power modulation of azimuth dimension according to claim 1, is characterized in that: protruding optimization process algorithm adopts l ₁norm optimization algorithm, selects BP algorithm.

5. the microwave imaging method based on the random power modulation of azimuth dimension according to claim 1, is characterized in that: the information that the observed result that each radar emission signal irradiation obtains has comprised all targets in whole observation area.

6. the microwave imaging method based on the random power modulation of azimuth dimension according to claim 1, it is characterized in that: for closely, wide visual angle imaging system, utilize the described random power modulation of the azimuth dimension that transmits, the radar observation result of take is carried out protruding optimization process with the Stochastic Modulation content that transmits as common input quantity.