SUBSURFACE IMAGING RADAR

Abstract

A method and system for obtaining SAR images with reduced or eliminated surface clutter to detect subsurface targets, the method comprising the following steps: -selecting a first frequency and an incidence angle for the radar signal such that the ratio of surface backscattering to subsurface target backscattering is significantly larger for vertical polarization than for horizontal -obtaining vertically and horizontally polarized SAR images based on the same SAR path exploiting the selected first frequency and viewing angle -weighting and differencing the vertically and horizontally polarized SAR images so that the surface backscattering completely cancels between the two images and only the combination of the target backscattering components remains.

Claims

1. A method of removing surface clutter in SAR radar imaging of subsurface targets, said method comprising the steps of: selecting a first frequency and an incidence angle (χ0) for a radar signal such that the ratio of surface backscattering to subsurface target backscattering is significantly larger for vertical polarization than for horizontal polarization; simultaneously obtaining vertically and horizontally polarized SAR images, the obtained SAR images being coincident except for differences in the polarization of each, the obtained SAR images exploiting the selected first frequency and incidence angle for a vertically polarized and a horizontally polarized radar signal; and weighting and differencing, via a control unit, the vertically and horizontally polarized SAR images so that the surface backscattering completely cancels between the two images and only the combination of the target backscattering components remains.

2. The method according to claim 1, wherein the first frequency of the radar signal and the incidence angle are chosen such that the wavelength of the vertically polarized radar signal is at least one of greater than or equal to the surface roughness.

3. The method according to claim 1, wherein the incidence angle is chosen to be as low as possible without shadows arising.

4. The method according to claim 1, wherein the incidence angle is chosen to be larger than zero and less than the Brewster angle, wherein an incidence angle of zero is horizontal incidence.

5. The method according to claim 1, wherein the horizontal and vertically polarized radar signals are generated by a horizontal and a vertical antenna that conduct registrations in a so called ping-pong mode.

6. The method according to claim 1, wherein the first frequency is between 25 and 500 MHz.

7. The method according to claim 6, wherein the first frequency is between 130 and 360 MHz.

8. The method according to claim 1, wherein transmitting and receiving components are provided and configured to work within a range of 25 m to 5000 m.

9. The method according to claim 1, wherein transmitting and receiving components are provided and configured to work within a range of 100 m to 500 m.

10. The method according to claim 1, wherein an adaptive minimum energy method is used to weight and difference the vertically and horizontally polarized SAR images so that the surface backscattering cancels between the two images and only the combination of the target backscattering components remains.

11. The method according to claim 10, wherein the method comprises the following steps: with the aid of an aircraft in flight (205): obtaining (210) a HH complex SAR image FH(x,y); obtaining (215) a VV complex SAR image Fv(x,y); selecting (305, 350), within the images, an area T that appears to be homogenous; calculating (310) a min such that: $E=\underset{T}{\int \int }-{.Math.F_H\left(x,y\right)-{\gamma }_{\min }.Math.F_V\left(x,y\right).Math.}^2.Math.\mathrm{dxdy}=\min$ forming (320) a ground clutter suppressed SAR image AF(x,y) by forming the expression:
ΔF(x,y)=F.sub.H(x,y)−.sub.minF.sub.V(x,y).

12. The method according to claim 11, wherein subsurface targets subsequently are detected (325) by applying e.g. CFAR thresholding, ICD or CCD methods on ΔF.

13. A SAR system for providing SAR images having removed surface clutter to improve detection of subsurface targets, said system comprising: a selection unit configured for selecting a first frequency and an incidence angle (χ0) for a radar signal such that the ratio of surface backscattering to subsurface target backscattering is significantly larger for vertical polarization than for horizontal polarization; a linear combiner configured, via a linear combiner control unit, for: simultaneously obtaining vertically and horizontally polarized SAR images, the obtained SAR images being coincident except for differences in the polarization of each, the obtained SAR images exploiting the selected first frequency and incidence angle for a vertically polarized and a horizontally polarized radar signal; and weighting and differencing the vertically and horizontally polarized SAR images so that the surface backscattering completely cancels between the two images and only the combination of the target backscattering components remains.

14. The system according to claim 13, wherein the selection unit is further configured for selecting a homogenous test area T as input to a gamma-min finding unit.

15. The system according to claim 13, further comprising a target detector and a display unit for detecting and visualizing detected targets to an operator.

16. The system according to claim 13, further comprising a ping-pong control unit connected to the transmitters to make the transmitters send in ping-pong mode.

Description

BRIEF DESCRIPTION OF THE FIGURES

[0084] The invention and its specific embodiment will now be described in detail with the aid of the following drawings of which

[0085] FIG. 1 is a view of a helicopter with a low frequency synthetic aperture radar equipped with antennas of different polarity

[0086] FIG. 2a is a flowchart of a general method of detecting subsurface targets using SAR

[0087] FIG. 2b is a flowchart of an analytical method of detecting subsurface targets using SAR

[0088] FIG. 3a is a flowchart of an adaptive method of detecting subsurface targets using SAR

[0089] FIG. 3b is a more detailed flowchart of the adaptive method of FIG. 3a

[0090] FIG. 4a shows an example of complex valued response from target and ground surface.

[0091] FIG. 4b shows the responses of FIG. 4a re-weighted by selecting weighting coefficients so as to cancel the ground surface response.

[0092] FIG. 4c shows a diagram of intensity [dB] of radar signal vs., incidence angle using SPM theory and Fresnel reflection coefficients to calculate target and surface response as a function of incidence angle and to see the net attenuation of the target response when the surface response is cancelled. It is observed that the attenuation is relatively independent of incidence angle.

[0093] FIG. 5 shows relations to radar cross section or reflectivities at different polarizations and for surface and target scattering elements, measured at any point in intensity SAR images.

[0094] FIG. 6 shows, in a block diagram representation, a SAR system for obtaining SAR images having removed surface clutter to improve detection of subsurface targets.

DETAILED DESCRIPTION OF VARIOUS EMBODIMENTS

Definitions

[0095] The following terms will be used with the associated meanings throughout this document if not otherwise explicitly stated.

[0096] Surface roughness; surface roughness is a measure of roughness of a ground surface; there are two well established criteria a surface roughness: [0097] Rayleigh Criterion: if .sub.Δh<.sub.λ/8 cos θ, the surface is smooth [0098] Fraunhofer Criterion: if .sub.Δh<.sub.λ/32 cos θ, then the surface is smooth [0099] where .sub.Δh: standard deviation of surface roughness [0100] .sub.λ: wavelength [0101] θ: incident angle

[0102] Incident angle; incident angle is the angle between longitudinal direction of incident radar signal and the average normal direction to the ground surface;

[0103] Ping-pong mode; a radar system having a first and a second combined transmitting and receiving antenna and accompanying transmitters and receivers, can be made to operate in ping-pong mode, i.e., the transmitter of the second antenna does not send until the receiver of the first antenna has received an echo from a signal transmitted by the transmitter of the first antenna, and vice versa.

General

[0104] A low frequency SAR radar is arranged to provide a horizontally polarized channel as well as a vertically polarized channel. The channels are arranged to be in line with respect to the direction of flight in order to each provide a SAR image pixel by pixel completely coincident except for the difference in polarization, i.e., if the same polarization had been used, there had been two entirely identical images.

[0105] Open land often has a roughness in which average height differences over a distance of one to a few meters is only a fraction of that distance. Radar wavelength of low frequency radar is of the order of one to a few meters. It is known, and a consequence of Maxwell's equations, that when roughness in this manner is small compared to the wavelength of the radar signal, backscattering of the radar signal with vertical polarization, is much stronger than backscattering of a horizontal signal. There is a relationship between backscatter at vertical and horizontal polarization at these particular conditions, which substantially depends on the incident angle, and only weakly depends on the dielectric constant, and do not depend on either the roughness or wavelength. This fact implies that a radar image of the ground surface will be very nearly identical for the horizontal and vertical polarization, with the only but significant difference that the vertically polarized image has much higher intensity.

[0106] If there are radar targets below the surface, these will also be found in the two SAR images. Radar strength of underground targets will vary between the channels but not according to the same mathematical laws as the surface reflexes. For underground targets the conditions are guided by Fresnel reflection coefficients, which entails that the vertical polarization provides a greater intensity. The intensity difference is however less for underground targets than for backscattering from the surface.

[0107] Because environmental conditions at ground surface and below facilitates it in the above taught manner, backscattering from the ground surface can be eliminated in the SAR image by seeking a linear combination of the differently polarized images. This involves to arrange to assign backscatter from the surface the same amplitude but opposite sign as the backscatter being differently polarized. Backscattering from the underground object will thereby be reduced, but only to a level that can be accepted. Compensation for this reduction is achieved by employing the radar system at a shorter distance, with the crucial advantage that competing surface clutter thereby to a great deal have been eliminated.

[0108] Because surface clutter in many cases is the main reason why subsurface targets cannot be distinguished, the method disclosed in the present application should be of great importance in applications intended to identify subsurface targets.

System Overview

[0109] The PCD algorithm requires a horizontal polarization transmitted and received (HH) and a vertical polarization transmitted and received (VV) SAR image of the ground which from every aspect of data collection are as similar as possible. Thus: [0110] Antennas may have a common phase center or a phase centre displaced along the flight axis. In the latter case phase centers should be adjusted to a common phase center with the separation between the two taken into account in SAR processing motion compensation; [0111] A realization may either be based on a common radar transceiver toggling between the H- and V-polarization antennas or one channel for either and operating in parallel. The latter case has however the drawback of picking up any unwanted cross polarization response.

[0112] FIG. 1 shows a helicopter 100 fitted with a low frequency SAR system equipped with H-polarization and V-polarization antennas 110, 115, 120, 125 for intertwined HH and VV SAR images.

[0113] FIG. 2a shows a flowchart of a general method of detecting subsurface targets using SAR. During flight 205 horizontal polarization SAR image 2 0 and vertical polarization SAR image 215 are obtained. The two images are linearly combined 220 by the use of certain method(s) to remove ground response and thereby accentuate underground/subsurface response. Subsequently subsurface targets may be detected manually from display picture or detected 230 with the aid of applying e.g. CFAR thresholding, ICD or CCD methods.

Mathematic Formulas

[0114] This section discloses polarimetry formulas for surface and target backscattering modification to semi-transparent surface

[0115] According to the small perturbation model (SPM) in the theory of electromagnetic rough surface scattering, the complex valued (including phase) HH and VV SAR images has the following structure (below index of refraction n may be assumed real-imaginary part affects very little for relevant soils, incidence angle χ.sub.0)

F.sub.X(x,y)=c.sub.H,gf.sub.g(x,y)+c.sub.H,tf.sub.t(x,y)   (I)

F.sub.V(x,y)=c.sub.V,gf.sub.g(x,y)+c.sub.V,tf.sub.t(x,y)   (II)

wherein [0116] F.sub.H(x,y) is the horizontal polarization SAR image [0117] F.sub.V((x,y) is the vertical polarization SAR image [0118] C.sub.H,g is a horizontal polarization specific SPM rough surface backscattering coefficient [0119] C.sub.V,g is a vertical polarization specific SPM rough surface backscattering coefficient [0120] f.sub.g (x,y) js SAR image contribution from rough surface [0121] C.sub.H,t is a horizontal polarization specific 2-way amplitude transmission loss equals Polarization specific 1-way power transmission loss [0122] C.sub.V,t is a vertical polarization specific 2-way amplitude transmission loss equals Polarization specific 1-way power transmission loss [0123] F.sub.t(x,y) is SAR image contribution from subsurface targets

[0124] Further, polarization specific SPM rough surface backscattering coefficients have the following structure:

[00002]$c_{H,g}=\frac{\left(n^2-1\right)\left[{\sin }^2.Math.{\chi }_0-n^2\left(1+{\sin }^2.Math.{\chi }_0\right)\right]}{{\left(n^2.Math.\cos .Math..Math.{\chi }_0+\sqrt{n^2-{\sin }^2.Math.{\chi }_0}\right)}^2}.Math.a_0$$c_{V,g}=\frac{\cos .Math..Math.{\chi }_0-\sqrt{n^2-{\sin }^2.Math.{\chi }_0}}{\cos .Math..Math.{\chi }_0+\sqrt{n^2-{\sin }^2.Math.{\chi }_0}}.Math.a_0$

[0125] More, further polarization specific 2-way amplitude transmission loss equals polarization specific 1-way power transmission loss

[00003]$c_{H,t}=1-{\left(\frac{\cos .Math..Math.{\chi }_0-\sqrt{n^2-{\sin }^2.Math.{\chi }_0}}{\cos .Math..Math.{\chi }_0+\sqrt{n^2-{\sin }^2.Math.{\chi }_0}}\right)}^2$$c_{V,t}=1-{\left(\frac{\sqrt{n^2-{\sin }^2.Math.{\chi }_0}-n^2.Math.\cos .Math..Math.{\chi }_0}{\sqrt{n^2-{\sin }^2.Math.{\chi }_0}+n^2.Math.\cos .Math..Math.{\chi }_0}\right)}^2$

wherein [0126] n is index of refraction (may be assumed real) [0127] χ.sub.0 is incidence angle [0128] and expressions squared in the two equations above is Fresnel reflection coefficient.

[0129] In this context it could be noted that “dense” equals n=5.5 and “light” equals n=3, which summarizes variability of most dry soils, at frequencies about 100 MHz.

Polarimetric Change Detection Principle

[0130] The present invention provides a method for creating a so called polarimetric change image. Such a polarimetric change image is obtained in two main steps. The steps efficiently remove the ground response but keeps the subsurface response. FIG. 2b shows a flowchart of such a method of detecting subsurface targets using SAR. It may be called the “analytical” method.

[0131] The main steps, in addition to obtaining horizontally and vertically polarized images, and forming 255, 260 coefficients as described above, are:

[0132] 1. Multiplying 265 the second equation with quotient . . . C.sub.Hg/C.sub.Vg. . .

F.sub.H(x,y)=c.sub.H,gf.sub.g(x,y)+c.sub.H,tf.sub.t(x,y)

F.sub.V(x,y)=c.sub.V,gf.sub.g(x,y)+c.sub.V,tf.sub.t(x,y)

[0133] 2. Subtracting 265 the two equations from each other forming a polarimetric change image ΔF(x,y) also called a ground clutter suppressed SAR image.

[00004]$\begin{array}{cc}\Delta .Math..Math.F\left(x,y\right)=F_H\left(x,y\right)-\frac{c_{H,g}}{c_{V,g}}.Math.F_V\left(x,y\right)=c_{H,t}\left(1-\frac{c_{H,g}}{c_{V,g}}.Math.\frac{c_{V,t}}{c_{H,t}}\right).Math.f_t\left(x,y\right)& \left(\mathrm{III}\right)\end{array}$

[0134] These steps result in a desired cancellation of ground response and also in a change in subsurface target response.

Interpretation of PCD with Respect to Radar Cross Section

[0135] Relations to radar cross section or reflectivities at different polarizations and for surface and target scattering elements, measured at any point in intensity SAR images will be explained in the following.

PCD Target Attenuation Understood

[0136] The independent ratio of H- and V-polarization responses from ground surface and target can be used to suppress the latter at the price of a certain attenuation affecting the target response.

[0137] FIG. 4a shows an example of complex valued response from target and ground surface. Incidence is from above left, and dotted arrow in first quadrant and unbroken line arrow in third quadrant represents target polarization difference due to diffuse surface reflection. Unbroken line arrow in second quadrant and dotted line arrow in fourth quadrant represents ground polarization difference due to refraction loss caused by Fresnel specular reflection coefficients.

[0138] FIG. 4b shows the responses of FIG. 4a re-weighted by selecting weighting coefficients so as to cancel the ground surface response.

[0139] FIG. 4c shows a diagram of intensity [dB] of radar signal vs. incidence angle using SPM theory and Fresnel reflection coefficients to calculate target and surface response as a function of incidence angle and to see the net attenuation of the target response when the surface response is cancelled. It is observed that the attenuation is relatively independent of incidence angle, as also can be seen from the equation below.

[00005]$\hspace{1em}\begin{array}{c}\begin{array}{c}\mathrm{atten}=.Math.\frac{1}{\sqrt{{\sigma }_{H-t}}}\left[\sqrt{{\sigma }_{H-g}}+\sqrt{{\sigma }_{H-t}}-\frac{\sqrt{{\sigma }_{H-g}}}{\sqrt{{\sigma }_{V-g}}}.Math.\left(\sqrt{{\sigma }_{V-g}}+\sqrt{{\sigma }_{V-t}}\right)\right]\\ =.Math.1-\frac{\sqrt{{\sigma }_{H-g}}}{\sqrt{{\sigma }_{V-g}}}.Math.\frac{\sqrt{{\sigma }_{V-t}}}{\sqrt{{\sigma }_{H-t}}}=1-\sqrt{\frac{{\sigma }_{V-t}/{\sigma }_{V-g}}{{\sigma }_{H-t}/{\sigma }_{H-g}}}=1-\sqrt{\frac{{\sigma }_{V-t}/{\sigma }_{H-t}}{{\sigma }_{V-g}/{\sigma }_{H-g}}}\end{array}\end{array}$

[0140] FIG. 5 shows relations to radar cross section or reflectivities at different polarizations and for surface and target scattering elements, measured at any point in intensity SAR images.

Adaptive PCD Algorithm

[0141] The present application discloses two basic methods for detecting subsurface targets [0142] one deterministic/analytic method, as described above, relying on the fact that ground index of refraction is known and [0143] one adaptive method, not requiring such knowledge but assuming index of refraction being the same over an area.

[0144] Whilst attenuation of the subsurface target response is almost independent of index of refraction, the surface V to H ratio depends significantly on the index of refraction, see FIG. 4a; this strongly favors the adaptive method.

[0145] FIGS. 3a and 3b shows a flowchart of the adaptive method. The method relies on that subsurface targets are sporadic and that they will not energy-wise affect the SAR image. The method comprises the following steps: [0146] selecting 330, 335 suitable SAR frequency and incidence angle such that the ratio of surface backscattering to subsurface target backscattering is significantly larger for vertical polarization than for horizontal; [0147] with the aid of an aircraft in flight (205); [0148] obtaining 210, 340 a HH complex SAR image F.sub.H(x,y); [0149] obtaining 215, 345 a VV complex SAR image F.sub.V(x,y); [0150] Selecting 305, 350 a homogenous test area T around a potential target location; [0151] Finding 310, 355 a γmin such that an energy function.sup.E formed as the integral over the area.sup.T of the square of the difference between the horizontal polarization SAR image and the product of γmin and the vertical polarization SAR image, is minimized;

[00006]$E=\underset{T}{\int \int }.Math.{.Math.F_H\left(x,y\right)-{\gamma }_{\min }.Math.F_V\left(x,y\right).Math.}^2.Math.\mathrm{dxdy}=\min$

forming 320, 360 a ground clutter suppressed SAR image ΔF(x,y) by forming the difference between the horizontal polarization SAR image and the product of γmin and the vertical polarization SAR image;

ΔF(x,y)=F.sub.H(x,y)−γ.sub.minF.sub.V(x,y)

[0152] Subsequently subsurface targets may be detected 325 by applying e.g. CFAR thresholding, ICD or CCD methods on ΔF.

System

[0153] FIG. 6 shows a block diagram of a system for detection of subsurface targets using one of or both methods as described above. In the following first transmitter, first antenna, first duplexer and first receiver are for horizontal polarization signals, while second transmitter, second antenna, second duplexer and second receiver are for vertical polarization signals.

[0154] The system comprises first chain for obtaining a horizontally polarized radar image, i.e., a first transmitter 615, a first antenna 605, a first duplexer 610 and a first receiver 620. Further it comprises a second chain for obtaining a vertically polarized radar image, i.e., a second transmitter 635, second antenna 625, second duplexer 630 and second receiver 640.

[0155] The system further comprises an incidence angle selection unit 650 for providing incidence angle to a analytical coefficient calculator 655, which calculates coefficients

C.sub.H,g, C.sub.V,g, C.sub.H,t, C.sub.V,t

as explained above.

[0156] Further the system comprises a linear combiner control unit 660 for controlling a linear combiner 665 to linearly combine the images from the first receiver 620 and the second receiver 640 to form a ground clutter suppressed SAR image ΔF(x,y) by forming the difference between the horizontal polarization SAR image and the product of γmin or C.sub.H,t/C.sub.V,t and the vertical polarization SAR image, as selected by an operator and conveyed by the linear combiner control unit 660.

[0157] The system also comprises a gamma-min γmin finder unit 670, for finding and providing to the linear combiner, a γmin according to what has been explained for minimizing the energy function E according to the adaptive method as explained above. The system also has a selection unit 675 for selecting a homogenous test area T as input to gamma-min finding unit 670.

[0158] The system may further be provided with a target detector 680 and a display unit 685 for detecting and visualizing detected targets to an operator.