Study on Imaging Algorithm of Missile-Borne MMW ... - Semantic Scholar
STUDY ON IMAGING ALGORITHM OF MISSILE-BORNE MMW SAR FOR GROUND TARGET WENCHONG XIE
WENFENG SUN
YONGLIANG WANG
KEY RESEARCH LAB., WUHAN RADAR ACADEMY, WUHAN, HUBEI, 430010, P.R.CHINA TEL.: (86-27)82898023 E-MAIL: [email protected]
ABSTRACT In this paper, in light of the imaging characteristics of missile-borne MMW spotlight SAR for ground target, some main problems including the imaging resolution and choice of pulse repetition frequency are discussed. Then a new scheme of motion compensation which consists of range alignment, phase adjustment and compensation for the movement with non-constant velocity is presented. Image reconstruction algorithm by using superresolution spectrum estimation is given based on linear spectrum extrapolation technique. Finally the simulation results prove its effectivity.
1
INTRODUCTION
Active MMW guidance is one of important aspects in radar precision guidance technique. Because of the improvement of resolution in MMW band, guidance radar can gain shape and structure information relative to target property by means of imaging 1-D or 2-D image of target, then fulfill target recognition and selection of attack points. As is well known, the radial range axis projection of target scatterer points will lead to information destroyed and loss, so target’s structure information offered by 1-D range profile is limited. With the rapid improvement of the performance of modern digital signal processing apparatus, the real-time problem in information processing that limits the application of SAR technique to the field of radar guidance is being solved. Study on the imaging algorithm of missile-borne MMW SAR for ground target has important practical meaning. Compared with air-borne SAR, missile-borne SAR has three important characteristics including the high flight speed, the straight movement with non-constant speed and the big squint angle, which bring forward crucial requirement for the performance of real time, the precision of motion compensation and the azimuth resolution of imaging algorithm. Because of the fact that the imaging of missile-borne SAR for ground target proceeds after target are found and tracked and the radar antenna always points to target during the imaging process, spotlight SAR mode can be assumed. For the missile movement is much quicker than the conventional ground target movement, we can assume ground target is stationary relative to the ground. Based on above consideration a novel imaging
0-7803-7750-8/03/$17.00 ©2003 IEEE.
algorithm of missile-borne MMW SAR for ground target is proposed in this paper. In the end the validity of algorithm is verified by the simulation experiment.
2 ANALYSIS OF IMAGING RESOLUTION AND PARAMETER SELECTION In the missile attack processing for ground target guidance radar based on missile need high resolution so that missile can accurately hit target or some special parts of target. For ground vehicle target, we think ρ a = ρ r = 0.5m can meet with the imaging request. We know the expression of range resolution ρ r is: c (1) 2B where c is the light velocity, B is the transmitter bandwidth. So ρ r = 0.5m need the transmitter bandwidth is 300MHz at least, and it can be realized by pulse compression technique. For the requirement of ρ a = 0.5 m, it can be met with by SAR technique only. Let us consider the geometry relation between missile and imaging target area depicted in Fig. 1, where v and a respectively is missile velocity and acceleration, α and β respectively is the angle between velocity and LOS(light of sight) and the angle between acceleration and LOS, ϕ is the incidence angle of LOS relative to the
ρr =
ground, and R (t ) is the instantaneous distance between radar and target. After missile is transmitted, firstly missile searches, locks and tracks ground target automatically through transmitting narrowband signal, and secondly when missile descends to an extent height and the distance between missile and target attains an extent value, endmost guidance starts, and missile transmits broad band signal, imaging processing using SAR technique is doing. Seeker measures the angle velocity between missile and LOS, then brings control signal according to prescriptive guidance mode. The signal is transferred to missile automatic steering apparatus, and control missile helm to deflect an extent angle, and make missile flight according to needed flight track. Because of the non-constant velocity movement of missile in the course of endmost guidance, it makes that missile flights along arc, LOS aims to target and an angle always exits between the missile flight direction and LOS, so that spotlight SAR mode
ICIP 2003
image can be adopted. The instantaneous distance between missile and target is: 1 R (t ) = R0 − v cos α ⋅ t − a cos β ⋅ t 2 (2) 2 where R0 is the initial distance between radar and target. Ts is assumed as one imaging processing interval, and the missile deflexion angle relative to target is assumed as ∆θ during Ts , the expression of ∆θ is: 1 a sin β ⋅ Ts2 2 ∆θ ≈ (3) R0 Commonly missile flight track is limited by maximal normal over loading, so that the missile deflexion angle can not be too big in one imaging processing interval. The expression of azimuth resolution is [1]: k λ ∆θ ρ a ≈ k a λ 4 sin( ) ≈ a (4) 2 2∆θ where k a is antenna weight along azimuth, and λ is radar wavelength. So v sin α ⋅ Ts −
Ts =
v sin α − v 2 sin 2 α − 2a sin β ⋅ R0 ⋅ (k a λ 2 ρ a )
(5) a sin β During one imaging processing interval, the target range walk relative to radar is: 1 ∆R = v cos α ⋅ Ts + a cos β ⋅ Ts2 (6) 2 In the imaging processing of SAR, the echo utilized in azimuth processing is sampled by pulse repetition frequency in fact. In order that echo of all the target area after being sampled is not overlapping, the selection of pulse repetition frequency should have a floor level. It is: Da fr ≥ (7) Ts ⋅ ρ a where D a is the size of azimuth imaging area. Above expression shows that the quantity of transmitting pulse should be bigger than the resolution cells included in azimuth imaging area. A set of model parameters is selected, for example, R0 = 8km, v = 600m/s, α = 15°, a = 10m/s2, β = 5°, λ = 8mm. The antenna azimuth weight is not considered, so k a = 1 , and the size of imaging area is 100m × 100m. So we can gain ∆θ ≈ 0.008 rad, T s = 421.6ms, f r ≥ 484.7 Hz, ∆R = 234 m. In actual system the missile movement can be effected by many factors such as airstream, and disorder induced by these factors makes antenna does not always point to imaging area. In addition, though small ∆θ leads to short imaging time, it can make that target radial motion relative to radar is far bigger than half range
resolution cell during short imaging interval as a result of rapid missile velocity, and the effect of weight in one imaging interval leads to the fall of azimuth resolution. All the factors cause that azimuth resolution is far more than range resolution. From the expression of (5), we can gain that azimuth resolution can be improved by extending imaging processing interval TS properly. But because the deflexion angle of seeker can not be too big in one imaging interval, it makes TS should not be too long. So with prolonging TS we can improve azimuth resolution by means of signal processing. On the other hand, under the condition of considering the performance request of system synthetically, we can gain appropriate azimuth resolution matched with range resolution by decreasing range resolution properly. Finally thanks to the imaging processing amount of calculation increasing at geometric series accompanying the increase of sampling, the storage and calculation velocity of signal processing apparatus in seeker are limited and the missile requirement for real-time is very high, pulse repetition frequency of signal should be small to the best of our abilities, for example f r = 500 Hz.
3 MOTION COMPENSATION The imaging quality of SAR is decided by motion compensation at some extent. In application of missile MMW radar, missile’s radial non-constant velocity movement can lead to high-step nonlinear phase error changing with time which cause image blurry and aberrant. On the other hand, though rotation is used in azimuth imaging, it must meet with approximately even condition of sampling interval for rotation angle in spectrum extrapolation imaging algorithm based on FFT. Missile’s rotation is non-constant velocity movement due to missile’s maneuvering movement. So for every echo in repetition period, the rotation angle of target is different in the same interval, and it leads to non-even sampling interval for rotation angle and image’s defocus. In the next analysis of imaging motion compensation, we assume that missile motion parameters can be precisely gained by missile’s INS. So motion compensation can be performed straightway according to known parameters.
3.1 RANGE ALIGNMENT AND PHASE ADJUSTMENT When the amount of distance variety between radar and target is more than half range resolution cell, range alignment must be done. Because the phase changing with time induced by the relative movement is annexed to echo, and its one-step derivative is Doppler frequency which leads to the position of target excursion whole and image SNR’s falling, and its high-step derivative brings frequency modulation which causes image blurry. So
phase adjustment has direct effect on the quality of image. As shown by the expression of (2), after R (t ) is sampled by pulse repetition frequency f r , we can gain n 1 n − a cos β ⋅ ( ) 2 fr 2 fr n = 0,1, " , N − 1 (8) where N = Ts ⋅ f r , so the range and phase delay of the pulse at n relative to the pulse at 1 respectively are: n 1 n ∆R (n) = R0 − R(n) = v cos α ⋅ + a cos β ⋅ ( ) 2 (9) fr 2 fr R (n) = R0 − v cos α ⋅
e j∆ϕ = exp[− j
4π
λ
∆R (n)] = exp[− j 2π
4π
λ
v cos α ⋅
n − fr
n 2 ) ] (10) fr λ Range alignment can be realized by transferring echo some resolution cells forewords. Phase adjustment can be completed by means of multiplying the whole echo by a j
a cos β ⋅ (
phase compensation term e − j∆ϕ .
3.2 NON-CONSTANT VELOCITY MOTION COMPENSATION Spectrum extrapolation imaging algorithm based on FFT requires 2-D sampling interval of frequency and rotation angle should be uniform. For frequency and rotation angle, sample points of echo data at polar coordinate locate on the polar coordinate grids in the small angle pie slice, and its data array relative to right angle coordinate is not of uniform interval. So coordinate transformation must be performed before imaging, that is a resampling operation converts the input signal, sampled on a polar array, to this desired rectangular data array. If missile rotation after the first-step compensation is motion with constant velocity, coordinate transformation can be completed by means of two-step interpolations in range and azimuth. But for the missile non-constant velocity motion, its velocity is non-constant after the first-step compensation. Because missile’s rotation parameters are known under this paper’s background, we can calculate the missile rotation angle during every repetition interval, then the practical grid position of echo data at polar coordinate can be decided. Compensation for irregular rotation is made up of two steps, firstly azimuth data that its position has been decided at arc is transformed to data arranged at grads grid after interpolation and resampling. Secondly the irregular data is interpolated at azimuth, then sampling makes that grads grid is transformed to rectangle grid which is uniform at azimuth and range. Finally data at rectangle grid is imaged based on spectrum extrapolation algorithm.
4 SUPERRESOLUTION SPECTRUM
ESTIMATION IN AZIMUTH AND IMAGING ALGORITHM For missile-borne SAR imaging, we can assumed that target is stationary during one imaging interval because missile velocity is far faster that target velocity. On the premise the imaging problem of missile-borne MMW SAR for ground target can be regarded as ISAR imaging after motion compensation, and radar is assumed stationary and target moves surrounding radar in one imaging interval. High range resolution in radar can be gained easily by transmitting broad band signal. From the analysis of the 2nd section, seeker deflexion angle in one imaging interval is very small and target’s range walk relative to radar is far bigger than half range resolution cell, so that simple imaging algorithm based on FFT does not be applied. Superresolution imaging algorithm can gain high azimuth resolution corresponding with range resolution under the condition of small deflexion angle. At the same time, considering the real-time problem, this paper adopts superresolution imaging algorithm based on linear spectrum extrapolation technique [2]. For spectrum extrapolation algorithm, the step value of forecast filter is calculated automatically based on the rule of FPE. Then the filter coefficient is calculated using Burg algorithm that is improved by means of automatic data weight way. In addition, due to target spectrum which is gained in rotation target imaging locates on an annulus spectrum field, the point expand function of imaging system has very high sidelobe, so that the image sidelobe should be restrained by adding Hamming window in order to improve the quality of image. The signal processing flow chart of missile-borne SAR is shown in Fig. 2.
5 EXPERIMENTAL SIMULATION AND SUMMARY We assumed radar transmits chirp signal. Signal parameters respectively are timewidth T = 1µs , bandwidth B = 500 MHz and wavelength λ = 8 mm. Other parameters respectively are the initial imaging distance R0 = 8 km, missile flight velocity v = 600 m/s, the angle between velocity and LOS α = 15° , missile flight acceleration a = 10 m/s2, the angle between acceleration and LOS β = 5° and pulse repetition frequency f r = 500Hz. In this paper we adopt the simulation model of tank as shown in the Fig.3, and the reconstructing image using this paper’s algorithm is shown in Fig.4. We can recognize the outline of tank clearly although its position having a little excursion. The result of simulation shows that imaging that adopts the way combining radial high range resolution with azimuth superresolution spectrum estimation is
successful. If we consider the motion of ground target during one imaging interval, the imaging model has not only the ingredient of SAR (missile motion), but also the ingredient of ISAR (target motion). Because the motion parameters of ground target is unknown, imaging becomes complex greatly. A novel imaging algorithm using hybrid
SAR-ISAR technique is proposed to image for ocean-going ship in paper [3]. But the missile motion is different from plane motion, so the consideration on imaging scheme is different, too. So study on the imaging problem of missile MMW SAR for ground target will be the main work in the future.
REFERENCE
[2] Rao M.Nuthalapati. “High Resolution Reconstruction of ISAR Images”. IEEE Trans AES, Vol.AES-28, pp. 462-472, 1992. [3] A.Damini and G.E.Haslam. “SAR/ISAR Ship Imaging: Theoretical Analysis and Practical Results”. EUSAR’96. Germany, pp. 443~446, 1996.
[1]
Ausherman D. A. et al. “Developments in Radar Imaging”. IEEE Trans AES, Vol.AES-20, pp. 363-398, 1988.
Non-constant
Phase adjustment
rotation
Compensation
Azimuth spectrum
Range FFT
extrapolation
Image reconstructing
Motion compensation
Fig.2 Imaging processing flow chart of missile-borne SAR
a
β
10
3
v
Missile
7
α
5
1
R(t )
ϕ
3.7m
8 4
Target center
6 Imaging area
2
9 7.0m
Fig.1 The imaging geometry relation of missile-borne SAR for ground target
Fig. 3 Tank model
azimuth range Fig.4 The SAR imaging result of this paper’s algorithm for tank model
Image output
I\Q signal
Range aligment
WENFENG SUN
YONGLIANG WANG
KEY RESEARCH LAB., WUHAN RADAR ACADEMY, WUHAN, HUBEI, 430010, P.R.CHINA TEL.: (86-27)82898023 E-MAIL: [email protected]
ABSTRACT In this paper, in light of the imaging characteristics of missile-borne MMW spotlight SAR for ground target, some main problems including the imaging resolution and choice of pulse repetition frequency are discussed. Then a new scheme of motion compensation which consists of range alignment, phase adjustment and compensation for the movement with non-constant velocity is presented. Image reconstruction algorithm by using superresolution spectrum estimation is given based on linear spectrum extrapolation technique. Finally the simulation results prove its effectivity.
1
INTRODUCTION
Active MMW guidance is one of important aspects in radar precision guidance technique. Because of the improvement of resolution in MMW band, guidance radar can gain shape and structure information relative to target property by means of imaging 1-D or 2-D image of target, then fulfill target recognition and selection of attack points. As is well known, the radial range axis projection of target scatterer points will lead to information destroyed and loss, so target’s structure information offered by 1-D range profile is limited. With the rapid improvement of the performance of modern digital signal processing apparatus, the real-time problem in information processing that limits the application of SAR technique to the field of radar guidance is being solved. Study on the imaging algorithm of missile-borne MMW SAR for ground target has important practical meaning. Compared with air-borne SAR, missile-borne SAR has three important characteristics including the high flight speed, the straight movement with non-constant speed and the big squint angle, which bring forward crucial requirement for the performance of real time, the precision of motion compensation and the azimuth resolution of imaging algorithm. Because of the fact that the imaging of missile-borne SAR for ground target proceeds after target are found and tracked and the radar antenna always points to target during the imaging process, spotlight SAR mode can be assumed. For the missile movement is much quicker than the conventional ground target movement, we can assume ground target is stationary relative to the ground. Based on above consideration a novel imaging
0-7803-7750-8/03/$17.00 ©2003 IEEE.
algorithm of missile-borne MMW SAR for ground target is proposed in this paper. In the end the validity of algorithm is verified by the simulation experiment.
2 ANALYSIS OF IMAGING RESOLUTION AND PARAMETER SELECTION In the missile attack processing for ground target guidance radar based on missile need high resolution so that missile can accurately hit target or some special parts of target. For ground vehicle target, we think ρ a = ρ r = 0.5m can meet with the imaging request. We know the expression of range resolution ρ r is: c (1) 2B where c is the light velocity, B is the transmitter bandwidth. So ρ r = 0.5m need the transmitter bandwidth is 300MHz at least, and it can be realized by pulse compression technique. For the requirement of ρ a = 0.5 m, it can be met with by SAR technique only. Let us consider the geometry relation between missile and imaging target area depicted in Fig. 1, where v and a respectively is missile velocity and acceleration, α and β respectively is the angle between velocity and LOS(light of sight) and the angle between acceleration and LOS, ϕ is the incidence angle of LOS relative to the
ρr =
ground, and R (t ) is the instantaneous distance between radar and target. After missile is transmitted, firstly missile searches, locks and tracks ground target automatically through transmitting narrowband signal, and secondly when missile descends to an extent height and the distance between missile and target attains an extent value, endmost guidance starts, and missile transmits broad band signal, imaging processing using SAR technique is doing. Seeker measures the angle velocity between missile and LOS, then brings control signal according to prescriptive guidance mode. The signal is transferred to missile automatic steering apparatus, and control missile helm to deflect an extent angle, and make missile flight according to needed flight track. Because of the non-constant velocity movement of missile in the course of endmost guidance, it makes that missile flights along arc, LOS aims to target and an angle always exits between the missile flight direction and LOS, so that spotlight SAR mode
ICIP 2003
image can be adopted. The instantaneous distance between missile and target is: 1 R (t ) = R0 − v cos α ⋅ t − a cos β ⋅ t 2 (2) 2 where R0 is the initial distance between radar and target. Ts is assumed as one imaging processing interval, and the missile deflexion angle relative to target is assumed as ∆θ during Ts , the expression of ∆θ is: 1 a sin β ⋅ Ts2 2 ∆θ ≈ (3) R0 Commonly missile flight track is limited by maximal normal over loading, so that the missile deflexion angle can not be too big in one imaging processing interval. The expression of azimuth resolution is [1]: k λ ∆θ ρ a ≈ k a λ 4 sin( ) ≈ a (4) 2 2∆θ where k a is antenna weight along azimuth, and λ is radar wavelength. So v sin α ⋅ Ts −
Ts =
v sin α − v 2 sin 2 α − 2a sin β ⋅ R0 ⋅ (k a λ 2 ρ a )
(5) a sin β During one imaging processing interval, the target range walk relative to radar is: 1 ∆R = v cos α ⋅ Ts + a cos β ⋅ Ts2 (6) 2 In the imaging processing of SAR, the echo utilized in azimuth processing is sampled by pulse repetition frequency in fact. In order that echo of all the target area after being sampled is not overlapping, the selection of pulse repetition frequency should have a floor level. It is: Da fr ≥ (7) Ts ⋅ ρ a where D a is the size of azimuth imaging area. Above expression shows that the quantity of transmitting pulse should be bigger than the resolution cells included in azimuth imaging area. A set of model parameters is selected, for example, R0 = 8km, v = 600m/s, α = 15°, a = 10m/s2, β = 5°, λ = 8mm. The antenna azimuth weight is not considered, so k a = 1 , and the size of imaging area is 100m × 100m. So we can gain ∆θ ≈ 0.008 rad, T s = 421.6ms, f r ≥ 484.7 Hz, ∆R = 234 m. In actual system the missile movement can be effected by many factors such as airstream, and disorder induced by these factors makes antenna does not always point to imaging area. In addition, though small ∆θ leads to short imaging time, it can make that target radial motion relative to radar is far bigger than half range
resolution cell during short imaging interval as a result of rapid missile velocity, and the effect of weight in one imaging interval leads to the fall of azimuth resolution. All the factors cause that azimuth resolution is far more than range resolution. From the expression of (5), we can gain that azimuth resolution can be improved by extending imaging processing interval TS properly. But because the deflexion angle of seeker can not be too big in one imaging interval, it makes TS should not be too long. So with prolonging TS we can improve azimuth resolution by means of signal processing. On the other hand, under the condition of considering the performance request of system synthetically, we can gain appropriate azimuth resolution matched with range resolution by decreasing range resolution properly. Finally thanks to the imaging processing amount of calculation increasing at geometric series accompanying the increase of sampling, the storage and calculation velocity of signal processing apparatus in seeker are limited and the missile requirement for real-time is very high, pulse repetition frequency of signal should be small to the best of our abilities, for example f r = 500 Hz.
3 MOTION COMPENSATION The imaging quality of SAR is decided by motion compensation at some extent. In application of missile MMW radar, missile’s radial non-constant velocity movement can lead to high-step nonlinear phase error changing with time which cause image blurry and aberrant. On the other hand, though rotation is used in azimuth imaging, it must meet with approximately even condition of sampling interval for rotation angle in spectrum extrapolation imaging algorithm based on FFT. Missile’s rotation is non-constant velocity movement due to missile’s maneuvering movement. So for every echo in repetition period, the rotation angle of target is different in the same interval, and it leads to non-even sampling interval for rotation angle and image’s defocus. In the next analysis of imaging motion compensation, we assume that missile motion parameters can be precisely gained by missile’s INS. So motion compensation can be performed straightway according to known parameters.
3.1 RANGE ALIGNMENT AND PHASE ADJUSTMENT When the amount of distance variety between radar and target is more than half range resolution cell, range alignment must be done. Because the phase changing with time induced by the relative movement is annexed to echo, and its one-step derivative is Doppler frequency which leads to the position of target excursion whole and image SNR’s falling, and its high-step derivative brings frequency modulation which causes image blurry. So
phase adjustment has direct effect on the quality of image. As shown by the expression of (2), after R (t ) is sampled by pulse repetition frequency f r , we can gain n 1 n − a cos β ⋅ ( ) 2 fr 2 fr n = 0,1, " , N − 1 (8) where N = Ts ⋅ f r , so the range and phase delay of the pulse at n relative to the pulse at 1 respectively are: n 1 n ∆R (n) = R0 − R(n) = v cos α ⋅ + a cos β ⋅ ( ) 2 (9) fr 2 fr R (n) = R0 − v cos α ⋅
e j∆ϕ = exp[− j
4π
λ
∆R (n)] = exp[− j 2π
4π
λ
v cos α ⋅
n − fr
n 2 ) ] (10) fr λ Range alignment can be realized by transferring echo some resolution cells forewords. Phase adjustment can be completed by means of multiplying the whole echo by a j
a cos β ⋅ (
phase compensation term e − j∆ϕ .
3.2 NON-CONSTANT VELOCITY MOTION COMPENSATION Spectrum extrapolation imaging algorithm based on FFT requires 2-D sampling interval of frequency and rotation angle should be uniform. For frequency and rotation angle, sample points of echo data at polar coordinate locate on the polar coordinate grids in the small angle pie slice, and its data array relative to right angle coordinate is not of uniform interval. So coordinate transformation must be performed before imaging, that is a resampling operation converts the input signal, sampled on a polar array, to this desired rectangular data array. If missile rotation after the first-step compensation is motion with constant velocity, coordinate transformation can be completed by means of two-step interpolations in range and azimuth. But for the missile non-constant velocity motion, its velocity is non-constant after the first-step compensation. Because missile’s rotation parameters are known under this paper’s background, we can calculate the missile rotation angle during every repetition interval, then the practical grid position of echo data at polar coordinate can be decided. Compensation for irregular rotation is made up of two steps, firstly azimuth data that its position has been decided at arc is transformed to data arranged at grads grid after interpolation and resampling. Secondly the irregular data is interpolated at azimuth, then sampling makes that grads grid is transformed to rectangle grid which is uniform at azimuth and range. Finally data at rectangle grid is imaged based on spectrum extrapolation algorithm.
4 SUPERRESOLUTION SPECTRUM
ESTIMATION IN AZIMUTH AND IMAGING ALGORITHM For missile-borne SAR imaging, we can assumed that target is stationary during one imaging interval because missile velocity is far faster that target velocity. On the premise the imaging problem of missile-borne MMW SAR for ground target can be regarded as ISAR imaging after motion compensation, and radar is assumed stationary and target moves surrounding radar in one imaging interval. High range resolution in radar can be gained easily by transmitting broad band signal. From the analysis of the 2nd section, seeker deflexion angle in one imaging interval is very small and target’s range walk relative to radar is far bigger than half range resolution cell, so that simple imaging algorithm based on FFT does not be applied. Superresolution imaging algorithm can gain high azimuth resolution corresponding with range resolution under the condition of small deflexion angle. At the same time, considering the real-time problem, this paper adopts superresolution imaging algorithm based on linear spectrum extrapolation technique [2]. For spectrum extrapolation algorithm, the step value of forecast filter is calculated automatically based on the rule of FPE. Then the filter coefficient is calculated using Burg algorithm that is improved by means of automatic data weight way. In addition, due to target spectrum which is gained in rotation target imaging locates on an annulus spectrum field, the point expand function of imaging system has very high sidelobe, so that the image sidelobe should be restrained by adding Hamming window in order to improve the quality of image. The signal processing flow chart of missile-borne SAR is shown in Fig. 2.
5 EXPERIMENTAL SIMULATION AND SUMMARY We assumed radar transmits chirp signal. Signal parameters respectively are timewidth T = 1µs , bandwidth B = 500 MHz and wavelength λ = 8 mm. Other parameters respectively are the initial imaging distance R0 = 8 km, missile flight velocity v = 600 m/s, the angle between velocity and LOS α = 15° , missile flight acceleration a = 10 m/s2, the angle between acceleration and LOS β = 5° and pulse repetition frequency f r = 500Hz. In this paper we adopt the simulation model of tank as shown in the Fig.3, and the reconstructing image using this paper’s algorithm is shown in Fig.4. We can recognize the outline of tank clearly although its position having a little excursion. The result of simulation shows that imaging that adopts the way combining radial high range resolution with azimuth superresolution spectrum estimation is
successful. If we consider the motion of ground target during one imaging interval, the imaging model has not only the ingredient of SAR (missile motion), but also the ingredient of ISAR (target motion). Because the motion parameters of ground target is unknown, imaging becomes complex greatly. A novel imaging algorithm using hybrid
SAR-ISAR technique is proposed to image for ocean-going ship in paper [3]. But the missile motion is different from plane motion, so the consideration on imaging scheme is different, too. So study on the imaging problem of missile MMW SAR for ground target will be the main work in the future.
REFERENCE
[2] Rao M.Nuthalapati. “High Resolution Reconstruction of ISAR Images”. IEEE Trans AES, Vol.AES-28, pp. 462-472, 1992. [3] A.Damini and G.E.Haslam. “SAR/ISAR Ship Imaging: Theoretical Analysis and Practical Results”. EUSAR’96. Germany, pp. 443~446, 1996.
[1]
Ausherman D. A. et al. “Developments in Radar Imaging”. IEEE Trans AES, Vol.AES-20, pp. 363-398, 1988.
Non-constant
Phase adjustment
rotation
Compensation
Azimuth spectrum
Range FFT
extrapolation
Image reconstructing
Motion compensation
Fig.2 Imaging processing flow chart of missile-borne SAR
a
β
10
3
v
Missile
7
α
5
1
R(t )
ϕ
3.7m
8 4
Target center
6 Imaging area
2
9 7.0m
Fig.1 The imaging geometry relation of missile-borne SAR for ground target
Fig. 3 Tank model
azimuth range Fig.4 The SAR imaging result of this paper’s algorithm for tank model
Image output
I\Q signal
Range aligment
